* config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
[official-gcc.git] / gcc / ChangeLog
blob5dc8ed03385d0eb296366926274e5acca487df63
1 2013-09-12  DJ Delorie  <dj@redhat.com>
3         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
4         between SP and FP.
5         (rl78_expand_epilogue): Likewise.
7 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
9         PR middle-end/58335
10         * lra-eliminations.c (remove_reg_equal_offset_note): New.
11         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
12         pointer elimination with using remove_reg_equal_offset_note.
14 2013-09-12  DJ Delorie  <dj@redhat.com>
16         * config/msp430/: New port.
17         * config.gcc (msp430): Added.
18         * doc/invoke.texi: Document MSP430 options.
19         * doc/install.texi: Document msp430-elf
20         * doc/md.texi: Document msp430-elf
21         * doc/contrib.texi: Document msp430-elf
23         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
24         PSImode.
26 2013-09-12  Martin Jambor  <mjambor@suse.cz>
28         PR ipa/58389
29         * ipa-prop.c (remove_described_reference): Give up if the edge in the
30         reference descriptor is NULL.
31         (ipa_edge_removal_hook): If owning a reference descriptor, set its
32         edge to NULL.
34 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
36         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
37         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
38         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
39         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
40         make_temp_ssa_name): move to tree-ssanames.h
41         * tree-ssa-alias.h: Move prototype.
42         * tree-ssa.h: Include tree-ssanames.h.
43         * tree-ssanames.c (FREE_SSANAMES): Move to here.
44         * tree-ssanames.h: New.  Move items from tree-flow*.h
45         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
47 2013-09-12  Richard Biener  <rguenther@suse.de>
49         PR tree-optimization/58404
50         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
51         propagate non-invariant addresses into dereferences wrapped
52         in component references.
54 2013-09-12  Richard Biener  <rguenther@suse.de>
56         PR tree-optimization/58402
57         * passes.def: Move pass_late_warn_uninitialized later.
59 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
61         * tree-ssa.h: New.  Move content from tree-flow.h and
62         tree-flow-inline.h.
63         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
64         Move prototypes belonging to tree-ssa.c.
65         * tree-flow-inline.h (redirect_edge_var_map_def, 
66         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
67         tree-ssa.h.
68         * gimple.h: Adjust prototypes.
69         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move to...
70         * gimple.c (useless_type_conversion_p, types_compatible_p):  Here.
71         * tree.h: Move prototype to tree-ssa.h.
72         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
73         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
74         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
75         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
76         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
77         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
78         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
79         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
80         graphite-blocking.c, graphite-clast-to-gimple.c, graphite-dependences.c,
81         graphite-interchange.c, graphite-optimize-isl.c, graphite-poly.c,
82         graphite-scop-detection.c, graphite-sese-to-poly.c, graphite.c,
83         ipa-cp.c, ipa-inline-analysis.c, ipa-inline-transform.c, ipa-inline.c,
84         ipa-prop.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa-utils.c,
85         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
86         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
87         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
88         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
89         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
90         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
91         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
92         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
93         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
94         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
95         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
96         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
97         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
98         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
99         value-prof.c, var-tracking.c,
100         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
102 2013-09-12  Richard Biener  <rguenther@suse.de>
104         PR tree-optimization/58396
105         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
106         (build_rdg): Take a loop-nest parameter, fix memleaks.
107         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
109 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
111         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
112         for SLM.
114 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
116         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
117         The FMA instruction names should have a 'v' prefix.
119 2013-09-12  Richard Biener  <rguenther@suse.de>
121         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
122         (dot_rdg): Use popen instead of system in optional code.
123         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
124         (already_processed_vertex_p): Adjust.
125         (has_anti_or_output_dependence, predecessor_has_mem_write,
126         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
127         rdg_flag_uses): Remove.
128         (rdg_flag_vertex): Simplify.
129         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
130         remove recursion.
131         (build_rdg_partition_for_component): Process the first vertex
132         of a component only.
133         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
135 2013-09-12  Alan Modra  <amodra@gmail.com>
137         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
139 2013-09-11  DJ Delorie  <dj@redhat.com>
140             Nick Clifton  <nickc@redhat.com>
142         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
143         (rl78_stack_based_mem): New.
144         * config/rl78/constraints.md (Iv08): New.
145         (Iv16): New.
146         (Iv24): New.
147         (Is09): New.
148         (Is17): New.
149         (Is25): New.
150         (ISsi): New.
151         (IShi): New.
152         (ISqi): New.
153         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
154         (movhi): Likewise.
155         (movsi): Change from expand to insn-and-split.
156         (ashrsi3): Clobber AX.
157         (lshrsi3): New.
158         (ashlsi3): New.
159         (cbranchsi4): New.
160         * config/rl78/rl78.md (CC_REG): Fix.
161         (addsi3): Allow memory and immediate operands.
162         (addsi3_internal): Split into...
163         (addsi3_internal_virt): ...new, and ...
164         (addsi3_internal_real): ...new.
165         (subsi): New.
166         (subsi3_internal_virt): New.
167         (subsi3_internal_real): New.
168         (mulsi3): Add memory operand.
169         (mulsi3_rl78): Likewise.
170         (mulsi3_g13): Likewise.
171         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
172         (cbranchqi4_real): Add more constraint options.
173         (cbranchhi4_real): Expand pattern.
174         (cbranchhi4_real_signed): New.
175         (cbranchhi4_real_inverted): New.
176         (cbranchsi4_real_lt): New.
177         (cbranchsi4_real_ge): New.
178         (cbranchsi4_real_signed): New.
179         (cbranchsi4_real): New.
180         (peephole2): New.
181         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
182         constant shifts.
183         (lshrsi3_virt): Likewise.
184         (ashlsi3_virt): Likewise.
185         (cbranchqi4_virt_signed): New.
186         (cbranchhi4_virt_signed): New.
187         (cbranchsi4_virt): New.
188         * config/rl78/rl78.c: Whitespace fixes throughout.
189         (move_elim_pass): New.
190         (pass_data_rl78_move_elim): New.
191         (pass_rl78_move_elim): New.
192         (make_pass_rl78_move_elim): New.
193         (rl78_devirt_info): Run devirt earlier.
194         (rl78_move_elim_info): New.
195         (rl78_asm_file_start): Register it.
196         (rl78_split_movsi): New.
197         (rl78_as_legitimate_address): Allow virtual base registers when
198         appropriate.
199         (rl78_addr_space_convert): Remove spurious debug stuff.
200         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
201         (rl78_print_operand): More cases for not printing '#'.
202         (rl78_expand_compare): Remove most of the logic.
203         (content_memory): New.
204         (clear_content_memory): New.
205         (get_content_index): New.
206         (get_content_name): New.
207         (display_content_memory): New.
208         (update_content): New.
209         (record_content): New.
210         (already_contains): New.
211         (insn_ok_now): Re-recog insns with virtual registers.
212         (add_postponed_content_update): New.
213         (process_postponed_content_update): New.
214         (gen_and_emit_move): New.
215         (transcode_memory_rtx): Record new location content.
216         Use gen_and_emit_move.
217         (force_into_acc): New.
218         (move_to_acc): Use gen_and_emit_move.
219         (move_from_acc): Likewise.
220         (move_acc_to_reg): Likewise.
221         (move_to_x): Likewise.
222         (move_to_hl): Likewise.
223         (move_to_de): Likewise.
224         (rl78_alloc_physical_registers_op1): Record location content.
225         (has_constraint): New.
226         (rl78_alloc_physical_registers_op2): Record location content.
227         Optimize use of HL.
228         (rl78_alloc_physical_registers_ro1): Likewise.
229         (rl78_alloc_physical_registers_cmp): Likewise.
230         (rl78_alloc_physical_registers_umul): Likewise.
231         (rl78_alloc_address_registers_macax): New.
232         (rl78_alloc_physical_registers): Initialize and set location
233         content memory as needed.
234         (rl78_reorg): Make sure split2 is called.
235         (rl78_rtx_costs): New.
237 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
239         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
240         for (not (neg ...)) and (neg (not ...)) cases.
242 2013-09-11  Richard Biener  <rguenther@suse.de>
244         PR middle-end/58377
245         * passes.def: Split critical edges before late uninit warning passes.
246         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
248 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
250         PR tree-optimization/58385
251         * fold-const.c (build_range_check): If both low and high are NULL,
252         use omit_one_operand_loc to preserve exp side-effects.
254 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
256         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
258 2013-09-11  Richard Biener  <rguenther@suse.de>
260         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
261         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
262         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
263         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
264         stmts_from_loop, known_dependences_p, build_empty_rdg,
265         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
266         * tree-loop-distribution.c: ... here.
267         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
268         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
269         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
270         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
271         * tree-loop-distribution.c: ... here.
272         * tree-loop-distribution.c: Include gimple-pretty-print.h.
273         (struct partition_s): Add loops member.
274         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
275         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
276         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
278 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
279             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
280             Sergey Lega  <sergey.s.lega@intel.com>
281             Anna Tikhonova  <anna.tikhonova@intel.com>
282             Ilya Tocar  <ilya.tocar@intel.com>
283             Andrey Turetskiy  <andrey.turetskiy@intel.com>
284             Ilya Verbin  <ilya.verbin@intel.com>
285             Kirill Yukhin  <kirill.yukhin@intel.com>
286             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
288         * config/i386/constraints.md (k): New.
289         (Yk): Ditto.
290         * config/i386/i386.c (const regclass_map): Add new mask registers.
291         (dbx_register_map): Ditto.
292         (dbx64_register_map): Ditto.
293         (svr4_dbx_register_map): Ditto.
294         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
295         disabled.
296         (ix86_preferred_reload_class): Disable constants for mask registers.
297         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
298         (ix86_hard_regno_mode_ok): Support new mask registers.
299         (x86_order_regs_for_local_alloc): Ditto.
300         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
301         (FIXED_REGISTERS): Add new mask registers.
302         (CALL_USED_REGISTERS): Ditto.
303         (REG_ALLOC_ORDER): Ditto.
304         (VALID_MASK_REG_MODE): New.
305         (FIRST_MASK_REG): Ditto.
306         (LAST_MASK_REG): Ditto.
307         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
308         (MAYBE_MASK_CLASS_P): New.
309         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
310         (REG_CLASS_CONTENTS): Ditto.
311         (MASK_REGNO_P): New.
312         (ANY_MASK_REG_P): Ditto.
313         (HI_REGISTER_NAMES): Add new mask registers.
314         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
315         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
316         mask registers.
317         (attribute "type"): Add mskmov, msklog.
318         (attribute "length_immediate"): Support them.
319         (attribute "memory"): Ditto.
320         (attribute "prefix_0f"): Ditto.
321         (*movhi_internal): Support new mask registers.
322         (*movqi_internal): Ditto.
323         (define_split): Split out clobber pattern is a logic
324         insn on mask registers.
325         (*k<logic><mode>): New.
326         (*andhi_1): Extend to support mask regs.
327         (*andqi_1): Extend to support mask regs.
328         (kandn<mode>): New.
329         (define_split): Split and-not to and and not if operands
330         are not mask regs.
331         (*<code><mode>_1): Separate HI mode to new pattern...
332         (*<code>hi_1): This.
333         (*<code>qi_1): Extend to support mask regs.
334         (kxnor<mode>): New.
335         (kortestzhi): Ditto.
336         (kortestchi): Ditto.
337         (kunpckhi): Ditto.
338         (*one_cmpl<mode>2_1): Remove HImode and handle it...
339         (*one_cmplhi2_1): ...Here, now with mask registers support.
340         (*one_cmplqi2_1): Support new mask registers.
341         (HI/QImode arithmetics splitter): Don't split if mask registers
342         are used.
343         (HI/QImode not splitter): Ditto.
344         * config/i386/predicated.md (mask_reg_operand): New.
345         (general_reg_operand): Ditto.
347 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
349         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
350         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
352 2013-09-10  Jeff Law  <law@redhat.com>
354         PR tree-optimization/58380
355         * tree-ssa-threadupdate.c (thread_block): Recognize another case
356         of threading through a buried loop header.
358         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
359         return value for single successor case.
361 2013-09-10  Jan Hubicka  <jh@suse.cz>
363         * ipa-devirt.c (ipa_devirt): Enable with LTO.
365 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
367         PR target/58361
368         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
369         support conditional execution.
370         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
372 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
374         * lra.c (lra): Clear lra_optional_reload_pseudos before every
375         constraint pass.
376         * lra-constraints.c (curr_insn_transform): Switch on optional
377         reloads.  Check destination too to check move insn.
378         (undo_optional_reloads): Add check that the original peudo did not
379         changed its allocation and the optional reload was inherited on
380         last inheritance pass.  Break loop after deciding to keep optional
381         reload.
382         (lra_undo_inheritance): Add check that inherited pseudo still in
383         memory.
385 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
387         * config/aarch64/aarch64.md (generic_sched): New.
388         * config/aarch64/aarch64-generic.md (load): Make conditional
389         on generic_sched attribute.
390         (nonload): Likewise.
392 2013-09-10  Jan Hubicka  <jh@suse.cz>
394         * lto-cgraph.c: Include ipa-utils.h.
395         (compute_ltrans_boundary): Also add possible targets into the boundary.
397 2013-09-10  Jan Hubicka  <jh@suse.cz>
399         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
400         VAR_DECL of vtable rather than full expression.
402 2013-09-10  Jan Hubicka  <jh@suse.cz>
403             Paolo Carlini  <paolo.carlini@oracle.com>
405         * cgraphunit.c (analyze_functions): Save input_location, set it
406         to UNKNOWN_LOCATION and restore it at the end.
408 2013-09-10  Martin Jambor  <mjambor@suse.cz>
410         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
411         represented by a thunk.
413 2013-09-10  Jeff Law  <law@redhat.com>
415         PR tree-optimization/58343
416         * tree-ssa-threadupdate.c (thread_block): Identify and disable
417         jump threading requests through loop headers buried in the middle
418         of a jump threading path.
420         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
421         in return value/type.
423 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
425         PR rtl-optimization/58365
426         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
427         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
428         it differs.
430 2013-09-10  Richard Biener  <rguenther@suse.de>
432         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
433         * tree-data-ref.c (create_rdg_vertices): Collect all data
434         references, signal failure to the caller, use data-ref API.
435         (build_rdg): Compute data references only once.  Maintain lifetime
436         of data references and data dependences from within RDG.
437         (free_rdg): Free dependence relations.
438         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
439         inventing extra dependences.
440         (distribute_loop): Update for RDG API changes.
442 2013-09-10  Kai Tietz  <ktietz@redhat.com>
444         * doc/invoke.texi (fms-extensions): Document changed
445         behavior for ms-abi targets.
446         * config/i386/i386.c (ix86_option_override_internal):
447         Set default value of option -fms-extension for ms-abi targets.
449 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
451         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
453 2013-09-10  Alan Modra  <amodra@gmail.com>
455         PR target/58330
456         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
458 2013-09-10  Alan Modra  <amodra@gmail.com>
460         * config/rs6000/predicates.md (add_cint_operand): New.
461         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
462         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
463         using add_cint_operand.
464         (largetoc_high_plus_aix): Likewise.
466 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
468         PR tree-optimization/58364
469         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
470         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
471         the current range can't be an unconditional true or false.
473 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
475         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
477 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
479         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
481 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
483         PR c++/43452
484         * doc/invoke.texi (-Wdelete-incomplete): Document it.
486 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
488         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
489         NO_REGS for immediate that can't be moved directly into FP_REGS.
491 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
493         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
494         comparison with negated operand.
495         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
496         RTL form.
498 2013-09-09  Richard Biener  <rguenther@suse.de>
500         PR middle-end/58326
501         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
502         of a subloop record all its block as affecting loop-closed SSA form.
504 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
506         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
507         of an rtx/bitpos pair.
508         (store_fixed_bit_field): Update accordingly.
510 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
512         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
513         GEN_INT.
514         * builtins.c (expand_errno_check): Likewise.
515         * dwarf2cfi.c (init_return_column_size): Likewise.
516         * except.c (sjlj_mark_call_sites): Likewise.
517         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
518         * lra-constraints.c (emit_inc): Likewise.
519         * ree.c (combine_set_extension): Likewise.
520         * regmove.c (fixup_match_2): Likewise.
521         * reload1.c (inc_for_reload): Likewise.
523 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
525         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
526         (force_to_mode, simplify_shift_const_1, simplify_comparison):
527         Use gen_int_mode with the mode of the associated simplify_* call.
528         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
529         * expmed.c (expand_shift_1): Likewise.
530         * function.c (instantiate_virtual_regs_in_insn): Likewise.
531         * loop-iv.c (iv_number_of_iterations): Likewise.
532         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
533         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
535 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
537         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
538         of the associated expand_* call.
539         (asan_emit_stack_protection): Likewise.
540         * builtins.c (round_trampoline_addr): Likewise.
541         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
542         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
543         (emit_store_flag): Likewise.
544         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
545         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
546         Likewise.
547         * function.c (instantiate_virtual_regs_in_insn): Likewise.
548         * ifcvt.c (noce_try_store_flag_constants): Likewise.
549         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
550         * modulo-sched.c (generate_prolog_epilog): Likewise.
551         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
552         (expand_ctz, expand_ffs, expand_unop): Likewise.
554 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
556         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
557         of the associated gen_rtx_* call.
558         * caller-save.c (init_caller_save): Likewise.
559         * combine.c (find_split_point, make_extraction): Likewise.
560         (make_compound_operation): Likewise.
561         * dwarf2out.c (mem_loc_descriptor): Likewise.
562         * explow.c (plus_constant, probe_stack_range): Likewise.
563         * expmed.c (expand_mult_const): Likewise.
564         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
565         * reload1.c (init_reload): Likewise.
566         * valtrack.c (cleanup_auto_inc_dec): Likewise.
567         * var-tracking.c (adjust_mems): Likewise.
568         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
569         rather than gen_rtx_fmt_ee.
571 2013-09-09  Jan Hubicka  <jh@suse.cz>
573         PR middle-end/58294
574         * value-prof.c (gimple_ic): Copy also abnormal edges.
576 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
578         * asan.c (asan_shadow_cst): Use gen_int_mode.
580 2013-09-08  Jan Hubicka  <jh@suse.cz>
582         * ipa-profile.c: Add toplevel comment.
583         (ipa_propagate_frequency_1): Be more conservative when profile is read.
584         (contains_hot_call_p): New function.
585         (ipa_propagate_frequency): Set frequencies based on counts when
586         profile is read.
587         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
588         profile; do not tamper with profile after inlining if it is read.
590 2013-09-08  Jan Hubicka  <jh@suse.cz>
592         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
593         speculative edges.
595 2013-09-08  Jan Hubicka  <jh@suse.cz>
597         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
598         summary generation.
600 2013-09-08  Jeff Law  <law@redhat.com>
602         PR bootstrap/58340
603         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
604         of 'found'.
606 2013-09-08  Andi Kleen  <ak@linux.intel.com>
608         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
610 2013-09-08  Jan Hubicka  <jh@suse.cz>
612         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
614 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
616         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
617         for non-debug insns.
618         * lra.c (new_insn_reg): Take the containing insn as a parameter.
619         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
620         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
621         accordingly.
623 2013-09-08  Jan Hubicka  <jh@suse.cz>
625         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
626         targets and devirtualize to BUILT_IN_UNREACHABLE.
627         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
628         * ipa.c (walk_polymorphic_call_targets): New function.
629         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
630         functions; use the new timevar.
631         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
632         was removed from the program.
633         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
634         not consider it in the walk when its vtable is dead.
635         (possible_polymorphic_call_targets_1): Pass anonymous flag to
636         record_binfo.
637         (devirt_variable_node_removal_hook): New function.
638         (possible_polymorphic_call_targets): Also register
639         devirt_variable_node_removal_hook.
640         (ipa_devirt): Do not do non-speculative devirtualization.
641         (gate_ipa_devirt): One execute if devirtualizing speculatively.
643 2013-09-08  Jan Hubicka  <jh@suse.cz>
645         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
646         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
647         varpool_remove_variable_insertion_hook): Declare.
648         * varpool.c (varpool_node_hook_list): New structure.
649         (first_varpool_node_removal_hook,
650         first_varpool_variable_insertion_hook): New variables.
651         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
652         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
653         varpool_remove_variable_insertion_hook,
654         varpool_call_variable_insertion_hooks): New functions.
655         (varpool_remove_node): Use it.
657 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
659         PR c++/54941
660         * diagnostic.c (diagnostic_build_prefix): When s.file is
661         "<built-in>" don't output line and column numbers.
663 2013-09-06  Jan Hubicka  <jh@suse.cz>
665         * cgraphunit.c (expand_thunk): Get body before touching arguments.
666         * lto-streamer-out.c: Stream thunks, too.
667         * lto-streamer-in.c (input_function): Pop cfun here
668         (lto_read_body): Instead of here.
670 2013-09-06  Caroline Tice  <cmtice@google.com>
672         * doc/install.texi: Add documentation for the --enable-vtable-verify
673         and the --disable-libvtv configure options.
675 2013-09-06  Jeff Law  <law@redhat.com>
677         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
678         edge implied equivalences into successor phis.
680 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
682         * resource.c (mark_referenced_resources): Handle COND_EXEC.
684 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
686         * resource.c (mark_target_live_regs): Compute resources taking
687         into account if a call is predicated or not.
689 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
691         * toplev.c (output_stack_usage): Be prepared for suffixes created by
692         the compiler in the function names.
694 2013-09-06  Jan Hubicka  <jh@suse.cz>
696         PR middle-end/58094
697         * ipa-inline.c (has_caller_p): New function.
698         (want_inline_function_to_all_callers_p): Use it.
699         (sum_callers, inline_to_all_callers): Break out from ...
700         (ipa_inline): ... here.
702 2013-09-06  Jan Hubicka  <jh@suse.cz>
704         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
705         only when AVX is enabled.
707 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
709         * config/aarch64/aarch64.md
710         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
711         is fpsimd_<load/store>2.
712         (load_pair<mode>): Likewise.
713         (store_pair<mode>): Likewise.
715 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
717         * config/arm/types.md (type): Add "mrs" type.
718         * config/aarch64/aarch64.md
719         (aarch64_load_tp_hard): Make type "mrs".
720         * config/arm/arm.md
721         (load_tp_hard): Make type "mrs".
722         * config/arm/cortex-a15.md: Update with new attributes.
723         * config/arm/cortex-a5.md: Update with new attributes.
724         * config/arm/cortex-a53.md: Update with new attributes.
725         * config/arm/cortex-a7.md: Update with new attributes.
726         * config/arm/cortex-a8.md: Update with new attributes.
727         * config/arm/cortex-a9.md: Update with new attributes.
728         * config/arm/cortex-m4.md: Update with new attributes.
729         * config/arm/cortex-r4.md: Update with new attributes.
730         * config/arm/fa526.md: Update with new attributes.
731         * config/arm/fa606te.md: Update with new attributes.
732         * config/arm/fa626te.md: Update with new attributes.
733         * config/arm/fa726te.md: Update with new attributes.
735 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
737         * config/aarch64/aarch64.md
738         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
739         (*movtf_aarch64): Likewise.
740         * config/arm/arm.md
741         (thumb1_movdi_insn): Use "multiple" for type where more than one
742         instruction is used for a move.
743         (*arm32_movhf): Likewise.
744         (*thumb_movdf_insn): Likewise.
746 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
748         * config/arm/types.md (type): Rename fcpys to fmov.
749         * config/arm/vfp.md
750         (*arm_movsi_vfp): Rename type fcpys as fmov.
751         (*thumb2_movsi_vfp): Likewise
752         (*movhf_vfp_neon): Likewise
753         (*movhf_vfp): Likewise
754         (*movsf_vfp): Likewise
755         (*thumb2_movsf_vfp): Likewise
756         (*movsfcc_vfp): Likewise
757         (*thumb2_movsfcc_vfp): Likewise
758         * config/aarch64/aarch64-simd.md
759         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
760         * config/aarch64/aarch64.md
761         (*movsi_aarch64): Replace type mov_reg with fmovs.
762         (*movdi_aarch64): Likewise
763         (*movsf_aarch64): Likewise
764         (*movdf_aarch64): Likewise
765         * config/arm/arm.c
766         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
767         * config/arm/iwmmxt.md
768         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
769         * config/arm/arm1020e.md: Update with new attributes.
770         * config/arm/cortex-a15-neon.md: Update with new attributes.
771         * config/arm/cortex-a5.md: Update with new attributes.
772         * config/arm/cortex-a53.md: Update with new attributes.
773         * config/arm/cortex-a7.md: Update with new attributes.
774         * config/arm/cortex-a8-neon.md: Update with new attributes.
775         * config/arm/cortex-a9.md: Update with new attributes.
776         * config/arm/cortex-m4-fpu.md: Update with new attributes.
777         * config/arm/cortex-r4f.md: Update with new attributes.
778         * config/arm/marvell-pj4.md: Update with new attributes.
779         * config/arm/vfp11.md: Update with new attributes.
781 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
783         * config/aarch64/aarch64.md
784         (*madd<mode>): Fix type attribute.
785         (*maddsi_uxtw): Likewise.
786         (*msub<mode>): Likewise.
787         (*msubsi_uxtw): Likewise.
788         (<su_optab>maddsidi4): Likewise.
789         (<su_optab>msubsidi4): Likewise.
791 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
793         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
794         * config/arm/arm.md (core_cycles): Remove fdiv.
795         * config/arm/vfp.md:
796         (*sqrtsf2_vfp): Update for attribute changes.
797         (*sqrtdf2_vfp): Likewise.
798         * config/aarch64/aarch64.md:
799         (sqrt<mode>2): Update for attribute changes.
800         * config/arm/arm1020e.md: Update with new attributes.
801         * config/arm/cortex-a15-neon.md: Update with new attributes.
802         * config/arm/cortex-a5.md: Update with new attributes.
803         * config/arm/cortex-a53.md: Update with new attributes.
804         * config/arm/cortex-a7.md: Update with new attributes.
805         * config/arm/cortex-a8-neon.md: Update with new attributes.
806         * config/arm/cortex-a9.md: Update with new attributes.
807         * config/arm/cortex-m4-fpu.md: Update with new attributes.
808         * config/arm/cortex-r4f.md: Update with new attributes.
809         * config/arm/marvell-pj4.md: Update with new attributes.
810         * config/arm/vfp11.md: Update with new attributes.
812 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
814         * config/arm/types.md
815         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
816         * config/aarch64/aarch64.md
817         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
818         new attributes.
819         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
820         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
821         (float<GPI:mode><GPF:mode>2): Likewise.
822         * config/arm/vfp.md
823         (*truncsisf2_vfp): Update with new attributes.
824         (*truncsidf2_vfp): Likewise.
825         (fixuns_truncsfsi2): Likewise.
826         (fixuns_truncdfsi2): Likewise.
827         (*floatsisf2_vfp): Likewise.
828         (*floatsidf2_vfp): Likewise.
829         (floatunssisf2): Likewise.
830         (floatunssidf2): Likewise.
831         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
832         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
833         * config/arm/arm1020e.md: Update with new attributes.
834         * config/arm/cortex-a15-neon.md: Update with new attributes.
835         * config/arm/cortex-a5.md: Update with new attributes.
836         * config/arm/cortex-a53.md: Update with new attributes.
837         * config/arm/cortex-a7.md: Update with new attributes.
838         * config/arm/cortex-a8-neon.md: Update with new attributes.
839         * config/arm/cortex-a9.md: Update with new attributes.
840         * config/arm/cortex-m4-fpu.md: Update with new attributes.
841         * config/arm/cortex-r4f.md: Update with new attributes.
842         * config/arm/marvell-pj4.md: Update with new attributes.
843         * config/arm/vfp11.md: Update with new attributes.
845 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
847         * config/aarch64/arm_neon.h
848         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
849         (vqtbx<1,2,3,4><q>_s8): Likewise.
851 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
853         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
854         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
855         (add<mode>3): Add type attribute.
856         (add<mode>3): Likewise.
857         (usadd<mode>3): Likewise.
858         (ssadd<mode>3): Likewise.
859         (sub<mode>3): Likewise.
860         (sub<mode>3): Likewise.
861         (ussub<mode>3): Likewise.
862         (sssub<mode>3): Likewise.
863         (ssmulsa3): Likewise.
864         (usmulusa3): Likewise.
865         (arm_usatsihi): Likewise.
866         * config/arm/vfp.md
867         (*movdi_vfp): Add types for all instructions.
868         (*movdi_vfp_cortexa8): Likewise.
869         (*movhf_vfp_neon): Likewise.
870         (*movhf_vfp): Likewise.
871         (*movdf_vfp): Likewise.
872         (*thumb2_movdf_vfp): Likewise.
873         (*thumb2_movdfcc_vfp): Likewise.
874         * config/arm/arm.md: Add type attribute to all insn patterns.
875         (*thumb1_adddi3): Add type attribute.
876         (*arm_adddi3): Likewise.
877         (*adddi_sesidi_di): Likewise.
878         (*adddi_zesidi_di): Likewise.
879         (*thumb1_addsi3): Likewise.
880         (addsi3_compare0): Likewise.
881         (*addsi3_compare0_scratch): Likewise.
882         (*compare_negsi_si): Likewise.
883         (cmpsi2_addneg): Likewise.
884         (*addsi3_carryin_<optab>): Likewise.
885         (*addsi3_carryin_alt2_<optab>): Likewise.
886         (*addsi3_carryin_clobercc_<optab>): Likewise.
887         (*subsi3_carryin): Likewise.
888         (*subsi3_carryin_const): Likewise.
889         (*subsi3_carryin_compare): Likewise.
890         (*subsi3_carryin_compare_const): Likewise.
891         (*arm_subdi3): Likewise.
892         (*thumb_subdi3): Likewise.
893         (*subdi_di_zesidi): Likewise.
894         (*subdi_di_sesidi): Likewise.
895         (*subdi_zesidi_di): Likewise.
896         (*subdi_sesidi_di): Likewise.
897         (*subdi_zesidi_ze): Likewise.
898         (thumb1_subsi3_insn): Likewise.
899         (*arm_subsi3_insn): Likewise.
900         (*anddi3_insn): Likewise.
901         (*anddi_zesidi_di): Likewise.
902         (*anddi_sesdi_di): Likewise.
903         (*ne_zeroextracts): Likewise.
904         (*ne_zeroextracts): Likewise.
905         (*ite_ne_zeroextr): Likewise.
906         (*ite_ne_zeroextr): Likewise.
907         (*anddi_notdi_di): Likewise.
908         (*anddi_notzesidi): Likewise.
909         (*anddi_notsesidi): Likewise.
910         (andsi_notsi_si): Likewise.
911         (thumb1_bicsi3): Likewise.
912         (*iordi3_insn): Likewise.
913         (*iordi_zesidi_di): Likewise.
914         (*iordi_sesidi_di): Likewise.
915         (*thumb1_iorsi3_insn): Likewise.
916         (*xordi3_insn): Likewise.
917         (*xordi_zesidi_di): Likewise.
918         (*xordi_sesidi_di): Likewise.
919         (*arm_xorsi3): Likewise.
920         (*andsi_iorsi3_no): Likewise.
921         (*smax_0): Likewise.
922         (*smax_m1): Likewise.
923         (*arm_smax_insn): Likewise.
924         (*smin_0): Likewise.
925         (*arm_smin_insn): Likewise.
926         (*arm_umaxsi3): Likewise.
927         (*arm_uminsi3): Likewise.
928         (*minmax_arithsi): Likewise.
929         (*minmax_arithsi_): Likewise.
930         (*satsi_<SAT:code>): Likewise.
931         (arm_ashldi3_1bit): Likewise.
932         (arm_ashrdi3_1bit): Likewise.
933         (arm_lshrdi3_1bit): Likewise.
934         (*arm_negdi2): Likewise.
935         (*thumb1_negdi2): Likewise.
936         (*arm_negsi2): Likewise.
937         (*thumb1_negsi2): Likewise.
938         (*negdi_extendsid): Likewise.
939         (*negdi_zero_extend): Likewise.
940         (*arm_abssi2): Likewise.
941         (*thumb1_abssi2): Likewise.
942         (*arm_neg_abssi2): Likewise.
943         (*thumb1_neg_abss): Likewise.
944         (one_cmpldi2): Likewise.
945         (extend<mode>di2): Likewise.
946         (*compareqi_eq0): Likewise.
947         (*arm_extendhisi2addsi): Likewise.
948         (*arm_movdi): Likewise.
949         (*thumb1_movdi_insn): Likewise.
950         (*arm_movt): Likewise.
951         (*thumb1_movsi_insn): Likewise.
952         (pic_add_dot_plus_four): Likewise.
953         (pic_add_dot_plus_eight): Likewise.
954         (tls_load_dot_plus_eight): Likewise.
955         (*thumb1_movhi_insn): Likewise.
956         (*thumb1_movsf_insn): Likewise.
957         (*movdf_soft_insn): Likewise.
958         (*thumb_movdf_insn): Likewise.
959         (cbranchsi4_insn): Likewise.
960         (cbranchsi4_scratch): Likewise.
961         (*negated_cbranchsi4): Likewise.
962         (*tbit_cbranch): Likewise.
963         (*tlobits_cbranch): Likewise.
964         (*tstsi3_cbranch): Likewise.
965         (*cbranchne_decr1): Likewise.
966         (*addsi3_cbranch): Likewise.
967         (*addsi3_cbranch_scratch): Likewise.
968         (*arm_cmpdi_insn): Likewise.
969         (*arm_cmpdi_unsig): Likewise.
970         (*arm_cmpdi_zero): Likewise.
971         (*thumb_cmpdi_zero): Likewise.
972         (*deleted_compare): Likewise.
973         (*mov_scc): Likewise.
974         (*mov_negscc): Likewise.
975         (*mov_notscc): Likewise.
976         (*cstoresi_eq0_thumb1_insn): Likewise.
977         (cstoresi_nltu_thumb1): Likewise.
978         (cstoresi_ltu_thu): Likewise.
979         (thumb1_addsi3_addgeu): Likewise.
980         (*arm_jump): Likewise.
981         (*thumb_jump): Likewise.
982         (*check_arch2): Likewise.
983         (arm_casesi_internal): Likewise.
984         (thumb1_casesi_dispatch): Likewise.
985         (*arm_indirect_jump): Likewise.
986         (*thumb1_indirect_jump): Likewise.
987         (nop): Likewise.
988         (*and_scc): Likewise.
989         (*ior_scc): Likewise.
990         (*compare_scc): Likewise.
991         (*cond_move): Likewise.
992         (*cond_arith): Likewise.
993         (*cond_sub): Likewise.
994         (*cmp_ite0): Likewise.
995         (*cmp_ite1): Likewise.
996         (*cmp_and): Likewise.
997         (*cmp_ior): Likewise.
998         (*ior_scc_scc): Likewise.
999         (*ior_scc_scc_cmp): Likewise.
1000         (*and_scc_scc): Likewise.
1001         (*and_scc_scc_cmp): Likewise.
1002         (*and_scc_scc_nod): Likewise.
1003         (*negscc): Likewise.
1004         (movcond_addsi): Likewise.
1005         (movcond): Likewise.
1006         (*ifcompare_plus_move): Likewise.
1007         (*if_plus_move): Likewise.
1008         (*ifcompare_move_plus): Likewise.
1009         (*if_move_plus): Likewise.
1010         (*ifcompare_arith_arith): Likewise.
1011         (*if_arith_arith): Likewise.
1012         (*ifcompare_arith_move): Likewise.
1013         (*if_arith_move): Likewise.
1014         (*ifcompare_move_arith): Likewise.
1015         (*if_move_arith): Likewise.
1016         (*ifcompare_move_not): Likewise.
1017         (*if_move_not): Likewise.
1018         (*ifcompare_not_move): Likewise.
1019         (*if_not_move): Likewise.
1020         (*ifcompare_shift_move): Likewise.
1021         (*if_shift_move): Likewise.
1022         (*ifcompare_move_shift): Likewise.
1023         (*if_move_shift): Likewise.
1024         (*ifcompare_shift_shift): Likewise.
1025         (*ifcompare_not_arith): Likewise.
1026         (*ifcompare_arith_not): Likewise.
1027         (*if_arith_not): Likewise.
1028         (*ifcompare_neg_move): Likewise.
1029         (*if_neg_move): Likewise.
1030         (*ifcompare_move_neg): Likewise.
1031         (*if_move_neg): Likewise.
1032         (prologue_thumb1_interwork): Likewise.
1033         (*cond_move_not): Likewise.
1034         (*sign_extract_onebit): Likewise.
1035         (*not_signextract_onebit): Likewise.
1036         (stack_tie): Likewise.
1037         (align_4): Likewise.
1038         (align_8): Likewise.
1039         (consttable_end): Likewise.
1040         (consttable_1): Likewise.
1041         (consttable_2): Likewise.
1042         (consttable_4): Likewise.
1043         (consttable_8): Likewise.
1044         (consttable_16): Likewise.
1045         (*thumb1_tablejump): Likewise.
1046         (prefetch): Likewise.
1047         (force_register_use): Likewise.
1048         (thumb_eh_return): Likewise.
1049         (load_tp_hard): Likewise.
1050         (load_tp_soft): Likewise.
1051         (tlscall): Likewise.
1052         (*arm_movtas_ze): Likewise.
1053         (*arm_rev): Likewise.
1054         (*arm_revsh): Likewise.
1055         (*arm_rev16): Likewise.
1056         * config/arm/thumb2.md
1057         (*thumb2_smaxsi3): Likewise.
1058         (*thumb2_sminsi3): Likewise.
1059         (*thumb32_umaxsi3): Likewise.
1060         (*thumb2_uminsi3): Likewise.
1061         (*thumb2_negdi2): Likewise.
1062         (*thumb2_abssi2): Likewise.
1063         (*thumb2_neg_abss): Likewise.
1064         (*thumb2_movsi_insn): Likewise.
1065         (tls_load_dot_plus_four): Likewise.
1066         (*thumb2_movhi_insn): Likewise.
1067         (*thumb2_mov_scc): Likewise.
1068         (*thumb2_mov_negs): Likewise.
1069         (*thumb2_mov_negs): Likewise.
1070         (*thumb2_mov_nots): Likewise.
1071         (*thumb2_mov_nots): Likewise.
1072         (*thumb2_movsicc_): Likewise.
1073         (*thumb2_movsfcc_soft_insn): Likewise.
1074         (*thumb2_indirect_jump): Likewise.
1075         (*thumb2_and_scc): Likewise.
1076         (*thumb2_ior_scc): Likewise.
1077         (*thumb2_ior_scc_strict_it): Likewise.
1078         (*thumb2_cond_move): Likewise.
1079         (*thumb2_cond_arith): Likewise.
1080         (*thumb2_cond_ari): Likewise.
1081         (*thumb2_cond_sub): Likewise.
1082         (*thumb2_negscc): Likewise.
1083         (*thumb2_movcond): Likewise.
1084         (thumb2_casesi_internal): Likewise.
1085         (thumb2_casesi_internal_pic): Likewise.
1086         (*thumb2_alusi3_short): Likewise.
1087         (*thumb2_mov<mode>_shortim): Likewise.
1088         (*thumb2_addsi_short): Likewise.
1089         (*thumb2_subsi_short): Likewise.
1090         (thumb2_addsi3_compare0): Likewise.
1091         (*thumb2_cbz): Likewise.
1092         (*thumb2_cbnz): Likewise.
1093         (*thumb2_one_cmplsi2_short): Likewise.
1094         (*thumb2_negsi2_short): Likewise.
1095         (*orsi_notsi_si): Likewise.
1096         * config/arm/arm1020e.md: Update with new attributes.
1097         * config/arm/arm1026ejs.md: Update with new attributes.
1098         * config/arm/arm1136jfs.md: Update with new attributes.
1099         * config/arm/arm926ejs.md: Update with new attributes.
1100         * config/arm/cortex-a15.md: Update with new attributes.
1101         * config/arm/cortex-a5.md: Update with new attributes.
1102         * config/arm/cortex-a53.md: Update with new attributes.
1103         * config/arm/cortex-a7.md: Update with new attributes.
1104         * config/arm/cortex-a8.md: Update with new attributes.
1105         * config/arm/cortex-a9.md: Update with new attributes.
1106         * config/arm/cortex-m4.md: Update with new attributes.
1107         * config/arm/cortex-r4.md: Update with new attributes.
1108         * config/arm/fa526.md: Update with new attributes.
1109         * config/arm/fa606te.md: Update with new attributes.
1110         * config/arm/fa626te.md: Update with new attributes.
1111         * config/arm/fa726te.md: Update with new attributes.
1113 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1115         * config/aarch64/aarch64-simd.md
1116         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1117         <vwx> iterator to ensure correct register choice.
1118         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1119         (aarch64_sqdmull_n<mode>): Likewise.
1120         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1121         * config/aarch64/arm_neon.h
1122         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1123         (vml<as><q>_n_<su>16): Likewise.
1124         (vml<as>l_high_lane<q>_<su>16): Likewise.
1125         (vml<as>l_high_n_<su>16): Likewise.
1126         (vml<as>l_lane<q>_<su>16): Likewise.
1127         (vml<as>l_n_<su>16): Likewise.
1128         (vmul<q>_lane<q>_<su>16): Likewise.
1129         (vmul<q>_n_<su>16): Likewise.
1130         (vmull_lane<q>_<su>16): Likewise.
1131         (vmull_n_<su>16): Likewise.
1132         (vmull_high_lane<q>_<su>16): Likewise.
1133         (vmull_high_n_<su>16): Likewise.
1134         (vqrdmulh<q>_n_s16): Likewise.
1136 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1138         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1139         have the correct lane parameter.
1141 2013-09-06  Richard Biener <rguenther@suse.de>
1143         * cfganal.c (control_dependences::~control_dependences):
1144         Properly free all of the vector.
1146 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1148         PR target/58269
1149         * config/i386/i386.c (ix86_conditional_register_usage):
1150         Proper initialize extended SSE registers.
1152 2013-09-06  Jan Hubicka  <jh@suse.cz>
1154         PR tree-optimization/58311
1155         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1157 2013-09-06  Jan Hubicka  <jh@suse.cz>
1159         * Makefile.in (tree-sra.o): Update dependencies.
1160         * tree-sra.c: Include ipa-utils.h
1161         (scan_function): Use recursive_call_p.
1162         (has_caller_p): New function.
1163         (cgraph_for_node_and_aliases): Count also callers of aliases.
1165 2013-09-06  Jan Hubicka  <jh@suse.cz>
1167         PR middle-end/58094
1168         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1169         * tree-tailcall.c: Include ipa-utils.h.
1170         (find_tail_calls): Use it.
1171         * ipa-pure-const.c (check_call): Likewise.
1172         * ipa-utils.c (recursive_call_p): New function.
1173         * ipa-utils.h (recursive_call_p): Dclare.
1174         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1175         (symtab_semantically_equivalent_p): New function.
1176         * Makefile.in (tree-tailcall.o): Update dependencies.
1178 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1180         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1181         non-inlinable part.
1183 2013-09-06  Richard Biener  <rguenther@suse.de>
1185         * lto-streamer.h (lto_global_var_decls): Remove.
1186         * Makefile.in (OBJS): Remove lto-symtab.o.
1187         (lto-symtab.o): Remove.
1188         (GTFILES): Remove lto-symtab.c
1189         * lto-symtab.c: Move to lto/
1191 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1193         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1194         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1195         (UNSPEC_FPINT_RINT): New constant definitions.
1196         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1197         definition with 2 attributes.
1198         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1199         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1200         definitions.
1202 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1204         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1205         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1206         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1207         attribute to "z196_alone".
1208         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1209         "zEC12_simple".
1211 2013-09-06  Richard Biener  <rguenther@suse.de>
1213         * basic-block.h (class control_dependences): New.
1214         * tree-ssa-dce.c (control_dependence_map): Remove.
1215         (cd): New global.
1216         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1217         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1218         find_pdom, find_control_dependence, find_all_control_dependences):
1219         Move to cfganal.c.
1220         (mark_control_dependent_edges_necessary,
1221         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1222         tree_dce_done, perform_tree_ssa_dce): Adjust.
1223         * cfganal.c (set_control_dependence_map_bit,
1224         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1225         find_all_control_dependences): Move from tree-ssa-dce.c and
1226         implement as methods of control_dependences class.
1227         (control_dependences::control_dependences): New.
1228         (control_dependences::~control_dependences): Likewise.
1229         (control_dependences::get_edges_dependent_on): Likewise.
1230         (control_dependences::get_edge): Likewise.
1232 2013-09-04  Jan Hubicka  <jh@suse.cz>
1234         * tree.c (types_same_for_odr): Drop overactive check.
1235         * ipa-devirt.c (hash_type_name): Likewise.
1237 2013-09-04  Jan Hubicka  <jh@suse.cz>
1239         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1240         (analyze_functions): ... here.
1242 2013-09-04  Jan Hubicka  <jh@suse.cz>
1244         PR middle-end/58201
1245         * cgraphunit.c (analyze_functions): Clear AUX fields
1246         after processing; initialize assembler name has.
1248 2013-09-05  Jeff Law  <law@redhat.com>
1250         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1251         from thread_around_empty_block.  Record threading path into PATH.
1252         Recurse if threading through the initial block is successful.
1253         (thread_across_edge): Corresponding changes to slightly simplify.
1255 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1257         * config/aarch64/aarch64.md
1258         (type): Remove frecpe, frecps, frecpx.
1259         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1260         fix to be a TARGET_SIMD instruction.
1261         (aarch64_frecps): Remove.
1262         * config/aarch64/aarch64-simd.md
1263         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1264         (aarch64_frecps<mode>): Handle all float/vector of float modes.
1266 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1267             Sofiane Naci  <sofiane.naci@arm.com>
1269         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1270         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1271         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1272         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1273         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
1274         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1275         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
1276         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
1277         into "clz, "rbit".  Rename "shift" to "shift_imm".
1278         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1279         changes.  Update for attribute changes all occurrences of arlo_* and
1280         shift* types.
1281         * config/arm/arm-fixed.md: Update for attribute changes
1282         all occurrences of arlo_* types.
1283         * config/arm/thumb2.md: Update for attribute changes all occurrences
1284         of arlo_* types.
1285         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
1286         (cortexa7_older_only): Likewise.
1287         (cortexa7_younger):  Likewise.
1288         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1289         (1020alu_shift_op): Likewise.
1290         (1020alu_shift_reg_op): Likewise.
1291         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1292         (alu_shift_op): Likewise.
1293         (alu_shift_reg_op): Likewise.
1294         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1295         (11_alu_shift_op): Likewise.
1296         (11_alu_shift_reg_op): Likewise.
1297         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1298         (9_alu_shift_reg_op): Likewise.
1299         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1300         attribute changes.
1301         (cortex_a15_alu_shift): Likewise.
1302         (cortex_a15_alu_shift_reg): Likewise.
1303         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1304         attribute changes.
1305         (cortex_a5_alu_shift): Likewise.
1306         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1307         attribute changes.
1308         (cortex_a53_alu_shift): Likewise.
1309         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1310         attribute changes.
1311         (cortex_a7_alu_reg): Likewise.
1312         (cortex_a7_alu_shift): Likewise.
1313         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1314         attribute changes.
1315         (cortex_a8_alu_shift): Likewise.
1316         (cortex_a8_alu_shift_reg): Likewise.
1317         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1318         (cortex_a9_dp_shift): Likewise.
1319         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1320         attribute changes.
1321         * config/arm/cortex-r4.md
1322         (cortex_r4_alu): Update for attribute changes.
1323         (cortex_r4_mov): Likewise.
1324         (cortex_r4_alu_shift_reg): Likewise.
1325         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1326         (526_alu_shift_op): Likewise.
1327         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1328         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1329         (626te_alu_shift_op): Likewise.
1330         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1331         (726te_alu_shift_op): Likewise.
1332         (726te_alu_shift_reg_op): Likewise.
1333         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1334         (mp626_alu_shift_op): Likewise.
1335         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1336         (pj4_alu_conds): Likewise.
1337         (pj4_shift): Likewise.
1338         (pj4_shift_conds): Likewise.
1339         (pj4_alu_shift): Likewise.
1340         (pj4_alu_shift_conds): Likewise.
1341         * config/aarch64/aarch64.md: Update for attribute change
1342         all occurrences of arlo_* and shift* types.
1344 2013-09-05  Mike Stump  <mikestump@comcast.net>
1346         * tree.h: Move documentation for tree_function_decl to tree-core.h
1347         with the declaration.
1349 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
1351         PR target/58139
1352         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1353         looking for widest mode.
1355 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
1357         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1359 2013-09-05  Richard Biener  <rguenther@suse.de>
1361         PR tree-optimization/58137
1362         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1363         Do not create vectors of pointers.
1364         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1365         types for the components of the vector initializer.
1366         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1367         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1369 2013-09-05  Martin Jambor  <mjambor@suse.cz>
1371         * ipa-prop.c (remove_described_reference): Accept missing references,
1372         return false if that hppens, otherwise return true.
1373         (cgraph_node_for_jfunc): New function.
1374         (try_decrement_rdesc_refcount): Likewise.
1375         (try_make_edge_direct_simple_call): Use them.
1376         (ipa_edge_removal_hook): Remove references from rdescs.
1377         (ipa_edge_duplication_hook): Clone rdescs and their references
1378         when the new edge has the same caller as the old one.
1379         * cgraph.c (cgraph_resolve_speculation): Remove speculative
1380         reference before removing any edges.
1382 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
1384         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1385         initial store.
1386         * thumb2.md (thumb2_storewb_parisi): New pattern.
1388 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
1390         * config/aarch64/aarch64-option-extensions.def: Add
1391         AARCH64_OPT_EXTENSION of 'crc'.
1392         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1393         (AARCH64_ISA_CRC): Ditto.
1394         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1395         description of the CRC extension.
1397 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1399         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1400         * config/rs6000/linux.h: Ditto.
1401         * alpha/linux.h: Ditto.
1402         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1403         no_c99_libc_has_function.
1404         * config/c6x/uclinux-elf.h: Ditto.
1405         * config/lm32/uclinux-elf.h: Ditto.
1406         * config/m68k/uclinux.h: Ditto.
1407         * config/moxie/uclinux.h: Ditto.
1408         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1409         (crisv32-*-linux*, cris-*-linux*): Ditto.
1410         * config/bfin/bfin.c: Include "tm_p.h".
1412 2013-09-05  Richard Biener  <rguenther@suse.de>
1414         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1415         check for a definition without a basic-block.
1417 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1418             Sofiane Naci <sofiane.naci@arm.com>
1420         * config/aarch64/aarch64.md
1421         (*movti_aarch64): Rename r_2_f and f_2_r.
1422         (*movsf_aarch64): Likewise.
1423         (*movdf_aarch64): Likewise.
1424         (*movtf_aarch64): Likewise.
1425         (aarch64_movdi_<mode>low): Likewise.
1426         (aarch64_movdi_<mode>high): Likewise.
1427         (aarch64_mov<mode>high_di): Likewise.
1428         (aarch64_mov<mode>low_di): Likewise.
1429         (aarch64_movtilow_tilow): Likewise.
1430         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
1431         values to config/arm/types.md
1432         (attribute "conds"): Update for attribute change.
1433         (anddi3_insn): Likewise.
1434         (iordi3_insn): Likewise.
1435         (xordi3_insn): Likewise.
1436         (one_cmpldi2): Likewise.
1437         * config/arm/types.md (type): Add Neon types.
1438         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1439         use "type" attribute.
1440         (movmisalign<mode>_neon_store): Likewise.
1441         (movmisalign<mode>_neon_load): Likewise.
1442         (vec_set<mode>_internal): Likewise.
1443         (vec_setv2di_internal): Likewise.
1444         (vec_extract<mode>): Likewise.
1445         (vec_extractv2di): Likewise.
1446         (add<mode>3_neon): Likewise.
1447         (adddi3_neon): Likewise.
1448         (sub<mode>3_neon): Likewise.
1449         (subdi3_neon): Likewise.
1450         (mul<mode>3_neon): Likewise.
1451         (mul<mode>3add<mode>_neon): Likewise.
1452         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1453         (fma<VCVTF:mode>4)): Likewise.
1454         (fma<VCVTF:mode>4_intrinsic): Likewise.
1455         (fmsub<VCVTF:mode>4)): Likewise.
1456         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1457         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1458         (ior<mode>3): Likewise.
1459         (and<mode>3): Likewise.
1460         (anddi3_neon): Likewise.
1461         (orn<mode>3_neon): Likewise.
1462         (orndi3_neon): Likewise.
1463         (bic<mode>3_neon): Likewise.
1464         (bicdi3_neon): Likewise.
1465         (xor<mode>3): Likewise.
1466         (one_cmpl<mode>2): Likewise.
1467         (abs<mode>2): Likewise.
1468         (neg<mode>2): Likewise.
1469         (umin<mode>3_neon): Likewise.
1470         (umax<mode>3_neon): Likewise.
1471         (smin<mode>3_neon): Likewise.
1472         (smax<mode>3_neon): Likewise.
1473         (vashl<mode>3): Likewise.
1474         (vashr<mode>3_imm): Likewise.
1475         (vlshr<mode>3_imm): Likewise.
1476         (ashl<mode>3_signed): Likewise.
1477         (ashl<mode>3_unsigned): Likewise.
1478         (neon_load_count): Likewise.
1479         (ashldi3_neon_noclobber): Likewise.
1480         (signed_shift_di3_neon): Likewise.
1481         (unsigned_shift_di3_neon): Likewise.
1482         (ashrdi3_neon_imm_noclobber): Likewise.
1483         (lshrdi3_neon_imm_noclobber): Likewise.
1484         (widen_ssum<mode>3): Likewise.
1485         (widen_usum<mode>3): Likewise.
1486         (quad_halves_<code>v4si): Likewise.
1487         (quad_halves_<code>v4sf): Likewise.
1488         (quad_halves_<code>v8hi): Likewise.
1489         (quad_halves_<code>v16qi): Likewise.
1490         (reduc_splus_v2di): Likewise.
1491         (neon_vpadd_internal<mode>): Likewise.
1492         (neon_vpsmin<mode>): Likewise.
1493         (neon_vpsmax<mode>): Likewise.
1494         (neon_vpumin<mode>): Likewise.
1495         (neon_vpumax<mode>): Likewise.
1496         (ss_add<mode>_neon): Likewise.
1497         (us_add<mode>_neon): Likewise.
1498         (ss_sub<mode>_neon): Likewise.
1499         (us_sub<mode>_neon): Likewise.
1500         (neon_vadd<mode>_unspec): Likewise.
1501         (neon_vaddl<mode>): Likewise.
1502         (neon_vaddw<mode>): Likewise.
1503         (neon_vhadd<mode>): Likewise.
1504         (neon_vqadd<mode>): Likewise.
1505         (neon_vaddhn<mode>): Likewise.
1506         (neon_vmul<mode>): Likewise.
1507         (neon_vmla<mode>): Likewise.
1508         (neon_vmlal<mode>): Likewise.
1509         (neon_vmls<mode>): Likewise.
1510         (neon_vmlsl<mode>): Likewise.
1511         (neon_vqdmulh<mode>): Likewise.
1512         (neon_vqdmlal<mode>): Likewise.
1513         (neon_vqdmlsl<mode>): Likewise.
1514         (neon_vmull<mode>): Likewise.
1515         (neon_vqdmull<mode>): Likewise.
1516         (neon_vsub<mode>_unspec): Likewise.
1517         (neon_vsubl<mode>): Likewise.
1518         (neon_vsubw<mode>): Likewise.
1519         (neon_vqsub<mode>): Likewise.
1520         (neon_vhsub<mode>): Likewise.
1521         (neon_vsubhn<mode>): Likewise.
1522         (neon_vceq<mode>): Likewise.
1523         (neon_vcge<mode>): Likewise.
1524         (neon_vcgeu<mode>): Likewise.
1525         (neon_vcgt<mode>): Likewise.
1526         (neon_vcgtu<mode>): Likewise.
1527         (neon_vcle<mode>): Likewise.
1528         (neon_vclt<mode>): Likewise.
1529         (neon_vcage<mode>): Likewise.
1530         (neon_vcagt<mode>): Likewise.
1531         (neon_vtst<mode>): Likewise.
1532         (neon_vabd<mode>): Likewise.
1533         (neon_vabdl<mode>): Likewise.
1534         (neon_vaba<mode>): Likewise.
1535         (neon_vabal<mode>): Likewise.
1536         (neon_vmax<mode>): Likewise.
1537         (neon_vmin<mode>): Likewise.
1538         (neon_vpaddl<mode>): Likewise.
1539         (neon_vpadal<mode>): Likewise.
1540         (neon_vpmax<mode>): Likewise.
1541         (neon_vpmin<mode>): Likewise.
1542         (neon_vrecps<mode>): Likewise.
1543         (neon_vrsqrts<mode>): Likewise.
1544         (neon_vqabs<mode>): Likewise.
1545         (neon_vqneg<mode>): Likewise.
1546         (neon_vcls<mode>): Likewise.
1547         (clz<mode>2): Likewise.
1548         (popcount<mode>2): Likewise.
1549         (neon_vrecpe): Likewise.
1550         (neon_vrsqrte): Likewise.
1551         (neon_vget_lane<mode>_sext_internal): Likewise.
1552         (neon_vget_lane<mode>_zext_internal): Likewise.
1553         (neon_vdup_n<mode>): Likewise.
1554         (neon_vdup_nv2di): Likewise.
1555         (neon_vdpu_lane<mode>_internal): Likewise.
1556         (neon_vswp<mode>): Likewise.
1557         (float<mode><V_cvtto>2): Likewise.
1558         (floatuns<mode><V_cvtto>2): Likewise.
1559         (fix_trunc<mode><V_cvtto>)2): Likewise
1560         (fixuns_trunc<mode><V_cvtto)2): Likewise.
1561         (neon_vcvt<mode>): Likewise.
1562         (neon_vcvtv4sfv4hf): Likewise.
1563         (neon_vcvtv4hfv4sf): Likewise.
1564         (neon_vcvt_n<mode>): Likewise.
1565         (neon_vmovn<mode>): Likewise.
1566         (neon_vqmovn<mode>): Likewise.
1567         (neon_vqmovun<mode>): Likewise.
1568         (neon_vmovl<mode>): Likewise.
1569         (neon_vmul_lane<mode>): Likewise.
1570         (neon_vmull_lane<mode>): Likewise.
1571         (neon_vqdmull_lane<mode>): Likewise.
1572         (neon_vqdmulh_lane<mode>): Likewise.
1573         (neon_vmla_lane<mode>): Likewise.
1574         (neon_vmlal_lane<mode>): Likewise.
1575         (neon_vqdmlal_lane<mode>): Likewise.
1576         (neon_vmls_lane<mode>): Likewise.
1577         (neon_vmlsl_lane<mode>): Likewise.
1578         (neon_vqdmlsl_lane<mode>): Likewise.
1579         (neon_vext<mode>): Likewise.
1580         (neon_vrev64<mode>): Likewise.
1581         (neon_vrev32<mode>): Likewise.
1582         (neon_vrev16<mode>): Likewise.
1583         (neon_vbsl<mode>_internal): Likewise.
1584         (neon_vshl<mode>): Likewise.
1585         (neon_vqshl<mode>): Likewise.
1586         (neon_vshr_n<mode>): Likewise.
1587         (neon_vshrn_n<mode>): Likewise.
1588         (neon_vqshrn_n<mode>): Likewise.
1589         (neon_vqshrun_n<mode>): Likewise.
1590         (neon_vshl_n<mode>): Likewise.
1591         (neon_vqshl_n<mode>): Likewise.
1592         (neon_vqshlu_n<mode>): Likewise.
1593         (neon_vshll_n<mode>): Likewise.
1594         (neon_vsra_n<mode>): Likewise.
1595         (neon_vsri_n<mode>): Likewise.
1596         (neon_vsli_n<mode>): Likewise.
1597         (neon_vtbl1v8qi): Likewise.
1598         (neon_vtbl2v8qi): Likewise.
1599         (neon_vtbl3v8qi): Likewise.
1600         (neon_vtbl4v8qi): Likewise.
1601         (neon_vtbx1v8qi): Likewise.
1602         (neon_vtbx2v8qi): Likewise.
1603         (neon_vtbx3v8qi): Likewise.
1604         (neon_vtbx4v8qi): Likewise.
1605         (neon_vtrn<mode>_internal): Likewise.
1606         (neon_vzip<mode>_internal): Likewise.
1607         (neon_vuzp<mode>_internal): Likewise.
1608         (neon_vld1<mode>): Likewise.
1609         (neon_vld1_lane<mode>): Likewise.
1610         (neon_vld1_dup<mode>): Likewise.
1611         (neon_vld1_dupv2di): Likewise.
1612         (neon_vst1<mode>): Likewise.
1613         (neon_vst1_lane<mode>): Likewise.
1614         (neon_vld2<mode>): Likewise.
1615         (neon_vld2_lane<mode>): Likewise.
1616         (neon_vld2_dup<mode>): Likewise.
1617         (neon_vst2<mode>): Likewise.
1618         (neon_vst2_lane<mode>): Likewise.
1619         (neon_vld3<mode>): Likewise.
1620         (neon_vld3qa<mode>): Likewise.
1621         (neon_vld3qb<mode>): Likewise.
1622         (neon_vld3_lane<mode>): Likewise.
1623         (neon_vld3_dup<mode>): Likewise.
1624         (neon_vst3<mode>): Likewise.
1625         (neon_vst3qa<mode>): Likewise.
1626         (neon_vst3qb<mode>): Likewise.
1627         (neon_vst3_lane<mode>): Likewise.
1628         (neon_vld4<mode>): Likewise.
1629         (neon_vld4qa<mode>): Likewise.
1630         (neon_vld4qb<mode>): Likewise.
1631         (neon_vld4_lane<mode>): Likewise.
1632         (neon_vld4_dup<mode>): Likewise.
1633         (neon_vst4<mode>): Likewise.
1634         (neon_vst4qa<mode>): Likewise.
1635         (neon_vst4qb<mode>): Likewise.
1636         (neon_vst4_lane<mode>): Likewise.
1637         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1638         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1639         (neon_vec_<US>mult_lo_<mode>): Likewise.
1640         (neon_vec_<US>mult_hi_<mode>): Likewise.
1641         (neon_vec_<US>shiftl_<mode>): Likewise.
1642         (neon_unpack<US>_<mode>): Likewise.
1643         (neon_vec_<US>mult_<mode>): Likewise.
1644         (vec_pack_trunc_<mode>): Likewise.
1645         (neon_vec_pack_trunk_<mode>): Likewise.
1646         (neon_vabd<mode>_2): Likewise.
1647         (neon_vabd<mode>_3): Likewise.
1648         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
1649         (thumb2_movsi_vfp): Likewise.
1650         (movdi_vfp): Likewise.
1651         (movdi_vfp_cortexa8): Likewise.
1652         (movhf_vfp_neon): Likewise.
1653         (movhf_vfp): Likewiwse.
1654         (movsf_vfp): Likewiwse.
1655         (thumb2_movsf_vfp): Likewiwse.
1656         (movdf_vfp): Likewise.
1657         (thumb2_movdf_vfp): Likewise.
1658         (movsfcc_vfp): Likewise.
1659         (thumb2_movsfcc_vfp): Likewise.
1660         (movdfcc_vfp): Likewise.
1661         (thumb2_movdfcc_vfp): Likewise.
1662         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
1663         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
1664         (v10_v2c): Likewise.
1665         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
1666         attribute change.
1667         (cortex_a15_neon_int_2): Likewise.
1668         (cortex_a15_neon_int_3): Likewise.
1669         (cortex_a15_neon_int_4): Likewise.
1670         (cortex_a15_neon_int_5): Likewise.
1671         (cortex_a15_neon_vqneg_vqabs): Likewise.
1672         (cortex_a15_neon_vmov): Likewise.
1673         (cortex_a15_neon_vaba): Likewise.
1674         (cortex_a15_neon_vaba_qqq): Likewise.
1675         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1676         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1677         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
1678         scalar_64_32_long_scalar): Likewise.
1679         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1680         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1681         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1682         lotype_qdd_64_32_long): Likewise.
1683         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1684         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1685         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1686         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1687         (cortex_a15_neon_shift_1): Likewise.
1688         (cortex_a15_neon_shift_2): Likewise.
1689         (cortex_a15_neon_shift_3): Likewise.
1690         (cortex_a15_neon_vshl_ddd): Likewise.
1691         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1692         (cortex_a15_neon_vsra_vrsra): Likewise.
1693         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1694         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1695         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1696         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1697         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1698         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1699         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1700         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1701         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1702         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1703         (cortex_a15_neon_bp_simple): Likewise.
1704         (cortex_a15_neon_bp_2cycle): Likewise.
1705         (cortex_a15_neon_bp_3cycle): Likewise.
1706         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1707         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1708         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1709         (cortex_a15_neon_vld2_4_regs): Likewise.
1710         (cortex_a15_neon_vld3_vld4): Likewise.
1711         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1712         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1713         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
1714         (cortex_a15_neon_vst3_vst4): Likewise.
1715         (cortex_a15_neon_vld1_vld2_lane): Likewise.
1716         (cortex_a15_neon_vld3_vld4_lane" 10
1717         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1718         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1719         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1720         (cortex_a15_neon_ldm_2): Likewise.0
1721         (cortex_a15_neon_stm_2): Likewise.
1722         (cortex_a15_neon_mcr): Likewise.
1723         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1724         (cortex_a15_neon_mrc): Likewise.
1725         (cortex_a15_neon_mrrc): Likewise.
1726         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
1727         change.
1728         (cortex_a15_alu_shift): Likewise.
1729         (cortex_a15_alu_shift_reg): Likewise.
1730         (cortex_a15_mult32): Likewise.
1731         (cortex_a15_mult64): Likewise.
1732         (cortex_a15_block): Likewise.
1733         (cortex_a15_branch): Likewise.
1734         (cortex_a15_load1): Likewise.
1735         (cortex_a15_load3): Likewise.
1736         (cortex_a15_store1): Likewise.
1737         (cortex_a15_store3): Likewise.
1738         (cortex_a15_call): Likewise.
1739         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
1740         (cortex_a5_f2r): Likewise.
1741         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
1742         change.
1743         (cortex_a53_f2r): Likewise.
1744         * config/arm/cortex-a7.md
1745         (cortex_a7_branch): Update for attribute change.
1746         (cortex_a7_call): Likewise.
1747         (cortex_a7_alu_imm): Likewise.
1748         (cortex_a7_alu_reg): Likewise.
1749         (cortex_a7_alu_shift): Likewise.
1750         (cortex_a7_mul): Likewise.
1751         (cortex_a7_load1): Likewise.
1752         (cortex_a7_store1): Likewise.
1753         (cortex_a7_load2): Likewise.
1754         (cortex_a7_store2): Likewise.
1755         (cortex_a7_load3): Likewise.
1756         (cortex_a7_store3): Likewise.
1757         (cortex_a7_load4): Likewise.
1758         (cortex_a7_store4): Likewise.
1759         (cortex_a7_fpalu): Likewise.
1760         (cortex_a7_fconst): Likewise.
1761         (cortex_a7_fpmuls): Likewise.
1762         (cortex_a7_neon_mul): Likewise.
1763         (cortex_a7_fpmacs): Likewise.
1764         (cortex_a7_neon_mla: Likewise.
1765         (cortex_a7_fpmuld: Likewise.
1766         (cortex_a7_fpmacd: Likewise.
1767         (cortex_a7_fpfmad: Likewise.
1768         (cortex_a7_fdivs: Likewise.
1769         (cortex_a7_fdivd: Likewise.
1770         (cortex_a7_r2f: Likewise.
1771         (cortex_a7_f2r: Likewise.
1772         (cortex_a7_f_flags: Likewise.
1773         (cortex_a7_f_loads: Likewise.
1774         (cortex_a7_f_loadd: Likewise.
1775         (cortex_a7_f_stores: Likewise.
1776         (cortex_a7_f_stored: Likewise.
1777         (cortex_a7_neon): Likewise.
1778         * config/arm/cortex-a8-neon.md
1779         (cortex_a8_neon_mrc): Update for attribute change.
1780         (cortex_a8_neon_mrrc): Likewise.
1781         (cortex_a8_neon_int_1): Likewise.
1782         (cortex_a8_neon_int_2): Likewise.
1783         (cortex_a8_neon_int_3): Likewise.
1784         (cortex_a8_neon_int_4): Likewise.
1785         (cortex_a8_neon_int_5): Likewise.
1786         (cortex_a8_neon_vqneg_vqabs): Likewise.
1787         (cortex_a8_neon_vmov): Likewise.
1788         (cortex_a8_neon_vaba): Likewise.
1789         (cortex_a8_neon_vaba_qqq): Likewise.
1790         (cortex_a8_neon_vsma): Likewise.
1791         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1792         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1793         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1794         long_scalar): Likewise.
1795         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1796         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1797         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1798         long_scalar_qdd_64_32_long): Likewise.
1799         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1800         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1801         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1802         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1803         (cortex_a8_neon_shift_1): Likewise.
1804         (cortex_a8_neon_shift_2): Likewise.
1805         (cortex_a8_neon_shift_3): Likewise.
1806         (cortex_a8_neon_vshl_ddd): Likewise.
1807         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1808         (cortex_a8_neon_vsra_vrsra): Likewise.
1809         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1810         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1811         (cortex_a8_neon_fp_vsum): Likewise.
1812         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1813         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1814         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1815         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1816         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1817         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1818         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1819         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1820         (cortex_a8_neon_bp_simple): Likewise.
1821         (cortex_a8_neon_bp_2cycle): Likewise.
1822         (cortex_a8_neon_bp_3cycle): Likewise.
1823         (cortex_a8_neon_ldr): Likewise.
1824         (cortex_a8_neon_str): Likewise.
1825         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1826         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1827         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1828         (cortex_a8_neon_vld2_4_regs): Likewise.
1829         (cortex_a8_neon_vld3_vld4): Likewise.
1830         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1831         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1832         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1833         (cortex_a8_neon_vst3_vst4): Likewise.
1834         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1835         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1836         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1837         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1838         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1839         (cortex_a8_neon_mcr): Likewise.
1840         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1841         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
1842         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
1843         change.
1844         (ca9_neon_mrrc): Likewise.
1845         (cortex_a9_neon_int_1): Likewise.
1846         (cortex_a9_neon_int_2): Likewise.
1847         (cortex_a9_neon_int_3): Likewise.
1848         (cortex_a9_neon_int_4): Likewise.
1849         (cortex_a9_neon_int_5): Likewise.
1850         (cortex_a9_neon_vqneg_vqabs): Likewise.
1851         (cortex_a9_neon_vmov): Likewise.
1852         (cortex_a9_neon_vaba): Likewise.
1853         (cortex_a9_neon_vaba_qqq): Likewise.
1854         (cortex_a9_neon_vsma): Likewise.
1855         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1856         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1857         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1858         long_scalar): Likewise.
1859         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1860         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1861         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1862         long_scalar_qdd_64_32_long): Likewise.
1863         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1864         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1865         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1866         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1867         (cortex_a9_neon_shift_1): Likewise.
1868         (cortex_a9_neon_shift_2): Likewise.
1869         (cortex_a9_neon_shift_3): Likewise.
1870         (cortex_a9_neon_vshl_ddd): Likewise.
1871         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1872         (cortex_a9_neon_vsra_vrsra): Likewise.
1873         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1874         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1875         (cortex_a9_neon_fp_vsum): Likewise.
1876         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1877         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1878         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1879         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1880         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1881         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1882         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1883         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1884         (cortex_a9_neon_bp_simple): Likewise.
1885         (cortex_a9_neon_bp_2cycle): Likewise.
1886         (cortex_a9_neon_bp_3cycle): Likewise.
1887         (cortex_a9_neon_ldr): Likewise.
1888         (cortex_a9_neon_str): Likewise.
1889         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1890         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1891         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1892         (cortex_a9_neon_vld2_4_regs): Likewise.
1893         (cortex_a9_neon_vld3_vld4): Likewise.
1894         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1895         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1896         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1897         (cortex_a9_neon_vst3_vst4): Likewise.
1898         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1899         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1900         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1901         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1902         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1903         (cortex_a9_neon_mcr): Likewise.
1904         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1905         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
1906         (cortex_a9_fps): Likewise.
1907         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
1908         change.
1909         (cortex_m4_fmuls): Likewise.
1910         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
1911         change.
1912         (cortex_r4_mrc): Likewise.
1913         * config/arm/iterators.md: Update comment referring to neon_type.
1914         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
1915         (iwmmxt_movsi_insn): Likewise.
1916         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
1917         attribute change.
1918         (pj4_core_to_vfp): Likewise.
1919         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
1920         attribute change.
1921         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
1922         (vfp_fstore): Likewise.
1923         * doc/md.texi: Change references to neon_type to refer to type.
1925 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1927         * tree.h (DECL_BUILT_IN): Fix typo in comment.
1929 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
1931         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
1932         lglobl if not weak.
1934 2013-09-04  Easwaran Raman  <eraman@google.com>
1936         PR middle-end/57370
1937         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
1938         (build_and_add_sum): Use it.
1939         (appears_later_in_bb): Simplify code.
1941 2013-09-04  Teresa Johnson  <tejohnson@google.com>
1943         * dumpfile.c (dump_finish): Don't close stderr/stdout.
1945 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
1947         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
1949 2013-09-04  Jan Hubicka  <jh@suse.cz>
1951         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
1952         * ipa-devirt.c: Include diganostic.h
1953         (odr_type_d): Add types and types_set.
1954         (hash_type_name): Work for types with vtables during LTO.
1955         (odr_hasher::remove): Fix comment; destroy types_set.
1956         (add_type_duplicate): New function,
1957         (get_odr_type): Use it.
1958         (dump_type_inheritance_graph): Dump type duplicates.
1959         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
1960         graph.
1961         * tree.c (types_same_for_odr): Give exact answers on types with
1962         virtual tables.
1964 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1966         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
1967         explaining their differences.
1969 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
1971         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
1973 2013-09-03  Jeff Law  <law@redhat.com>
1975         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
1976         when not threading through a joiner block.  Pass joiner/no joiner
1977         state to register_jump_thread.
1978         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
1979         state from argument rather than implying on path length.
1980         Dump the entire jump thread path into debugging dump.
1981         * tree-flow.h (register_jump_thread): Update prototype.
1983 2013-08-29  Xinliang David Li  <davidxl@google.com>
1985         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1986         Remove a trivial gcc_assert.
1988 2013-08-29  Xinliang David Li  <davidxl@google.com>
1990         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
1991         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
1992         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1993         Delay base decl alignment adjustment.
1994         * tree-vectorizer.c (vect_destroy_datarefs): New function.
1995         * tree-vectorizer.h: New data structure.
1996         (set_dr_misalignment): New function.
1997         (dr_misalignment): Ditto.
1998         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
1999         (vectorizable_load): Ditto.
2000         (ensure_base_align): New function.
2001         (vectorize_loops): Add dbg_cnt support.
2002         (execute_vect_slp): Ditto.
2003         * dbgcnt.def: New debug counter.
2004         * Makefile: New dependency.
2006 2013-09-03  Meador Inge  <meadori@codesourcery.com>
2008         Revert:
2010         2013-08-30  Meador Inge  <meadori@codesourcery.com>
2012         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2014 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
2016         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2017         function descriptor.
2019 2013-09-03  Richard Biener  <rguenther@suse.de>
2021         * tree-affine.c (add_elt_to_tree): Fix association issue,
2022         avoid useless converts and make sure to always return a
2023         properly typed result.
2025 2013-09-03  Richard Biener  <rguenther@suse.de>
2027         PR middle-end/57656
2028         * fold-const.c (negate_expr_p): Fix division case.
2029         (negate_expr): Likewise.
2031 2013-09-03  Richard Biener  <rguenther@suse.de>
2033         PR lto/58285
2034         * tree-streamer-out.c: Include tm.h.
2035         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2037 2013-09-03  Jan Hubicka  <jh@suse.cz>
2039         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2041 2013-09-03  Alan Modra  <amodra@gmail.com>
2043         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2044         multilibs to big-endian target and vice versa.
2045         * config/rs6000/t-linux64: Use := assignment on all vars.
2046         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2047         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2048         * config/rs6000/t-linux64le: New file.
2049         * config/rs6000/t-linux64bele: New file.
2050         * config/rs6000/t-linux64lebe: New file.
2052 2013-09-02  Jan Hubicka  <jh@suse.cz>
2054         * ipa-inline-transform.c (inline_transform): Do not
2055         optimize_inline_calls when not optimizing.
2057 2013-09-02  Jan Hubicka  <jh@suse.cz>
2059         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2060         duplicated nodes for assembler names.
2061         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2062         hard registers from assembler name hash.
2064 2013-09-02  Jan Hubicka  <jh@suse.cz>
2066         * ipa-split.c (execute_split_functions): Split externally visible
2067         functions called once.
2069 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2071         PR ipa/58106
2072         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2073         linked list.  When finding the correct duplicate, also consider also
2074         the caller in additon to its inlined_to node.
2076 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2078         * config/aarch64/aarch64-simd-builtins.def
2079         (dup_lane_scalar): Remove.
2080         * config/aarch64/aarch64-simd.md
2081         (aarch64_simd_dup): Add 'w->w' alternative.
2082         (aarch64_dup_lane<mode>): Allow for VALL.
2083         (aarch64_dup_lane_scalar<mode>): Remove.
2084         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2085         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2086         (aarch64_get_lane_unsigned<mode>): Likewise.
2087         (aarch64_get_lane<mode>): Likewise.
2088         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2089         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2090         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2091         (VCON): Change container of V2SF.
2092         (vswap_width_name): Likewise.
2093         * config/aarch64/arm_neon.h
2094         (__aarch64_vdup_lane_any): New.
2095         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2096         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2097         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2099 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2101         PR middle-end/56382
2102         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2103         the source or the destination is a single hard register.
2105 2013-09-02  Richard Biener  <rguenther@suse.de>
2107         PR middle-end/57511
2108         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2109         non-linear SCEVs.
2111 2013-09-02  Richard Biener  <rguenther@suse.de>
2113         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2114         arithmetic to sizetype.
2116 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2118         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2119         Find auto-increment use both before and after candidate.
2121 2013-09-02  Marek Polacek  <polacek@redhat.com>
2123         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2125 2013-09-01  Jan Hubicka  <jh@suse.cz>
2127         * Makefile.in: Add ipa-profile.o
2128         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2129         * cgraph.c (struct cgraph_propagate_frequency_data,
2130         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2131         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2132         * cgraph.h (cgraph_propagate_frequency): Remove.
2133         * ipa-inline-analysis.c: Include ipa-utils.h;
2134         drop duplicated cfgloop.h.
2135         (inline_update_callee_summaries): Update.
2136         * ipa-profile.c: New file.
2137         * ipa-utils.h (ipa_propagate_frequency): Declare.
2138         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2139         data-streamer.h, value-prof.h.
2140         (symtab_remove_unreachable_nodes): Update profile.
2141         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2142         account_time_size, cmp_counts, dump_histogram,
2143         ipa_profile_generate_summary, ipa_profile_write_summary,
2144         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2145         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2146         Move to ipa-profile.c.
2148 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2150         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2152 2013-09-01  Jan Hubicka  <jh@suse.cz>
2154         * common.opt (fdevirtualize-speculatively): New function.
2155         * invoke.texi (fdevirtualize-speculatively): Document.
2156         * ipa-devirt.c: Include ipa-inline.h
2157         (likely_target_p): New function.
2158         (ipa_devirt): New function.
2159         (gate_ipa_devirt): New function.
2160         (pass_data_ipa_devirt): New static var.
2161         (pass_ipa_devirt): Likewise.
2162         (make_pass_ipa_devirt): New function.
2163         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2164         (common_handle_option): Disable devirtualization when
2165         value range profiling is available.
2166         * passes.def (pass_ipa_devirt): Add.
2167         * timever.def (TV_IPA_DEVIRT): New timevar.
2168         * tree-pass.h (make_pass_ipa_devirt):
2170 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2172         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2173         include sanitize(undefined).
2175 2013-08-31  Diego Novillo  <dnovillo@google.com>
2177         * Makefile.in (TREE_CORE_H): Define.
2178         (TREE_H): Use.
2179         (GTFILES): Add tree-core.h.
2180         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2181         size the array.
2182         * tree-core.h: New file.
2183         Move all data structures, enum, typedefs, global
2184         declarations and constants from ...
2185         * tree.h: ... here.
2187 2013-08-31  Jan Hubicka  <jh@suse.cz>
2189         * bulitins.c (expand_builtin): Do not early exit for gcov
2190         instrumented functions.
2192 2013-08-31  Marek Polacek  <polacek@redhat.com>
2194         * ubsan.c: Include tm_p.h.
2196 2013-08-31  Jan Hubicka  <jh@suse.cz>
2198         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2199         warning.
2201         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2202         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2203         * tree-cfg.c (verify_gimple_label): ... here.
2204         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2205         (ipa_merge_profiles): New function.
2206         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2207         (lto_input_function_body): Likewise.
2208         * ipa-utils.h (ipa_merge_profiles): Declare.
2209         * lto-streamer.h (lto_input_function_body): Update prototype.
2210         (emit_label_in_global_context_p): Remove.
2211         * lto-symtab.c: Include ipa-utils.h
2212         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2214 2013-08-31  Jan Hubicka  <jh@suse.cz>
2216         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2218 2013-08-31  Jan Hubicka  <jh@suse.cz>
2220         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2222 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2224         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2225         "cmp" RTX before signed_comparison_operator check to account
2226         for "code" changes.
2228 2013-08-30  Jan Hubicka  <jh@suse.cz>
2230         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2231         (detect_type_change_1): Rename to ...
2232         (detect_type_change): ... this one; early return on non-polymorphic
2233         types.
2234         (detect_type_change_ssa): Add comp_type parameter; update
2235         use of detect_type_change.
2236         (compute_complex_assign_jump_func): Add param_type parameter;
2237         update use of detect_type_change_ssa.
2238         (compute_complex_ancestor_jump_func): Likewise.
2239         (ipa_get_callee_param_type): New function.
2240         (ipa_compute_jump_functions_for_edge): Compute parameter type;
2241         update calls to the jump function computation functions.
2243 2013-08-30  Teresa Johnson  <tejohnson@google.com>
2244             Steven Bosscher  <steven@gcc.gnu.org>
2246         * cfgrtl.c (fixup_new_cold_bb): New routine.
2247         (commit_edge_insertions): Invoke fixup_partitions.
2248         (find_partition_fixes): New routine.
2249         (fixup_partitions): Ditto.
2250         (verify_hot_cold_block_grouping): Update comments.
2251         (rtl_verify_edges): Invoke find_partition_fixes.
2252         (rtl_verify_bb_pointers): Update comments.
2253         (rtl_verify_bb_layout): Ditto.
2254         * basic-block.h (probably_never_executed_edge_p): Declare.
2255         (fixup_partitions): Ditto.
2256         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2257         * bb-reorder.c (sanitize_hot_paths): New function.
2258         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2259         sanitize_hot_paths.
2260         * predict.c (probably_never_executed_edge_p): New routine.
2261         * cfg.c (check_bb_profile): Add partition insanity warnings.
2263 2013-08-30  Meador Inge  <meadori@codesourcery.com>
2265         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2267 2013-08-30  Marek Polacek  <polacek@redhat.com>
2269         * Makefile.in (ubsan.o): Add.
2270         (c-family/c-ubsan.o): Add.
2271         (builtins.o): Add ubsan.h dependency.
2272         * ubsan.h: New file.
2273         * ubsan.c: New file.
2274         * common.opt: Add -fsanitize=undefined option.
2275         (flag_sanitize): Add variable.
2276         (fsanitize=): Add option.  Add Driver.
2277         (fsanitize=thread): Remove option.
2278         (fsanitize=address): Likewise.
2279         (static-libubsan): New option.
2280         * doc/invoke.texi: Document the new flag and -static-libubsan.
2281         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2282         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2283         * builtin-attrs.def (ATTR_COLD): Define.
2284         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2285         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2286         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2287         * flag-types.h (sanitize_code): New enum.
2288         * opts.c (common_handle_option): Parse command line arguments
2289         of -fsanitize=.  Add -fsanitize=unreachable option.
2290         * varasm.c (get_variable_section): Adjust.
2291         (assemble_noswitch_variable): Likewise.
2292         (assemble_variable): Likewise.
2293         (output_constant_def_contents): Likewise.
2294         (categorize_decl_for_section): Likewise.
2295         (place_block_symbol): Likewise.
2296         (output_object_block): Likewise.
2297         * builtins.def: Likewise.
2298         * toplev.c (compile_file): Likewise.
2299         (process_options): Likewise.
2300         * cppbuiltin.c: Likewise.
2301         * tsan.c (tsan_pass): Likewise.
2302         (tsan_gate): Likewise.
2303         (tsan_gate_O0): Likewise.
2304         * cfgexpand.c (partition_stack_vars): Likewise.
2305         (expand_stack_vars): Likewise.
2306         (defer_stack_allocation): Likewise.
2307         (expand_used_vars): Likewise.
2308         * cfgcleanup.c (old_insns_match_p): Likewise.
2309         * asan.c (asan_finish_file): Likewise.
2310         (asan_instrument): Likewise.
2311         (gate_asan): Likewise.
2312         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2313         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2314         (asan_global_struct): Use pointer_sized_int_node instead
2315         calling build_nonstandard_integer_type.
2316         (initialize_sanitizer_builtins): Likewise.
2317         (asan_finish_file): Likewise.
2318         * gcc.c: Document %{%:function(args):X}.
2319         (static_spec_functions): Add sanitize.
2320         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2321         store funcval != NULL there.
2322         (do_spec_1): Adjust handle_spec_function caller.
2323         (handle_braces): Allow %:function(args) as condition.
2324         (sanitize_spec_function): New function.
2325         (ADD_STATIC_LIBUBSAN_LIBS): Define.
2326         (LIBUBSAN_SPEC): Likewise.
2327         (LIBUBSAN_EARLY_SPEC): Likewise.
2328         (SANITIZER_SPEC): Handle libubsan.
2329         (SANITIZER_EARLY_SPEC): Likewise.
2330         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2331         instead of fsanitize=address.
2332         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2333         instead of fsanitize=address*.
2334         * builtins.c: Include ubsan.h.
2335         (fold_builtin_0): Instrument __builtin_unreachable.
2336         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2337         instead of flag_asan.
2338         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2339         (pointer_sized_int_node): Define.
2340         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2342 2013-08-30  Mike Stump  <mikestump@comcast.net>
2344         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2345         with RE patterns.
2347 2013-08-29  Jan Hubicka  <jh@suse.cz>
2349         * cgraph.c (cgraph_function_body_availability): Handle weakref
2350         correctly.
2351         * passes.def: Remove pass_fixup_cfg.
2352         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2353         track when we need to remove functions.
2354         (gate_ipa_inline): Execute inlining always; add comment why.
2355         (pass_data_ipa_inline): Remove TODO_remove_functions.
2356         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2357         do not produce summaries.
2358         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2359         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2360         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2361         constant pool and vtable.
2363 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
2365         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2366         New arm_neon.h's internal macros to specify 64-bit constants.
2367         Avoid using stdint.h's macros.
2369 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2371         * recog.c (verify_changes): Verify that changes[i].old is non-zero
2372         before applying REG_P.
2374 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
2376         PR tree-optimization/58277
2377         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2378         after seeing too many stmts with vdef in between dombb and current
2379         bb, invalidate everything.
2381 2013-08-30  Richard Biener  <rguenther@suse.de>
2383         * fold-const.c (fold_single_bit_test): Fix overflow test.
2385 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
2387         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2388         and which can live in a register, always retrieve the value on entry.
2389         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2390         passed by invisible reference specially.
2391         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2392         (vt_add_function_parameter): Correctly deal with a parameter passed by
2393         invisible reference.
2395 2013-08-30  Jan Hubicka  <jh@suse.cz>
2397         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2399 2013-08-30  Richard Biener  <rguenther@suse.de>
2401         PR tree-optimization/58228
2402         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2403         allow invariant loads in nested loop vectorization.
2405 2013-08-30  Richard Biener  <rguenther@suse.de>
2407         PR tree-optimization/58223
2408         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2409         (has_anti_or_output_dependence): ... this and adjust to also
2410         look for output dependences.
2411         (mark_nodes_having_upstream_mem_writes): Adjust.
2412         (rdg_flag_uses): Likewise.
2414 2013-08-30  Richard Biener  <rguenther@suse.de>
2416         PR tree-optimization/58010
2417         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2418         assert that we have a loop-closed PHI.
2420 2013-08-29  Jan Hubicka  <jh@suse.cz>
2422         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2423         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2424         * lto-section-in.c (lto_free_function_in_decl_state): New function.
2425         (lto_free_function_in_decl_state_for_node): New function.
2427 2013-08-29  Xinliang David Li  <davidxl@google.com>
2429         * loop-unroll.c (report_unroll_peel): Minor message change.
2430         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2431         Emit alignment peeling message with default -fopt-info.
2432         (vect_loop_versioning): Emit loop version info message.
2433         * tree-vectorizer.c (vectorize_loops): Minor message change.
2434         (execute_vect_slp): Ditto.
2436 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
2438         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2439         of the clone from the DECL_NAME of the original function.
2441 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
2443         * passes.c (register_pass): Add overload.
2444         * tree-pass.h (register_pass): Forward declare it.  Add comment.
2446 2013-08-29  Jan Hubicka  <jh@suse.cz>
2448         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2449         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2450         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2451         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2452         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2453         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2454         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2455         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2457 2013-08-29  Teresa Johnson  <tejohnson@google.com>
2459         * dumpfile.c (dump_loc): Output column number.
2460         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2461         * doc/invoke.texi: Document optall -fopt-info flag.
2462         * profile.c (read_profile_edge_counts): Use new dump framework.
2463         (compute_branch_probabilities): Ditto.
2464         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2465         when pass not in any opt group.
2466         * pass_manager.h (pass_manager::get_pass_profile): New method.
2467         * value-prof.c (check_counter): Use new dump framework.
2468         (check_ic_target): Ditto.
2469         * coverage.c (get_coverage_counts): Ditto.
2470         (coverage_init): Setup new dump framework.
2472 2013-08-29  Richard Biener  <rguenther@suse.de>
2474         PR tree-optimization/58246
2475         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2476         handle the dominance check inside a basic-block.
2478 2013-08-29  Richard Biener  <rguenther@suse.de>
2480         PR middle-end/57287
2481         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2482         of default defs that appear in abnormal PHI nodes.
2484 2013-08-29  Richard Biener  <rguenther@suse.de>
2486         PR tree-optimization/57685
2487         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2488         single-use operands to avoid exponential complexity.
2490 2013-08-28  Dehao Chen  <dehao@google.com>
2492         * ipa-inline.c (edge_badness): Fix integer underflow.
2494 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
2496         * gtm-builtins.def (_ITM_free): Declare leaf.
2498 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2500         PR target/58067
2501         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2502         (*tls_local_dynamic_base_64_largepic): Likewise.
2503         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2504         Remove predicate from call operand.
2505         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2506         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2508 2013-08-28  Jeff Law  <law@redhat.com>
2510         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2511         checks for the number of predecessors and successors allowed.
2512         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2513         which require copying a joiner block if there is a request which
2514         is a subpath that requires no joiner block copying.
2516 2013-08-28  Jan Hubicka  <jh@suse.cz>
2518         * lto-streamer-out.c (DFS_write_tree_body): Drop
2519         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2520         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2521         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2522         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2523         Do not read DECL_ERROR_ISSUED.
2524         (unpack_ts_decl_with_vis_value_fields): Do not read
2525         DECL_DEFER_OUTPUT.
2526         (lto_input_ts_binfo_tree_pointers): Do not read
2527         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2528         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2529         write DECL_ERROR_ISSUED..
2530         (pack_ts_decl_with_vis_value_fields): Do not write
2531         DECL_DEFER_OUTPUT.
2532         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2533         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2534         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2535         * tree.h (tree_decl_common): Update comment.
2536         (DECL_ERROR_ISSUED): Remove.
2538 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2540         PR middle-end/58257
2541         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2543 2013-08-28  Jan Hubicka  <jh@suse.cz>
2545         * builtins.def (free): Declare leaf.
2547 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
2549         * gdbhooks.py: New.
2550         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2551         * configure: Regenerate.
2553 2013-08-27  Martin Jambor  <mjambor@suse.cz>
2555         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2556         (ipa_ancestor_jf_data): Likewise.
2557         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2558         (ipa_get_jf_pass_through_type_preserved): New function.
2559         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2560         (ipa_get_jf_ancestor_type_preserved): New function.
2561         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2562         (ipa_get_jf_ancestor_result): Likewise.
2563         (propagate_vals_accross_pass_through): Use
2564         ipa_get_jf_pass_through_result to do all the value mappings.
2565         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2566         type_preserved flag.
2567         (ipa_set_jf_cst_copy): New function.
2568         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2569         (ipa_set_jf_arith_pass_through): Likewise.
2570         (ipa_set_ancestor_jf): Likewise.
2571         (compute_complex_assign_jump_func): Set type_preserved instead of
2572         punting.
2573         (ipa_compute_jump_functions_for_edge): Likewise.
2574         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2575         (update_jump_functions_after_inlining): Update type_preserved.
2576         Explicitely create jump functions when combining one with pass_through.
2577         (ipa_write_jump_function): Stream the type_preserved flags.
2578         (ipa_read_jump_function): Likewise.
2580 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
2581             Aldy Hernandez  <aldyh@redhat.com>
2583         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
2584         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
2585         * function.h (struct function): Add has_force_vect_loops and
2586         has_simduid_loops.
2587         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
2588         * gimple.c (gimple_build_omp_critical): Add KIND argument and
2589         handle it.
2590         * gimple.def: Update CLAUSES comments.
2591         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
2592         (gimple_build_omp_for): Add argument to prototype.
2593         (gimple_omp_for_kind): New.
2594         (gimple_omp_for_set_kind): New.
2595         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
2596         GOVD_DATA_SHARE_CLASS.
2597         (enum omp_region_type): Add ORT_SIMD.
2598         (gimple_add_tmp_var): Handle ORT_SIMD.
2599         (gimplify_var_or_parm_decl): Same.
2600         (is_gimple_stmt): Same.
2601         (omp_firstprivatize_variable): Same.
2602         (omp_add_variable): Only use splay_tree_insert if lookup failed.
2603         (omp_notice_variable): Handle ORT_SIMD.
2604         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
2605         (omp_check_private): Handle ORT_SIMD.
2606         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2607         OMP_CLAUSE_SAFELEN.
2608         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
2609         Handle OMP_CLAUSE_LASTPRIVATE.
2610         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2611         OMP_CLAUSE_SAFELEN.
2612         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
2613         (gimplify_expr): Handle OMP_SIMD.
2614         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
2615         (expand_GOMP_SIMD_VF): New.
2616         (expand_GOMP_SIMD_LAST_LANE): New.
2617         * internal-fn.def (GOMP_SIMD_LANE): New.
2618         (GOMP_SIMD_VF): New.
2619         (GOMP_SIMD_LAST_LANE): New.
2620         * omp-low.c: Include target.h.
2621         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
2622         OMP_CLAUSE_SAFELEN.
2623         (check_omp_nesting_restrictions): Same.
2624         (omp_max_vf): New.
2625         (lower_rec_simd_input_clauses): New.
2626         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
2627         OMP_CLAUSE_LINEAR.
2628         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
2629         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
2630         (expand_omp_build_assign): New.
2631         (expand_omp_for_init_counts): New.
2632         (expand_omp_for_init_vars): New.
2633         (extract_omp_for_update_vars): New.
2634         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
2635         and rewrite accordingly.
2636         (expand_omp_simd): New.
2637         (expand_omp_for): Use expand_omp_simd.
2638         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
2639         (lower_omp_for): Do not lower the body.
2640         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
2641         in their own loops.
2642         * tree-flow.h (find_omp_clause): Remove prototype.
2643         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
2644         forcing vectorization of the loop, or if flag_tree_vectorize.
2645         (gate_tree_if_conversion): Similarly.
2646         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
2647         gimple_build_omp_for.
2648         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
2649         * tree-parloops (create_parallel_loop): Pass kind argument to
2650         gimple_build_omp_for.
2651         * tree-pretty-print.c (dump_omp_clause): Add cases for
2652         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
2653         OMP_CLAUSE__SIMDUID_.
2654         (dump_generic_node): Handle OMP_SIMD.
2655         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
2656         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
2657         unroll OMP_SIMD loops here.
2658         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
2659         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
2660         loop->safelen.
2661         (vect_analyze_data_refs): Handle simd loops.
2662         * tree-vect-loop.c (vectorizable_live_operation): Handle
2663         IFN_GOMP_SIMD*.
2664         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
2665         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
2666         (vectorizable_load): Same.
2667         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
2668         (struct simduid_to_vf): New.
2669         (simduid_to_vf::hash): New.
2670         (simduid_to-vf::equal): New.
2671         (struct simd_array_to_simduid): New.
2672         (simd_array_to_simduid::hash): New.
2673         (simd_array_to_simduid::equal): New.
2674         (adjust_simduid_builtins): New.
2675         (struct note_simd_array_uses_struct): New.
2676         (note_simd_array_uses_cb): New.
2677         (note_simd_array_uses): New.
2678         (vectorize_loops): Handle simd hints and adjust simd builtins
2679         accordingly.
2680         * tree-vectorizer.h (struct _stmt_vec_info): Add
2681         simd_lane_access_p field.
2682         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
2683         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
2684         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
2685         (omp_clause_code_name): Same.
2686         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
2687         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
2688         * tree.def (OMP_SIMD): New entry.
2689         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
2690         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
2691         (OMP_CLAUSE_DECL): Adjust range for new clauses.
2692         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
2693         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
2694         (OMP_CLAUSE_LINEAR_STEP): New.
2695         (OMP_CLAUSE_SAFELEN_EXPR): New.
2696         (OMP_CLAUSE__SIMDUID__DECL): New.
2697         (find_omp_clause): New prototype.
2699 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
2701         * config/i386/driver-i386.c (host_detect_local_cpu): Update
2702         Haswell processor detection.
2704 2013-08-27  Christian Widmer  <shadow@umbrox.de>
2706         PR target/57927
2707         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
2708         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
2709         AVX2 capable processors.
2711 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
2713         * config/aarch64/arm_neon.h: Replace all inline asm implementations
2714         of vget_low_* with implementations in terms of other intrinsics.
2716 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
2718         PR middle-end/57219
2719         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
2720         values to -1, 0 and 1.
2722 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
2724         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
2725         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
2726         (<optab><mode>3_insn): Remove.
2727         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
2728         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
2729         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
2730         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
2731         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
2732         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
2733         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
2734         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
2735         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
2736         (ror<mode>3_insn): Likewise.
2737         * config/aarch64/predicates.md (aarch64_simd_register): New.
2739 2013-08-27  Richard Biener  <rguenther@suse.de>
2741         PR tree-optimization/57521
2742         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
2743         one edge is non-critical.
2744         (find_phi_replacement_condition): Make sure to use a non-critical
2745         edge.  Cleanup and remove old bug workarounds.
2746         (bb_postdominates_preds): Remove.
2747         (if_convertible_loop_p_1): Do not compute post-dominators.
2748         (combine_blocks): Do not free post-dominators.
2749         (main_tree_if_conversion): Likewise.
2750         (pass_data_if_conversion): Add TODO_verify_ssa.
2752 2013-08-27  DJ Delorie  <dj@redhat.com>
2754         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
2756 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
2758         * function.c (assign_parm_find_data_types): Set passed_mode and
2759         nominal_mode to the TYPE_MODE of nominal_type for the built
2760         pointer type in case of the struct-pass-by-reference.
2762 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2764         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
2765         (UINT16_TYPE): Change default to "unsigned int".
2767         * config/avr/avr.opt (mfract-convert-truncate): New option.
2768         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
2769         is set, round negative fractional integers according to n1169
2770         when converting to integer types.
2772 2013-08-26  Jan Hubicka  <jh@suse.cz>
2774         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
2775         methods can not be called indirectly when their address is not taken.
2777 2013-08-26  Jan Hubicka  <jh@suse.cz>
2779         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
2780         ctor_for_folding.
2782 2013-08-26  Jan Hubicka  <jh@suse.cz>
2784         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
2785         can be unshared.
2787 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2789         * reload.c (find_valid_class): Allow classes that do not include
2790         FIRST_PSEUDO_REGISTER - 1.
2792 2013-08-26  Jan Hubicka  <jh@suse.cz>
2794         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
2795         fix edge count/frequency when speculation failed; fix type check
2796         for the direct call.
2798 2013-08-26  Jan Hubicka  <jh@suse.cz>
2800         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
2802 2013-08-26  Jan Hubicka  <jh@suse.cz>
2804         * ipa-inline-transform.c (inline_transform): Be ready for basic block
2805         to be changed by edge redirection.
2807 2013-08-26  Jan Hubicka  <jh@suse.cz>
2809         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
2810         formating; add sanity check.
2811         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
2812         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
2813         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
2814         (resolve_noninline_speculation): Update callee keys, too.
2816 2013-08-26  Jan Hubicka  <jh@suse.cz>
2818         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
2819         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
2821 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2823         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
2824         into proper place.
2826 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
2828         * config/i386/i386.c (ix86_debug_options): Remove prototype.
2829         (x86_64_elf_select_section): Ditto.
2830         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
2831         arguments.
2832         (ix86_pass_by_reference): Ditto.
2833         (output_set_got): Ditto.
2834         (ix86_unary_operator_ok): Ditto.
2835         (ix86_expand_builtin): Ditto.
2837 2013-08-23  Jan Hubicka  <jh@suse.cz>
2839         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
2841 2013-08-23  Jan Hubicka  <jh@suse.cz>
2843         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
2844         (tree_decl_with_vis): Add FINAL field.
2846 2013-08-23  Jeff Law  <law@redhat.com>
2848         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
2849         the debugging dump when the expression is fully redundant.
2851 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2853         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
2854         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
2855         * pretty-print.c (pp_formatted_text_data): Likewise.
2856         (pp_write_text_to_stream): Likewise.
2857         (pp_write_text_as_dot_label_to_stream): Likewise.
2858         (pp_append_r): Likewise.
2859         (pp_format): Likewise.
2860         (pp_flush): Likewise.
2861         (pp_clear_output_area): Likewise.
2862         (pp_append_text): Likewise.
2863         (pp_formatted_text): Likewise.
2864         (pp_remaining_character_count_for_line): Likewise.
2865         (pp_newline): Likewise.
2866         (pp_character): Likewise.
2867         (output_buffer::~output_buffer): Define.
2868         (pretty_printer::~pretty_printer): Destruct output buffer.
2869         * pretty-print.h (output_buffer::~output_buffer): Declare.
2870         (pretty_printer::~pretty_printer): Declare virtual.
2872 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
2874         PR other/57324
2875         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
2876         HOST_WIDE_INT_M1U): New macros.
2877         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
2878         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
2879         unsigned -1 for lshift.
2880         * cse.c (cse_insn): Likewise.
2881         * double-int.c (rshift_double, lshift_double): Likewise.
2882         * builtins.c (fold_builtin_bitop): Likewise.
2883         * combine.c (force_to_mode): Likewise.
2884         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
2885         * simplify-rtx.c (simplify_const_unary_operation,
2886         simplify_const_binary_operation): Likewise.
2887         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
2888         check_va_list_escapes): Likewise.
2889         * rtlanal.c (nonzero_bits1): Likewise.
2890         * expmed.c (expand_smod_pow2): Likewise.
2891         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
2893 2013-08-23  Jan Hubicka  <jh@suse.cz>
2895         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
2896         as having address taken.
2898 2013-08-23  Jan Hubicka  <jh@suse.cz>
2900         * ipa-utils.h (method_class_type): Declare.
2901         * ipa-devirt.c (method_class_type): Export.
2903         * cgraphunit.c (analyze_functions): Do basic devirtualization;
2904         do not walk base classes of anonymous types.
2906 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
2908         PR rtl-optimization/58220
2909         PR regression/58221
2910         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
2911         handle SEQUENCE insns properly.
2913 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2915         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
2916         definition.
2917         (pp_newline_and_indent): Likewise.
2918         (pp_separate_with): Likewise.
2919         * pretty-print.c (pp_newline_and_flush): Define.
2920         (pp_newline_and_indent): Likewise.
2921         (pp_separate_with): Likewise.
2923 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2925         PR target/58218
2926         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
2927         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
2929 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2931         * config/i386/predicates.md (ext_sse_reg_operand): New.
2932         * config/i386/i386.md (*movti_internal): Use
2933         predicate to determine if EVEX is needed.
2934         (*movsi_internal): Ditto.
2935         (*movdf_internal): Ditto.
2936         (*movsf_internal): Ditto.
2937         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
2939 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2941         PR tree-optimization/58209
2942         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
2943         (find_tail_calls): Give up for pointer result types if m is non-NULL.
2944         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
2945         emit POINTER_PLUS_EXPR.
2946         (create_tailcall_accumulator): For pointer result type accumulate in
2947         sizetype type.
2949 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
2951         * configure.ac: Add backslashes missing from the last change.
2952         * configure: Regenerate.
2954 2013-08-22  Jan Hubicka  <jh@susue.cz>
2956         * ipa.c (function_and_variable_visibility): First remember function
2957         was global and then make it local.
2959 2013-08-22  Julian Brown  <julian@codesourcery.com>
2961         * configure.ac: Add aarch64 to list of arches which use "nop" in
2962         debug_line test.
2963         * configure: Regenerate.
2965 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2967         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
2968         gnu_libc_has_function.
2969         * config/s390/tpf.h: Likewise.
2971 2013-08-22  Jan Hubicka  <jh@susue.cz>
2973         * timevar.c (validate_phases): Add cast.
2975 2013-08-22  Jan Hubicka  <jh@susue.cz>
2977         * timevar.c (validate_phases): Use size_t for memory.
2978         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
2980 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2982         * pretty-print.h (output_buffer::output_buffer): Declare.
2983         (pretty_printer::pretty_printer): Likewise.
2984         (pp_construct): Remove.
2985         * pretty-print.c (output_buffer::output_buffer): Define.
2986         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
2987         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
2988         (print_gimple_expr): Likewise.
2989         (print_gimple_seq): Likewise.
2990         (gimple_dump_bb): Likewise.
2991         * sched-vis.c (dump_value_slim): Likewise.
2992         (dump_insn_slim): Likewise.
2993         (dump_rtl_slim): Likewise.
2994         (str_pattern_slim): Likewise.
2995         * tree-mudflap.c (mf_varname_tree): Likewise.
2996         * graph.c (print_graph_cfg): Likewise.
2997         (start_graph_dump): Likewise.
2998         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
2999         placement-new.
3000         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
3001         pretty printer initialization.
3002         * coretypes.h (diagnostic_context): Remove superflous type alias
3003         declaration.
3004         (pretty_printer): Likewise.  Declare directly as a class.
3005         (pretty_print_info): Remove declaration as class.
3006         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3007         and pp_clear_output_area.
3008         (asan_add_global): Likewise.
3010 2013-08-22  Jan Hubicka  <jh@suse.cz>
3012         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3013         * ipa-utils.h (update_type_inheritance_graph): Declare.
3014         (possible_polymorphic_call_target_p): Declare.
3015         (possible_polymorphic_call_target_p): New.
3016         * ipa-devirt.c: Update toplevel comments.
3017         (cached_polymorphic_call_targets): Move up.
3018         (odr_type_d): Move ID down.
3019         (polymorphic_type_binfo_p): Update comment.
3020         (odr_hasher::remove): Likewise;
3021         (get_odr_type): Set anonymous_namespace.
3022         (dump_odr_type): Dump it.
3023         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3024         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3025         (record_binfo): Add comment.
3026         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3027         built.
3028         (devirt_node_removal_hook): Do not iCE when cache is freed.
3029         (possible_polymorphic_call_target_p): New predicate.
3030         (update_type_inheritance_graph): New function.
3032 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3033             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3034             Sergey Lega  <sergey.s.lega@intel.com>
3035             Anna Tikhonova  <anna.tikhonova@intel.com>
3036             Ilya Tocar  <ilya.tocar@intel.com>
3037             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3038             Ilya Verbin  <ilya.verbin@intel.com>
3039             Kirill Yukhin  <kirill.yukhin@intel.com>
3040             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3042         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3043         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3044         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3045         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3046         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3047         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3048         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3049         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3050         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3051         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3052         OPT_mavx512pf, OPT_mavx512er cases.
3053         * config/i386/constraints.md (v): New constraint.
3054         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3055         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3056         (bit_AVX512CD): New.
3057         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3058         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3059         * config/i386/i386-c.c (ix86_target_macros_internal):
3060         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3061         __AVX512PF__.
3062         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3063         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3064         * config/i386/i386.c (regclass_map, dbx_register_map)
3065         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3066         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3067         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3068         -mavx512pf options.
3069         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3070         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3071         -mavx512pf options.  Fix formatting.
3072         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3073         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3074         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3075         -mavx512cd, -mavx512pf options.
3076         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3077         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3078         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3079         ALL_SSE_REGS.
3080         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3081         (ix86_set_reg_reg_cost): Ditto.
3082         (x86_order_regs_for_local_alloc): Ditto.
3083         (MAX_VECT_LEN): Extend to 64-byte.
3084         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3085         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3086         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3087         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3088         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3089         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3090         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3091         (SSE_REG_MODE_P): Support new modes.
3092         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3093         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3094         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3095         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3096         (REG_CLASS_CONTENTS): Add new registers.
3097         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3098         (EXT_REX_SSE_REGNO_P): New.
3099         (HI_REGISTER_NAMES): Add new registers.
3100         * config/i386/i386.md: Define constants for new registers.
3101         (mode): Add new 512-bit modes.
3102         (prefix): Support evex prefix.
3103         (isa): Support avx512f, noavx512f, fma_avx512f.
3104         (ssemodesuffix): Add new 512-bit modes.
3105         (movxi): New.
3106         (*movxi_internal_avx512f): Ditto.
3107         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3108         Support MODE_XI.
3109         (*movsi_internal): Likewise.
3110         (*movdf_internal): Likewise.
3111         (*movsf_internal): Likewise.
3112         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3113         (<code><mode>3): Likewise.
3114         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3115         New.
3116         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3117         with the new constraint "v".
3118         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3119         modes.
3120         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3121         with the new constraint "v".
3122         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3123         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3124         (*<plusminus_insn><mode>3): Likewise.
3125         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3126         (*mul<mode>3): Likewise.
3127         (<sse>_vmmul<mode>3): Likewise.
3128         (<sse>_div<mode>3): Likewise.
3129         (<sse>_vmdiv<mode>3): Likewise.
3130         (<sse>_sqrt<mode>2): Likewise.
3131         (<sse>_vmsqrt<mode>2): Likewise.
3132         (*<code><mode>3_finite): Likewise.
3133         (*<code><mode>3) <smaxmin>: Likewise.
3134         (<sse>_vm<code><mode>3): Likewise.
3135         (*<code><mode>3) <any_logic>: Likewise.
3136         (*fma_fmadd_<mode>): Likewise.
3137         (*fma_fmsub_<mode>): Likewise.
3138         (*fma_fnmadd_<mode>): Likewise.
3139         (*fma_fnmsub_<mode>): Likewise.
3140         (*fma_fmaddsub_<mode>): Likewise.
3141         (*fma_fmsubadd_<mode>): Likewise.
3142         (*fmai_fmadd_<mode>): Likewise.
3143         (*fmai_fmsub_<mode>): Likewise.
3144         (*fmai_fnmadd_<mode>): Likewise.
3145         (*fmai_fnmsub_<mode>): Likewise.
3146         (sse_cvtsi2ss): Likewise.
3147         (sse_cvtsi2ssq): Likewise.
3148         (sse_cvtss2si): Likewise.
3149         (sse_cvtss2si_2): Likewise.
3150         (sse_cvtss2siq): Likewise.
3151         (sse_cvtss2siq_2): Likewise.
3152         (sse_cvttss2si): Likewise.
3153         (sse_cvtss2siq_2): Likewise.
3154         (float<sseintvecmodelower><mode>2): Likewise.
3155         (sse2_cvtsd2si_2): Likewise.
3156         (sse2_cvtsd2siq_2): Likewise.
3157         (*<plusminus_insn><mode>3): Likewise.
3158         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3159         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3160         (ashr<mode>3): Likewise.
3161         (<shift_insn><mode>3): Likewise.
3162         (avx2_<code><mode>3): Likewise.
3163         (*avx2_<code><mode>3): Likewise.
3164         (*andnot<mode>3): Likewise.
3165         (*<code><mode>3) <any_logic>: Likewise.
3166         (abs<mode>2): Likewise.
3167         (avx2_permvar<mode>): Likewise.
3168         (avx2_perm<mode>_1): Likewise.
3169         (*avx_vpermilp<mode>): Likewise.
3170         (avx_vpermilvar<mode>3): Likewise.
3171         (avx2_ashrv<mode>): Likewise.
3172         (avx2_<shift_insn>v<mode>): Likewise.
3173         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3174         -mavx512cd.
3175         * doc/rtl.texi: Document XImode.
3177 2013-08-21  Jeff Law  <law@redhat.com>
3179         * tree-flow.h (register_jump_thread): Pass vector of edges
3180         instead of each important edge.
3181         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3182         thread path into a vector and pass that to register_jump_thread.
3183         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3184         passed in edge vector to the current 3-edge form.
3186         Revert:
3187         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3189         * dce.c (fini_dce): Call df_analyze again just in case
3190         delete_unmarked_insns removed anything.
3192 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3194         * reload.h (struct reg_equivs): Rename to ..
3195         (struct reg_equivs_s): .. this.
3197 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3199         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3201 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3203         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3205 2013-08-21  Jeff Law  <law@redhat.com>
3207         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3208         simplify assignments too.  If the RHS collapses to a singleton
3209         range, then return the value for the range.
3211 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3213         * config/i386/sse.md (V16): Rename to...
3214         (VMOVE): this.
3215         (mov<mode>): Update iterator name.
3216         (*mov<mode>_internal): Ditto.
3217         (push<mode>1): Ditto.
3218         (movmisalign<mode>): Ditto.
3220 2013-08-20  Jan Hubicka  <jh@suse.cz>
3222         PR bootstrap/58186
3223         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3224         entry for direct edges.
3225         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3227 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3229         Revert my last two changes, r201865 and r201864:
3231         Revert r201865:
3232         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3234         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3235         instances can own GC refs.
3237         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3238         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3239         (gcc::context::gt_pch_nx): Likewise.
3240         (gcc::context::gt_pch_nx):  Likewise.
3241         * ggc.h (gt_ggc_mx <T>): New.
3242         (gt_pch_nx_with_op <T>): New.
3243         (gt_pch_nx <T>): New.
3244         * passes.c (opt_pass::gt_ggc_mx): New.
3245         (opt_pass::gt_pch_nx): New.
3246         (opt_pass::gt_pch_nx_with_op): New.
3247         (pass_manager::gt_ggc_mx): New.
3248         (pass_manager::gt_pch_nx): New.
3249         (pass_manager::gt_pch_nx_with_op): New.
3250         (pass_manager::operator new): Use
3251         ggc_internal_cleared_alloc_stat rather than xcalloc.
3252         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3253         (pass_manager::gt_ggc_mx): New.
3254         (pass_manager::gt_pch_nx): New.
3255         (pass_manager::gt_pch_nx_with_op): New.
3256         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3257         (opt_pass::operator new): New.
3258         (opt_pass::gt_ggc_mx): New.
3259         (opt_pass::gt_pch_nx): New.
3260         (opt_pass::gt_pch_nx_with_op): New.
3262         Revert r201864:
3263         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3265         * Makefile.in (GTFILES): Add context.h.
3266         * context.c (gcc::context::operator new): New.
3267         (gcc::context::gt_ggc_mx): New.
3268         (gcc::context::gt_pch_nx): New.
3269         (gcc::context::gt_pch_nx): New.
3270         * context.h (gcc::context): Add GTY((user)) marking.
3271         (gcc::context::operator new): New.
3272         (gcc::context::gt_ggc_mx): New.
3273         (gcc::context::gt_pch_nx): New.
3274         (gcc::context::gt_pch_nx): New.
3275         (g): Add GTY marking.
3276         (gt_ggc_mx (gcc::context *)): New.
3277         (gt_pch_nx (gcc::context *)): New.
3278         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3279         void *cookie)): New.
3280         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3282 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3284         * dce.c (fini_dce): Call df_analyze again just in case
3285         delete_unmarked_insns removed anything.
3287 2013-08-20  Teresa Johnson  <tejohnson@google.com>
3289         PR rtl-optimizations/57451
3290         * final.c (reemit_insn_block_notes): Prevent lexical blocks
3291         from crossing split section boundaries.
3293 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3295         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3296         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3297         with MULTLIB_DEFAULTS.
3299 2013-08-20  Nick Clifton  <nickc@redhat.com>
3301         * target.def (narrow_volatile_bitfield): Note that the default
3302         value is false, not !TARGET_STRICT_ALIGN.
3303         * doc/tm.texi: Regenerate.
3305 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
3307         Fix LIB_SPEC for systems without libpthread.
3309         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3310         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3311         for Android.
3312         * config/i386/linux-common.h: Likewise.
3313         * config/mips/linux-common.h: Likewise.
3315 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
3317         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3318         checks.
3320 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3322         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3323         instances can own GC refs.
3325         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3326         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3327         (gcc::context::gt_pch_nx): Likewise.
3328         (gcc::context::gt_pch_nx):  Likewise.
3329         * ggc.h (gt_ggc_mx <T>): New.
3330         (gt_pch_nx_with_op <T>): New.
3331         (gt_pch_nx <T>): New.
3332         * passes.c (opt_pass::gt_ggc_mx): New.
3333         (opt_pass::gt_pch_nx): New.
3334         (opt_pass::gt_pch_nx_with_op): New.
3335         (pass_manager::gt_ggc_mx): New.
3336         (pass_manager::gt_pch_nx): New.
3337         (pass_manager::gt_pch_nx_with_op): New.
3338         (pass_manager::operator new): Use
3339         ggc_internal_cleared_alloc_stat rather than xcalloc.
3340         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3341         (pass_manager::gt_ggc_mx): New.
3342         (pass_manager::gt_pch_nx): New.
3343         (pass_manager::gt_pch_nx_with_op): New.
3344         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3345         (opt_pass::operator new): New.
3346         (opt_pass::gt_ggc_mx): New.
3347         (opt_pass::gt_pch_nx): New.
3348         (opt_pass::gt_pch_nx_with_op): New.
3350 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3352         * Makefile.in (GTFILES): Add context.h.
3353         * context.c (gcc::context::operator new): New.
3354         (gcc::context::gt_ggc_mx): New.
3355         (gcc::context::gt_pch_nx): New.
3356         (gcc::context::gt_pch_nx): New.
3357         * context.h (gcc::context): Add GTY((user)) marking.
3358         (gcc::context::operator new): New.
3359         (gcc::context::gt_ggc_mx): New.
3360         (gcc::context::gt_pch_nx): New.
3361         (gcc::context::gt_pch_nx): New.
3362         (g): Add GTY marking.
3363         (gt_ggc_mx (gcc::context *)): New.
3364         (gt_pch_nx (gcc::context *)): New.
3365         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3366         void *cookie)): New.
3367         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3369 2013-08-20  Alan Modra  <amodra@gmail.com>
3371         PR target/57865
3372         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3373         (rs6000_emit_epilogue): Likewise.
3375 2013-08-19  Dehao Chen  <dehao@google.com>
3377         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3379 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
3380             Jakub Jelinek  <jakub@redhat.com>
3382         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3383         (BUILT_IN_FABSD64): Likewise.
3384         (BUILT_IN_FABSD128): Likewise.
3385         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3386         (fold_builtin_1): Likewise.
3387         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3388         destination and source operands.
3389         (*abstd2_fpr): Likewise.
3390         (*nabstd2_fpr): Likewise.
3392 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
3394         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3395         JUMP_P and INSN_P.
3397 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
3399         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3400         incomplete.
3402 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3404         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3405         * builtins.c (default_libc_has_function): New.
3406         (gnu_libc_has_function): Ditto.
3407         (no_c99_libc_has_function): Ditto.
3408         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3409         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3410         (fold_builtin_sincos): Likewise.
3411         (fold_builtin_cexp): Likewise.
3412         * builtins.def (DEF_C94_BUILTIN): Likewise.
3413         (DEF_C99_BUILTIN): Likewise.
3414         (DEF_C99_C90RES_BUILTIN): Likewise.
3415         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3416         definitions to using this define.
3417         * config/darwin-protos.h (darwin_libc_has_function): New.
3418         * config/darwin.c (darwin_libc_has_function): Ditto.
3419         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3420         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3421         * config/darwin.h: Ditto.
3422         * config/elfos.h: Ditto.
3423         * config/freebsd.h: Ditto.
3424         * config/i386/cygming.h: Ditto.
3425         * config/i386/djgpp.h: Ditto.
3426         * config/i386/i386-interix.h: Ditto.
3427         * config/microblaze/microblaze.h: Ditto.
3428         * config/mmix/mmix.h: Ditto.
3429         * config/gnu-user.h: Ditto.
3430         * config/ia64/hpux.h: Ditto.
3431         * config/pa/pa-hpux.h: Ditto.
3432         * config/pdp11/pdp11.h: Ditto.
3433         * config/picochip/picochip.h: Ditto.
3434         * config/linux.h: Ditto.
3435         * config/netbsd.h: Ditto.
3436         * config/openbsd.h: Ditto.
3437         * config/rs6000/aix43.h: Ditto.
3438         * config/rs6000/aix51.h: Ditto.
3439         * config/rs6000/aix52.h: Ditto.
3440         * config/rs6000/aix53.h: Ditto.
3441         * config/rs6000/aix61.h: Ditto.
3442         * config/rs6000/darwin.h: Ditto.
3443         * config/rs6000/linux.h: Ditto.
3444         * config/rs6000/linux64.h: Ditto.
3445         * config/s390/tpf.h: Ditto.
3446         * config/sol2-10.h: Ditto.
3447         * config/sol2.h: Ditto.
3448         * config/vms/vms.h: Ditto.
3449         * config/vxworks.h: Ditto.
3450         * config/linux-android.c (linux_android_libc_has_function):
3451         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3452         * config/linux-protos.h (linux_android_libc_has_function):
3453         New declaration.
3454         * config/i386/i386.c (ix86_libc_has_function): New.
3455         * config/i386/i386-protos.h
3456         (ix86_libc_has_function): New declaration.
3457         * config/i386/i386.md
3458         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3459         ("isinf<mode>2): Likewise.
3460         * convert.c (convert_to_integer): Using new target hook
3461         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3462         TARGET_C99_FUNCTIONS.
3463         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3464         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3465         * coretypes.h (function_class): New enum for different
3466         classes of functions.
3467         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3468         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3469         (TARGET_HAS_SINCOS): Likewise.
3470         (TARGET_LIBC_HAS_FUNCTION): New.
3471         * doc/tm.texi: Regenerated.
3472         * targhooks.h (default_libc_has_function): New declaration.
3473         (no_c99_libc_has_function): Ditto.
3474         (gnu_libc_has_function): Ditto.
3475         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3476         and TARGET_HAS_SINCOS.
3478 2013-08-18  Jan Hubicka  <jh@suse.cz>
3480         * Makeifle-in (ipa-devirt.o): New.
3481         (GTFILES): Add ipa-utils.h and ipa-devirt.c
3482         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3483         (analyze_functions): Look into possible targets of polymorphic call.
3484         * dumpfile.c (dump_files): Add type-inheritance dump.
3485         * dumpfile.h (TDI_inheritance): New.
3486         * ipa-devirt.c: New file.
3487         * ipa-utils.h (odr_type_d): Forward declare.
3488         (odr_type): New type.
3489         (build_type_inheritance_graph): Declare.
3490         (possible_polymorphic_call_targets): Declare and introduce inline
3491         variant when only edge is pased.
3492         (dump_possible_polymorphic_call_targets): Likewise.
3493         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3494         * tree.c (type_in_anonymous_namespace_p): Break out from ...
3495         (types_same_for_odr): ... here.
3496         * tree.h (type_in_anonymous_namespace_p): Declare.
3498 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
3500         PR tree-optimization/58006
3501         * tree-parloops.c (take_address_of): Don't ICE if get_name
3502         returns NULL.
3503         (eliminate_local_variables_stmt): Remove clobber stmts.
3505 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
3507         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3508         error message is issued for an alias to undefined symbol.
3510 2013-08-18  Jan Hubicka  <jh@suse.cz>
3512         * cgraph.c (cgraph_create_indirect_edge): Discover
3513         polymorphic calls and record basic info into indirect_info.
3514         * gimple-fold.c (gimple_fold_call): When doing BINFO based
3515         devirtualization, ignore objc function calls.
3516         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3517         call with no parm index info.
3518         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3519         * tree.c (virtual_method_call_p): New function.
3520         * tree.h (virtual_method_call_p): Declare.
3522 2013-08-16  Jan Hubicka  <jh@suse.cz>
3524         PR middle-end/58179
3525         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3527 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
3529         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3530         attribute.
3532 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3534         * gengtype.c (type_for_name): Add special-case support for
3535         locating types within the "gcc::" namespace.
3536         (open_base_files): Emit a "using namespace gcc" directive.
3538 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3540         PR target/58160
3541         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3542         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3544         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3545         array instead of each individual operand as a separate argument.
3546         (emit_fusion_gpr_load): Likewise.
3547         (expand_fusion_gpr_load): Add new function declaration.
3549         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3550         signature to have the operands passed as an array, instead of as
3551         separate arguments.  Allow ZERO_EXTEND to be in the memory
3552         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
3553         depend on the register live/dead flags when peepholes are run.
3554         (expand_fusion_gpr_load): New function to be called from the
3555         peephole2 pass, to change the register that addis sets to be the
3556         target register.
3557         (emit_fusion_gpr_load): Change the calling signature to have the
3558         operands passed as an array, instead of as separate arguments.
3559         Allow ZERO_EXTEND to be in the memory address, and also
3560         SIGN_EXTEND if -mpower8-fusion-sign.
3562         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3563         unspec enumeration.
3564         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3565         adjust the register addis loads up in the peephole2 pass.  Do not
3566         depend on the register live/dead state when the peephole pass is done.
3568 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3570         * gengtype.c (create_user_defined_type): Ensure that the kind
3571         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3572         declaration is seen before the GTY((user)) marking.
3574 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3576         PR target/58105
3577         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
3579 2013-08-16  Jan Hubicka  <jh@suse.cz>
3581         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
3582         arugment expected_type.
3583         (gimple_fold_call): Use it.
3584         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
3585         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
3586         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
3587         (try_make_edge_direct_virtual_call): Likewise.
3588         * tree.c (obj_type_ref_class): New.
3589         * tree.h (obj_type_ref_class): Use it.
3591 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3593         * sched-vis.c (rtl_slim_pp_initialized): Remove.
3594         (rtl_slim_pp): Likewise.
3595         (init_rtl_slim_pretty_print): Likewise.
3596         (dump_value_slim):  Don't call it.  Use local pretty printer.
3597         (dump_insn_slim): Likewise.
3598         (dump_rtl_slim): Likewise.
3599         (str_pattern_slim): Likewise.
3600         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
3601         Simplify.
3603 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
3605         PR tree-optimization/58164
3606         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
3607         walk gimple_goto_dest of GIMPLE_GOTO.
3609         PR tree-optimization/58165
3610         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
3611         bi_call must be the last stmt in a bb, don't split_block, instead
3612         use fallthru edge from it and give up if there is none.
3613         Release conds vector when returning early.
3615 2013-08-14  Xinliang David Li  <davidxl@google.com>
3617         * config/i386/i386.c (ix86_option_override_internal):
3618         Remove unused variable and field.
3620 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3622         PR target/57949
3623         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
3624         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
3625         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
3626         and Linux, correct BLKmode alignment when 128-bit alignment is
3627         required and compatibility flag is not set.
3628         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
3629         for zero-size arguments when compatibility flag is not set.
3631 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3633         PR tree-optimization/58145
3634         * tree-sra.c (build_ref_for_offset): If prev_base has
3635         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
3637 2013-08-14  Xinliang David Li  <davidxl@google.com>
3639         * config/i386/i386.c (ix86_option_override_internal):
3640         Fix uninitialized variable error.
3642 2013-08-14  Xinliang David Li  <davidxl@google.com>
3644         * config/i386/i386.opt: Define two new options.
3645         * config/i386/x86-tune.def: Add arch selector field in macros.
3646         * config/i386/i386.h: Adjust macro definition.
3647         * config/i386/i386.c (ix86_option_override_internal):
3648         Refactor the code.
3649         (parse_mtune_ctrl_str): New function.
3650         (set_ix86_tune_features): New function.
3651         (ix86_function_specific_restore): Call the new helper function.
3653 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
3655         PR rtl-optimization/57662
3656         * sel-sched.c (code_motion_process_successors): When the current insn
3657         is removed after the recursive traversal, break from the loop.
3658         Add comments and debug printouts.
3660 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3661             Alexandre Oliva  <aoliva@redhat.com>
3663         PR target/58067
3664         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
3665         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
3666         there also UNSPEC_PLTOFF.
3668 2013-08-14  Marek Polacek  <polacek@redhat.com>
3670         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
3671         NUM_CONDITIONS bit positions.
3673 2013-08-13  Cary Coutant  <ccoutant@google.com>
3675         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
3676         (attr_checksum): Hash vector contents instead of pointer.
3677         (attr_checksum_ordered): Likewise.
3679 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
3681         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
3682         when Pmode != word_mode.  Add length_address attribute.
3683         (sse3_monitor_<mode>): Merge from sse3_monitor and
3684         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
3685         Pmode != word_mode.  Update insn length attribute.
3686         * config/i386/i386.c (ix86_option_override_internal): Update
3687         ix86_gen_monitor selection for merged sse3_monitor insn.
3689 2013-08-13  Julian Brown  <julian@codesourcery.com>
3691         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
3692         perform invalid legitimization on greater-than-word-size modes for
3693         TARGET_E500_DOUBLE.
3695 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
3697         * ira.c (setup_class_translate_array): Use aclass instead of cl
3698         for classes not fully covered by allocno classes.
3700 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
3702         PR tree-optimization/57661
3703         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
3704         * tree-inline.c (tree_function_versioning): Initialize it.
3705         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
3706         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
3707         that is not being copied.
3709         PR sanitizer/56417
3710         * asan.c (instrument_strlen_call): Fix typo in comment.
3711         Use char * type even for the lhs of POINTER_PLUS_EXPR.
3713 2013-08-13  Steve Ellcey  <sellcey@mips.com>
3715         * config/mips/mips.md (prefetch): Use lw instead of ld on
3716         loongson in 32bit mode.
3718 2013-08-13  Nick Clifton  <nickc@redhat.com>
3720         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
3722 2013-08-13  Jan Hubicka  <jh@suse.cz>
3724         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
3725         introduced edge; fix typo in sanity check.
3726         (cgraph_resolve_speculation): Export; improve diagnostic.
3727         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
3728         speculation at type mismatch.
3729         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
3730         (cgraph_resolve_speculation): Declare.
3731         (symtab_can_be_discarded): New function.
3732         * value-prof.c (gimple_ic_transform): Remove actual transform code.
3733         * ipa-inline-transform.c (speculation_removed): New global var.
3734         (clone_inlined_nodes): See if speculation can be removed.
3735         (inline_call): If speculations was removed, we growths may not match.
3736         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
3737         (speculation_useful_p): New function.
3738         (resolve_noninline_speculation): New function.
3739         (inline_small_functions): Resolve useless speculations.
3740         * ipa-inline.h (speculation_useful_p): Declare
3741         * ipa.c (can_replace_by_local_alias): Simplify.
3742         (ipa_profile): Produce speculative calls in non-lto, too;
3743         add simple cost model; produce local aliases.
3745 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
3747         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
3748         PASS_MANAGER_H.
3750 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
3752         * config/i386/i386.c (ix86_function_versions): Use error + inform.
3754 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
3756         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
3757         iterator instead of X87MODEF.
3759 2013-08-12  Perez Read  <netfirewall@gmail.com>
3761         PR target/58132
3762         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
3763         operand 0 for intel asm alternative.
3764         (*movabs<mode>_2): Ditto for operand 1.
3766 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
3768         * config/aarch64/arm_none.h
3769         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
3771 2013-08-12  Nick Clifton  <nickc@redhat.com>
3773         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
3775 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
3777         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
3778         expand for QI/HImode operand to produce more effictive code for
3779         unsigned char(short) --> float(double) conversion.
3781 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
3783         * doc/invoke.texi: Mention that -ftls-model does not force the final
3784         model.
3786 2013-08-12  Marek Polacek  <polacek@redhat.com>
3787             Marc Glisse  <marc.glisse@inria.fr>
3789         PR tree-optimization/57980
3790         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
3791         when creating -1 constant.
3793 2013-08-10  Jan Hubicka  <jh@suse.cz>
3795         Workaround binutils PR14342.
3796         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
3797         (gimple_init_edge_profiler): Likewise.
3798         (gimple_gen_ic_func_profiler): Likewise.
3800 2013-08-09  Jan Hubicka  <jh@suse.cz>
3802         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
3804 2013-08-09  Xinliang David Li  <davidxl@google.com>
3806         * config/i386/stringop.def: New file.
3807         * config/i386/stringop.opt: New file.
3808         * config/i386/i386-opts.h: Include stringopt.def.
3809         * config/i386/i386.opt: Include stringopt.opt.
3810         * config/i386/i386.c (ix86_option_override_internal):
3811         Override default size based stringop inline strategies with options.
3812         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
3813         New function.
3815 2013-08-09  Jan Hubicka  <jh@suse.cz>
3817         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
3819 2013-08-09  Jan Hubicka  <jh@suse.cz>
3821         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
3822         CGRAPH_FREQ_MAX.
3823         (dump_cgraph_node): Dump profile-id.
3824         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
3825         and common_target_probability.
3826         * lto-cgraph.c (lto_output_edge): Stream common targets.
3827         (lto_output_node): Stream profile ids.
3828         (input_node): Stream profile ids.
3829         (input_edge): Stream common targets.
3830         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
3831         * ipa.c: Include value-prof.h
3832         (ipa_profile_generate_summary): Turn indirect call statement histograms
3833         into common targets.
3834         (ipa_profile): Turn common targets into speculative edges.
3836 2013-08-09  Jan Hubicka  <jh@suse.cz>
3838         * cgraph.h (cgraph_node): Add profile_id.
3839         * value-prof.c (cgraph_node_map): Turn into pointer_map.
3840         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
3841         (del_node_map): Update.
3842         (find_func_by_funcdef_no): Replace by ...
3843         (find_func_by_profile_id): ... this one.
3844         (gimple_ic_transform): Do not remove useful histograms when
3845         speculation is not done; dump info when indirect call removal
3846         can happen at LTO.
3847         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
3848         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
3849         (__gcov_indirect_call_profiler_v2): .. this one.
3850         * profile.h (init_node_map): Update.
3851         * coverage.c (coverage_compute_profile_id): New function.
3852         * coverage.h (coverage_compute_profile_id): Declare.
3853         * tree-profile.c (init_ic_make_global_vars): Make
3854         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
3855         (gimple_init_edge_profiler): Update prototype of
3856         __gcov_indirect_call_profiler.
3857         (gimple_gen_ic_func_profiler): Simplify.
3858         (tree_profiling): Use init_node_map
3860 2013-08-09  Jan Hubicka  <jh@suse.cz>
3862         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
3863         non-speculative refs.
3864         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
3865         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
3866         (cgraph_set_call_stmt): Likewise.
3867         (cgraph_create_edge_1): Fix release checking compilatoin;
3868         clear lto_stmt_uid.
3869         (cgraph_free_edge): Free indirect info.
3870         (cgraph_turn_edge_to_speculative): New function.
3871         (cgraph_speculative_call_info): New function.
3872         (cgraph_make_edge_direct): Return direct edge; handle speculation.
3873         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
3874         (dump_cgraph_node): Dump speculation.
3875         (verify_edge_count_and_frequency): Accept speculative edges.
3876         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
3877         (verify_cgraph_node): Handle speculation.
3878         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
3879         (cgraph_set_call_stmt): Update prototype.
3880         (cgraph_make_edge_direct): Update prototype.
3881         (cgraph_speculative_call_info): Declare.
3882         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
3883         to change; update call of ipa_find_references.
3884         * ipa-ref.c (ipa_record_reference): Fix return value; clear
3885         lto_stmt_uid and speculative flags.
3886         (ipa_dump_references): Dump speculation.
3887         (ipa_clone_references): Clone speculative flag.
3888         (ipa_clone_referring): Likewise.
3889         (ipa_clone_ref): New function.
3890         (ipa_find_reference): Look into lto_stmt_uids
3891         (ipa_clear_stmts_in_references): Do not clear speculative calls.
3892         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
3893         (ipa_find_reference): Update declaration.
3894         (ipa_clone_ref): Declare.
3895         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
3896         stream speculative flag.
3897         (lto_output_ref): Stream statements uids and speculation.
3898         (input_ref): Likewise.
3899         (input_edge): Stream speuclation.
3900         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
3901         (cgraph_set_call_stmt_including_clones): Handle speculation.
3902         * ipa-inline.c (heap_edge_removal_hook): New function.
3903         (inline_small_functions): Register it.
3904         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
3905         also initialize refs.
3906         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
3907         edge to change.
3908         (try_make_edge_direct_simple_call): Likewise.
3909         (try_make_edge_direct_simple_call): Likewise.
3910         (update_indirect_edges_after_inlining): Likewise.
3911         (remove_described_reference): Look proper lto_stmt_uid.
3912         (propagate_controlled_uses): Likewise.
3913         (propagate_controlled_uses): Liekwise.
3914         * tree-inline.c (copy_bb): Copy speculative edges.
3915         (redirect_all_calls): New function.
3916         (copy_cfg_body): Do redirection after loop info is updated.
3917         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
3919 2013-08-09  Jan Hubicka  <jh@suse.cz>
3921         * lto-streamer-out.c (output_function): Renumber PHIs.
3922         * lto-streamer-in.c (input_function): Likewise.
3924 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
3926         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
3927         (get_lane_unsigned): Likewise.
3928         (dup_lane_scalar): Likewise.
3929         (get_lane): enable for VALL.
3930         * config/aarch64/aarch64-simd.md
3931         (aarch64_dup_lane_scalar<mode>): Remove.
3932         (aarch64_get_lane_signed<mode>): Likewise.
3933         (aarch64_get_lane_unsigned<mode>): Likewise.
3934         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
3935         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
3936         (aarch64_get_lane<mode>): Enable for all vector modes.
3937         (aarch64_get_lanedi): Remove misleading constraints.
3938         * config/aarch64/arm_neon.h
3939         (__aarch64_vget_lane_any): Define.
3940         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
3941         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
3942         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
3943         * config/aarch64/iterators.md (VDQQH): New.
3944         (VDQQHS): Likewise.
3945         (vwcore): Likewise.
3947 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
3949         * configure.ac: Add GAS check for LEON instructions on SPARC.
3950         * configure: Regenerate.
3951         * config.in: Likewise.
3952         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
3953         sparc*-*-* block.
3954         * config/sparc/sparc.opt (LEON, LEON3): New masks.
3955         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
3956         for LEON or LEON3.
3957         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
3958         (AS_LEON_FLAG): New macro.
3959         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
3960         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
3961         Deal with LEON and LEON3 for the memory model.
3962         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
3963         (atomic_compare_and_swap<mode>_1): Likewise.
3964         (*atomic_compare_and_swap<mode>_1): Likewise.
3966 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3968         * config/arm/neon.md (vcond): Fix floating-point vector
3969         comparisons against 0.
3971 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
3973         * lra-constraints.c (emit_spill_move): Remove assert.
3974         (process_alt_operands): Add more debugging
3975         output.  Increase reject for spilling into memory.  Decrease
3976         reject for reloading scratch.
3977         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
3979 2013-08-08  Steve Ellcey  <sellcey@mips.com>
3981         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
3982         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
3983         micromips incompatible.  Add nan2008.
3984         (MULTILIB_DIRNAMES): Add nan2008.
3985         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3986         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
3987         and micromips incompatible.  Add nan2008.
3988         (MULTILIB_DIRNAMES): Add nan2008.
3989         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3991 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
3993         PR rtl-optimization/58079
3994         * combine.c (combine_simplify_rtx): Avoid using SUBST if
3995         simplify_comparison has widened a comparison with an integer.
3997 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3999         * config/arm/neon.md (movmisalign<mode>): Disable when we
4000         don't allow unaligned accesses.
4001         (*movmisalign<mode>_neon_store): Likewise.
4002         (*movmisalign<mode>_neon_load): Likewise.
4003         (*movmisalign<mode>_neon_store): Likewise.
4004         (*movmisalign<mode>_neon_load): Likewise.
4006 2013-08-08  Jan Hubicka  <jh@suse.cz>
4008         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4009         (make_pass_rebuild_cgraph_edges): Also clear references.
4010         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4011         * ipa-inline-transform.c (inline_transform): Remove all references
4012         after inlining.
4013         * cgraphunit.c (expand_function): Remove all references after
4014         expansion.
4015         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4016         (ipa_find_reference): Rewrite to iterator.
4017         (remove_stmt_references): Likewise.
4018         (ipa_clear_stmts_in_references): New function.
4019         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4020         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4021         clear references.
4022         * ipa-split.c (split_function): Remove references in split function.
4024 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4026         PR target/57431
4027         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4028         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4030 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4032         PR target/56979
4033         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4034         suggested mode for the assignment isn't compatible with the
4035         registers required.
4037 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4039         PR target/58065
4040         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4042 2013-08-07  Xinliang David Li  <davidxl@google.com>
4044         * config/i386/i386.opt: New option -mtune-ctrl=.
4045         * config/i386/x86-tune.def: New file.
4046         * config/i386/i386.h: include x86-tune.def.
4047         * config/i386/i386.c (ix86_option_override_internal):
4048         Parsing -mtune-ctrl= option and set tune features.
4050 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4052         PR other/12081
4053         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4054         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4055         to gen_2arg_fn_t.
4057 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4059         * rtl.h (update_alignments): Declare.
4060         * final.c (grow_label_align): New function extracted from...
4061         (shorten_branches): ...here.  Call it.
4062         (update_alignments): New function.
4063         * reorg.c (sibling_labels): New variable.
4064         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4065         the new label along with it onto the sibling_labels vector.
4066         (fill_simple_delay_slots): Adjust call to get_label_before.
4067         (fill_slots_from_thread): Likewise.
4068         (relax_delay_slots): Likewise.
4069         (make_return_insns): Likewise.
4070         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4072 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4074         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4075         document its semantics.
4076         (diagnostic_report_diagnostic): Adjust accordingly.
4078 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4080         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4081         (sparc_option_override): ...and port to new C++ pass API.
4082         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4084 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4086         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4088 2013-08-06  Caroline Tice  <cmtice@google.com>
4090         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4091         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4092         * tree-pass.h: Add pass_vtable_verify.
4093         * varasm.c (assemble_variable): Add code to properly set the comdat
4094         section and name for the .vtable_map_vars section.
4095         (assemble_vtyv_preinit_initializer): New function.
4096         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4097         LINK_ONCE flag.
4098         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4099         * vtable-verify.c: New file.
4100         * vtable-verify.h: New file.
4101         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4102         initialiation levels.
4103         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4104         * passes.def: Insert pass_vtable_verify.
4105         * aclocal.m4: Reorder includes.
4106         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4107         -fvtv-counts options.
4108         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4109         as appropriate, if -fvtable-verify=... is used.
4110         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4111         -fvtable-verify=... is used.
4112         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4113         (vtable-verify.o): Add new build rule.
4114         (GTFILES): Add vtable-verify.c to list.
4115         * common.opt (fvtable-verify=): New flag.
4116         (vtv_priority): Values for fvtable-verify= flag.
4117         (fvtv-counts): New flag.
4118         (fvtv-debug): New flag.
4119         * tree.h (save_vtable_map_decl): New extern function decl.
4121 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4123         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4124         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4125         (pass_data_rl78_devirt): ...new pass_data instance and...
4126         (make_pass_rl78_devirt): ...new function.
4127         (rl78_asm_file_start): Port pass registration to new C++ API.
4129 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4131         * coretypes.h (rtl_opt_pass): Add.
4132         (gcc::context): Add.
4133         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4134         (epiphany_init): Port to new C++ pass API.
4135         (epiphany_optimize_mode_switching): Likewise.
4136         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4137         (pass_manager::get_pass_mode_switching): New.
4138         (pass_manager::get_pass_peephole2): New.
4139         * mode-switching.c (pass_mode_switching): Add clone method.
4140         * recog.c (pass_peephole2): Add clone method.
4141         (pass_split_all_insns): Add clone method.
4143 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4145         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4146         (mips_option_override): ...here, porting to new C++ API for passes.
4148 2013-08-06  Jan Hubicka  <jh@suse.cz>
4150         * cgraph.c (cgraph_get_body): New function based on lto.c
4151         implementation.
4152         * cgraph.h (cgraph_get_body): Declare.
4153         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4154         LTO paths.
4155         * cgraphunit.c (expand_function): Get body prior expanding.
4156         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4157         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4158         really need.
4159         * passes.c (do_per_function_toporder): Get body.
4160         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4161         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4163 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4165         PR fortran/57987
4166         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4167         is not re-finalized.  Rename second parameter to no_collect.
4169 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4171         PR middle-end/58041
4172         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4173         MEM_REF has proper alignment information.
4175 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4177         PR other/12081
4178         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4179         class insn_gen_fn.
4180         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4181         rtx (*) (rtx, ...) with insn_gen_fn.
4182         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4183         insn_gen_fn::stored_funcptr.  Add initializer braces.
4185 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4187         Rewrite how instances of passes are cloned to remove assumptions
4188         about their sizes (thus allowing pass subclasses to have
4189         additional data fields, albeit non-GC-managed ones at this point).
4191         * passes.c (make_pass_instance): Now that passes have clone
4192         methods, rewrite this function to eliminate XNEW and memcpy
4193         calls that used hardcoded sizes.  Since this function no longer
4194         creates pass instances, rename it to...
4195         (add_pass_instance): ...this.  Document the old way that passes were
4196         numbered and flagged, and rework this function to continue using it.
4197         (next_pass_1): Add an initial_pass argument for use by
4198         add_pass_instance.
4199         (position_pass): When adding multiple instances of a pass, use
4200         the pass's clone method, rather than relying on the XNEW/memcpy
4201         within the former make_pass_instance (now add_pass_instance).
4202         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4203         the initial instance of the current pass within the pass manager.
4205 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4207         This is the automated part of the conversion of passes from C
4208         structs to C++ classes.
4210         Patch autogenerated by refactor_passes.py from
4211         https://github.com/davidmalcolm/gcc-refactoring-scripts
4212         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4214         * asan.c (pass_asan): Convert from a global struct to a subclass of
4215         gimple_opt_pass along with...
4216         (pass_data_asan): ...new pass_data instance and...
4217         (make_pass_asan): ...new function.
4218         (pass_asan_O0): Convert from a global struct to a subclass of
4219         gimple_opt_pass along with...
4220         (pass_data_asan_O0): ...new pass_data instance and...
4221         (make_pass_asan_O0): ...new function.
4222         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4223         subclass of rtl_opt_pass along with...
4224         (pass_data_inc_dec): ...new pass_data instance and...
4225         (make_pass_inc_dec): ...new function.
4226         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4227         a subclass of rtl_opt_pass along with...
4228         (pass_data_reorder_blocks): ...new pass_data instance and...
4229         (make_pass_reorder_blocks): ...new function.
4230         (pass_duplicate_computed_gotos): Convert from a global struct to a
4231         subclass of rtl_opt_pass along with...
4232         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4233         (make_pass_duplicate_computed_gotos): ...new function.
4234         (pass_partition_blocks): Convert from a global struct to a subclass of
4235         rtl_opt_pass along with...
4236         (pass_data_partition_blocks): ...new pass_data instance and...
4237         (make_pass_partition_blocks): ...new function.
4238         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4239         struct to a subclass of rtl_opt_pass along with...
4240         (pass_data_branch_target_load_optimize1): ...new pass_data instance
4241         and...
4242         (make_pass_branch_target_load_optimize1): ...new function.
4243         (pass_branch_target_load_optimize2): Convert from a global struct to a
4244         subclass of rtl_opt_pass along with...
4245         (pass_data_branch_target_load_optimize2): ...new pass_data instance
4246         and...
4247         (make_pass_branch_target_load_optimize2): ...new function.
4248         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4249         of rtl_opt_pass along with...
4250         (pass_data_jump): ...new pass_data instance and...
4251         (make_pass_jump): ...new function.
4252         (pass_jump2): Convert from a global struct to a subclass of
4253         rtl_opt_pass along with...
4254         (pass_data_jump2): ...new pass_data instance and...
4255         (make_pass_jump2): ...new function.
4256         * cfgexpand.c (pass_expand): Convert from a global struct to a
4257         subclass of rtl_opt_pass along with...
4258         (pass_data_expand): ...new pass_data instance and...
4259         (make_pass_expand): ...new function.
4260         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4261         of rtl_opt_pass along with...
4262         (pass_data_free_cfg): ...new pass_data instance and...
4263         (make_pass_free_cfg): ...new function.
4264         (pass_into_cfg_layout_mode): Convert from a global struct to a
4265         subclass of rtl_opt_pass along with...
4266         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4267         (make_pass_into_cfg_layout_mode): ...new function.
4268         (pass_outof_cfg_layout_mode): Convert from a global struct to a
4269         subclass of rtl_opt_pass along with...
4270         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4271         (make_pass_outof_cfg_layout_mode): ...new function.
4272         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4273         struct to a subclass of gimple_opt_pass along with...
4274         (pass_data_build_cgraph_edges): ...new pass_data instance and...
4275         (make_pass_build_cgraph_edges): ...new function.
4276         (pass_rebuild_cgraph_edges): Convert from a global struct to a
4277         subclass of gimple_opt_pass along with...
4278         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4279         (make_pass_rebuild_cgraph_edges): ...new function.
4280         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4281         subclass of gimple_opt_pass along with...
4282         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4283         and...
4284         (make_pass_remove_cgraph_callee_edges): ...new function.
4285         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4286         struct to a subclass of rtl_opt_pass along with...
4287         (pass_data_stack_adjustments): ...new pass_data instance and...
4288         (make_pass_stack_adjustments): ...new function.
4289         * combine.c (pass_combine): Convert from a global struct to a subclass
4290         of rtl_opt_pass along with...
4291         (pass_data_combine): ...new pass_data instance and...
4292         (make_pass_combine): ...new function.
4293         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4294         global struct to a subclass of rtl_opt_pass along with...
4295         (pass_data_compare_elim_after_reload): ...new pass_data instance
4296         and...
4297         (make_pass_compare_elim_after_reload): ...new function.
4298         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4299         of rtl_opt_pass along with...
4300         (pass_data_rtl_cprop): ...new pass_data instance and...
4301         (make_pass_rtl_cprop): ...new function.
4302         * cse.c (pass_cse): Convert from a global struct to a subclass of
4303         rtl_opt_pass along with...
4304         (pass_data_cse): ...new pass_data instance and...
4305         (make_pass_cse): ...new function.
4306         (pass_cse2): Convert from a global struct to a subclass of
4307         rtl_opt_pass along with...
4308         (pass_data_cse2): ...new pass_data instance and...
4309         (make_pass_cse2): ...new function.
4310         (pass_cse_after_global_opts): Convert from a global struct to a
4311         subclass of rtl_opt_pass along with...
4312         (pass_data_cse_after_global_opts): ...new pass_data instance and...
4313         (make_pass_cse_after_global_opts): ...new function.
4314         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4315         of rtl_opt_pass along with...
4316         (pass_data_ud_rtl_dce): ...new pass_data instance and...
4317         (make_pass_ud_rtl_dce): ...new function.
4318         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4319         rtl_opt_pass along with...
4320         (pass_data_fast_rtl_dce): ...new pass_data instance and...
4321         (make_pass_fast_rtl_dce): ...new function.
4322         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4323         a subclass of rtl_opt_pass along with...
4324         (pass_data_df_initialize_opt): ...new pass_data instance and...
4325         (make_pass_df_initialize_opt): ...new function.
4326         (pass_df_initialize_no_opt): Convert from a global struct to a
4327         subclass of rtl_opt_pass along with...
4328         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4329         (make_pass_df_initialize_no_opt): ...new function.
4330         (pass_df_finish): Convert from a global struct to a subclass of
4331         rtl_opt_pass along with...
4332         (pass_data_df_finish): ...new pass_data instance and...
4333         (make_pass_df_finish): ...new function.
4334         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4335         rtl_opt_pass along with...
4336         (pass_data_rtl_dse1): ...new pass_data instance and...
4337         (make_pass_rtl_dse1): ...new function.
4338         (pass_rtl_dse2): Convert from a global struct to a subclass of
4339         rtl_opt_pass along with...
4340         (pass_data_rtl_dse2): ...new pass_data instance and...
4341         (make_pass_rtl_dse2): ...new function.
4342         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4343         subclass of rtl_opt_pass along with...
4344         (pass_data_dwarf2_frame): ...new pass_data instance and...
4345         (make_pass_dwarf2_frame): ...new function.
4346         * except.c (pass_set_nothrow_function_flags): Convert from a global
4347         struct to a subclass of rtl_opt_pass along with...
4348         (pass_data_set_nothrow_function_flags): ...new pass_data instance
4349         and...
4350         (make_pass_set_nothrow_function_flags): ...new function.
4351         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4352         subclass of rtl_opt_pass along with...
4353         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4354         and...
4355         (make_pass_convert_to_eh_region_ranges): ...new function.
4356         * final.c (pass_compute_alignments): Convert from a global struct to a
4357         subclass of rtl_opt_pass along with...
4358         (pass_data_compute_alignments): ...new pass_data instance and...
4359         (make_pass_compute_alignments): ...new function.
4360         (pass_final): Convert from a global struct to a subclass of
4361         rtl_opt_pass along with...
4362         (pass_data_final): ...new pass_data instance and...
4363         (make_pass_final): ...new function.
4364         (pass_shorten_branches): Convert from a global struct to a subclass of
4365         rtl_opt_pass along with...
4366         (pass_data_shorten_branches): ...new pass_data instance and...
4367         (make_pass_shorten_branches): ...new function.
4368         (pass_clean_state): Convert from a global struct to a subclass of
4369         rtl_opt_pass along with...
4370         (pass_data_clean_state): ...new pass_data instance and...
4371         (make_pass_clean_state): ...new function.
4372         * function.c (pass_instantiate_virtual_regs): Convert from a global
4373         struct to a subclass of rtl_opt_pass along with...
4374         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4375         (make_pass_instantiate_virtual_regs): ...new function.
4376         (pass_leaf_regs): Convert from a global struct to a subclass of
4377         rtl_opt_pass along with...
4378         (pass_data_leaf_regs): ...new pass_data instance and...
4379         (make_pass_leaf_regs): ...new function.
4380         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4381         subclass of rtl_opt_pass along with...
4382         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4383         and...
4384         (make_pass_thread_prologue_and_epilogue): ...new function.
4385         (pass_match_asm_constraints): Convert from a global struct to a
4386         subclass of rtl_opt_pass along with...
4387         (pass_data_match_asm_constraints): ...new pass_data instance and...
4388         (make_pass_match_asm_constraints): ...new function.
4389         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4390         subclass of rtl_opt_pass along with...
4391         (pass_data_rtl_fwprop): ...new pass_data instance and...
4392         (make_pass_rtl_fwprop): ...new function.
4393         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4394         rtl_opt_pass along with...
4395         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4396         (make_pass_rtl_fwprop_addr): ...new function.
4397         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4398         rtl_opt_pass along with...
4399         (pass_data_rtl_pre): ...new pass_data instance and...
4400         (make_pass_rtl_pre): ...new function.
4401         (pass_rtl_hoist): Convert from a global struct to a subclass of
4402         rtl_opt_pass along with...
4403         (pass_data_rtl_hoist): ...new pass_data instance and...
4404         (make_pass_rtl_hoist): ...new function.
4405         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4406         subclass of gimple_opt_pass along with...
4407         (pass_data_lower_cf): ...new pass_data instance and...
4408         (make_pass_lower_cf): ...new function.
4409         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4410         from a global struct to a subclass of gimple_opt_pass along with...
4411         (pass_data_strength_reduction): ...new pass_data instance and...
4412         (make_pass_strength_reduction): ...new function.
4413         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4414         of rtl_opt_pass along with...
4415         (pass_data_rtl_ifcvt): ...new pass_data instance and...
4416         (make_pass_rtl_ifcvt): ...new function.
4417         (pass_if_after_combine): Convert from a global struct to a subclass of
4418         rtl_opt_pass along with...
4419         (pass_data_if_after_combine): ...new pass_data instance and...
4420         (make_pass_if_after_combine): ...new function.
4421         (pass_if_after_reload): Convert from a global struct to a subclass of
4422         rtl_opt_pass along with...
4423         (pass_data_if_after_reload): ...new pass_data instance and...
4424         (make_pass_if_after_reload): ...new function.
4425         * init-regs.c (pass_initialize_regs): Convert from a global struct to
4426         a subclass of rtl_opt_pass along with...
4427         (pass_data_initialize_regs): ...new pass_data instance and...
4428         (make_pass_initialize_regs): ...new function.
4429         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4430         of ipa_opt_pass_d along with...
4431         (pass_data_ipa_cp): ...new pass_data instance and...
4432         (make_pass_ipa_cp): ...new function.
4433         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4434         global struct to a subclass of gimple_opt_pass along with...
4435         (pass_data_inline_parameters): ...new pass_data instance and...
4436         (make_pass_inline_parameters): ...new function.
4437         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4438         subclass of gimple_opt_pass along with...
4439         (pass_data_early_inline): ...new pass_data instance and...
4440         (make_pass_early_inline): ...new function.
4441         (pass_ipa_inline): Convert from a global struct to a subclass of
4442         ipa_opt_pass_d along with...
4443         (pass_data_ipa_inline): ...new pass_data instance and...
4444         (make_pass_ipa_inline): ...new function.
4445         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4446         struct to a subclass of gimple_opt_pass along with...
4447         (pass_data_local_pure_const): ...new pass_data instance and...
4448         (make_pass_local_pure_const): ...new function.
4449         (pass_ipa_pure_const): Convert from a global struct to a subclass of
4450         ipa_opt_pass_d along with...
4451         (pass_data_ipa_pure_const): ...new pass_data instance and...
4452         (make_pass_ipa_pure_const): ...new function.
4453         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4454         to a subclass of ipa_opt_pass_d along with...
4455         (pass_data_ipa_reference): ...new pass_data instance and...
4456         (make_pass_ipa_reference): ...new function.
4457         * ipa-split.c (pass_split_functions): Convert from a global struct to
4458         a subclass of gimple_opt_pass along with...
4459         (pass_data_split_functions): ...new pass_data instance and...
4460         (make_pass_split_functions): ...new function.
4461         (pass_feedback_split_functions): Convert from a global struct to a
4462         subclass of gimple_opt_pass along with...
4463         (pass_data_feedback_split_functions): ...new pass_data instance and...
4464         (make_pass_feedback_split_functions): ...new function.
4465         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4466         global struct to a subclass of simple_ipa_opt_pass along with...
4467         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4468         instance and...
4469         (make_pass_ipa_function_and_variable_visibility): ...new function.
4470         (pass_ipa_free_inline_summary): Convert from a global struct to a
4471         subclass of simple_ipa_opt_pass along with...
4472         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4473         (make_pass_ipa_free_inline_summary): ...new function.
4474         (pass_ipa_whole_program_visibility): Convert from a global struct to a
4475         subclass of ipa_opt_pass_d along with...
4476         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4477         and...
4478         (make_pass_ipa_whole_program_visibility): ...new function.
4479         (pass_ipa_profile): Convert from a global struct to a subclass of
4480         ipa_opt_pass_d along with...
4481         (pass_data_ipa_profile): ...new pass_data instance and...
4482         (make_pass_ipa_profile): ...new function.
4483         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4484         ipa_opt_pass_d along with...
4485         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4486         (make_pass_ipa_cdtor_merge): ...new function.
4487         * ira.c (pass_ira): Convert from a global struct to a subclass of
4488         rtl_opt_pass along with...
4489         (pass_data_ira): ...new pass_data instance and...
4490         (make_pass_ira): ...new function.
4491         (pass_reload): Convert from a global struct to a subclass of
4492         rtl_opt_pass along with...
4493         (pass_data_reload): ...new pass_data instance and...
4494         (make_pass_reload): ...new function.
4495         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4496         subclass of rtl_opt_pass along with...
4497         (pass_data_cleanup_barriers): ...new pass_data instance and...
4498         (make_pass_cleanup_barriers): ...new function.
4499         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4500         of rtl_opt_pass along with...
4501         (pass_data_loop2): ...new pass_data instance and...
4502         (make_pass_loop2): ...new function.
4503         (pass_rtl_loop_init): Convert from a global struct to a subclass of
4504         rtl_opt_pass along with...
4505         (pass_data_rtl_loop_init): ...new pass_data instance and...
4506         (make_pass_rtl_loop_init): ...new function.
4507         (pass_rtl_loop_done): Convert from a global struct to a subclass of
4508         rtl_opt_pass along with...
4509         (pass_data_rtl_loop_done): ...new pass_data instance and...
4510         (make_pass_rtl_loop_done): ...new function.
4511         (pass_rtl_move_loop_invariants): Convert from a global struct to a
4512         subclass of rtl_opt_pass along with...
4513         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4514         (make_pass_rtl_move_loop_invariants): ...new function.
4515         (pass_rtl_unswitch): Convert from a global struct to a subclass of
4516         rtl_opt_pass along with...
4517         (pass_data_rtl_unswitch): ...new pass_data instance and...
4518         (make_pass_rtl_unswitch): ...new function.
4519         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4520         subclass of rtl_opt_pass along with...
4521         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4522         and...
4523         (make_pass_rtl_unroll_and_peel_loops): ...new function.
4524         (pass_rtl_doloop): Convert from a global struct to a subclass of
4525         rtl_opt_pass along with...
4526         (pass_data_rtl_doloop): ...new pass_data instance and...
4527         (make_pass_rtl_doloop): ...new function.
4528         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4529         a subclass of rtl_opt_pass along with...
4530         (pass_data_lower_subreg): ...new pass_data instance and...
4531         (make_pass_lower_subreg): ...new function.
4532         (pass_lower_subreg2): Convert from a global struct to a subclass of
4533         rtl_opt_pass along with...
4534         (pass_data_lower_subreg2): ...new pass_data instance and...
4535         (make_pass_lower_subreg2): ...new function.
4536         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4537         struct to a subclass of ipa_opt_pass_d along with...
4538         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4539         (make_pass_ipa_lto_gimple_out): ...new function.
4540         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4541         of ipa_opt_pass_d along with...
4542         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4543         (make_pass_ipa_lto_finish_out): ...new function.
4544         * mode-switching.c (pass_mode_switching): Convert from a global struct
4545         to a subclass of rtl_opt_pass along with...
4546         (pass_data_mode_switching): ...new pass_data instance and...
4547         (make_pass_mode_switching): ...new function.
4548         * modulo-sched.c (pass_sms): Convert from a global struct to a
4549         subclass of rtl_opt_pass along with...
4550         (pass_data_sms): ...new pass_data instance and...
4551         (make_pass_sms): ...new function.
4552         * omp-low.c (pass_expand_omp): Convert from a global struct to a
4553         subclass of gimple_opt_pass along with...
4554         (pass_data_expand_omp): ...new pass_data instance and...
4555         (make_pass_expand_omp): ...new function.
4556         (pass_lower_omp): Convert from a global struct to a subclass of
4557         gimple_opt_pass along with...
4558         (pass_data_lower_omp): ...new pass_data instance and...
4559         (make_pass_lower_omp): ...new function.
4560         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4561         of gimple_opt_pass along with...
4562         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4563         (make_pass_diagnose_omp_blocks): ...new function.
4564         * passes.c (pass_early_local_passes): Convert from a global struct to
4565         a subclass of simple_ipa_opt_pass along with...
4566         (pass_data_early_local_passes): ...new pass_data instance and...
4567         (make_pass_early_local_passes): ...new function.
4568         (pass_all_early_optimizations): Convert from a global struct to a
4569         subclass of gimple_opt_pass along with...
4570         (pass_data_all_early_optimizations): ...new pass_data instance and...
4571         (make_pass_all_early_optimizations): ...new function.
4572         (pass_all_optimizations): Convert from a global struct to a subclass
4573         of gimple_opt_pass along with...
4574         (pass_data_all_optimizations): ...new pass_data instance and...
4575         (make_pass_all_optimizations): ...new function.
4576         (pass_all_optimizations_g): Convert from a global struct to a subclass
4577         of gimple_opt_pass along with...
4578         (pass_data_all_optimizations_g): ...new pass_data instance and...
4579         (make_pass_all_optimizations_g): ...new function.
4580         (pass_rest_of_compilation): Convert from a global struct to a subclass
4581         of rtl_opt_pass along with...
4582         (pass_data_rest_of_compilation): ...new pass_data instance and...
4583         (make_pass_rest_of_compilation): ...new function.
4584         (pass_postreload): Convert from a global struct to a subclass of
4585         rtl_opt_pass along with...
4586         (pass_data_postreload): ...new pass_data instance and...
4587         (make_pass_postreload): ...new function.
4588         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
4589         subclass of rtl_opt_pass along with...
4590         (pass_data_gcse2): ...new pass_data instance and...
4591         (make_pass_gcse2): ...new function.
4592         * postreload.c (pass_postreload_cse): Convert from a global struct to
4593         a subclass of rtl_opt_pass along with...
4594         (pass_data_postreload_cse): ...new pass_data instance and...
4595         (make_pass_postreload_cse): ...new function.
4596         * predict.c (pass_profile): Convert from a global struct to a subclass
4597         of gimple_opt_pass along with...
4598         (pass_data_profile): ...new pass_data instance and...
4599         (make_pass_profile): ...new function.
4600         (pass_strip_predict_hints): Convert from a global struct to a subclass
4601         of gimple_opt_pass along with...
4602         (pass_data_strip_predict_hints): ...new pass_data instance and...
4603         (make_pass_strip_predict_hints): ...new function.
4604         * recog.c (pass_peephole2): Convert from a global struct to a subclass
4605         of rtl_opt_pass along with...
4606         (pass_data_peephole2): ...new pass_data instance and...
4607         (make_pass_peephole2): ...new function.
4608         (pass_split_all_insns): Convert from a global struct to a subclass of
4609         rtl_opt_pass along with...
4610         (pass_data_split_all_insns): ...new pass_data instance and...
4611         (make_pass_split_all_insns): ...new function.
4612         (pass_split_after_reload): Convert from a global struct to a subclass
4613         of rtl_opt_pass along with...
4614         (pass_data_split_after_reload): ...new pass_data instance and...
4615         (make_pass_split_after_reload): ...new function.
4616         (pass_split_before_regstack): Convert from a global struct to a
4617         subclass of rtl_opt_pass along with...
4618         (pass_data_split_before_regstack): ...new pass_data instance and...
4619         (make_pass_split_before_regstack): ...new function.
4620         (pass_split_before_sched2): Convert from a global struct to a subclass
4621         of rtl_opt_pass along with...
4622         (pass_data_split_before_sched2): ...new pass_data instance and...
4623         (make_pass_split_before_sched2): ...new function.
4624         (pass_split_for_shorten_branches): Convert from a global struct to a
4625         subclass of rtl_opt_pass along with...
4626         (pass_data_split_for_shorten_branches): ...new pass_data instance
4627         and...
4628         (make_pass_split_for_shorten_branches): ...new function.
4629         * ree.c (pass_ree): Convert from a global struct to a subclass of
4630         rtl_opt_pass along with...
4631         (pass_data_ree): ...new pass_data instance and...
4632         (make_pass_ree): ...new function.
4633         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
4634         subclass of rtl_opt_pass along with...
4635         (pass_data_stack_regs): ...new pass_data instance and...
4636         (make_pass_stack_regs): ...new function.
4637         (pass_stack_regs_run): Convert from a global struct to a subclass of
4638         rtl_opt_pass along with...
4639         (pass_data_stack_regs_run): ...new pass_data instance and...
4640         (make_pass_stack_regs_run): ...new function.
4641         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
4642         subclass of rtl_opt_pass along with...
4643         (pass_data_cprop_hardreg): ...new pass_data instance and...
4644         (make_pass_cprop_hardreg): ...new function.
4645         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
4646         subclass of rtl_opt_pass along with...
4647         (pass_data_reginfo_init): ...new pass_data instance and...
4648         (make_pass_reginfo_init): ...new function.
4649         * regmove.c (pass_regmove): Convert from a global struct to a subclass
4650         of rtl_opt_pass along with...
4651         (pass_data_regmove): ...new pass_data instance and...
4652         (make_pass_regmove): ...new function.
4653         * regrename.c (pass_regrename): Convert from a global struct to a
4654         subclass of rtl_opt_pass along with...
4655         (pass_data_regrename): ...new pass_data instance and...
4656         (make_pass_regrename): ...new function.
4657         * reorg.c (pass_delay_slots): Convert from a global struct to a
4658         subclass of rtl_opt_pass along with...
4659         (pass_data_delay_slots): ...new pass_data instance and...
4660         (make_pass_delay_slots): ...new function.
4661         (pass_machine_reorg): Convert from a global struct to a subclass of
4662         rtl_opt_pass along with...
4663         (pass_data_machine_reorg): ...new pass_data instance and...
4664         (make_pass_machine_reorg): ...new function.
4665         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
4666         of rtl_opt_pass along with...
4667         (pass_data_sched): ...new pass_data instance and...
4668         (make_pass_sched): ...new function.
4669         (pass_sched2): Convert from a global struct to a subclass of
4670         rtl_opt_pass along with...
4671         (pass_data_sched2): ...new pass_data instance and...
4672         (make_pass_sched2): ...new function.
4673         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
4674         to a subclass of rtl_opt_pass along with...
4675         (pass_data_stack_ptr_mod): ...new pass_data instance and...
4676         (make_pass_stack_ptr_mod): ...new function.
4677         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
4678         to a subclass of rtl_opt_pass along with...
4679         (pass_data_rtl_store_motion): ...new pass_data instance and...
4680         (make_pass_rtl_store_motion): ...new function.
4681         * tracer.c (pass_tracer): Convert from a global struct to a subclass
4682         of gimple_opt_pass along with...
4683         (pass_data_tracer): ...new pass_data instance and...
4684         (make_pass_tracer): ...new function.
4685         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
4686         to a subclass of gimple_opt_pass along with...
4687         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
4688         (make_pass_diagnose_tm_blocks): ...new function.
4689         (pass_lower_tm): Convert from a global struct to a subclass of
4690         gimple_opt_pass along with...
4691         (pass_data_lower_tm): ...new pass_data instance and...
4692         (make_pass_lower_tm): ...new function.
4693         (pass_tm_init): Convert from a global struct to a subclass of
4694         gimple_opt_pass along with...
4695         (pass_data_tm_init): ...new pass_data instance and...
4696         (make_pass_tm_init): ...new function.
4697         (pass_tm_mark): Convert from a global struct to a subclass of
4698         gimple_opt_pass along with...
4699         (pass_data_tm_mark): ...new pass_data instance and...
4700         (make_pass_tm_mark): ...new function.
4701         (pass_tm_edges): Convert from a global struct to a subclass of
4702         gimple_opt_pass along with...
4703         (pass_data_tm_edges): ...new pass_data instance and...
4704         (make_pass_tm_edges): ...new function.
4705         (pass_tm_memopt): Convert from a global struct to a subclass of
4706         gimple_opt_pass along with...
4707         (pass_data_tm_memopt): ...new pass_data instance and...
4708         (make_pass_tm_memopt): ...new function.
4709         (pass_ipa_tm): Convert from a global struct to a subclass of
4710         simple_ipa_opt_pass along with...
4711         (pass_data_ipa_tm): ...new pass_data instance and...
4712         (make_pass_ipa_tm): ...new function.
4713         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
4714         subclass of gimple_opt_pass along with...
4715         (pass_data_call_cdce): ...new pass_data instance and...
4716         (make_pass_call_cdce): ...new function.
4717         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
4718         subclass of gimple_opt_pass along with...
4719         (pass_data_build_cfg): ...new pass_data instance and...
4720         (make_pass_build_cfg): ...new function.
4721         (pass_split_crit_edges): Convert from a global struct to a subclass of
4722         gimple_opt_pass along with...
4723         (pass_data_split_crit_edges): ...new pass_data instance and...
4724         (make_pass_split_crit_edges): ...new function.
4725         (pass_warn_function_return): Convert from a global struct to a
4726         subclass of gimple_opt_pass along with...
4727         (pass_data_warn_function_return): ...new pass_data instance and...
4728         (make_pass_warn_function_return): ...new function.
4729         (pass_warn_function_noreturn): Convert from a global struct to a
4730         subclass of gimple_opt_pass along with...
4731         (pass_data_warn_function_noreturn): ...new pass_data instance and...
4732         (make_pass_warn_function_noreturn): ...new function.
4733         (pass_warn_unused_result): Convert from a global struct to a subclass
4734         of gimple_opt_pass along with...
4735         (pass_data_warn_unused_result): ...new pass_data instance and...
4736         (make_pass_warn_unused_result): ...new function.
4737         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
4738         a subclass of gimple_opt_pass along with...
4739         (pass_data_merge_phi): ...new pass_data instance and...
4740         (make_pass_merge_phi): ...new function.
4741         * tree-complex.c (pass_lower_complex): Convert from a global struct to
4742         a subclass of gimple_opt_pass along with...
4743         (pass_data_lower_complex): ...new pass_data instance and...
4744         (make_pass_lower_complex): ...new function.
4745         (pass_lower_complex_O0): Convert from a global struct to a subclass of
4746         gimple_opt_pass along with...
4747         (pass_data_lower_complex_O0): ...new pass_data instance and...
4748         (make_pass_lower_complex_O0): ...new function.
4749         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
4750         subclass of gimple_opt_pass along with...
4751         (pass_data_lower_eh): ...new pass_data instance and...
4752         (make_pass_lower_eh): ...new function.
4753         (pass_refactor_eh): Convert from a global struct to a subclass of
4754         gimple_opt_pass along with...
4755         (pass_data_refactor_eh): ...new pass_data instance and...
4756         (make_pass_refactor_eh): ...new function.
4757         (pass_lower_resx): Convert from a global struct to a subclass of
4758         gimple_opt_pass along with...
4759         (pass_data_lower_resx): ...new pass_data instance and...
4760         (make_pass_lower_resx): ...new function.
4761         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
4762         of gimple_opt_pass along with...
4763         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
4764         (make_pass_lower_eh_dispatch): ...new function.
4765         (pass_cleanup_eh): Convert from a global struct to a subclass of
4766         gimple_opt_pass along with...
4767         (pass_data_cleanup_eh): ...new pass_data instance and...
4768         (make_pass_cleanup_eh): ...new function.
4769         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
4770         to a subclass of simple_ipa_opt_pass along with...
4771         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
4772         (make_pass_ipa_lower_emutls): ...new function.
4773         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
4774         a subclass of gimple_opt_pass along with...
4775         (pass_data_if_conversion): ...new pass_data instance and...
4776         (make_pass_if_conversion): ...new function.
4777         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
4778         subclass of gimple_opt_pass along with...
4779         (pass_data_build_ssa): ...new pass_data instance and...
4780         (make_pass_build_ssa): ...new function.
4781         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
4782         global struct to a subclass of gimple_opt_pass along with...
4783         (pass_data_loop_distribution): ...new pass_data instance and...
4784         (make_pass_loop_distribution): ...new function.
4785         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
4786         subclass of gimple_opt_pass along with...
4787         (pass_data_mudflap_1): ...new pass_data instance and...
4788         (make_pass_mudflap_1): ...new function.
4789         (pass_mudflap_2): Convert from a global struct to a subclass of
4790         gimple_opt_pass along with...
4791         (pass_data_mudflap_2): ...new pass_data instance and...
4792         (make_pass_mudflap_2): ...new function.
4793         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
4794         subclass of gimple_opt_pass along with...
4795         (pass_data_mudflap_1): ...new pass_data instance and...
4796         (make_pass_mudflap_1): ...new function.
4797         (pass_mudflap_2): Convert from a global struct to a subclass of
4798         gimple_opt_pass along with...
4799         (pass_data_mudflap_2): ...new pass_data instance and...
4800         (make_pass_mudflap_2): ...new function.
4801         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
4802         gimple_opt_pass along with...
4803         (pass_data_nrv): ...new pass_data instance and...
4804         (make_pass_nrv): ...new function.
4805         (pass_return_slot): Convert from a global struct to a subclass of
4806         gimple_opt_pass along with...
4807         (pass_data_return_slot): ...new pass_data instance and...
4808         (make_pass_return_slot): ...new function.
4809         * tree-object-size.c (pass_object_sizes): Convert from a global struct
4810         to a subclass of gimple_opt_pass along with...
4811         (pass_data_object_sizes): ...new pass_data instance and...
4812         (make_pass_object_sizes): ...new function.
4813         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
4814         global struct to a subclass of gimple_opt_pass along with...
4815         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
4816         and...
4817         (make_pass_cleanup_cfg_post_optimizing): ...new function.
4818         (pass_fixup_cfg): Convert from a global struct to a subclass of
4819         gimple_opt_pass along with...
4820         (pass_data_fixup_cfg): ...new pass_data instance and...
4821         (make_pass_fixup_cfg): ...new function.
4822         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
4823         (make_pass_mudflap_1): ...new function.
4824         (pass_mudflap_2): Replace declaration with that of...
4825         (make_pass_mudflap_2): ...new function.
4826         (pass_asan): Replace declaration with that of...
4827         (make_pass_asan): ...new function.
4828         (pass_asan_O0): Replace declaration with that of...
4829         (make_pass_asan_O0): ...new function.
4830         (pass_tsan): Replace declaration with that of...
4831         (make_pass_tsan): ...new function.
4832         (pass_tsan_O0): Replace declaration with that of...
4833         (make_pass_tsan_O0): ...new function.
4834         (pass_lower_cf): Replace declaration with that of...
4835         (make_pass_lower_cf): ...new function.
4836         (pass_refactor_eh): Replace declaration with that of...
4837         (make_pass_refactor_eh): ...new function.
4838         (pass_lower_eh): Replace declaration with that of...
4839         (make_pass_lower_eh): ...new function.
4840         (pass_lower_eh_dispatch): Replace declaration with that of...
4841         (make_pass_lower_eh_dispatch): ...new function.
4842         (pass_lower_resx): Replace declaration with that of...
4843         (make_pass_lower_resx): ...new function.
4844         (pass_build_cfg): Replace declaration with that of...
4845         (make_pass_build_cfg): ...new function.
4846         (pass_early_tree_profile): Replace declaration with that of...
4847         (make_pass_early_tree_profile): ...new function.
4848         (pass_cleanup_eh): Replace declaration with that of...
4849         (make_pass_cleanup_eh): ...new function.
4850         (pass_sra): Replace declaration with that of...
4851         (make_pass_sra): ...new function.
4852         (pass_sra_early): Replace declaration with that of...
4853         (make_pass_sra_early): ...new function.
4854         (pass_early_ipa_sra): Replace declaration with that of...
4855         (make_pass_early_ipa_sra): ...new function.
4856         (pass_tail_recursion): Replace declaration with that of...
4857         (make_pass_tail_recursion): ...new function.
4858         (pass_tail_calls): Replace declaration with that of...
4859         (make_pass_tail_calls): ...new function.
4860         (pass_tree_loop): Replace declaration with that of...
4861         (make_pass_tree_loop): ...new function.
4862         (pass_tree_loop_init): Replace declaration with that of...
4863         (make_pass_tree_loop_init): ...new function.
4864         (pass_lim): Replace declaration with that of...
4865         (make_pass_lim): ...new function.
4866         (pass_tree_unswitch): Replace declaration with that of...
4867         (make_pass_tree_unswitch): ...new function.
4868         (pass_predcom): Replace declaration with that of...
4869         (make_pass_predcom): ...new function.
4870         (pass_iv_canon): Replace declaration with that of...
4871         (make_pass_iv_canon): ...new function.
4872         (pass_scev_cprop): Replace declaration with that of...
4873         (make_pass_scev_cprop): ...new function.
4874         (pass_empty_loop): Replace declaration with that of...
4875         (make_pass_empty_loop): ...new function.
4876         (pass_record_bounds): Replace declaration with that of...
4877         (make_pass_record_bounds): ...new function.
4878         (pass_graphite): Replace declaration with that of...
4879         (make_pass_graphite): ...new function.
4880         (pass_graphite_transforms): Replace declaration with that of...
4881         (make_pass_graphite_transforms): ...new function.
4882         (pass_if_conversion): Replace declaration with that of...
4883         (make_pass_if_conversion): ...new function.
4884         (pass_loop_distribution): Replace declaration with that of...
4885         (make_pass_loop_distribution): ...new function.
4886         (pass_vectorize): Replace declaration with that of...
4887         (make_pass_vectorize): ...new function.
4888         (pass_slp_vectorize): Replace declaration with that of...
4889         (make_pass_slp_vectorize): ...new function.
4890         (pass_complete_unroll): Replace declaration with that of...
4891         (make_pass_complete_unroll): ...new function.
4892         (pass_complete_unrolli): Replace declaration with that of...
4893         (make_pass_complete_unrolli): ...new function.
4894         (pass_parallelize_loops): Replace declaration with that of...
4895         (make_pass_parallelize_loops): ...new function.
4896         (pass_loop_prefetch): Replace declaration with that of...
4897         (make_pass_loop_prefetch): ...new function.
4898         (pass_iv_optimize): Replace declaration with that of...
4899         (make_pass_iv_optimize): ...new function.
4900         (pass_tree_loop_done): Replace declaration with that of...
4901         (make_pass_tree_loop_done): ...new function.
4902         (pass_ch): Replace declaration with that of...
4903         (make_pass_ch): ...new function.
4904         (pass_ccp): Replace declaration with that of...
4905         (make_pass_ccp): ...new function.
4906         (pass_phi_only_cprop): Replace declaration with that of...
4907         (make_pass_phi_only_cprop): ...new function.
4908         (pass_build_ssa): Replace declaration with that of...
4909         (make_pass_build_ssa): ...new function.
4910         (pass_build_alias): Replace declaration with that of...
4911         (make_pass_build_alias): ...new function.
4912         (pass_build_ealias): Replace declaration with that of...
4913         (make_pass_build_ealias): ...new function.
4914         (pass_dominator): Replace declaration with that of...
4915         (make_pass_dominator): ...new function.
4916         (pass_dce): Replace declaration with that of...
4917         (make_pass_dce): ...new function.
4918         (pass_dce_loop): Replace declaration with that of...
4919         (make_pass_dce_loop): ...new function.
4920         (pass_cd_dce): Replace declaration with that of...
4921         (make_pass_cd_dce): ...new function.
4922         (pass_call_cdce): Replace declaration with that of...
4923         (make_pass_call_cdce): ...new function.
4924         (pass_merge_phi): Replace declaration with that of...
4925         (make_pass_merge_phi): ...new function.
4926         (pass_split_crit_edges): Replace declaration with that of...
4927         (make_pass_split_crit_edges): ...new function.
4928         (pass_pre): Replace declaration with that of...
4929         (make_pass_pre): ...new function.
4930         (pass_profile): Replace declaration with that of...
4931         (make_pass_profile): ...new function.
4932         (pass_strip_predict_hints): Replace declaration with that of...
4933         (make_pass_strip_predict_hints): ...new function.
4934         (pass_lower_complex_O0): Replace declaration with that of...
4935         (make_pass_lower_complex_O0): ...new function.
4936         (pass_lower_complex): Replace declaration with that of...
4937         (make_pass_lower_complex): ...new function.
4938         (pass_lower_vector): Replace declaration with that of...
4939         (make_pass_lower_vector): ...new function.
4940         (pass_lower_vector_ssa): Replace declaration with that of...
4941         (make_pass_lower_vector_ssa): ...new function.
4942         (pass_lower_omp): Replace declaration with that of...
4943         (make_pass_lower_omp): ...new function.
4944         (pass_diagnose_omp_blocks): Replace declaration with that of...
4945         (make_pass_diagnose_omp_blocks): ...new function.
4946         (pass_expand_omp): Replace declaration with that of...
4947         (make_pass_expand_omp): ...new function.
4948         (pass_expand_omp_ssa): Replace declaration with that of...
4949         (make_pass_expand_omp_ssa): ...new function.
4950         (pass_object_sizes): Replace declaration with that of...
4951         (make_pass_object_sizes): ...new function.
4952         (pass_strlen): Replace declaration with that of...
4953         (make_pass_strlen): ...new function.
4954         (pass_fold_builtins): Replace declaration with that of...
4955         (make_pass_fold_builtins): ...new function.
4956         (pass_stdarg): Replace declaration with that of...
4957         (make_pass_stdarg): ...new function.
4958         (pass_early_warn_uninitialized): Replace declaration with that of...
4959         (make_pass_early_warn_uninitialized): ...new function.
4960         (pass_late_warn_uninitialized): Replace declaration with that of...
4961         (make_pass_late_warn_uninitialized): ...new function.
4962         (pass_cse_reciprocals): Replace declaration with that of...
4963         (make_pass_cse_reciprocals): ...new function.
4964         (pass_cse_sincos): Replace declaration with that of...
4965         (make_pass_cse_sincos): ...new function.
4966         (pass_optimize_bswap): Replace declaration with that of...
4967         (make_pass_optimize_bswap): ...new function.
4968         (pass_optimize_widening_mul): Replace declaration with that of...
4969         (make_pass_optimize_widening_mul): ...new function.
4970         (pass_warn_function_return): Replace declaration with that of...
4971         (make_pass_warn_function_return): ...new function.
4972         (pass_warn_function_noreturn): Replace declaration with that of...
4973         (make_pass_warn_function_noreturn): ...new function.
4974         (pass_cselim): Replace declaration with that of...
4975         (make_pass_cselim): ...new function.
4976         (pass_phiopt): Replace declaration with that of...
4977         (make_pass_phiopt): ...new function.
4978         (pass_forwprop): Replace declaration with that of...
4979         (make_pass_forwprop): ...new function.
4980         (pass_phiprop): Replace declaration with that of...
4981         (make_pass_phiprop): ...new function.
4982         (pass_tree_ifcombine): Replace declaration with that of...
4983         (make_pass_tree_ifcombine): ...new function.
4984         (pass_dse): Replace declaration with that of...
4985         (make_pass_dse): ...new function.
4986         (pass_nrv): Replace declaration with that of...
4987         (make_pass_nrv): ...new function.
4988         (pass_rename_ssa_copies): Replace declaration with that of...
4989         (make_pass_rename_ssa_copies): ...new function.
4990         (pass_sink_code): Replace declaration with that of...
4991         (make_pass_sink_code): ...new function.
4992         (pass_fre): Replace declaration with that of...
4993         (make_pass_fre): ...new function.
4994         (pass_check_data_deps): Replace declaration with that of...
4995         (make_pass_check_data_deps): ...new function.
4996         (pass_copy_prop): Replace declaration with that of...
4997         (make_pass_copy_prop): ...new function.
4998         (pass_vrp): Replace declaration with that of...
4999         (make_pass_vrp): ...new function.
5000         (pass_uncprop): Replace declaration with that of...
5001         (make_pass_uncprop): ...new function.
5002         (pass_return_slot): Replace declaration with that of...
5003         (make_pass_return_slot): ...new function.
5004         (pass_reassoc): Replace declaration with that of...
5005         (make_pass_reassoc): ...new function.
5006         (pass_rebuild_cgraph_edges): Replace declaration with that of...
5007         (make_pass_rebuild_cgraph_edges): ...new function.
5008         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5009         (make_pass_remove_cgraph_callee_edges): ...new function.
5010         (pass_build_cgraph_edges): Replace declaration with that of...
5011         (make_pass_build_cgraph_edges): ...new function.
5012         (pass_local_pure_const): Replace declaration with that of...
5013         (make_pass_local_pure_const): ...new function.
5014         (pass_tracer): Replace declaration with that of...
5015         (make_pass_tracer): ...new function.
5016         (pass_warn_unused_result): Replace declaration with that of...
5017         (make_pass_warn_unused_result): ...new function.
5018         (pass_diagnose_tm_blocks): Replace declaration with that of...
5019         (make_pass_diagnose_tm_blocks): ...new function.
5020         (pass_lower_tm): Replace declaration with that of...
5021         (make_pass_lower_tm): ...new function.
5022         (pass_tm_init): Replace declaration with that of...
5023         (make_pass_tm_init): ...new function.
5024         (pass_tm_mark): Replace declaration with that of...
5025         (make_pass_tm_mark): ...new function.
5026         (pass_tm_memopt): Replace declaration with that of...
5027         (make_pass_tm_memopt): ...new function.
5028         (pass_tm_edges): Replace declaration with that of...
5029         (make_pass_tm_edges): ...new function.
5030         (pass_split_functions): Replace declaration with that of...
5031         (make_pass_split_functions): ...new function.
5032         (pass_feedback_split_functions): Replace declaration with that of...
5033         (make_pass_feedback_split_functions): ...new function.
5034         (pass_strength_reduction): Replace declaration with that of...
5035         (make_pass_strength_reduction): ...new function.
5036         (pass_ipa_lower_emutls): Replace declaration with that of...
5037         (make_pass_ipa_lower_emutls): ...new function.
5038         (pass_ipa_function_and_variable_visibility): Replace declaration with
5039         that of...
5040         (make_pass_ipa_function_and_variable_visibility): ...new function.
5041         (pass_ipa_tree_profile): Replace declaration with that of...
5042         (make_pass_ipa_tree_profile): ...new function.
5043         (pass_early_local_passes): Replace declaration with that of...
5044         (make_pass_early_local_passes): ...new function.
5045         (pass_ipa_whole_program_visibility): Replace declaration with that
5046         of...
5047         (make_pass_ipa_whole_program_visibility): ...new function.
5048         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5049         (make_pass_ipa_lto_gimple_out): ...new function.
5050         (pass_ipa_increase_alignment): Replace declaration with that of...
5051         (make_pass_ipa_increase_alignment): ...new function.
5052         (pass_ipa_inline): Replace declaration with that of...
5053         (make_pass_ipa_inline): ...new function.
5054         (pass_ipa_free_lang_data): Replace declaration with that of...
5055         (make_pass_ipa_free_lang_data): ...new function.
5056         (pass_ipa_free_inline_summary): Replace declaration with that of...
5057         (make_pass_ipa_free_inline_summary): ...new function.
5058         (pass_ipa_cp): Replace declaration with that of...
5059         (make_pass_ipa_cp): ...new function.
5060         (pass_ipa_reference): Replace declaration with that of...
5061         (make_pass_ipa_reference): ...new function.
5062         (pass_ipa_pure_const): Replace declaration with that of...
5063         (make_pass_ipa_pure_const): ...new function.
5064         (pass_ipa_pta): Replace declaration with that of...
5065         (make_pass_ipa_pta): ...new function.
5066         (pass_ipa_lto_finish_out): Replace declaration with that of...
5067         (make_pass_ipa_lto_finish_out): ...new function.
5068         (pass_ipa_tm): Replace declaration with that of...
5069         (make_pass_ipa_tm): ...new function.
5070         (pass_ipa_profile): Replace declaration with that of...
5071         (make_pass_ipa_profile): ...new function.
5072         (pass_ipa_cdtor_merge): Replace declaration with that of...
5073         (make_pass_ipa_cdtor_merge): ...new function.
5074         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5075         of...
5076         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5077         (pass_init_datastructures): Replace declaration with that of...
5078         (make_pass_init_datastructures): ...new function.
5079         (pass_fixup_cfg): Replace declaration with that of...
5080         (make_pass_fixup_cfg): ...new function.
5081         (pass_expand): Replace declaration with that of...
5082         (make_pass_expand): ...new function.
5083         (pass_instantiate_virtual_regs): Replace declaration with that of...
5084         (make_pass_instantiate_virtual_regs): ...new function.
5085         (pass_rtl_fwprop): Replace declaration with that of...
5086         (make_pass_rtl_fwprop): ...new function.
5087         (pass_rtl_fwprop_addr): Replace declaration with that of...
5088         (make_pass_rtl_fwprop_addr): ...new function.
5089         (pass_jump): Replace declaration with that of...
5090         (make_pass_jump): ...new function.
5091         (pass_jump2): Replace declaration with that of...
5092         (make_pass_jump2): ...new function.
5093         (pass_lower_subreg): Replace declaration with that of...
5094         (make_pass_lower_subreg): ...new function.
5095         (pass_cse): Replace declaration with that of...
5096         (make_pass_cse): ...new function.
5097         (pass_fast_rtl_dce): Replace declaration with that of...
5098         (make_pass_fast_rtl_dce): ...new function.
5099         (pass_ud_rtl_dce): Replace declaration with that of...
5100         (make_pass_ud_rtl_dce): ...new function.
5101         (pass_rtl_dce): Replace declaration with that of...
5102         (make_pass_rtl_dce): ...new function.
5103         (pass_rtl_dse1): Replace declaration with that of...
5104         (make_pass_rtl_dse1): ...new function.
5105         (pass_rtl_dse2): Replace declaration with that of...
5106         (make_pass_rtl_dse2): ...new function.
5107         (pass_rtl_dse3): Replace declaration with that of...
5108         (make_pass_rtl_dse3): ...new function.
5109         (pass_rtl_cprop): Replace declaration with that of...
5110         (make_pass_rtl_cprop): ...new function.
5111         (pass_rtl_pre): Replace declaration with that of...
5112         (make_pass_rtl_pre): ...new function.
5113         (pass_rtl_hoist): Replace declaration with that of...
5114         (make_pass_rtl_hoist): ...new function.
5115         (pass_rtl_store_motion): Replace declaration with that of...
5116         (make_pass_rtl_store_motion): ...new function.
5117         (pass_cse_after_global_opts): Replace declaration with that of...
5118         (make_pass_cse_after_global_opts): ...new function.
5119         (pass_rtl_ifcvt): Replace declaration with that of...
5120         (make_pass_rtl_ifcvt): ...new function.
5121         (pass_into_cfg_layout_mode): Replace declaration with that of...
5122         (make_pass_into_cfg_layout_mode): ...new function.
5123         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5124         (make_pass_outof_cfg_layout_mode): ...new function.
5125         (pass_loop2): Replace declaration with that of...
5126         (make_pass_loop2): ...new function.
5127         (pass_rtl_loop_init): Replace declaration with that of...
5128         (make_pass_rtl_loop_init): ...new function.
5129         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5130         (make_pass_rtl_move_loop_invariants): ...new function.
5131         (pass_rtl_unswitch): Replace declaration with that of...
5132         (make_pass_rtl_unswitch): ...new function.
5133         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5134         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5135         (pass_rtl_doloop): Replace declaration with that of...
5136         (make_pass_rtl_doloop): ...new function.
5137         (pass_rtl_loop_done): Replace declaration with that of...
5138         (make_pass_rtl_loop_done): ...new function.
5139         (pass_web): Replace declaration with that of...
5140         (make_pass_web): ...new function.
5141         (pass_cse2): Replace declaration with that of...
5142         (make_pass_cse2): ...new function.
5143         (pass_df_initialize_opt): Replace declaration with that of...
5144         (make_pass_df_initialize_opt): ...new function.
5145         (pass_df_initialize_no_opt): Replace declaration with that of...
5146         (make_pass_df_initialize_no_opt): ...new function.
5147         (pass_reginfo_init): Replace declaration with that of...
5148         (make_pass_reginfo_init): ...new function.
5149         (pass_inc_dec): Replace declaration with that of...
5150         (make_pass_inc_dec): ...new function.
5151         (pass_stack_ptr_mod): Replace declaration with that of...
5152         (make_pass_stack_ptr_mod): ...new function.
5153         (pass_initialize_regs): Replace declaration with that of...
5154         (make_pass_initialize_regs): ...new function.
5155         (pass_combine): Replace declaration with that of...
5156         (make_pass_combine): ...new function.
5157         (pass_if_after_combine): Replace declaration with that of...
5158         (make_pass_if_after_combine): ...new function.
5159         (pass_ree): Replace declaration with that of...
5160         (make_pass_ree): ...new function.
5161         (pass_partition_blocks): Replace declaration with that of...
5162         (make_pass_partition_blocks): ...new function.
5163         (pass_match_asm_constraints): Replace declaration with that of...
5164         (make_pass_match_asm_constraints): ...new function.
5165         (pass_regmove): Replace declaration with that of...
5166         (make_pass_regmove): ...new function.
5167         (pass_split_all_insns): Replace declaration with that of...
5168         (make_pass_split_all_insns): ...new function.
5169         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5170         (make_pass_fast_rtl_byte_dce): ...new function.
5171         (pass_lower_subreg2): Replace declaration with that of...
5172         (make_pass_lower_subreg2): ...new function.
5173         (pass_mode_switching): Replace declaration with that of...
5174         (make_pass_mode_switching): ...new function.
5175         (pass_sms): Replace declaration with that of...
5176         (make_pass_sms): ...new function.
5177         (pass_sched): Replace declaration with that of...
5178         (make_pass_sched): ...new function.
5179         (pass_ira): Replace declaration with that of...
5180         (make_pass_ira): ...new function.
5181         (pass_reload): Replace declaration with that of...
5182         (make_pass_reload): ...new function.
5183         (pass_clean_state): Replace declaration with that of...
5184         (make_pass_clean_state): ...new function.
5185         (pass_branch_prob): Replace declaration with that of...
5186         (make_pass_branch_prob): ...new function.
5187         (pass_value_profile_transformations): Replace declaration with that
5188         of...
5189         (make_pass_value_profile_transformations): ...new function.
5190         (pass_postreload_cse): Replace declaration with that of...
5191         (make_pass_postreload_cse): ...new function.
5192         (pass_gcse2): Replace declaration with that of...
5193         (make_pass_gcse2): ...new function.
5194         (pass_split_after_reload): Replace declaration with that of...
5195         (make_pass_split_after_reload): ...new function.
5196         (pass_branch_target_load_optimize1): Replace declaration with that
5197         of...
5198         (make_pass_branch_target_load_optimize1): ...new function.
5199         (pass_thread_prologue_and_epilogue): Replace declaration with that
5200         of...
5201         (make_pass_thread_prologue_and_epilogue): ...new function.
5202         (pass_stack_adjustments): Replace declaration with that of...
5203         (make_pass_stack_adjustments): ...new function.
5204         (pass_peephole2): Replace declaration with that of...
5205         (make_pass_peephole2): ...new function.
5206         (pass_if_after_reload): Replace declaration with that of...
5207         (make_pass_if_after_reload): ...new function.
5208         (pass_regrename): Replace declaration with that of...
5209         (make_pass_regrename): ...new function.
5210         (pass_cprop_hardreg): Replace declaration with that of...
5211         (make_pass_cprop_hardreg): ...new function.
5212         (pass_reorder_blocks): Replace declaration with that of...
5213         (make_pass_reorder_blocks): ...new function.
5214         (pass_branch_target_load_optimize2): Replace declaration with that
5215         of...
5216         (make_pass_branch_target_load_optimize2): ...new function.
5217         (pass_leaf_regs): Replace declaration with that of...
5218         (make_pass_leaf_regs): ...new function.
5219         (pass_split_before_sched2): Replace declaration with that of...
5220         (make_pass_split_before_sched2): ...new function.
5221         (pass_compare_elim_after_reload): Replace declaration with that of...
5222         (make_pass_compare_elim_after_reload): ...new function.
5223         (pass_sched2): Replace declaration with that of...
5224         (make_pass_sched2): ...new function.
5225         (pass_stack_regs): Replace declaration with that of...
5226         (make_pass_stack_regs): ...new function.
5227         (pass_stack_regs_run): Replace declaration with that of...
5228         (make_pass_stack_regs_run): ...new function.
5229         (pass_df_finish): Replace declaration with that of...
5230         (make_pass_df_finish): ...new function.
5231         (pass_compute_alignments): Replace declaration with that of...
5232         (make_pass_compute_alignments): ...new function.
5233         (pass_duplicate_computed_gotos): Replace declaration with that of...
5234         (make_pass_duplicate_computed_gotos): ...new function.
5235         (pass_variable_tracking): Replace declaration with that of...
5236         (make_pass_variable_tracking): ...new function.
5237         (pass_free_cfg): Replace declaration with that of...
5238         (make_pass_free_cfg): ...new function.
5239         (pass_machine_reorg): Replace declaration with that of...
5240         (make_pass_machine_reorg): ...new function.
5241         (pass_cleanup_barriers): Replace declaration with that of...
5242         (make_pass_cleanup_barriers): ...new function.
5243         (pass_delay_slots): Replace declaration with that of...
5244         (make_pass_delay_slots): ...new function.
5245         (pass_split_for_shorten_branches): Replace declaration with that of...
5246         (make_pass_split_for_shorten_branches): ...new function.
5247         (pass_split_before_regstack): Replace declaration with that of...
5248         (make_pass_split_before_regstack): ...new function.
5249         (pass_convert_to_eh_region_ranges): Replace declaration with that
5250         of...
5251         (make_pass_convert_to_eh_region_ranges): ...new function.
5252         (pass_shorten_branches): Replace declaration with that of...
5253         (make_pass_shorten_branches): ...new function.
5254         (pass_set_nothrow_function_flags): Replace declaration with that of...
5255         (make_pass_set_nothrow_function_flags): ...new function.
5256         (pass_dwarf2_frame): Replace declaration with that of...
5257         (make_pass_dwarf2_frame): ...new function.
5258         (pass_final): Replace declaration with that of...
5259         (make_pass_final): ...new function.
5260         (pass_rtl_seqabstr): Replace declaration with that of...
5261         (make_pass_rtl_seqabstr): ...new function.
5262         (pass_release_ssa_names): Replace declaration with that of...
5263         (make_pass_release_ssa_names): ...new function.
5264         (pass_early_inline): Replace declaration with that of...
5265         (make_pass_early_inline): ...new function.
5266         (pass_inline_parameters): Replace declaration with that of...
5267         (make_pass_inline_parameters): ...new function.
5268         (pass_update_address_taken): Replace declaration with that of...
5269         (make_pass_update_address_taken): ...new function.
5270         (pass_convert_switch): Replace declaration with that of...
5271         (make_pass_convert_switch): ...new function.
5272         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5273         to a subclass of simple_ipa_opt_pass along with...
5274         (pass_data_ipa_tree_profile): ...new pass_data instance and...
5275         (make_pass_ipa_tree_profile): ...new function.
5276         * tree-sra.c (pass_sra_early): Convert from a global struct to a
5277         subclass of gimple_opt_pass along with...
5278         (pass_data_sra_early): ...new pass_data instance and...
5279         (make_pass_sra_early): ...new function.
5280         (pass_sra): Convert from a global struct to a subclass of
5281         gimple_opt_pass along with...
5282         (pass_data_sra): ...new pass_data instance and...
5283         (make_pass_sra): ...new function.
5284         (pass_early_ipa_sra): Convert from a global struct to a subclass of
5285         gimple_opt_pass along with...
5286         (pass_data_early_ipa_sra): ...new pass_data instance and...
5287         (make_pass_early_ipa_sra): ...new function.
5288         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5289         subclass of gimple_opt_pass along with...
5290         (pass_data_ccp): ...new pass_data instance and...
5291         (make_pass_ccp): ...new function.
5292         (pass_fold_builtins): Convert from a global struct to a subclass of
5293         gimple_opt_pass along with...
5294         (pass_data_fold_builtins): ...new pass_data instance and...
5295         (make_pass_fold_builtins): ...new function.
5296         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5297         subclass of gimple_opt_pass along with...
5298         (pass_data_copy_prop): ...new pass_data instance and...
5299         (make_pass_copy_prop): ...new function.
5300         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5301         global struct to a subclass of gimple_opt_pass along with...
5302         (pass_data_rename_ssa_copies): ...new pass_data instance and...
5303         (make_pass_rename_ssa_copies): ...new function.
5304         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5305         subclass of gimple_opt_pass along with...
5306         (pass_data_dce): ...new pass_data instance and...
5307         (make_pass_dce): ...new function.
5308         (pass_dce_loop): Convert from a global struct to a subclass of
5309         gimple_opt_pass along with...
5310         (pass_data_dce_loop): ...new pass_data instance and...
5311         (make_pass_dce_loop): ...new function.
5312         (pass_cd_dce): Convert from a global struct to a subclass of
5313         gimple_opt_pass along with...
5314         (pass_data_cd_dce): ...new pass_data instance and...
5315         (make_pass_cd_dce): ...new function.
5316         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5317         subclass of gimple_opt_pass along with...
5318         (pass_data_dominator): ...new pass_data instance and...
5319         (make_pass_dominator): ...new function.
5320         (pass_phi_only_cprop): Convert from a global struct to a subclass of
5321         gimple_opt_pass along with...
5322         (pass_data_phi_only_cprop): ...new pass_data instance and...
5323         (make_pass_phi_only_cprop): ...new function.
5324         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5325         subclass of gimple_opt_pass along with...
5326         (pass_data_dse): ...new pass_data instance and...
5327         (make_pass_dse): ...new function.
5328         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5329         a subclass of gimple_opt_pass along with...
5330         (pass_data_forwprop): ...new pass_data instance and...
5331         (make_pass_forwprop): ...new function.
5332         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5333         struct to a subclass of gimple_opt_pass along with...
5334         (pass_data_tree_ifcombine): ...new pass_data instance and...
5335         (make_pass_tree_ifcombine): ...new function.
5336         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5337         subclass of gimple_opt_pass along with...
5338         (pass_data_ch): ...new pass_data instance and...
5339         (make_pass_ch): ...new function.
5340         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5341         subclass of gimple_opt_pass along with...
5342         (pass_data_tree_loop): ...new pass_data instance and...
5343         (make_pass_tree_loop): ...new function.
5344         (pass_tree_loop_init): Convert from a global struct to a subclass of
5345         gimple_opt_pass along with...
5346         (pass_data_tree_loop_init): ...new pass_data instance and...
5347         (make_pass_tree_loop_init): ...new function.
5348         (pass_lim): Convert from a global struct to a subclass of
5349         gimple_opt_pass along with...
5350         (pass_data_lim): ...new pass_data instance and...
5351         (make_pass_lim): ...new function.
5352         (pass_tree_unswitch): Convert from a global struct to a subclass of
5353         gimple_opt_pass along with...
5354         (pass_data_tree_unswitch): ...new pass_data instance and...
5355         (make_pass_tree_unswitch): ...new function.
5356         (pass_predcom): Convert from a global struct to a subclass of
5357         gimple_opt_pass along with...
5358         (pass_data_predcom): ...new pass_data instance and...
5359         (make_pass_predcom): ...new function.
5360         (pass_vectorize): Convert from a global struct to a subclass of
5361         gimple_opt_pass along with...
5362         (pass_data_vectorize): ...new pass_data instance and...
5363         (make_pass_vectorize): ...new function.
5364         (pass_graphite): Convert from a global struct to a subclass of
5365         gimple_opt_pass along with...
5366         (pass_data_graphite): ...new pass_data instance and...
5367         (make_pass_graphite): ...new function.
5368         (pass_graphite_transforms): Convert from a global struct to a subclass
5369         of gimple_opt_pass along with...
5370         (pass_data_graphite_transforms): ...new pass_data instance and...
5371         (make_pass_graphite_transforms): ...new function.
5372         (pass_check_data_deps): Convert from a global struct to a subclass of
5373         gimple_opt_pass along with...
5374         (pass_data_check_data_deps): ...new pass_data instance and...
5375         (make_pass_check_data_deps): ...new function.
5376         (pass_iv_canon): Convert from a global struct to a subclass of
5377         gimple_opt_pass along with...
5378         (pass_data_iv_canon): ...new pass_data instance and...
5379         (make_pass_iv_canon): ...new function.
5380         (pass_scev_cprop): Convert from a global struct to a subclass of
5381         gimple_opt_pass along with...
5382         (pass_data_scev_cprop): ...new pass_data instance and...
5383         (make_pass_scev_cprop): ...new function.
5384         (pass_record_bounds): Convert from a global struct to a subclass of
5385         gimple_opt_pass along with...
5386         (pass_data_record_bounds): ...new pass_data instance and...
5387         (make_pass_record_bounds): ...new function.
5388         (pass_complete_unroll): Convert from a global struct to a subclass of
5389         gimple_opt_pass along with...
5390         (pass_data_complete_unroll): ...new pass_data instance and...
5391         (make_pass_complete_unroll): ...new function.
5392         (pass_complete_unrolli): Convert from a global struct to a subclass of
5393         gimple_opt_pass along with...
5394         (pass_data_complete_unrolli): ...new pass_data instance and...
5395         (make_pass_complete_unrolli): ...new function.
5396         (pass_parallelize_loops): Convert from a global struct to a subclass
5397         of gimple_opt_pass along with...
5398         (pass_data_parallelize_loops): ...new pass_data instance and...
5399         (make_pass_parallelize_loops): ...new function.
5400         (pass_loop_prefetch): Convert from a global struct to a subclass of
5401         gimple_opt_pass along with...
5402         (pass_data_loop_prefetch): ...new pass_data instance and...
5403         (make_pass_loop_prefetch): ...new function.
5404         (pass_iv_optimize): Convert from a global struct to a subclass of
5405         gimple_opt_pass along with...
5406         (pass_data_iv_optimize): ...new pass_data instance and...
5407         (make_pass_iv_optimize): ...new function.
5408         (pass_tree_loop_done): Convert from a global struct to a subclass of
5409         gimple_opt_pass along with...
5410         (pass_data_tree_loop_done): ...new pass_data instance and...
5411         (make_pass_tree_loop_done): ...new function.
5412         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5413         struct to a subclass of gimple_opt_pass along with...
5414         (pass_data_cse_reciprocals): ...new pass_data instance and...
5415         (make_pass_cse_reciprocals): ...new function.
5416         (pass_cse_sincos): Convert from a global struct to a subclass of
5417         gimple_opt_pass along with...
5418         (pass_data_cse_sincos): ...new pass_data instance and...
5419         (make_pass_cse_sincos): ...new function.
5420         (pass_optimize_bswap): Convert from a global struct to a subclass of
5421         gimple_opt_pass along with...
5422         (pass_data_optimize_bswap): ...new pass_data instance and...
5423         (make_pass_optimize_bswap): ...new function.
5424         (pass_optimize_widening_mul): Convert from a global struct to a
5425         subclass of gimple_opt_pass along with...
5426         (pass_data_optimize_widening_mul): ...new pass_data instance and...
5427         (make_pass_optimize_widening_mul): ...new function.
5428         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5429         subclass of gimple_opt_pass along with...
5430         (pass_data_phiopt): ...new pass_data instance and...
5431         (make_pass_phiopt): ...new function.
5432         (pass_cselim): Convert from a global struct to a subclass of
5433         gimple_opt_pass along with...
5434         (pass_data_cselim): ...new pass_data instance and...
5435         (make_pass_cselim): ...new function.
5436         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5437         subclass of gimple_opt_pass along with...
5438         (pass_data_phiprop): ...new pass_data instance and...
5439         (make_pass_phiprop): ...new function.
5440         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5441         subclass of gimple_opt_pass along with...
5442         (pass_data_pre): ...new pass_data instance and...
5443         (make_pass_pre): ...new function.
5444         (pass_fre): Convert from a global struct to a subclass of
5445         gimple_opt_pass along with...
5446         (pass_data_fre): ...new pass_data instance and...
5447         (make_pass_fre): ...new function.
5448         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5449         subclass of gimple_opt_pass along with...
5450         (pass_data_reassoc): ...new pass_data instance and...
5451         (make_pass_reassoc): ...new function.
5452         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5453         subclass of gimple_opt_pass along with...
5454         (pass_data_sink_code): ...new pass_data instance and...
5455         (make_pass_sink_code): ...new function.
5456         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5457         subclass of gimple_opt_pass along with...
5458         (pass_data_strlen): ...new pass_data instance and...
5459         (make_pass_strlen): ...new function.
5460         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5461         struct to a subclass of gimple_opt_pass along with...
5462         (pass_data_build_alias): ...new pass_data instance and...
5463         (make_pass_build_alias): ...new function.
5464         (pass_build_ealias): Convert from a global struct to a subclass of
5465         gimple_opt_pass along with...
5466         (pass_data_build_ealias): ...new pass_data instance and...
5467         (make_pass_build_ealias): ...new function.
5468         (pass_ipa_pta): Convert from a global struct to a subclass of
5469         simple_ipa_opt_pass along with...
5470         (pass_data_ipa_pta): ...new pass_data instance and...
5471         (make_pass_ipa_pta): ...new function.
5472         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5473         subclass of gimple_opt_pass along with...
5474         (pass_data_uncprop): ...new pass_data instance and...
5475         (make_pass_uncprop): ...new function.
5476         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5477         global struct to a subclass of gimple_opt_pass along with...
5478         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5479         (make_pass_late_warn_uninitialized): ...new function.
5480         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5481         to a subclass of gimple_opt_pass along with...
5482         (pass_data_init_datastructures): ...new pass_data instance and...
5483         (make_pass_init_datastructures): ...new function.
5484         (pass_early_warn_uninitialized): Convert from a global struct to a
5485         subclass of gimple_opt_pass along with...
5486         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5487         (make_pass_early_warn_uninitialized): ...new function.
5488         (pass_update_address_taken): Convert from a global struct to a
5489         subclass of gimple_opt_pass along with...
5490         (pass_data_update_address_taken): ...new pass_data instance and...
5491         (make_pass_update_address_taken): ...new function.
5492         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5493         struct to a subclass of gimple_opt_pass along with...
5494         (pass_data_release_ssa_names): ...new pass_data instance and...
5495         (make_pass_release_ssa_names): ...new function.
5496         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5497         subclass of gimple_opt_pass along with...
5498         (pass_data_stdarg): ...new pass_data instance and...
5499         (make_pass_stdarg): ...new function.
5500         * tree-switch-conversion.c (pass_convert_switch): Convert from a
5501         global struct to a subclass of gimple_opt_pass along with...
5502         (pass_data_convert_switch): ...new pass_data instance and...
5503         (make_pass_convert_switch): ...new function.
5504         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5505         to a subclass of gimple_opt_pass along with...
5506         (pass_data_tail_recursion): ...new pass_data instance and...
5507         (make_pass_tail_recursion): ...new function.
5508         (pass_tail_calls): Convert from a global struct to a subclass of
5509         gimple_opt_pass along with...
5510         (pass_data_tail_calls): ...new pass_data instance and...
5511         (make_pass_tail_calls): ...new function.
5512         * tree-vect-generic.c (pass_lower_vector): Convert from a global
5513         struct to a subclass of gimple_opt_pass along with...
5514         (pass_data_lower_vector): ...new pass_data instance and...
5515         (make_pass_lower_vector): ...new function.
5516         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5517         gimple_opt_pass along with...
5518         (pass_data_lower_vector_ssa): ...new pass_data instance and...
5519         (make_pass_lower_vector_ssa): ...new function.
5520         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5521         to a subclass of gimple_opt_pass along with...
5522         (pass_data_slp_vectorize): ...new pass_data instance and...
5523         (make_pass_slp_vectorize): ...new function.
5524         (pass_ipa_increase_alignment): Convert from a global struct to a
5525         subclass of simple_ipa_opt_pass along with...
5526         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5527         (make_pass_ipa_increase_alignment): ...new function.
5528         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5529         gimple_opt_pass along with...
5530         (pass_data_vrp): ...new pass_data instance and...
5531         (make_pass_vrp): ...new function.
5532         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5533         subclass of simple_ipa_opt_pass along with...
5534         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5535         (make_pass_ipa_free_lang_data): ...new function.
5536         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5537         gimple_opt_pass along with...
5538         (pass_data_tsan): ...new pass_data instance and...
5539         (make_pass_tsan): ...new function.
5540         (pass_tsan_O0): Convert from a global struct to a subclass of
5541         gimple_opt_pass along with...
5542         (pass_data_tsan_O0): ...new pass_data instance and...
5543         (make_pass_tsan_O0): ...new function.
5544         * var-tracking.c (pass_variable_tracking): Convert from a global
5545         struct to a subclass of rtl_opt_pass along with...
5546         (pass_data_variable_tracking): ...new pass_data instance and...
5547         (make_pass_variable_tracking): ...new function.
5548         * web.c (pass_web): Convert from a global struct to a subclass of
5549         rtl_opt_pass along with...
5550         (pass_data_web): ...new pass_data instance and...
5551         (make_pass_web): ...new function.
5552         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5553         declaration with that of...
5554         (make_pass_mode_switch_use): ...new function.
5555         (pass_resolve_sw_modes): Replace declaration with that of...
5556         (make_pass_resolve_sw_modes): ...new function.
5557         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5558         from a global struct to a subclass of rtl_opt_pass along with...
5559         (pass_data_mode_switch_use): ...new pass_data instance and...
5560         (make_pass_mode_switch_use): ...new function.
5561         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5562         from a global struct to a subclass of rtl_opt_pass along with...
5563         (pass_data_resolve_sw_modes): ...new pass_data instance and...
5564         (make_pass_resolve_sw_modes): ...new function.
5565         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5566         struct to a subclass of rtl_opt_pass along with...
5567         (pass_data_insert_vzeroupper): ...new pass_data instance and...
5568         (make_pass_insert_vzeroupper): ...new function.
5569         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5570         global struct to a subclass of rtl_opt_pass along with...
5571         (pass_data_work_around_errata): ...new pass_data instance and...
5572         (make_pass_work_around_errata): ...new function.
5573         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5574         struct to a subclass of rtl_opt_pass along with...
5575         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
5576         (make_pass_mips_machine_reorg2): ...new function.
5578 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5580         * passes.c (pass_manager::operator new): New.
5582 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5584         Handwritten part of conversion of passes to C++ classes.
5586         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
5587         (toplev.o): Add dep on PASS_MANAGER_H.
5588         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
5589         of early local pases to reflect this moving from a global to a
5590         member of gcc::pass_manager.
5591         (cgraph_add_new_function): Likewise.
5592         * lto-cgraph.c (lto_output_node): Update for conversion of
5593         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
5594         * passes.c (opt_pass::clone): New.
5595         (opt_pass::gate): New.
5596         (opt_pass::execute): New.
5597         (opt_pass::opt_pass): New.
5598         (pass_manager::execute_early_local_passes): New.
5599         (pass_manager::execute_pass_mode_switching): new.
5600         (finish_optimization_passes): Convert to...
5601         (pass_manager::finish_optimization_passes): ...this.
5602         (finish_optimization_passes): Update for conversion of passes to
5603         C++ classes.
5604         (register_dump_files_1): Use has_gate since we cannot portably
5605         check a vtable entry against NULL.
5606         (dump_one_pass): Likewise.
5607         (ipa_write_summaries_2): Likewise.
5608         (ipa_write_optimization_summaries_1): Likewise.
5609         (ipa_read_summaries_1): Likewise.
5610         (ipa_read_optimization_summaries_1): Likewise.
5611         (execute_ipa_stmt_fixups): Likewise.
5612         (pass_manager::pass_manager): Rewrite pass-creation, invoking
5613         pass-creation functions rather than wiring up globals, and
5614         storing the results in fields of pass_manager generated using
5615         pass-instances.def.
5616         (pass_manager::dump_profile_report): Update for conversion of
5617         passes to C++ classes.
5618         (pass_manager::execute_ipa_summary_passes): Likewise.
5619         (execute_one_ipa_transform_pass): Likewise.
5620         (execute_one_pass): Use has_gate and has_execute since we cannot
5621         portably check a vtable entry against NULL.
5622         * pass_manager.h (pass_manager::finish_optimization_passes): New.
5623         (pass_manager): Use pass-instances.def to add fields for the
5624         various pass instances.
5625         * toplev.c (finalize): Update for move of
5626         finish_optimization_passes to a method of gcc::pass_manager.
5627         * toplev.h (finish_optimization_passes): Move to method of class
5628         pass_manager.
5629         * tree-pass.h (struct pass_data): New.
5630         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
5631         (opt_pass::gate): Convert to virtual function.
5632         (opt_pass::~opt_pass): New.
5633         (opt_pass::clone): New.
5634         (opt_pass::execute): Convert to virtual function.
5635         (opt_pass::opt_pass): New.
5636         (opt_pass::ctxt_): new.
5637         (gimple_opt_pass): Convert to subclass of opt_pass.
5638         (gimple_opt_pass::gimple_opt_pass): New.
5639         (rtl_opt_pass): Convert to subclass of opt_pass.
5640         (rtl_opt_pass::rtl_opt_pass): New.
5641         (ipa_opt_pass_d): Convert to subclass of opt_pass.
5642         (ipa_opt_pass_d::ipa_opt_pass_d): New.
5643         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
5644         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
5645         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
5646         invocation of pass_mode_switching to reflect this moving from a
5647         global to a member of gcc::pass_manager.
5648         (ix86_option_override): Rework how pass_insert_vzeroupper is
5649         added to the pass_manager to reflect autogenerated changes.
5650         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
5652 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
5654         PR rtl-optimization/57708
5655         * recog.c (peep2_find_free_register): Validate all regs in a
5656         multi-reg mode.
5658 2013-08-05  Jan Hubicka  <jh@suse.cz>
5660         PR lto/57602
5661         * cgraph.c (verify_cgraph_node): Accept local flags from other
5662         partitions.
5663         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
5664         (function_and_variable_visibility): Likewise.
5665         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
5667 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5669         * graph.c (init_graph_slim_pretty_print): Remove.
5670         (print_graph_cfg): Do not call it.  Use local pretty printer.
5671         (start_graph_dump): Likewise.
5673 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5675         * gimple-pretty-print.c (buffer): Remove.
5676         (initialized): Likewise.
5677         (maybe_init_pretty_print): Likewise.
5678         (print_gimple_stmt): Do not call it.  Use non-static local
5679         pretty_printer variable.
5680         (print_gimple_expr): Likewise.
5681         (print_gimple_seq): Likewise.
5682         (gimple_dump_bb): Likewise.
5684 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5686         * asan.c (asan_pp): Remove.
5687         (asan_pp_initialized): Likewise.
5688         (asan_pp_initialize): Likewise.
5689         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
5690         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
5691         (asan_add_global): Likewise.
5693 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5695         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
5696         * diagnostic.h (diagnostic_flush_buffer): Adjust.
5697         * pretty-print.c (pp_formatted_text_data): Likewise.
5698         (pp_indent): Rename from pp_base_indent.
5699         (pp_format): Rename from pp_base_format.
5700         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
5701         (pp_format_verbatim): Rename from pp_base_format_verbatim.
5702         (pp_flush): Rename from pp_base_flush.
5703         (pp_set_line_maximum_length): Rename from
5704         pp_base_set_line_maximum_length.
5705         (pp_clear_output_area): Rename from pp_base_clear_output_area.
5706         (pp_set_prefix): Rename from pp_base_set_prefix.
5707         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
5708         (pp_emit_prefix): Rename from pp_base_emit_prefix.
5709         (pp_append_text): Rename from pp_base_append_text.
5710         (pp_formatted_text): Rename from pp_base_formatted_text.
5711         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
5712         (pp_remaining_character_count_for_line): Rename from
5713         pp_base_remaining_character_count_for_line.
5714         (pp_newline): Rename from pp_base_newline.
5715         (pp_character): Rename from pp_base_character.
5716         (pp_string): Rename from pp_base_string.
5717         (pp_maybe_space): Rename from pp_base_maybe_space.
5718         * asan.c (asan_pp_string): Adjust.
5719         (asan_emit_stack_protection): Likewise.
5720         (asan_add_global): Likewise.
5721         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
5722         * tree-mudflap.c (mf_varname_tree): Likewise.
5723         * tree-pretty-print.c (pp_tree_identifier): Rename from
5724         pp_base_tree_identifier.
5725         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
5726         Declare as function.
5728 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5730         * pretty-print.h (pp_bar_bar): New.
5731         (pp_ampersand_ampersand): Likewise.
5732         (pp_less_equal): Likewise.
5733         (pp_greater_equal): Likewise.
5734         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
5735         printer functions instead of pp_string or operators and punctuators.
5736         (dump_gimple_call): Likewise.
5737         (dump_gimple_omp_for): Likewise.
5738         (dump_gimple_transaction): Likewise.
5739         (dump_gimple_phi): Likewise.
5740         (pp_gimple_stmt_1): Likewise.
5741         * sched-vis.c (print_insn): Likewise.
5742         * tree-mudflap.c (mf_varname_tree): Likewise.
5743         * tree-pretty-print.c (dump_block_node): Likewise.
5744         (dump_generic_node): Likewise.
5746 2013-08-02  Jan Hubicka  <jh@suse.cz>
5748         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
5749         boundaries.
5750         * lto-streamer-out.c (tree_is_indexable): Results decls and
5751         parm decls are not indexable.
5752         (DFS_write_tree_body): Do not follow args and results.
5753         (hash_tree): Likewise.
5754         (output_functions): Rearrange so struct function is needed
5755         only when real body is output; be able to also ouptut abstract
5756         functions; output DECL_ARGUMENTS and DECL_RESULT.
5757         (lto_output): When not in WPA, ale store abstract functions.
5758         (write_symbol): Do not care about RESULT_DECL.
5759         (output_symbol_p): Handle correctly sbtract decls.
5760         * lto-streamer-in.c (input_function): Rearrange so struct
5761         function can be NULL at entry; allow streaming of
5762         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
5763         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
5764         sanity check during LTO.
5765         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
5766         RESULT_DECl and DECL_ARGUMENTS.
5767         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
5768         Likewise.
5770 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5772         * pretty-print.h (pp_underscore): New.
5773         (pp_comma): Tidy.
5774         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
5775         printer functions instead of pp_character.
5776         (dump_binary_rhs): Likewise.
5777         (dump_ternary_rhs): Likewise.
5778         (dump_gimple_call_args): Likewise.
5779         (pp_points_to_solution): Likewise.
5780         (dump_gimple_call): Likewise.
5781         (dump_gimple_switch): Likewise.
5782         (dump_gimple_cond): Likewise.
5783         (dump_gimple_bind): Likewise.
5784         (dump_gimple_try): Likewise.
5785         (dump_gimple_omp_for): Likewise.
5786         (dump_gimple_omp_continue): Likewise.
5787         (dump_gimple_omp_single): Likewise.
5788         (dump_gimple_omp_sections): Likewise.
5789         (dump_gimple_omp_block): Likewise.
5790         (dump_gimple_omp_critical): Likewise.
5791         (dump_gimple_transaction): Likewise.
5792         (dump_gimple_asm): Likewise.
5793         (dump_gimple_phi): Likewise.
5794         (dump_gimple_omp_parallel): Likewise.
5795         (dump_gimple_omp_task): Likewise.
5796         (dump_gimple_omp_atomic_load): Likewise.
5797         (dump_gimple_omp_atomic_store): Likewise.
5798         (dump_gimple_mem_ops): Likewise.
5799         (pp_gimple_stmt_1): Likewise.
5800         (pp_cfg_jump): Likewise.
5801         (dump_implicit_edges): Likewise.
5802         (gimple_dump_bb_for_graph): Likewise.
5803         * graph.c (draw_cfg_node): Likewise.
5804         * langhooks.c (lhd_print_error_function): Likewise.
5805         * sched-vis.c (print_exp): Likewise.
5806         (print_value): Likewise.
5807         (print_pattern): Likewise.
5808         (print_insn): Likewise.
5809         (rtl_dump_bb_for_graph): Likewise.
5810         * tree-pretty-print.c (dump_function_declaration): Likewise.
5811         (dump_array_domain): Likewise.
5812         (dump_omp_clause): Likewise.
5813         (dump_location): Likewise.
5814         (dump_generic_node): Likewise.
5815         (print_struct_decl): Likewise.
5816         * diagnostic.c (diagnostic_show_locus): Use pp_space.
5818 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5820         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
5821         candidate table when replacing a candidate statement.
5822         (replace_rhs_if_not_dup): Likewise.
5823         (replace_one_candidate): Likewise.
5825 2013-08-02  Jan Hubicka  <jh@suse.cz>
5826             Martin Liska  <marxin.liska@gmail.com>
5828         * cgraphunit.c (add_new_function): Fix logic when adding from
5829         late IPA pass.
5830         (assemble_thunk): Rename to ...
5831         (expand_thunk); .. this one; export; get it working with
5832         general functions; make produced gimple valid.
5833         * cgraph.h (expand_thunk): Declare.
5835 2013-08-02  Jan Hubicka  <jh@suse.cz>
5837         * ipa-cp.c (gather_context_independent_values): Use
5838         ipa_get_param_move_cost.
5839         (get_replacement_map): Remove PARAM; move parameter folding
5840         into tree-inline.c
5841         (create_specialized_node): Update.
5842         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
5843         assert that we have gimple body; update move_cost.
5844         (count_formal_params): Assert that we have gimple body.
5845         (ipa_dump_param): New function.
5846         (ipa_alloc_node_params): Break out from ...
5847         (ipa_initialize_node_params): ... here.
5848         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
5849         (ipa_write_node_info): Stream move costs.
5850         (ipa_read_node_info): Read move costs.
5851         (ipa_update_after_lto_read): Do not recompute node params.
5852         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
5853         (ipa_get_param): Check we are not in WPA.
5854         (ipa_get_param_move_cost): New.
5855         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
5856         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
5857         parm numbers to be present.
5859 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5861         PR rtl-optimization/58048
5862         * lra-constraints.c (process_alt_operands): Don't check asm
5863         operand on register.
5865 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
5867         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
5868         the implied StoreLoad barrier for atomic operations if before.
5870 2013-08-02  Jan Hubicka  <jh@suse.cz>
5871             Martin Liska  <marxin.liska@gmail.com>
5873         * cgraph.c (cgraph_function_body_availability): Do not check
5874         cgraph flags.
5875         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
5876         symtab_node_availability): Declare.
5877         * ipa.c (can_replace_by_local_alias): New.
5878         (function_and_variable_visibility): Use it.
5879         * symtab.c (symtab_for_node_and_aliases,
5880         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
5882 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5884         PR rtl-optimization/57963
5885         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
5886         (lra_constraints): Use them.
5888 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
5890         * config/arm/types.md (define_attr "type"): Add "load_acq"
5891         and "store_rel".
5892         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
5893         changes.
5894         (cortex_a53_store1): Likewise.
5896 2013-08-01  Jan Hubicka  <jh@suse.cz>
5898         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
5899         partitions are not needed.
5901 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
5903         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
5904         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
5905         MAYBE_NON_Q_CLASS_P where appropriate.
5907 2013-08-01  Jan Hubicka  <jh@suse.cz>
5909         * cgraph.h (release_function_body): Declare.
5910         * tree.c (free_lang_data_in_decl): Free, parameters and return values
5911         of unused delcarations.
5913 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5915         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
5916         RTL form when subtracting a constant.
5918 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5920         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
5921         Generate canonical plus rtx with negated immediate instead of minus
5922         where appropriate.
5923         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
5925 2013-08-01  Jan Hubicka  <jh@suse.cz>
5927         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
5928         (cgraph_release_function_body): Likewise.
5929         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
5930         * cgraph.h (cgrpah_node): Rename abstract_and_needed
5931         to used_as_abstract_origin.
5932         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
5933         symbols used as abstract origins.
5934         * cgraphunit.c (analyze_functions): Update.
5935         * ipa.c (symtab_remove_unreachable_nodes): Recompute
5936         used_as_abstract_origin.
5937         * tree-inline.c (tree_function_versioning): Update
5938         used_as_abstract_origin; be ready for DECL_RESULT and
5939         DECL_ARGUMENTS to be NULL.
5941         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
5942         for abstract functions.
5943         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
5944         real symbols.
5946 2013-08-01  Jan Hubicka  <jh@suse.cz>
5948         * profile.c (compute_value_histograms): Fix thinko.
5950 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5952         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
5953         aarch-common-protos.h to extra_headers.
5954         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5955         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
5956         * config/aarch64/t-aarch64 (aarch-common.o): Define.
5958 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5960         * config/aarch64/aarch64.md (define_attr "type"): Delete.
5961         Include "../arm/types.md".  Define "type" attribute for all patterns.
5962         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
5963         attribute changes.
5965 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
5967         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
5968         to support power8 load fusion.
5969         (fusion_gpr_mem_load): Likewise.
5971         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
5973         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
5974         declarations for power8 load fusion.
5975         (emit_fusion_gpr_load): Likewise.
5977         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
5978         tuning for power8, turn on fusion mode by default.  Turn on sign
5979         extending fusion mode if normal fusion mode is on, and we are at
5980         -O2 or -O3.
5981         (fusion_gpr_load_p): New function, return true if we can fuse an
5982         addis instruction with a dependent load to a GPR.
5983         (emit_fusion_gpr_load): Emit the instructions for power8 load
5984         fusion to GPRs.
5986         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
5987         (VSX load fusion peepholes): New peepholes to fuse together an
5988         addi instruction with a VSX load instruction.
5990         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
5991         peepholes to fuse an addis instruction with a load to a GPR base
5992         register.  If we are supporting sign extending fusions, convert
5993         sign extending loads to zero extending loads and add an explicit
5994         sign extension.
5996 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
5998         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
5999         aarch-common-protos.h to extra_headers.
6000         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6001         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
6002         (arm_early_store_addr_dep): Likewise.
6003         (arm_no_early_alu_shift_dep): Likewise.
6004         (arm_no_early_alu_shift_value_dep): Likewise.
6005         (arm_no_early_mul_dep): Likewise.
6006         (arm_no_early_store_addr_dep): Likewise.
6007         (arm_mac_accumulator_is_mul_result): Likewise.
6008         (arm_mac_accumulator_is_result): Likewise.
6009         * config/arm/aarch-common.c: ... here.  New file.
6010         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6011         here to ...
6012         (arm_early_store_addr_dep): Likewise.
6013         (arm_no_early_alu_shift_dep): Likewise.
6014         (arm_no_early_alu_shift_value_dep): Likewise.
6015         (arm_no_early_mul_dep): Likewise.
6016         (arm_no_early_store_addr_dep): Likewise.
6017         (arm_mac_accumulator_is_mul_result): Likewise.
6018         (arm_mac_accumulator_is_result): Likewise.
6019         * config/arm/aarch-common-protos.h: ... here.  New file.
6020         * config/arm/t-arm (aarch-common.o): Define.
6022 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6024         * config/arm/arm.md: Include new file "types.md".
6025         (define_attr "type"): Move from here to ...
6026         (define_attr "mul32"): Likewise.
6027         (define_attr "mul64"): Likewise.
6028         * config/arm/types.md: ... here.  New file.
6030 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6032         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6033         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6035 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6037         * gen-pass-instances.awk: Fix offset of substr().
6039 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6041         * Makefile.in (pass-instances.def): New.
6042         (passes.o): Replace dependency on passes.def with one on
6043         pass-instances.def
6045         * gen-pass-instances.awk: New.
6047         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6048         rather than passes.def, updating local definition of NEXT_PASS
6049         macro to add an extra NUM parameter (currently unused).
6051 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6053         * Makefile.in (PASS_MANAGER_H): New.
6054         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6055         (passes.o): Likewise.
6056         (statistics.o): Likewise.
6057         (cgraphunit.o): Likewise.
6058         (context.o): Depend on PASS_MANAGER_H.
6060         * pass_manager.h: New.
6062         * cgraphunit.c (cgraph_add_new_function): Update for moves
6063         of globals to fields of pass_manager.
6064         (analyze_function): Likewise.
6065         (expand_function): Likewise.
6066         (ipa_passes): Likewise.
6067         (compile): Likewise.
6069         * context.c (context::context): New.
6070         * context.h  (context::context): New.
6071         (context::get_passes): New.
6072         (context::passes_): New.
6074         * lto-cgraph.c (input_node): Update for moves of globals to
6075         fields of pass_manager.
6077         * passes.c (all_passes): Remove, in favor of a field of the
6078         same name within the new class pass_manager.
6079         (all_small_ipa_passes): Likewise.
6080         (all_lowering_passes): Likewise.
6081         (all_regular_ipa_passes): Likewise.
6082         (all_late_ipa_passes): Likewise.
6083         (all_lto_gen_passes): Likewise.
6084         (passes_by_id): Likewise.
6085         (passes_by_id_size): Likewise.
6086         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6087         the new class pass_manager.
6088         (set_pass_for_id): Convert to...
6089         (pass_manager::set_pass_for_id): ...method.
6090         (get_pass_for_id): Convert to...
6091         (pass_manager::get_pass_for_id): ...method.
6092         (register_one_dump_file): Move body of implementation into...
6093         (pass_manager::register_one_dump_file): ...here.
6094         (register_dump_files_1): Convert to...
6095         (pass_manager::register_dump_files_1): ...method.
6096         (register_dump_files): Convert to...
6097         (pass_manager::register_dump_files): ...method.
6098         (create_pass_tab): Update for moves of globals to fields of
6099         pass_manager.
6100         (dump_passes): Move body of implementation into...
6101         (pass_manager::dump_passes): ...here.
6102         (register_pass): Move body of implementation into...
6103         (pass_manager::register_pass): ...here.
6104         (init_optimization_passes): Convert into...
6105         (pass_manager::pass_manager): ...constructor for new
6106         pass_manager class, and initialize the pass_lists array.
6107         (check_profile_consistency): Update for moves of globals to
6108         fields of pass_manager.
6109         (dump_profile_report): Move body of implementation into...
6110         (pass_manager::dump_profile_report): ...here.
6111         (ipa_write_summaries_1): Update for moves of pass lists from
6112         being globals to fields of pass_manager.
6113         (ipa_write_optimization_summaries): Likewise.
6114         (ipa_read_summaries):  Likewise.
6115         (ipa_read_optimization_summaries): Likewise.
6116         (execute_all_ipa_stmt_fixups): Likewise.
6118         * statistics.c (statistics_fini): Update for moves of globals to
6119         fields of pass_manager.
6121         * toplev.c (general_init): Replace call to
6122         init_optimization_passes with construction of the pass_manager
6123         instance.
6125         * tree-pass.h (all_passes): Remove, in favor of a field of the
6126         same name within the new class pass_manager.
6127         (all_small_ipa_passes): Likewise.
6128         (all_lowering_passes): Likewise.
6129         (all_regular_ipa_passes): Likewise.
6130         (all_lto_gen_passes): Likewise.
6131         (all_late_ipa_passes): Likewise.
6132         (passes_by_id): Likewise.
6133         (passes_by_id_size): Likewise.
6134         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6135         the new class pass_manager.
6136         (get_pass_for_id): Remove.
6138 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6140         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6141         configs.
6143 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6145         * arm.md (mulhi3): New expand pattern.
6147 2013-07-30  Jan Hubicka  <jh@suse.cz>
6148             Martin Liska  <marxin.liska@gmail.com>
6150         * profile.c (compute_value_histograms): Do not ICE when
6151         there is mismatch only on some counters.
6153 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6155         PR rtl-optimization/57637
6156         * function.c (move_insn_for_shrink_wrap): Also check the
6157         GEN set of the LIVE problem for the liveness analysis
6158         if it exists, otherwise give up.
6160 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6162         PR tree-optimization/57993
6163         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6164         replaced statement in the candidate table.
6165         (phi_add_costs): Return infinite cost when the hidden basis does
6166         not dominate all phis on which the candidate is dependent.
6167         (replace_one_candidate): Record replaced statement in the
6168         candidate table.
6170 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6172         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6173         (ashlv2si3): New expander.
6174         (*ashlv2si3_i): New define_insn_and_split.
6175         * predicates.md (float_operation): Allow patterns with three
6176         basic sub-patterns.
6178         PR rtl-optimization/58021
6179         * mode-switching.c (create_pre_exit): Always split off preceding
6180         insns if we are not at the basic block head.
6182 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6184         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6185         (UCLIBC_DYNAMIC_LINKER): New macro.
6186         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6187         `-mnan=2008'.
6188         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6189         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6190         `-mnan=2008'.
6191         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6192         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6193         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6194         for SF and DF modes.  Use ieee_quad_format for TF mode.
6195         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6196         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6197         (mips_option_override): Handle `-mnan=legacy'.
6198         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6199         `-mabs=2008' and `-mnan=2008'.
6200         (OPTION_DEFAULT_SPECS): Add "nan" default.
6201         (ASM_SPEC): Handle `-mnan='.
6202         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6203         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6204         comment accordingly.
6205         (neg<mode>2): Likewise.
6206         * config/mips/mips.opt (mabs, mnan): New options.
6207         * doc/install.texi (Configuration): Document `--with-nan=' option.
6208         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6209         `-mnan=' options.
6210         (MIPS Options): Document them.
6211         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6212         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6213         * configure: Regenerate.
6214         * config.in: Regenerate.
6216 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6218         * config/i386/i386.md (float post-reload splitters): Do not check
6219         for subregs of SSE registers.
6221 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6222             H.J. Lu  <hongjiu.lu@intel.com>
6224         PR target/57954
6225         PR target/57988
6226         * config/i386/i386.md (post-reload splitter
6227         to avoid partial SSE reg dependency stalls): New pattern.
6229 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6231         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6233 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6235         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6236         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6238 2013-07-26  Cary Coutant  <ccoutant@google.com>
6240         * dwarf2out.c (die_checksum_ordered): Don't include template
6241         instantiations in signature.
6242         (is_template_parameter): New function.
6243         (is_template_instantiation): New function.
6244         (generate_skeleton_bottom_up): Don't include template instantiations
6245         in type unit DIE.
6246         (generate_skeleton): Likewise.
6247         (break_out_comdat_types): Move recursive call to break out nested
6248         types earlier.
6249         (prune_unused_types_mark_generic_parms_dies): Call
6250         is_template_parameter.
6252 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
6254         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6255         uses vector registers.
6256         * config/aarch64/iterators.md: Add attributes rtn and vas.
6258 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6259             Richard Earnshaw  <richard.earnshaw@arm.com>
6261         * combine.c (simplify_comparison): Re-canonicalize operands
6262         where appropriate.
6263         * config/arm/arm.md (movcond_addsi): New splitter.
6265 2013-07-25  Sterling Augustine  <saugustine@google.com>
6267         * dwarf2out.c (size_of_pubnames): Move code to...
6268         (include_pubname_in_output): ...here.  New.
6269         (want_pubnames): Rearrange.
6270         (output_pubnames): Call include_pubname_in_output.  Move assertion.
6272 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6274         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6276 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6278         PR target/38836
6279         * doc/extend.texi: Remove obsolete builtins. Fix
6280         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6282 2013-07-25  Jan Hubicka  <jh@suse.cz>
6284         * cgraph.c (release_function_body): Break out from ...
6285         (cgraph_release_function_body): ... this one; also release DECL_RESULT
6286         and DECL_ARGUMENTS.
6287         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6288         old_tree in the map.
6289         (create_specialized_node): Update.
6290         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6291         into index.
6292         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6293         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6294         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6295         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6296         DECL_RESULT.
6298 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6300         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6301         addsi3_carryin_alt2_<optab>): Correct output template.
6303 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6305         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6306         Adjust for arm_restrict_it.
6307         Remove trailing whitespace.
6309 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
6311         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6312         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6314         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6316 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
6318         PR rtl-optimization/57960
6319         * lra-constraints.c (process_alt_operands): Use the right mode
6320         when checking strict_low.
6322 2013-07-25  Jan Hubicka  <jh@suse.cz>
6324         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6325         * cgraph.c (cgraph_remove_node): Do not release function body
6326         when in cgraph streaming.
6327         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6328         in other partitions are not considered reachable; fix handling of
6329         clones.
6331 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6333         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6335 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6337         PR target/19599
6338         PR target/57731
6339         PR target/57837
6340         * config/arm/arm.md ("*sibcall_insn): Replace use of
6341         Ss with US. Adjust output for v5 and v4t.
6342         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6344         * config/arm/constraints.md ("Ss"): Rename to US.
6346 2013-07-25  Terry Guo  <terry.guo@arm.com>
6348         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6349         shift_add/shift_sub0/shift_sub1 RTXs.
6351 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
6352             Anton Blanchard  <anton@au1.ibm.com>
6354         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6355         (altivec_vpks<VI_char>ss): Likewise.
6356         (altivec_vpks<VI_char>us): Likewise.
6357         (altivec_vpku<VI_char>us): Likewise.
6358         (altivec_vpku<VI_char>um): Likewise.
6360 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
6362         Introduce context class.
6364         * Makefile.in (CONTEXT_H): New.
6365         (OBJS): Add context.o.
6366         (toplev.o): Add CONTEXT_H to dependencies.
6367         (context.o): New.
6369         * toplev.c (general_init): Create the singleton gcc::context instance.
6371         * context.c: New.
6373         * context.h: New.
6375 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
6377         PR rtl-optimization/57968
6378         * mode-switching.c (create_pre_exit): Allow instructions that
6379         don't set a return register to need a non-exit mode.
6381 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6382             Anton Blanchard  <anton@au1.ibm.com>
6384         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6385         operands to vperm for little endian.
6386         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6387         of lvsl to create the control mask for a vperm for little endian.
6389 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6390             Anton Blanchard  <anton@au1.ibm.com>
6392         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6393         two operands for little-endian.
6395 2013-07-23  Steve Ellcey  <sellcey@mips.com>
6397         * config/mips/mips.c (mips_case_values_threshold): New.
6398         (TARGET_CASE_VALUES_THRESHOLD): Define.
6400 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6401             Anton Blanchard  <anton@au1.ibm.com>
6403         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6404         selection of field for vector splat in little endian mode.
6406 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6408         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6409         expanders to rs6000.md.
6410         (ior<mode>3): Likewise.
6411         (and<mode>3): Likewise.
6412         (one_cmpl<mode>2): Likewise.
6413         (nor<mode>3): Likewise.
6414         (andc<mode>3): Likewise.
6415         (eqv<mode>3): Likewise.
6416         (nand<mode>3): Likewise.
6417         (orc<mode>3): Likewise.
6419         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6420         declaration.
6422         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6423         to split multi-word logical operations.
6424         (rs6000_split_logical_di): Likewise.
6425         (rs6000_split_logical): Likewise.
6427         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6428         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6429         and allow TImode operations in 32-bit.
6430         (vsx_and<mode>3_64bit): Likewise.
6431         (vsx_ior<mode>3_32bit): Likewise.
6432         (vsx_ior<mode>3_64bit): Likewise.
6433         (vsx_xor<mode>3_32bit): Likewise.
6434         (vsx_xor<mode>3_64bit): Likewise.
6435         (vsx_one_cmpl<mode>2_32bit): Likewise.
6436         (vsx_one_cmpl<mode>2_64bit): Likewise.
6437         (vsx_nor<mode>3_32bit): Likewise.
6438         (vsx_nor<mode>3_64bit): Likewise.
6439         (vsx_andc<mode>3_32bit): Likewise.
6440         (vsx_andc<mode>3_64bit): Likewise.
6441         (vsx_eqv<mode>3_32bit): Likewise.
6442         (vsx_eqv<mode>3_64bit): Likewise.
6443         (vsx_nand<mode>3_32bit): Likewise.
6444         (vsx_nand<mode>3_64bit): Likewise.
6445         (vsx_orc<mode>3_32bit): Likewise.
6446         (vsx_orc<mode>3_64bit): Likewise.
6448         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6449         logical types in GPRs.
6451         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6452         logical insns to rs6000.md, and allow TImode operations in
6453         32-bit.
6454         (altivec_ior<mode>3): Likewise.
6455         (altivec_xor<mode>3): Likewise.
6456         (altivec_one_cmpl<mode>2): Likewise.
6457         (altivec_nor<mode>3): Likewise.
6458         (altivec_andc<mode>3): Likewise.
6460         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6461         attributes for moving the 128-bit logical operations into
6462         rs6000.md.
6463         (BOOL_REGS_OUTPUT): Likewise.
6464         (BOOL_REGS_OP1): Likewise.
6465         (BOOL_REGS_OP2): Likewise.
6466         (BOOL_REGS_UNARY): Likewise.
6467         (BOOL_REGS_AND_CR0): Likewise.
6468         (one_cmpl<mode>2): Add support for DI logical operations on
6469         32-bit, splitting the operations to 32-bit.
6470         (anddi3): Likewise.
6471         (iordi3): Likewise.
6472         (xordi3): Likewise.
6473         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6474         changes to combine the 32/64-bit code, allow logical operations on
6475         TI mode in 32-bit, and to use similar match_operator patterns like
6476         scalar mode uses.  Combine the Altivec and VSX code for logical
6477         operations, and move it here.
6478         (ior<mode>3, 128-bit types): Likewise.
6479         (xor<mode>3, 128-bit types): Likewise.
6480         (one_cmpl<mode>3, 128-bit types): Likewise.
6481         (nor<mode>3, 128-bit types): Likewise.
6482         (andc<mode>3, 128-bit types): Likewise.
6483         (eqv<mode>3, 128-bit types): Likewise.
6484         (nand<mode>3, 128-bit types): Likewise.
6485         (orc<mode>3, 128-bit types): Likewise.
6486         (and<mode>3_internal): Likewise.
6487         (bool<mode>3_internal): Likewise.
6488         (boolc<mode>3_internal1): Likewise.
6489         (boolc<mode>3_internal2): Likewise.
6490         (boolcc<mode>3_internal1): Likewise.
6491         (boolcc<mode>3_internal2): Likewise.
6492         (eqv<mode>3_internal1): Likewise.
6493         (eqv<mode>3_internal2): Likewise.
6494         (one_cmpl1<mode>3_internal): Likewise.
6496 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6498         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6499         Rename flag_stack_usage to flag_stack_usage_info.
6501 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6503         * config/microblaze/sync.md: New file.
6504         * config/microblaze/microblaze.md: Include sync.md
6505         * config/microblaze/microblaze.c: Add print_operand 'y'.
6506         * config/microblaze/constraints.md: Add memory_contraint
6507         'Q' which is a single register.
6509 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
6511         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6513 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6515         * reload.c (find_reloads): Exit loop once we find this operand
6516         cannot be reloaded somehow for this alternative.
6518         * reload.c (find_reloads): Exit loop once we find a hard register.
6520         * rtlanal.c (computed_jump_p): Exit loop once we find label
6521         reference is used.
6523         * i386.c (ix86_pad_returns): Exit loop after setting replace.
6525         * cfgloopmanip.c (remove_path): Exit loop after setting
6526         irred_invalidated.
6528         * gensupport.c (subst_dup): Avoid loop if code is not
6529         MATCH_DUP nor MATCH_OP_DUP.
6531 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
6533         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6535 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6537         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6538         true for SP_REGNUM if mode == ptr_mode.
6539         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6540         with value R0_REGNUM + 31.
6542 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6544         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6545         pad pointer-typed argument downward.
6547 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6549         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6550         and __ILP32__ when the ILP32 model is in use.
6552 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6554         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6555         (aarch64_load_symref_appropriately): In the case of
6556         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6557         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6558         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6559         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6560         if 'mode' doesn't equal to 'ptr_mode'.
6561         (aarch64_output_mi_thunk): Add an assertion on the alignment of
6562         'vcall_offset'; change to call aarch64_emit_move differently depending
6563         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6564         to calculate the upper bound of 'vcall_offset'.
6565         (aarch64_cannot_force_const_mem): Change to also return true if
6566         mode != ptr_mode.
6567         (aarch64_legitimize_reload_address): In the case of large
6568         displacements, add new local variable 'xmode' and an assertion
6569         based on it; change to use 'xmode' to generate the new rtx and
6570         reload.
6571         (aarch64_asm_trampoline_template): Change to generate the template
6572         differently depending on TARGET_ILP32 or not; change to use
6573         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6574         (aarch64_trampoline_size): Removed.
6575         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
6576         and replace immediate literals with it.  Change to use 'ptr_mode'
6577         instead of 'DImode' and call convert_memory_address if the mode
6578         of 'fnaddr' doesn't equal to 'ptr_mode'.
6579         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
6580         to output symbol.
6581         (aarch64_elf_asm_destructor): Likewise.
6582         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
6583         on TARGET_ILP32 instead of aarch64_trampoline_size.
6584         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
6585         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
6586         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
6587         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
6588         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
6589         (storewb_pair<GPI:mode>_<P:mode>): ... this.
6590         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
6591         depending on the value of 'mode'.
6592         (add_losym_<mode>): New.
6593         (ldr_got_small_<mode>): New, based on ldr_got_small.
6594         (ldr_got_small): Remove.
6595         (ldr_got_small_sidi): New.
6596         * config/aarch64/iterators.md (P): New.
6597         (PTR): Change to 'ptr_mode' in the condition.
6599 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6601         * config.gcc (aarch64*-*-*): Support --with-abi.
6602         (aarch64*-*-elf): Support --with-multilib-list.
6603         (aarch64*-*-linux*): Likewise.
6604         (supported_defaults): Add abi to aarch64*-*-*.
6605         * configure.ac: Mention AArch64 for --with-multilib-list.
6606         * configure: Re-generated.
6607         * config/aarch64/biarchilp32.h: New file.
6608         * config/aarch64/biarchlp64.h: New file.
6609         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
6610         (ABI_SPEC): Ditto.
6611         (MULTILIB_DEFAULTS): Ditto.
6612         (DRIVER_SELF_SPECS): Ditto.
6613         (ASM_SPEC): Update to also substitute -mabi.
6614         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
6615         file whose name depends on -mabi= and -mbig-endian.
6616         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
6617         TARGET_ILP32.
6618         (POINTER_SIZE): New define.
6619         (POINTERS_EXTEND_UNSIGNED): Ditto.
6620         (enum aarch64_abi_type): New enumeration tag.
6621         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
6622         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
6623         (TARGET_ILP32): New define.
6624         * config/aarch64/aarch64.opt (mabi): New.
6625         (aarch64_abi): New.
6626         (ilp32, lp64): New values for -mabi.
6627         * config/aarch64/t-aarch64 (comma): New define.
6628         (MULTILIB_OPTIONS): Ditto.
6629         (MULTILIB_DIRNAMES): Ditto.
6630         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
6631         * doc/invoke.texi: Document -mabi for AArch64.
6633 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
6635         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
6637 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6638             Anton Blanchard  <anton@au1.ibm.com>
6640         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
6641         endianness when selecting field to splat.
6643 2013-07-22  Eric Christopher  <echristo@gmail.com>
6645         * dwarf2out.c (die_odr_checksum): New function to use
6646         CHECKSUM_ macros and ULEB128 for DIE tag.
6647         (generate_type_signature): Use.
6649 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
6651         * config.gcc (sparc*-*-*): Accept leon3 processor.
6652         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
6653         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
6654         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
6655         * config/sparc/sparc.opt (enum processor_type): Add leon3.
6656         (mfix-ut699): Adjust comment.
6657         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
6658         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
6659         (CPP_CPU_SPEC): Likewise.
6660         (ASM_CPU_SPEC): Likewise.
6661         * config/sparc/sparc.c (leon3_cost): New constant.
6662         (sparc_option_override): Add leon3 support.
6663         (mem_ref): New function.
6664         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
6665         (sparc_do_work_around_errata): Look into the instruction in the delay
6666         slot and adjust accordingly.  Add fix for the data cache nullify issues
6667         of the UT699.  Change insertion position for the NOP.
6668         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
6669         (leon3_load): New reservation.
6670         (leon_store): Bump latency to 2.
6671         (grfpu): New automaton.
6672         (grfpu_alu): New unit.
6673         (grfpu_ds): Likewise.
6674         (leon_fp_alu): Adjust.
6675         (leon_fp_mult): Delete.
6676         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
6677         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
6678         * config/sparc/sparc.md (cpu): Add leon3.
6679         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
6680         (swapsi): Likewise.
6681         (atomic_test_and_set): Likewise.
6682         (ldstub): Likewise.
6684 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
6686         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
6687         default for R5900 targets.
6688         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
6689         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
6690         * config/mips/mips.c (mips_option_override): Report an error for
6691         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
6692         for -march=r5900 -mhard-float.
6694 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6696         * df-problems.c (can_move_insns_across): Exit loop once we
6697         find a non-fixed, non-global register.
6699         * ipa-pure-const.c (propagate_nothrow): Exit loop after
6700         setting can_throw.
6702         * omega.c (omega_eliminate_red): Break after setting red_found.
6703         (omega_problem_has_red_equations): Similarly after setting found.
6704         (omega_query_variable): Similarly after setting coupled.
6706 2013-07-22  Marek Polacek  <polacek@redhat.com>
6708         * gimplify.c: Don't include gimple.h twice.
6710 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6712         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
6713         instead of TARGET_THUMB1.
6714         (Pz): New constraint.
6715         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
6716         encodings.
6717         (compare_negsi_si): Likewise.
6718         (compare_addsi2_op0): Likewise.
6719         (compare_addsi2_op1): Likewise.
6720         (addsi3_carryin_<optab>): Likewise.
6721         (addsi3_carryin_alt2_<optab>): Likewise.
6722         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
6723         for arm_restrict_it.
6724         (subsi3_carryin): Likewise.
6725         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
6726         (minmax_arithsi): Disable for arm_restrict_it.
6727         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
6728         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
6729         (satsi_<SAT:code>_shift): Likewise.
6730         (arm_shiftsi3): Add alternative for 16-bit encoding.
6731         (arm32_movhf): Disable for arm_restrict_it.
6732         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
6733         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
6735 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
6737         * config/arm/arm.md (attribute "insn"): Delete.
6738         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
6739         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
6740         (not_shiftsi): Update for attribute change.
6741         (not_shiftsi_compare0): Likewise.
6742         (not_shiftsi_compare0_scratch): Likewise.
6743         (arm_one_cmplsi2): Likewise.
6744         (thumb1_one_cmplsi2): Likewise.
6745         (notsi_compare0): Likewise.
6746         (notsi_compare0_scratch): Likewise.
6747         (thumb1_movdi_insn): Likewise.
6748         (arm_movsi_insn): Likewise.
6749         (movhi_insn_arch4): Likewise.
6750         (movhi_bytes): Likewise.
6751         (arm_movqi_insn): Likewise.
6752         (thumb1_movqi_insn): Likewise.
6753         (arm32_movhf): Likewise.
6754         (thumb1_movhf): Likewise.
6755         (arm_movsf_soft_insn): Likewise.
6756         (thumb1_movsf_insn): Likewise.
6757         (thumb_movdf_insn): Likewise.
6758         (movsicc_insn): Likewise.
6759         (movsfcc_soft_insn): Likewise.
6760         (and_scc): Likewise.
6761         (cond_move): Likewise.
6762         (if_move_not): Likewise.
6763         (if_not_move): Likewise.
6764         (if_shift_move): Likewise.
6765         (if_move_shift): Likewise.
6766         (if_shift_shift): Likewise.
6767         (if_not_arith): Likewise.
6768         (if_arith_not): Likewise.
6769         (cond_move_not): Likewise.
6770         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
6771         (neon_mov<mode>): Likewise.
6772         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
6773         (thumb2_movsi_vfp): Likewise.
6774         (movsf_vfp): Likewise.
6775         (thumb2_movsf_vfp): Likewise.
6776         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
6777         change.
6778         (cortexa7_older_only): Likewise.
6779         (cortexa7_younger): Likewise.
6780         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
6781         (1020alu_shift_op): Likewise.
6782         (1020alu_shift_reg_op): Likewise.
6783         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
6784         (alu_shift_op): Likewise.
6785         (alu_shift_reg_op): Likewise.
6786         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
6787         (11_alu_shift_op): Likewise.
6788         (11_alu_shift_reg_op): Likewise.
6789         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
6790         (9_alu_shift_reg_op): Likewise.
6791         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6792         change.
6793         (cortex_a15_alu_shift): Likewise.
6794         (cortex_a15_alu_shift_reg): Likewise.
6795         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
6796         (cortex_a5_alu_shift): Likewise.
6797         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
6798         change.
6799         (cortex_a53_alu_shift): Likewise.
6800         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
6801         change.
6802         (cortex_a7_alu_reg): Likewise.
6803         (cortex_a7_alu_shift): Likewise.
6804         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6805         (cortex_a8_alu_shift): Likewise.
6806         (cortex_a8_alu_shift_reg): Likewise.
6807         (cortex_a8_mov): Likewise.
6808         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
6809         (cortex_a9_dp_shift): Likewise.
6810         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
6811         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
6812         (cortex_r4_mov): Likewise.
6813         (cortex_r4_alu_shift): Likewise.
6814         (cortex_r4_alu_shift_reg): Likewise.
6815         * config/arm/fa526.md (526_alu_op): Update for attribute change.
6816         (526_alu_shift_op): Likewise.
6817         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
6818         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
6819         (626te_alu_shift_op): Likewise.
6820         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
6821         (726te_alu_op): Likewise.
6822         (726te_alu_shift_op): Likewise.
6823         (726te_alu_shift_reg_op): Likewise.
6824         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
6825         (mp626_alu_shift_op): Likewise.
6826         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
6827         (pj4_alu_e1_conds): Likewise.
6828         (pj4_alu): Likewise.
6829         (pj4_alu_conds): Likewise.
6830         (pj4_shift): Likewise.
6831         (pj4_shift_conds): Likewise.
6832         (pj4_alu_shift): Likewise.
6833         (pj4_alu_shift_conds): Likewise.
6835 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6837         * config/arm/predicates.md (shiftable_operator_strict_it):
6838         New predicate.
6839         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
6840         Disable cond_exec version for arm_restrict_it.
6841         (thumb2_smaxsi3): Convert to generate cond_exec.
6842         (thumb2_sminsi3): Likewise.
6843         (thumb32_umaxsi3): Likewise.
6844         (thumb2_uminsi3): Likewise.
6845         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
6846         (thumb2_neg_abssi2): Likewise.
6847         (thumb2_mov_scc): Add alternative for 16-bit encoding.
6848         (thumb2_movsicc_insn): Adjust alternatives.
6849         (thumb2_mov_negscc): Disable for arm_restrict_it.
6850         (thumb2_mov_negscc_strict_it): New pattern.
6851         (thumb2_mov_notscc_strict_it): New pattern.
6852         (thumb2_mov_notscc): Disable for arm_restrict_it.
6853         (thumb2_ior_scc): Likewise.
6854         (thumb2_ior_scc_strict_it): New pattern.
6855         (thumb2_cond_move): Adjust for arm_restrict_it.
6856         (thumb2_cond_arith): Disable for arm_restrict_it.
6857         (thumb2_cond_arith_strict_it): New pattern.
6858         (thumb2_cond_sub): Adjust for arm_restrict_it.
6859         (thumb2_movcond): Likewise.
6860         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
6861         (thumb2_zero_extendhisi2_v6): Likewise.
6862         (thumb2_zero_extendqisi2_v6): Likewise.
6863         (orsi_notsi_si): Likewise.
6864         (orsi_not_shiftsi_si): Likewise.
6866 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
6868         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
6869         instruction sequence is 1 byte shorter.
6871 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
6873         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
6874         it is not needed after split.
6876 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
6878         PR target/51784
6879         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
6880         second label for nonlocal goto receivers. Don't output pic base labels
6881         unless we're producing PIC; mark that action unreachable().
6882         (ix86_save_reg): If the function contains a nonlocal label, save the
6883         PIC base reg.
6884         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
6885         * config/darwin.c (emitted_pic_label_num): New GTY.
6886         (update_pic_label_number_if_needed): New.
6887         (machopic_output_function_base_name): Adjust for nonlocal receiver
6888         case.
6889         (machopic_should_output_picbase_label): New.
6890         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
6891         (nonlocal_goto_receiver): New insn and split.
6893 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
6895         * config/aarch64/aarch64-builtins.c
6896         (aarch64_fold_builtin): Fold abs in all modes.
6897         * config/aarch64/aarch64-simd-builtins.def
6898         (abs): Enable for all modes.
6899         * config/aarch64/arm_neon.h
6900         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
6901         (vabs_f64): Add missing intrinsic.
6903 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
6905         * config/aarch64/arm_neon.h (vabs_s64): New function
6907 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
6909         PR target/57516
6910         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
6911         * config/avr/avr.md (adjust_len): Add `round'.
6912         * config/avr/avr-protos.h (avr_out_round): New prototype.
6913         (avr_out_plus): Add `out_label' argument.
6914         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
6915         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
6916         Handle the case where `insn' is just a pattern.
6917         (avr_out_bitop): Handle the case where `insn' is just a pattern.
6918         (avr_out_round): New function.
6919         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
6921 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
6923         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6924         Add check for flag_stack_usage to handle -fstack-usage support
6926 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
6928         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
6929         interaction for new Power8 flags and VSX.
6931 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
6933         PR middle-end/57698
6934         * tree-inline.c (expand_call_inline): Emit errors during
6935         early_inlining only if optimization is not turned on.
6937 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6939         * passes.def: New.
6941         * passes.c (init_optimization_passes): Move the construction of
6942         the pass hierarchy into a new passes.def file.
6944         * Makefile.in (passes.o): Add dependency on passes.def.
6946 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6948         * passes.c (init_optimization_passes): Introduce macros for
6949         constructing the tree of passes (INSERT_PASSES_AFTER,
6950         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
6951         TERMINATE_PASS_LIST).
6953 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
6954             Wei Mi  <wmi@google.com>
6956         PR rtl-optimization/57878
6957         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
6958         top.
6959         (reload_pseudo_compare_func): Check nregs first for reload
6960         pseudos.
6962 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6964         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
6966 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
6968         * read-rtl.c (validate_const_int): Once an invalid character is
6969         seen, quit the loop.
6971         * gengtype.c (write_roots): Similarly once we find the "deletable"
6972         or "if_marked" option.
6974 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6976         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
6977         "xtab" and "sat".  Move value "clz" from here to ...
6978         (attriubte "type"): ... here.
6979         (satsi_<SAT:code>): Delete "insn" attribute.
6980         (satsi_<SAT:code>_shift): Likewise.
6981         (arm_zero_extendqisi2addsi): Likewise.
6982         (arm_extendqisi2addsi): Likewise.
6983         (clzsi2): Update for attribute changes.
6984         (rbitsi2): Likewise.
6985         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
6986         attribute.
6987         (arm_usatsihi): Likewise.
6988         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6990 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6992         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
6993         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
6994         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
6995         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
6996         in alphabetical order.
6997         (attribute "core_cycles"): Update for attribute changes.
6998         (arm_addsi3): Likewise.
6999         (addsi3_compare0): Likewise.
7000         (addsi3_compare0_scratch): Likewise.
7001         (addsi3_compare_op1): Likewise.
7002         (addsi3_compare_op2): Likewise.
7003         (compare_addsi2_op0): Likewise.
7004         (compare_addsi2_op1): Likewise.
7005         (addsi3_carryin_shift_<optab>): Likewise.
7006         (subsi3_carryin_shift): Likewise.
7007         (rsbsi3_carryin_shift): Likewise.
7008         (arm_subsi3_insn): Likewise.
7009         (subsi3_compare0): Likewise.
7010         (subsi3_compare): Likewise.
7011         (arm_andsi3_insn): Likewise.
7012         (thumb1_andsi3_insn): Likewise.
7013         (andsi3_compare0): Likewise.
7014         (andsi3_compare0_scratch): Likewise.
7015         (zeroextractsi_compare0_scratch
7016         (andsi_not_shiftsi_si): Likewise.
7017         (iorsi3_insn): Likewise.
7018         (iorsi3_compare0): Likewise.
7019         (iorsi3_compare0_scratch): Likewise.
7020         (arm_xorsi3): Likewise.
7021         (thumb1_xorsi3_insn): Likewise.
7022         (xorsi3_compare0): Likewise.
7023         (xorsi3_compare0_scratch): Likewise.
7024         (satsi_<SAT:code>_shift): Likewise.
7025         (rrx): Likewise.
7026         (arm_shiftsi3): Likewise.
7027         (shiftsi3_compare0): Likewise.
7028         (not_shiftsi): Likewise.
7029         (not_shiftsi_compare0): Likewise.
7030         (not_shiftsi_compare0_scratch): Likewise.
7031         (arm_one_cmplsi2): Likewise.
7032         (thumb_one_complsi2): Likewise.
7033         (notsi_compare0): Likewise.
7034         (notsi_compare0_scratch): Likewise.
7035         (thumb1_zero_extendhisi2): Likewise.
7036         (arm_zero_extendhisi2): Likewise.
7037         (arm_zero_extendhisi2_v6): Likewise.
7038         (arm_zero_extendhisi2addsi): Likewise.
7039         (thumb1_zero_extendqisi2): Likewise.
7040         (thumb1_zero_extendqisi2_v6): Likewise.
7041         (arm_zero_extendqisi2): Likewise.
7042         (arm_zero_extendqisi2_v6): Likewise.
7043         (arm_zero_extendqisi2addsi): Likewise.
7044         (thumb1_extendhisi2): Likewise.
7045         (arm_extendhisi2): Likewise.
7046         (arm_extendhisi2_v6): Likewise.
7047         (arm_extendqisi): Likewise.
7048         (arm_extendqisi_v6): Likewise.
7049         (arm_extendqisi2addsi): Likewise.
7050         (thumb1_extendqisi2): Likewise.
7051         (thumb1_movdi_insn): Likewise.
7052         (arm_movsi_insn): Likewise.
7053         (movsi_compare0): Likewise.
7054         (movhi_insn_arch4): Likewise.
7055         (movhi_bytes): Likewise.
7056         (arm_movqi_insn): Likewise.
7057         (thumb1_movqi_insn): Likewise.
7058         (arm32_movhf): Likewise.
7059         (thumb1_movhf): Likewise.
7060         (arm_movsf_soft_insn): Likewise.
7061         (thumb1_movsf_insn): Likewise.
7062         (movdf_soft_insn): Likewise.
7063         (thumb_movdf_insn): Likewise.
7064         (arm_cmpsi_insn): Likewise.
7065         (cmpsi_shiftsi): Likewise.
7066         (cmpsi_shiftsi_swp): Likewise.
7067         (arm_cmpsi_negshiftsi_si): Likewise.
7068         (movsicc_insn): Likewise.
7069         (movsfcc_soft_insn): Likewise.
7070         (arith_shiftsi): Likewise.
7071         (arith_shiftsi_compare0
7072         (arith_shiftsi_compare0_scratch
7073         (sub_shiftsi): Likewise.
7074         (sub_shiftsi_compare0
7075         (sub_shiftsi_compare0_scratch
7076         (and_scc): Likewise.
7077         (cond_move): Likewise.
7078         (if_plus_move): Likewise.
7079         (if_move_plus): Likewise.
7080         (if_move_not): Likewise.
7081         (if_not_move): Likewise.
7082         (if_shift_move): Likewise.
7083         (if_move_shift): Likewise.
7084         (if_shift_shift): Likewise.
7085         (if_not_arith): Likewise.
7086         (if_arith_not): Likewise.
7087         (cond_move_not): Likewise.
7088         (thumb1_ashlsi3): Set type attribute.
7089         (thumb1_ashrsi3): Likewise.
7090         (thumb1_lshrsi3): Likewise.
7091         (thumb1_rotrsi3): Likewise.
7092         (shiftsi3_compare0_scratch): Likewise.
7093         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7094         (neon_mov<mode>): Likewise.
7095         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7096         attribute changes.
7097         (thumb2_movsi_insn): Likewise.
7098         (thumb2_cmpsi_neg_shiftsi): Likewise.
7099         (thumb2_extendqisi_v6): Likewise.
7100         (thumb2_zero_extendhisi2_v6): Likewise.
7101         (thumb2_zero_extendqisi2_v6): Likewise.
7102         (thumb2_shiftsi3_short): Likewise.
7103         (thumb2_addsi3_compare0_scratch): Likewise.
7104         (orsi_not_shiftsi_si): Likewise.
7105         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7106         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7107         changes.
7108         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7109         (1020alu_shift_op): Likewise.
7110         (1020alu_shift_reg_op): Likewise.
7111         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7112         (alu_shift_op): Likewise.
7113         (alu_shift_reg_op): Likewise.
7114         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7115         (11_alu_shift_op): Likewise.
7116         (11_alu_shift_reg_op): Likewise.
7117         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7118         (9_alu_shift_reg_op): Likewise.
7119         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7120         changes.
7121         (cortex_a15_alu_shift): Likewise.
7122         (cortex_a15_alu_shift_reg): Likewise.
7123         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7124         changes.
7125         (cortex_a5_alu_shift): Likewise.
7126         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7127         changes.
7128         (cortex_a53_alu_shift): Likewise.
7129         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7130         changes.
7131         (cortex_a7_alu_reg): Likewise.
7132         (cortex_a7_alu_shift): Likewise.
7133         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7134         changes.
7135         (cortex_a8_alu_shift): Likewise.
7136         (cortex_a8_alu_shift_reg): Likewise.
7137         (cortex_a8_mov): Likewise.
7138         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7139         (cortex_a9_dp_shift): Likewise.
7140         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7141         changes.
7142         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7143         changes.
7144         (cortex_r4_mov): Likewise.
7145         (cortex_r4_alu_shift): Likewise.
7146         (cortex_r4_alu_shift_reg): Likewise.
7147         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7148         (526_alu_shift_op): Likewise.
7149         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7150         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7151         (626te_alu_shift_op): Likewise.
7152         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7153         (726te_alu_op): Likewise.
7154         (726te_alu_shift_op): Likewise.
7155         (726te_alu_shift_reg_op): Likewise.
7156         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7157         (mp626_alu_shift_op): Likewise.
7158         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7159         (pj4_alu_e1_conds): Likewise.
7160         (pj4_alu): Likewise.
7161         (pj4_alu_conds): Likewise.
7162         (pj4_shift): Likewise.
7163         (pj4_shift_conds): Likewise.
7164         (pj4_alu_shift): Likewise.
7165         (pj4_alu_shift_conds): Likewise.
7166         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7167         changes.
7168         (cortexa7_older_only): Likewise.
7169         (cortexa7_younger): Likewise.
7171 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7173         * ipa-pure-const.c (generate_summary): Rename to...
7174         (pure_const_generate_summary): ... this.
7176 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7178         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7180 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7182         PR target/57909
7183         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7184         usage in HI mode.
7186 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7188         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7189         enabled without -march=zEC12.
7190         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7191         flags to be set.
7193 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7195         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7196         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7197         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7198         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7199         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7200         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7201         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7202         ISA_HAS_FP4.
7203         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7204         and ISA_HAS_NMADD3_NMSUB3.
7205         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7206         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7207         (nmsub4<mode>, nmsub3<mode>): Likewise.
7208         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7210 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7212         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7213         TARGET_MIPS5400 checking.
7215 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7216             Peter Bergner  <bergner@vnet.ibm.com>
7218         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7219         registers in the comment.
7220         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7221         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7222         rather than FIRST_PSEUDO_REGISTERS.
7224 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7226         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7227         enable extra ISA flags with TARGET_HTM.
7229 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7231         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7232         Fix comment typos.
7234 2013-07-15  Cong Hou  <congh@google.com>
7236         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7237         in compare function for sorting.
7239 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
7241         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7242         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7243         * config/rs6000/rs6000.opt: Add -mhtm option.
7244         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7245         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7246         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7247         __HTM__ if the HTM instructions are available.
7248         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7249         htm_spr_reg_operand): New define_predicates.
7250         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7251         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7252         Include htm.md.
7253         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7254         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7255         HTM builtin functions.
7256         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7257         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7258         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7259         (rs6000_builtin_mask_calculate): Likewise.
7260         (rs6000_option_override_internal): Likewise.
7261         (bdesc_htm): Add new HTM builtin support.
7262         (htm_spr_num): New function.
7263         (htm_spr_regno): Likewise.
7264         (rs6000_htm_spr_icode): Likewise.
7265         (htm_expand_builtin): Likewise.
7266         (htm_init_builtins): Likewise.
7267         (rs6000_expand_builtin): Add support for HTM builtin functions.
7268         (rs6000_init_builtins): Likewise.
7269         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7270         option.
7271         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7272         (TARGET_HTM, MASK_HTM): Define macros.
7273         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7274         (FIXED_REGISTERS): Likewise.
7275         (CALL_USED_REGISTERS): Likewise.
7276         (CALL_REALLY_USED_REGISTERS): Likewise.
7277         (REG_ALLOC_ORDER): Likewise.
7278         (enum reg_class): Likewise.
7279         (REG_CLASS_NAMES): Likewise.
7280         (REG_CLASS_CONTENTS): Likewise.
7281         (REGISTER_NAMES): Likewise.
7282         (ADDITIONAL_REGISTER_NAMES): Likewise.
7283         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7284         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7285         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7286         * config/rs6000/htm.md: New file.
7287         * config/rs6000/htmintrin.h: New file.
7288         * config/rs6000/htmxlintrin.h: New file.
7290 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7292         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7293         Define SYMBOL_TINY_GOT, update comment.
7294         * config/aarch64/aarch64.c
7295         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7296         (aarch64_expand_mov_immediate): Likewise.
7297         (aarch64_print_operand): Likewise.
7298         (aarch64_classify_symbol): Likewise.
7299         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7300         (ldr_got_tiny): Define.
7302 2013-07-13  Tobias Grosser  <tobias@grosser.es>
7304         PR tree-optimization/54094
7305         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7306           scheduling dimension for the parallelism check from the polyhedral
7307           information in the AST.
7308         * graphite-dependences.c (carries_deps): Do not assume the schedule is
7309           in 2D + 1 form.
7311 2013-07-13  Jason Merrill  <jason@redhat.com>
7313         * print-tree.c (debug_vec_tree): Use debug_raw.
7314         (debug_raw (vec<tree, va_gc> &)): New.
7315         (debug_raw (vec<tree, va_gc> *)): New.
7316         * tree.h: Declare them.
7318 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
7320         * ifcvt.c (ifcvt_after_combine): New static variable.
7321         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7322         for size.
7323         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
7324         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7325         rest_of_handle_if_after_reload): Pass new argument for if_convert.
7327 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
7329         * config/mips/mips.c (mips_expand_call): Remove empty statement.
7331 2013-07-12  Michael Matz  <matz@suse.de>
7333         PR middle-end/55771
7334         * convert.c (convert_to_real): Reject non-float inner types.
7336 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
7338         * config/aarch64/aarch64-protos.h
7339         (aarch64_simd_immediate_valid_for_move): Remove.
7340         * config/aarch64/aarch64.c (simd_immediate_info): New member.
7341         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7342         cases.
7343         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7345 2013-07-11  Steve Ellcey  <sellcey@mips.com>
7347         * config/mips/mips.c (mips_conditional_register_usage): Do not
7348         use t[0-7] registers in MIPS16 mode when optimizing for size.
7350 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
7352         * config/i386/i386.c (dispatch_function_versions): Fix array
7353         indexing of function_version_info to match actual_versions.
7355 2013-07-11  Teresa Johnson  <tejohnson@google.com>
7357         * vec.h (struct va_gc): Move release out-of-line.
7358         (va_gc::release): Call ggc_free on released vec.
7360 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7362         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7363         Require GOT register as additional operand in UNSPEC.
7364         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7365         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7366         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7367         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7368         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7369         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7370         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7372 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
7374         PR target/57631
7375         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7376         name seen by assembler/linker rather if available.
7378 2013-07-11  Andreas Schwab  <schwab@suse.de>
7380         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7382 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
7384         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7386 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
7388         * doc/tm.texi.in: Move hook documentation to ....
7389         * target.def: ... here.
7391         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7392         text on @hook line.
7393         * doc/tm.texi: Regenerate.
7395 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
7397         PR c++/57869
7398         * doc/invoke.texi: Document -Wconditionally-supported.
7400 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7402         PR target/57844
7403         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7404         of my_fp.
7406 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7408         PR target/57506
7409         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7410         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7411         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7412         Remove duplicate devices.
7413         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7414         * config/avr/t-multilib: Regenerate.
7415         * config/avr/avr-tables.opt: Regenerate.
7416         * doc/avr-mmcu.texi: Regenerate.
7418 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7420         PR target/56987
7421         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7423 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
7425         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7426         the cost of MULT when optimizing for size.
7428 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7430         * config/cr16/cr16-protos.h: Don't include target.h.
7432 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
7434         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7435         adjust register size for TDmode and TFmode for VSX registers.
7437 2013-07-08  Kai Tietz  <ktietz@redhat.com>
7439         PR target/56892
7440         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7441         hook_bool_const_tree_true.
7443 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7445         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7446         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7447         * config/s390/s390.md: Define FPR*_REGNUM constants.
7448         Fix FPR2_REGNUM constant (18 -> 17).
7449         ("*trunc<BFP:mode><DFP_ALL:mode>2")
7450         ("*trunc<DFP_ALL:mode><BFP:mode>2")
7451         ("trunc<BFP:mode><DFP_ALL:mode>2")
7452         ("trunc<DFP_ALL:mode><BFP:mode>2")
7453         ("*extend<BFP:mode><DFP_ALL:mode>2")
7454         ("*extend<DFP_ALL:mode><BFP:mode>2")
7455         ("extend<BFP:mode><DFP_ALL:mode>2")
7456         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7457         FPR4_REGNUM.
7459 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
7461         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7463 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7465         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7466         and cfun_fpr_bit_p to cfun_fpr_save_p.
7467         (s390_frame_area, s390_register_info, s390_frame_info)
7468         (s390_emit_prologue, s390_emit_epilogue)
7469         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7470         register numbers.
7471         * config/s390/s390.h: Define *_REGNUM macros for floating point
7472         register numbers.
7474 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
7476         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7478 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
7480         PR rtl-optimization/57786
7481         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7482         and break out of the loop when it is set to false.
7484 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
7486         PR target/57819
7487         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7488         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7489         (const_int 63)) 0)).
7490         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7491         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7492         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7494         PR rtl-optimization/57829
7495         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7496         mask bits outside of mode are just sign-extension from mode to HWI.
7498 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7500         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7501         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7502         adjust_address instead of change_address to keep info about alignment.
7503         (emit_strmov): Remove.
7504         (emit_memmov): New function.
7505         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7506         (expand_movmem_epilogue): Likewise and return updated rtx for
7507         destination.
7508         (expand_constant_movmem_prologue): Likewise and return updated rtx for
7509         destination and source.
7510         (decide_alignment): Refactor, handle vector_loop.
7511         (ix86_expand_movmem): Likewise.
7512         (ix86_expand_setmem): Likewise.
7513         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7515 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
7517         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7518         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7520 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7522         * config/i386/sse.md (sse_movlhps): Change alternative 3
7523         of operand 2 to "m".
7525 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7527         PR target/57807
7528         * config/i386/sse.md (iptr): New mode attribute.
7529         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7530         (<sse>_vm<plusminus_insn><mode>3): Ditto.
7531         (<sse>_vmmul<mode>3): Ditto.
7532         (<sse>_vmdiv<mode>3): Ditto.
7533         (sse_vmrcpv4sf2): Ditto.
7534         (<sse>_vmsqrt<mode>2): Ditto.
7535         (sse_vmrsqrtv4sf2): Ditto.
7536         (<sse>_vm<code><mode>3): Ditto.
7537         (avx_vmcmp<mode>3): Ditto.
7538         (<sse>_vmmaskcmp<mode>3): Ditto.
7539         (<sse>_comi): Ditto.
7540         (<sse>_ucomi): Ditto.
7541         (*xop_vmfrcz_<mode>): Ditto.
7542         (*fmai_fmadd_<mode>): Ditto.
7543         (*fmai_fmsub_<mode>): Ditto.
7544         (*fmai_fnmadd_<mode>): Ditto.
7545         (*fmai_fnmsub_<mode>): Ditto.
7546         (*fma4i_vmfmadd_<mode>): Ditto.
7547         (*fma4i_vmfmsub_<mode>): Ditto.
7548         (*fma4i_vmfnmadd_<mode>): Ditto.
7549         (*fma4i_vmfnmsub_<mode>): Ditto.
7550         (*xop_vmfrcz_<mode>): Ditto.
7551         (sse_cvtps2pi): Ditto.
7552         (sse_cvttps2pi): Ditto.
7553         (sse_cvtss2si): Ditto.
7554         (sse_cvtss2si_2): Ditto.
7555         (sse_cvtss2siq_2): Ditto.
7556         (sse_cvttss2si): Ditto.
7557         (sse_cvttss2siq): Ditto.
7558         (sse_cvtsd2si): Ditto.
7559         (sse_cvtsd2si_2): Ditto.
7560         (sse_cvtsd2siq_2): Ditto.
7561         (sse_cvttsd2si): Ditto.
7562         (sse_cvttsd2siq): Ditto.
7563         (sse_cvtsd2ss): Ditto.
7564         (sse_cvtss2sd): Ditto.
7565         (avx2_pbroadcast<mode>): Ditto.
7566         (avx2_pbroadcast<mode>_1): Ditto.
7567         (*avx_vperm_broadcast_v4sf): Ditto.
7569         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7570         (sse_movlhps): Ditto.
7571         (sse_storehps): Ditto.
7572         (sse_loadhps): Ditto.
7573         (sse_storelps): Ditto.
7574         (sse_loadlps): Ditto.
7575         (*vec_concatv4sf): Ditto.
7576         (*vec_interleave_highv2df): Ditto.
7577         (*vec_interleave_lowv2df): Ditto.
7578         (*vec_extractv2df_1_sse): Ditto.
7579         (*vec_extractv2df_0_sse): Ditto.
7580         (sse2_storelpd): Ditto.
7581         (sse2_loadlpd): Ditto.
7582         (sse2_movsd): Ditto.
7583         (*vec_concatv4si): Ditto.
7584         (vec_concatv2di): Ditto.
7586         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
7587         for Intel asm dialect.
7588         (mmx_punpcklwd): Ditto.
7589         (mmx_punpckldq): Ditto.
7591         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
7592         for intel assembler dialect.
7594 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
7596         PR target/29776
7597         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
7598         for BUILT_IN_C{LZ,LRSB}*.
7599         * tree.h (CASE_INT_FN): Add FN##IMAX case.
7600         * tree-vrp.c (extract_range_basic): Handle
7601         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
7602         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
7603         fall thru to code calling set_value*.
7604         * builtins.c (expand_builtin): Remove *IMAX cases.
7605         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
7606         if width is bigger than 2*HWI.
7608 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
7610         PR rtl-optimization/55342
7611         * lra-int.h (lra_subreg_reload_pseudos): New.
7612         * lra.c: Add undoing optional reloads to the block diagram.
7613         (lra_subreg_reload_pseudos): New.
7614         (lra_optional_reload_pseudos): Change comments.
7615         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
7616         lra_optional_reload_pseudos after undo transformations.
7617         * lra-assigns.c (pseudo_prefix_title): New.
7618         (lra_setup_reg_renumber): Use it.
7619         (spill_for): Ditto.  Check subreg reload pseudos too.
7620         (assign_by_spills): Consider subreg reload pseudos too.
7621         * lra-constraints.c (simplify_operand_subreg): Use
7622         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
7623         (curr_insn_transform): Recognize and do optional reloads.
7624         (undo_optional_reloads): New.
7625         (lra_undo_inheritance): Call undo_optional_reloads.
7627 2013-07-05  Thomas Quinot  <quinot@adacore.com>
7629         * tree-complex.c (expand_complex_operations_1): Fix typo.
7631 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
7633         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
7634         (tune_params): New member 'const vec_costs'.
7635         * config/aarch64/aarch64.c (generic_vector_cost): New.
7636         (generic_tunings): New member 'generic_vector_cost'.
7637         (aarch64_builtin_vectorization_cost): New.
7638         (aarch64_add_stmt_cost): New.
7639         (TARGET_VECTORIZE_ADD_STMT_COST): New.
7640         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
7642 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
7644         PR target/57777
7645         * config/i386/predicates.md (vsib_address_operand): Disallow
7646         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
7648 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
7650         PR middle-end/55030
7651         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
7652         expand_builtin_setjmp_receiver.
7653         (expand_label): Adjust, call expand_builtin_setjmp_receiver
7654         with NULL for the label parameter.
7655         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
7656         the frame-pointer.  Adjust comments.
7657         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
7658         only if LABEL is non-NULL.
7660 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
7662         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
7663         (ARM_ABI_AAPCS64): Ditto.
7664         (arm_abi): Ditto.
7665         (ARM_DEFAULT_ABI): Ditto.
7667 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
7669         * config/aarch64/aarch64-builtins.c
7670         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
7671         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
7672         (st1): Likewise.
7673         * config/aarch64/aarch64-simd.md
7674         (aarch64_ld1<VALL:mode>): New.
7675         (aarch64_st1<VALL:mode>): Likewise.
7676         * config/aarch64/arm_neon.h
7677         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
7679 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
7681         * config/i386/i386.c (gate_insert_vzeroupper): Check if
7682         target ISA is AVX.
7683         (ix86_option_override_internal):Turn on all -mavx target flags by
7684         default as they are dependent on AVX anyway.
7686 2013-07-02  Cary Coutant  <ccoutant@google.com>
7688         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
7689         deterministic hash.
7690         (loc_checksum_ordered): Likewise.
7691         (hash_loc_operands): Remove inline keyword.
7693 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
7695         PR tree-optimization/57741
7696         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
7697         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
7698         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
7699         Allow REAL_CST step_exprs if flag_associative_math.
7700         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
7702 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7704         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
7706 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7708         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
7710 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7712         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
7713         encoding.
7714         (iorsi3_insn): Likewise.
7715         (arm_xorsi3): Likewise.
7717 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
7719         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
7720         to ...
7721         (attribute "type"): ... here, and prefix with "wmmx_".
7722         (attribute "core_cycles"): Update for attribute changes.
7723         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
7724         (tbcstv4hi): Likewise.
7725         (tbcstv2si): Likewise.
7726         (iwmmxt_iordi3): Likewise.
7727         (iwmmxt_xordi3): Likewise.
7728         (iwmmxt_anddi3): Likewise.
7729         (iwmmxt_nanddi3): Likewise.
7730         (iwmmxt_arm_movdi): Likewise.
7731         (iwmmxt_movsi_insn): Likewise.
7732         (mov<mode>_internal): Likewise.
7733         (and<mode>3_iwmmxt): Likewise.
7734         (ior<mode>3_iwmmxt): Likewise.
7735         (xor<mode>3_iwmmxt): Likewise.
7736         (add<mode>3_iwmmxt): Likewise.
7737         (ssaddv8qi3): Likewise.
7738         (ssaddv4hi3): Likewise.
7739         (ssaddv2si3): Likewise.
7740         (usaddv8qi3): Likewise.
7741         (usaddv4hi3): Likewise.
7742         (usaddv2si3): Likewise.
7743         (sub<mode>3_iwmmxt): Likewise.
7744         (sssubv8qi3): Likewise.
7745         (sssubv4hi3): Likewise.
7746         (sssubv2si3): Likewise.
7747         (ussubv8qi3): Likewise.
7748         (ussubv4hi3): Likewise.
7749         (ussubv2si3): Likewise.
7750         (mulv4hi3_iwmmxt): Likewise.
7751         (smulv4hi3_highpart): Likewise.
7752         (umulv4hi3_highpart): Likewise.
7753         (iwmmxt_wmacs): Likewise.
7754         (iwmmxt_wmacsz): Likewise.
7755         (iwmmxt_wmacu): Likewise.
7756         (iwmmxt_wmacuz): Likewise.
7757         (iwmmxt_clrdi): Likewise.
7758         (iwmmxt_clrv8qi): Likewise.
7759         (iwmmxt_clr4hi): Likewise.
7760         (iwmmxt_clr2si): Likewise.
7761         (iwmmxt_uavgrndv8qi3): Likewise.
7762         (iwmmxt_uavgrndv4hi3): Likewise.
7763         (iwmmxt_uavgv8qi3): Likewise.
7764         (iwmmxt_uavgv4hi3): Likewise.
7765         (iwmmxt_tinsrb): Likewise.
7766         (iwmmxt_tinsrh): Likewise.
7767         (iwmmxt_tinsrw): Likewise.
7768         (iwmmxt_textrmub): Likewise.
7769         (iwmmxt_textrmsb): Likewise.
7770         (iwmmxt_textrmuh): Likewise.
7771         (iwmmxt_textrmsh): Likewise.
7772         (iwmmxt_textrmw): Likewise.
7773         (iwmxxt_wshufh): Likewise.
7774         (eqv8qi3): Likewise.
7775         (eqv4hi3): Likewise.
7776         (eqv2si3): Likewise.
7777         (gtuv8qi3): Likewise.
7778         (gtuv4hi3): Likewise.
7779         (gtuv2si3): Likewise.
7780         (gtv8qi3): Likewise.
7781         (gtv4hi3): Likewise.
7782         (gtv2si3): Likewise.
7783         (smax<mode>3_iwmmxt): Likewise.
7784         (umax<mode>3_iwmmxt): Likewise.
7785         (smin<mode>3_iwmmxt): Likewise.
7786         (umin<mode>3_iwmmxt): Likewise.
7787         (iwmmxt_wpackhss): Likewise.
7788         (iwmmxt_wpackwss): Likewise.
7789         (iwmmxt_wpackdss): Likewise.
7790         (iwmmxt_wpackhus): Likewise.
7791         (iwmmxt_wpackwus): Likewise.
7792         (iwmmxt_wpackdus): Likewise.
7793         (iwmmxt_wunpckihb): Likewise.
7794         (iwmmxt_wunpckihh): Likewise.
7795         (iwmmxt_wunpckihw): Likewise.
7796         (iwmmxt_wunpckilb): Likewise.
7797         (iwmmxt_wunpckilh): Likewise.
7798         (iwmmxt_wunpckilw): Likewise.
7799         (iwmmxt_wunpckehub): Likewise.
7800         (iwmmxt_wunpckehuh): Likewise.
7801         (iwmmxt_wunpckehuw): Likewise.
7802         (iwmmxt_wunpckehsb): Likewise.
7803         (iwmmxt_wunpckehsh): Likewise.
7804         (iwmmxt_wunpckehsw): Likewise.
7805         (iwmmxt_wunpckelub): Likewise.
7806         (iwmmxt_wunpckeluh): Likewise.
7807         (iwmmxt_wunpckeluw): Likewise.
7808         (iwmmxt_wunpckelsb): Likewise.
7809         (iwmmxt_wunpckelsh): Likewise.
7810         (iwmmxt_wunpckelsw): Likewise.
7811         (ror<mode>3): Likewise.
7812         (ashr<mode>3_iwmmxt): Likewise.
7813         (lshr<mode>3_iwmmxt): Likewise.
7814         (ashl<mode>3_iwmmxt): Likewise.
7815         (ror<mode>3_di): Likewise.
7816         (ashr<mode>3_di): Likewise.
7817         (lshr<mode>3_di): Likewise.
7818         (ashl<mode>3_di): Likewise.
7819         (iwmmxt_wmadds): Likewise.
7820         (iwmmxt_wmaddu): Likewise.
7821         (iwmmxt_tmia): Likewise.
7822         (iwmmxt_tmiaph): Likewise.
7823         (iwmmxt_tmiabb): Likewise.
7824         (iwmmxt_tmiatb): Likewise.
7825         (iwmmxt_tmiabt): Likewise.
7826         (iwmmxt_tmiatt): Likewise.
7827         (iwmmxt_tmovmskb): Likewise.
7828         (iwmmxt_tmovmskh): Likewise.
7829         (iwmmxt_tmovmskw): Likewise.
7830         (iwmmxt_waccb): Likewise.
7831         (iwmmxt_wacch): Likewise.
7832         (iwmmxt_waccw): Likewise.
7833         (iwmmxt_waligni): Likewise.
7834         (iwmmxt_walignr): Likewise.
7835         (iwmmxt_walignr0): Likewise.
7836         (iwmmxt_walignr1): Likewise.
7837         (iwmmxt_walignr2): Likewise.
7838         (iwmmxt_walignr3): Likewise.
7839         (iwmmxt_wsadb): Likewise.
7840         (iwmmxt_wsadh): Likewise.
7841         (iwmmxt_wsadbz): Likewise.
7842         (iwmmxt_wsadhz): Likewise.
7843         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
7844         (iwmmxt_wabsdiffb): Likewise.
7845         (iwmmxt_wabsdiffh): Likewise.
7846         (iwmmxt_wabsdiffw): Likewise.
7847         (iwmmxt_waddsubhx): Likewise
7848         (iwmmxt_wsubaddhx): Likewise.
7849         (addc<mode>3): Likewise.
7850         (iwmmxt_avg4): Likewise.
7851         (iwmmxt_avg4r): Likewise.
7852         (iwmmxt_wmaddsx): Likewise.
7853         (iwmmxt_wmaddux): Likewise.
7854         (iwmmxt_wmaddsn): Likewise.
7855         (iwmmxt_wmaddun): Likewise.
7856         (iwmmxt_wmulwsm): Likewise.
7857         (iwmmxt_wmulwum): Likewise.
7858         (iwmmxt_wmulsmr): Likewise.
7859         (iwmmxt_wmulumr): Likewise.
7860         (iwmmxt_wmulwsmr): Likewise.
7861         (iwmmxt_wmulwumr): Likewise.
7862         (iwmmxt_wmulwl): Likewise.
7863         (iwmmxt_wqmulm): Likewise.
7864         (iwmmxt_wqmulwm): Likewise.
7865         (iwmmxt_wqmulmr): Likewise.
7866         (iwmmxt_wqmulwmr): Likewise.
7867         (iwmmxt_waddbhusm): Likewise.
7868         (iwmmxt_waddbhusl): Likewise.
7869         (iwmmxt_wqmiabb): Likewise.
7870         (iwmmxt_wqmiabt): Likewise.
7871         (iwmmxt_wqmiatb): Likewise.
7872         (iwmmxt_wqmiatt): Likewise.
7873         (iwmmxt_wqmiabbn): Likewise.
7874         (iwmmxt_wqmiabtn): Likewise.
7875         (iwmmxt_wqmiatbn): Likewise.
7876         (iwmmxt_wqmiattn): Likewise.
7877         (iwmmxt_wmiabb): Likewise.
7878         (iwmmxt_wmiabt): Likewise.
7879         (iwmmxt_wmiatb): Likewise.
7880         (iwmmxt_wmiatt): Likewise.
7881         (iwmmxt_wmiabbn): Likewise.
7882         (iwmmxt_wmiabtn): Likewise.
7883         (iwmmxt_wmiatbn): Likewise.
7884         (iwmmxt_wmiattn): Likewise.
7885         (iwmmxt_wmiawbb): Likewise.
7886         (iwmmxt_wmiawbt): Likewise.
7887         (iwmmxt_wmiawtb): Likewise.
7888         (iwmmxt_wmiawtt): Likewise.
7889         (iwmmxt_wmiawbbn): Likewise.
7890         (iwmmxt_wmiawbtn): Likewise.
7891         (iwmmxt_wmiawtbn): Likewise.
7892         (iwmmxt_wmiawttn): Likewise.
7893         (iwmmxt_wmerge): Likewise.
7894         (iwmmxt_tandc<mode>3): Likewise.
7895         (iwmmxt_torc<mode>3): Likewise.
7896         (iwmmxt_torvsc<mode>3): Likewise.
7897         (iwmmxt_textrc<mode>3): Likewise.
7898         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
7899         (wmmxt_pack): Likewise.
7900         (wmmxt_mult_c1): Likewise.
7901         (wmmxt_mult_c2): Likewise.
7902         (wmmxt_alu_c1): Likewise.
7903         (wmmxt_alu_c2): Likewise.
7904         (wmmxt_alu_c3): Likewise.
7905         (wmmxt_transfer_c1): Likewise.
7906         (wmmxt_transfer_c2): Likewise.
7907         (wmmxt_transfer_c3): Likewise.
7908         (marvell_f_iwmmxt_wstr): Likewise.
7909         (marvell_f_iwmmxt_wldr): Likewise.
7911 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
7913         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
7915 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7917         Revert:
7918         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7919         * lra-constraints.c (need_for_split_p): Check call used hard regs
7920         living through calls.
7922         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
7923         call used regs for call insn.
7925 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
7927         PR target/57736
7928         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
7929         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
7931 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7933         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
7935 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7937         * lra-constraints.c (need_for_split_p): Check call used hard regs
7938         living through calls.
7940 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
7942         PR target/57744
7943         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
7944         to tie with any other modes.  Eliminate Altivec vector mode tests,
7945         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
7946         code, to return 0 if testing MODE2 for a condition, if we've
7947         already tested MODE1 for the same condition.
7949 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7951         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
7952         layout.
7954 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7956         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7957         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
7959 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7961         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
7962         Define.
7963         (aarch64_symbolic_constant_p): Remove.
7964         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
7965         static.  Fix line length and white space.
7966         (aarch64_symbolic_constant_p): Remove.
7967         * config/aarch64/predicates.md (aarch64_valid_symref):
7968         Use aarch64_classify_symbol_expression.
7970 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7972         * config/arm/constraints.md (Ts): New constraint.
7973         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
7974         16-bit encodings.
7975         (compare_scc): Use "Ts" constraint for operand 0.
7976         (ior_scc_scc): Likewise.
7977         (and_scc_scc): Likewise.
7978         (and_scc_scc_nodom): Likewise.
7979         (ior_scc_scc_cmp): Likewise for operand 7.
7980         (and_scc_scc_cmp): Likewise.
7981         * config/arm/thumb2.md (thumb2_movsi_insn):
7982         Add alternatives for 16-bit encodings.
7983         (thumb2_movhi_insn): Likewise.
7984         (thumb2_movsicc_insn): Likewise.
7985         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
7986         (thumb2_negscc): Use "Ts" constraint.
7987         Move mvn instruction outside cond_exec block.
7988         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
7989         for 16-bit encodings.
7991 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7993         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
7994         encoding.
7995         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
7996         (mulsi3subsi): Likewise.
7997         (mulsidi3adddi): Likewise.
7998         (mulsidi3_v6): Likewise.
7999         (umulsidi3_v6): Likewise.
8000         (umulsidi3adddi_v6): Likewise.
8001         (smulsi3_highpart_v6): Likewise.
8002         (umulsi3_highpart_v6): Likewise.
8003         (mulhisi3tb): Likewise.
8004         (mulhisi3bt): Likewise.
8005         (mulhisi3tt): Likewise.
8006         (maddhisi4): Likewise.
8007         (maddhisi4tb): Likewise.
8008         (maddhisi4tt): Likewise.
8009         (maddhidi4): Likewise.
8010         (maddhidi4tb): Likewise.
8011         (maddhidi4tt): Likewise.
8012         (zeroextractsi_compare0_scratch): Likewise.
8013         (insv_zero): Likewise.
8014         (insv_t2): Likewise.
8015         (anddi_notzesidi_di): Likewise.
8016         (anddi_notsesidi_di): Likewise.
8017         (andsi_notsi_si): Likewise.
8018         (iordi_zesidi_di): Likewise.
8019         (xordi_zesidi_di): Likewise.
8020         (andsi_iorsi3_notsi): Likewise.
8021         (smax_0): Likewise.
8022         (smax_m1): Likewise.
8023         (smin_0): Likewise.
8024         (not_shiftsi): Likewise.
8025         (unaligned_loadsi): Likewise.
8026         (unaligned_loadhis): Likewise.
8027         (unaligned_loadhiu): Likewise.
8028         (unaligned_storesi): Likewise.
8029         (unaligned_storehi): Likewise.
8030         (extv_reg): Likewise.
8031         (extzv_t2): Likewise.
8032         (divsi3): Likewise.
8033         (udivsi3): Likewise.
8034         (arm_zero_extendhisi2addsi): Likewise.
8035         (arm_zero_extendqisi2addsi): Likewise.
8036         (compareqi_eq0): Likewise.
8037         (arm_extendhisi2_v6): Likewise.
8038         (arm_extendqisi2addsi): Likewise.
8039         (arm_movt): Likewise.
8040         (thumb2_ldrd): Likewise.
8041         (thumb2_ldrd_base): Likewise.
8042         (thumb2_ldrd_base_neg): Likewise.
8043         (thumb2_strd): Likewise.
8044         (thumb2_strd_base): Likewise.
8045         (thumb2_strd_base_neg): Likewise.
8046         (arm_negsi2): Add alternative for 16-bit encoding.
8047         (arm_one_cmplsi2): Likewise.
8049 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8051         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8052         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8053         (movdfcc): Likewise.
8054         * config/arm/vfp.md (*thumb2_movsf_vfp):
8055         Disable predication for arm_restrict_it.
8056         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8057         (*thumb2_movdfcc_vfp): Likewise.
8058         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8059         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8060         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8061         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8062         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8063         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8064         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8065         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8066         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8067         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8068         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8069         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8070         Disable predication for arm_restrict_it.
8072 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8074         * config/i386/bmiintrin.h (_bextr_u32): New.
8075         (_bextr_u64): Ditto.
8077 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8079         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8080         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8081         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8082         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8083         * config/mips/n32-elf.h: ...this new file.
8085 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8087         PR target/57224
8088         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8089         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8091 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8093         * config/mips/mips-tables.opt: Regenerate.
8094         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8095         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8096         * doc/invoke.texi: Add -m14kc.
8098 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8100         PR target/57623
8101         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8102         constraints of operand 1 and 2.
8104         PR target/57623
8105         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8106         to match RTL canonicalization.  Swap predicates and
8107         constraints of operand 1 and 2.
8109 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8111         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8112         Process OP_INOUT regs for splitting too.
8114 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8116         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8117         decl before the loop, initialize to NULL.
8118         (vectorizable_load): Initialize ptr_incr to NULL.
8120 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8122         PR lto/57208
8123         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8124         * ipa-ref.c (ipa_maybe_record_reference): New function.
8125         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8126         * ipa-cp.c (create_specialized_node): Record potential references from
8127         aggvals.
8128         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8130 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8132         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8133         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8134         'mode' to force_reg.
8135         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8136         (aarch64_expand_mov_immediate): Likewise.
8138 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8140         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8141         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8143 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8145         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8146         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8147         (struct machine_function): Add tbegin_p.
8148         (s390_canonicalize_comparison): Fold CC mode compares to
8149         conditional jump if possible.
8150         (s390_emit_jump): Return the emitted jump.
8151         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8152         Handle CCRAWmode compares.
8153         (s390_option_override): Default to -mhtm if available.
8154         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8155         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8156         FPRs instead of df_regs_ever_live_p.
8157         (s390_optimize_nonescaping_tx): New function.
8158         (s390_init_frame_layout): Extend clobbered_regs array to cover
8159         FPRs as well.
8160         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8161         (s390_expand_tbegin): New function.
8162         (enum s390_builtin): New enum definition.
8163         (code_for_builtin): New array definition.
8164         (s390_init_builtins): New function.
8165         (s390_expand_builtin): New function.
8166         (TARGET_INIT_BUILTINS): Define.
8167         (TARGET_EXPAND_BUILTIN): Define.
8168         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8169         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8170         (s390_alc_comparison): Likewise.
8171         * config/s390/s390-modes.def: Add CCRAWmode.
8172         * config/s390/s390.h (processor_flags): Add PF_TX.
8173         (TARGET_CPU_HTM): Define macro.
8174         (TARGET_HTM): Define macro.
8175         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8176         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8177         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8178         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8179         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8180         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8181         ("*cc_to_int"): Move up.
8182         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8183         constants other than 0.
8184         ("*ccraw_to_int"): New insn and splitter definition.
8185         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8186         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8187         ("tx_assist"): New expander.
8188         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8189         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8190         * config/s390/s390.opt: Add -mhtm option.
8191         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8192         * config/s390/htmxlintrin.h: New file.
8193         * config/s390/htmintrin.h: New file.
8194         * config/s390/s390intrin.h: New file.
8195         * doc/extend.texi: Document htm builtins.
8196         * config.gcc: Add the new header files to extra_headers.
8198 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8200         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8201         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8203 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8204             Pat Haugen  <pthaugen@us.ibm.com>
8205             Peter Bergner  <bergner@vnet.ibm.com>
8207         * config/rs6000/power8.md: New.
8208         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8209         setting for power8 entry.
8210         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8211         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8212         test for Power4/Power5 only.
8213         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8214         support.
8215         (force_new_group): Adjust comment.
8216         * config/rs6000/rs6000.md: Include power8.md.
8218 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8220         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8221         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8222         declaration.
8223         (tune_params): Update comment.
8224         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8225         (arm_max_conditional_execute): New function.
8226         (thumb2_final_prescan_insn): Use max_insn_skipped and
8227         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8229 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8231         PR tree-optimization/57705
8232         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8233         SSA_NAME step, provided that it is not defined inside the loop.
8234         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8235         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8237 2013-06-25  Martin Jambor  <mjambor@suse.cz>
8239         PR middle-end/57670
8240         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8241         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8242         calls in the dump.
8243         (ipa_note_param_call): Initialize member_ptr flag.
8244         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8245         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8246         (ipa_write_indirect_edge_info): Stream member_ptr flag.
8247         (ipa_read_indirect_edge_info): Likewise.
8249 2013-06-25  Richard Biener  <rguenther@suse.de>
8251         PR middle-end/56977
8252         * passes.c (init_optimization_passes): Move pass_fold_builtins
8253         and pass_dce earlier with -Og.
8255 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
8257         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8258         <BIT_FIELD_REF>: Remove trailing TAB.
8259         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8260         remove blank line.
8262 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8264         PR tree-optimization/57358
8265         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8266         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8267         (ipa_analyze_params_uses): Generate pessimistic info when true.
8269 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8271         PR tree-optimization/57539
8272         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8273         global.inlined_to of the new node to it.  All callers changed.
8274         * ipa-inline-transform.c (clone_inlined_nodes): New variable
8275         inlining_into, pass it to cgraph_clone_node.
8276         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8277         ipa_free_edge_args_substructures.
8278         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8279         rdesc linked list.  Do not assert rdesc edges have inlined caller.
8280         Assert we have found an rdesc in the rdesc list.
8282 2013-06-24  Richard Biener  <rguenther@suse.de>
8284         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8285         (pointer_set_lookup): Declare.
8286         (class pointer_map): New template class implementing a
8287         generic pointer to T map.
8288         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8289         pointer_map<T>::contains, pointer_map<T>::insert,
8290         pointer_map<T>::traverse): New functions.
8291         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8292         (pointer_set_lookup): New function.
8293         (pointer_set_contains): Use pointer_set_lookup.
8294         (pointer_set_insert): Likewise.
8295         (insert_aux): Remove.
8296         (struct pointer_map_t): Embed a pointer_set_t.
8297         (pointer_map_create): Adjust.
8298         (pointer_map_destroy): Likewise.
8299         (pointer_map_contains): Likewise.
8300         (pointer_map_insert): Likewise.
8301         (pointer_map_traverse): Likewise.
8302         * tree-streamer.h (struct streamer_tree_cache_d): Use a
8303         pointer_map<unsigned> instead of a pointer_map_t.
8304         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8305         (streamer_tree_cache_lookup): Likewise.
8306         (streamer_tree_cache_create): Likewise.
8307         (streamer_tree_cache_delete): Likewise.
8308         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8309         pointer_map<unsigned> instead of a pointer_map_t.
8310         (lto_init_tree_ref_encoder): Adjust.
8311         (lto_destroy_tree_ref_encoder): Likewise.
8312         * lto-section-out.c (lto_output_decl_index): Likewise.
8313         (lto_record_function_out_decl_state): Likewise.
8314         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8316 2013-06-24  Richard Biener  <rguenther@suse.de>
8318         PR tree-optimization/57488
8319         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8321 2013-06-24  Alan Modra  <amodra@gmail.com>
8323         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8324         (gen_easy_altivec_constant): Likewise.
8325         * config/rs6000/predicates.md (easy_vector_constant_add_self,
8326         easy_vector_constant_msb): Likewise.
8328 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
8330         PR target/57688
8331         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8332         add missing return true.
8334 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
8336         PR target/52483
8337         * config/sh/predicates.md (general_extend_operand): Invoke
8338         general_movsrc_operand for memory operands.
8339         (general_movsrc_operand): Allow reg+reg addressing, do not use
8340         general_operand for memory operands.
8342 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
8344         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
8345         when current target options does not apply.
8346         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8347         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8348         * config/i386/bmiintrin.h: Pass appropriate target
8349         attributes to header.
8350         * config/i386/mmintrin.h: Ditto.
8351         * config/i386/nmmintrin.h: Ditto.
8352         * config/i386/avx2intrin.h: Ditto.
8353         * config/i386/fxsrintrin.h: Ditto.
8354         * config/i386/tbmintrin.h: Ditto.
8355         * config/i386/xsaveintrin.h: Ditto.
8356         * config/i386/f16cintrin.h: Ditto.
8357         * config/i386/xtestintrin.h: Ditto.
8358         * config/i386/xsaveoptintrin.h: Ditto.
8359         * config/i386/bmi2intrin.h: Ditto.
8360         * config/i386/lzcntintrin.h: Ditto.
8361         * config/i386/smmintrin.h: Ditto.
8362         * config/i386/wmmintrin.h: Ditto.
8363         * config/i386/x86intrin.h: Remove all header include guards.
8364         * config/i386/prfchwintrin.h: Ditto.
8365         * config/i386/pmmintrin.h: Ditto.
8366         * config/i386/tmmintrin.h: Ditto.
8367         * config/i386/xmmintrin.h: Ditto.
8368         * config/i386/popcntintrin.h: Ditto.
8369         * config/i386/rdseedintrin.h: Ditto.
8370         * config/i386/ammintrin.h: Ditto.
8371         * config/i386/emmintrin.h: Ditto.
8372         * config/i386/immintrin.h: Remove all header include guards.
8373         * config/i386/fma4intrin.h: Ditto.
8374         * config/i386/lwpintrin.h: Ditto.
8375         * config/i386/xopintrin.h: Ditto.
8376         * config/i386/ia32intrin.h: Ditto.
8377         * config/i386/avxintrin.h: Ditto.
8378         * config/i386/rtmintrin.h: Ditto.
8379         * config/i386/fmaintrin.h: Ditto.
8380         * config/i386/mm3dnow.h: Ditto.
8382 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
8384         * common/config/i386/i386-common.c: Handle LZCNT.
8386 2013-06-22  Andi Kleen  <ak@linux.intel.com>
8388         * doc/extend.texi: Use __atomic_store_n instead of
8389         __atomic_store in HLE example.
8391 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
8393         * config/sh/sh.c: Remove <cstdlib> workaround.
8395 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8397         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
8399 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8401         * doc/extend.texi: Document that __atomic_clear and
8402           __atomic_test_and_set should only be used with bool.
8404 2013-06-20  Jan Hubicka  <jh@suse.cz>
8406         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8407         types_same_for_odr.
8408         * tree.c (decls_same_for_odr): New function.
8409         (same_for_edr): New function.
8410         (types_same_for_odr): New function.
8411         (get_binfo_at_offset): Use it.
8412         * tree.h (types_same_for_odr): Declare.
8414 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
8415             Jason Merrill  <jason@redhat.com>
8417         * system.h: Include <cstdlib> as well as <stdlib.h>.
8419 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
8421         PR target/57655
8422         * config/i386/i386.c (construct_container): Report error if
8423         long double is used with disabled x87 float returns.
8425 2013-06-20  Jan Hubicka  <jh@suse.cz>
8427         * lto-cgraph.c (input_symtab): Do not set cgraph state.
8429 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
8431         PR rtl-optimization/57425
8432         PR rtl-optimization/57569
8433         * alias.c (write_dependence_p): Remove parameters mem_mode and
8434         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
8435         Changed all callers.
8436         (canon_anti_dependence): Get comments and semantics in sync.
8437         Add parameter mem_canonicalized.  Changed all callers.
8438         * rtl.h (canon_anti_dependence): Update prototype.
8440 2013-06-20  Richard Biener  <rguenther@suse.de>
8442         * data-streamer-in.c (streamer_read_uhwi): Optimize single
8443         byte case, inline streamer_read_uchar and defer section
8444         overrun check.
8446 2013-06-20  Richard Biener  <rguenther@suse.de>
8448         PR tree-optimization/57584
8449         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8450         SSA names into the expanded expression that take part in
8451         abnormal coalescing.
8453 2013-06-19  Sharad Singhai  <singhai@google.com>
8455         * gcov.c (print_usage): Handle new option.
8456         (process_args): Ditto.
8457         (get_gcov_intermediate_filename): New function.
8458         (output_intermediate_file): New function.
8459         (output_gcov_file): New function
8460         (generate_results): Handle new option.
8461         (release_function): Relase demangled name.
8462         (read_graph_file): Handle demangled name.
8463         (output_lines): Ditto.
8464         * doc/gcov.texi: Document gcov intermediate format.
8466 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
8468         PR bootstrap/57604
8469         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8470         (lra_emit_add): Use the functions.  Add comment about Y as an
8471         address segment.
8473 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
8475         PR driver/57652
8476         * collect2.c (collect_atexit): New.
8477         (collect_exit): Delete.
8478         (main): Register collect_atexit with atexit.
8479         (collect_wait): Change collect_exit to exit.
8480         (do_wait): Same.
8481         * collect2.h (collect_exit): Delete.
8482         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8484 2013-06-19  Wei Mi  <wmi@google.com>
8486         PR rtl-optimization/57518
8487         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8488         if regno is used in paradoxical subreg.
8489         (update_equiv_regs): Check pdx_subregs[regno] before
8490         set a reg to be equivalent with a mem.
8492 2013-06-19  Matthias Klose  <doko@ubuntu.com>
8494         PR driver/57651
8495         * file-find.h (find_a_file): Add a mode parameter.
8496         * file-find.c (find_a_file): Likewise.
8497         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8498         with X_OK for the executables.
8499         * collect2.c (main): Call find_a_file with X_OK.
8501 2013-06-19  Steve Ellcey  <sellcey@mips.com>
8503         PR target/56942
8504         * config/mips/mips.md (casesi_internal_mips16_<mode>):
8505         Use NEXT_INSN instead of next_real_insn.
8507 2013-06-19  Jan Hubicka  <jh@suse.cz>
8509         * cgraph.h (const_value_known_p): Replace by ...
8510         (ctor_for_folding): .. this one.
8511         * cgraphunit.c (process_function_and_variable_attributes): Use it.
8512         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8513         * expr.c (expand_expr_real_1): Likewise.
8514         (string_constant): Likewise.
8515         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8516         * ipa.c (process_references): Likewise.
8517         (symtab_remove_unreachable_nodes): Likewise.
8518         * ipa-inline-analysis.c (param_change_prob): Likewise.
8519         * gimple-fold.c (canonicalize_constructor_val): Likewise.
8520         (get_base_constructor): Likwise.
8521         * varpool.c (varpool_remove_node): Likewise.
8522         (varpool_remove_initializer): LIkewise.
8523         (dump_varpool_node): LIkwise.
8524         (const_value_known_p): Rewrite to ...
8525         (ctor_for_folding): ... this one.
8527 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
8529         PR driver/57651
8530         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8531         PERSONALITY in $PATH derived prefixes.
8533 2013-06-19  Jeff Law  <law@redhat.com>
8535         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8536         in comment.
8538         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8539         (simplify_bitwise_binary): Use it to simpify certain binary ops on
8540         booleans.
8542 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
8544         * config/arm/vfp.md: Move VFP instruction classification documentation
8545         to ...
8546         * config/arm/arm.md: ... here.  Update instruction classification
8547         documentation.
8549 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
8551         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8552         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8553         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
8555 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
8557         PR target/57609
8558         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8559         with NEXT_INSN.  Use tablejump_p to check for jump table data
8560         insns.
8562 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
8564         PR c++/56544
8565         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8566         that now in C++ the value is correct per the C++ standards.
8568 2013-06-19  Richard Biener  <rguenther@suse.de>
8570         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8571         for global context.
8573 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8575         Revert:
8576         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8578         PR target/57609
8579         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8580         with next_active_insn.
8582 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
8584         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
8585         functions are inlined during failures to flag an error.
8586         * tree-inline.c (expand_call_inline): Allow the error to be flagged
8587         in early inline pass.
8589 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
8591         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
8592         in comments.
8594 2013-06-18  Julian Brown  <julian@codesourcery.com>
8596         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
8597         Permit virtual register pre-reload if !strict.
8598         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
8599         change.
8600         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
8601         prototype.
8602         * config/arm/neon.md (movmisalign<mode>): Use
8603         neon_perm_struct_or_reg_operand instead of
8604         neon_struct_or_register_operand.
8605         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
8606         neon_permissive_struct_operand instead of neon_struct_operand.
8607         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
8608         neon_vector_mem_operand.
8609         * config/arm/predicates.md (neon_struct_operand): Adjust call to
8610         neon_vector_mem_operand.
8611         (neon_permissive_struct_operand): New.
8612         (neon_struct_or_register_operand): Rename to...
8613         (neon_perm_struct_or_reg_operand): This. Adjust call to
8614         neon_vector_mem_operand.
8616 2013-06-18  Richard Biener  <rguenther@suse.de>
8618         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
8619         * lto-streamer.h: Include pointer-set.h.
8620         (struct lto_decl_slot): Remove.
8621         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
8622         Remove next_index entry.
8623         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8624         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8625         (lto_init_tree_ref_encoder): Adjust.
8626         (lto_destroy_tree_ref_encoder): Likewise.
8627         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8628         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8629         (lto_output_decl_index): Adjust.
8630         (lto_new_out_decl_state): Likewise.
8631         (lto_record_function_out_decl_state): Likewise.
8632         * lto-streamer-out.c (copy_function): Likewise.
8634 2013-06-18  Richard Biener  <rguenther@suse.de>
8636         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
8637         * cgraphunit.c: Include cfgloop.h.
8638         (init_lowered_empty_function): Initialize the loop tree.
8639         (assemble_thunk): Insert new BBs into loops.
8641 2013-06-18  Richard Biener  <rguenther@suse.de>
8643         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
8644         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
8645         the map from cache entry to cache index optional.
8646         (streamer_tree_cache_replace_tree): Adjust accordingly.
8647         (streamer_tree_cache_append): Likewise.
8648         (streamer_tree_cache_delete): Likewise.
8649         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
8650         streamer cache map from cache entry to cache index.
8651         * lto-streamer-out.c (create_output_block): Adjust.
8653 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
8655         * config/arm/arm.md (attribute "insn"): Move multiplication and
8656         division attributes to...
8657         (attribute "type"): ... here.  Remove mult.
8658         (attribute "mul32"): New attribute.
8659         (attribute "mul64"): Add umaal.
8660         (*arm_mulsi3): Update attributes.
8661         (*arm_mulsi3_v6): Likewise.
8662         (*thumb_mulsi3): Likewise.
8663         (*thumb_mulsi3_v6): Likewise.
8664         (*mulsi3_compare0): Likewise.
8665         (*mulsi3_compare0_v6): Likewise.
8666         (*mulsi_compare0_scratch): Likewise.
8667         (*mulsi_compare0_scratch_v6): Likewise.
8668         (*mulsi3addsi): Likewise.
8669         (*mulsi3addsi_v6): Likewise.
8670         (*mulsi3addsi_compare0): Likewise.
8671         (*mulsi3addsi_compare0_v6): Likewise.
8672         (*mulsi3addsi_compare0_scratch): Likewise.
8673         (*mulsi3addsi_compare0_scratch_v6): Likewise.
8674         (*mulsi3subsi): Likewise.
8675         (*mulsidi3adddi): Likewise.
8676         (*mulsi3addsi_v6): Likewise.
8677         (*mulsidi3adddi_v6): Likewise.
8678         (*mulsidi3_nov6): Likewise.
8679         (*mulsidi3_v6): Likewise.
8680         (*umulsidi3_nov6): Likewise.
8681         (*umulsidi3_v6): Likewise.
8682         (*umulsidi3adddi): Likewise.
8683         (*umulsidi3adddi_v6): Likewise.
8684         (*smulsi3_highpart_nov6): Likewise.
8685         (*smulsi3_highpart_v6): Likewise.
8686         (*umulsi3_highpart_nov6): Likewise.
8687         (*umulsi3_highpart_v6): Likewise.
8688         (mulhisi3): Likewise.
8689         (*mulhisi3tb): Likewise.
8690         (*mulhisi3bt): Likewise.
8691         (*mulhisi3tt): Likewise.
8692         (maddhisi4): Likewise.
8693         (*maddhisi4tb): Likewise.
8694         (*maddhisi4tt): Likewise.
8695         (maddhidi4): Likewise.
8696         (*maddhidi4tb): Likewise.
8697         (*maddhidi4tt): Likewise.
8698         (divsi3): Likewise.
8699         (udivsi3): Likewise.
8700         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
8701         (thumb2_mulsi_short_compare0): Likewise.
8702         (thumb2_mulsi_short_compare0_scratch): Likewise.
8703         * config/arm/arm1020e.md (1020mult1): Update attribute change.
8704         (1020mult2): Likewise.
8705         (1020mult3): Likewise.
8706         (1020mult4): Likewise.
8707         (1020mult5): Likewise.
8708         (1020mult6): Likewise.
8709         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
8710         change.
8711         (cortex_a15_mult64): Likewise.
8712         (cortex_a15_sdiv): Likewise.
8713         (cortex_a15_udiv): Likewise.
8714         * config/arm/arm1026ejs.md (mult1): Update attribute change.
8715         (mult2): Likewise.
8716         (mult3): Likewise.
8717         (mult4): Likewise.
8718         (mult5): Likewise.
8719         (mult6): Likewise.
8720         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
8721         (pj4_ir_div): Likewise.
8722         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
8723         (11_mult2): Likewise.
8724         (11_mult3): Likewise.
8725         (11_mult4): Likewise.
8726         (11_mult5): Likewise.
8727         (11_mult6): Likewise.
8728         (11_mult7): Likewise.
8729         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
8730         (cortex_a8_mla): Likewise.
8731         (cortex_a8_mull): Likewise.
8732         (cortex_a8_smulwy): Likewise.
8733         (cortex_a8_smlald): Likewise.
8734         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
8735         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
8736         (cortex_r4_mul_3): Likewise.
8737         (cortex_r4_mla_4): Likewise.
8738         (cortex_r4_mla_3): Likewise.
8739         (cortex_r4_smlald): Likewise.
8740         (cortex_r4_mull): Likewise.
8741         (cortex_r4_sdiv): Likewise.
8742         (cortex_r4_udiv): Likewise.
8743         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
8744         (cortex_a7_idiv): Likewise.
8745         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
8746         (9_mult2): Likewise.
8747         (9_mult3): Likewise.
8748         (9_mult4): Likewise.
8749         (9_mult5): Likewise.
8750         (9_mult6): Likewise.
8751         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
8752         (cortex_a53_sdiv): Likewise.
8753         (cortex_a53_udiv): Likewise.
8754         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
8755         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
8756         (mp626_mult2): Likewise.
8757         (mp626_mult3): Likewise.
8758         (mp626_mult4): Likewise.
8759         * config/arm/fa526.md (526_mult1): Update attribute change.
8760         (526_mult2): Likewise.
8761         * config/arm/arm-generic.md (mult): Update attribute change.
8762         (mult_ldsched_strongarm): Likewise.
8763         (mult_ldsched): Likewise.
8764         (multi_cycle): Likewise.
8765         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
8766         * config/arm/fa606te.md (606te_mult1): Update attribute change.
8767         (606te_mult2): Likewise.
8768         (606te_mult3): Likewise.
8769         (606te_mult4): Likewise.
8770         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
8771         (cortex_a9_mac16): Likewise.
8772         (cortex_a9_multiply): Likewise.
8773         (cortex_a9_mac): Likewise.
8774         (cortex_a9_multiply_long): Likewise.
8775         * config/arm/fa626te.md (626te_mult1): Update attribute change.
8776         (626te_mult2): Likewise.
8777         (626te_mult3): Likewise.
8778         (626te_mult4): Likewise.
8780 2013-06-18  Richard Biener  <rguenther@suse.de>
8782         PR lto/57334
8783         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
8785 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8787         PR target/57609
8788         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8789         with next_active_insn.
8791 2013-06-18  Alan Modra  <amodra@gmail.com>
8793         * config/rs6000/rs6000.h (enum data_align): New.
8794         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
8795         (DATA_ABI_ALIGNMENT): Define.
8796         (CONSTANT_ALIGNMENT): Correct comment.
8797         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
8798         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
8800 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
8802         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
8803         ATTRIBUTE_UNUSED marking.
8805 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
8807         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
8808         alternative and update.
8809         (aarch64_dup_lanedi): Delete.
8810         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
8811         * config/aarch64/aarch64-simd-builtins.def: Update.
8813 2013-06-17  Richard Biener  <rguenther@suse.de>
8815         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
8816         (lto_input_scc): Declare.
8817         (lto_input_tree_1): Likewise.
8818         (struct lto_stats_d): Add num_tree_bodies_output and
8819         num_pickle_refs_output.
8820         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
8821         (lto_read_tree_1): Split out from ...
8822         (lto_read_tree): ... this.
8823         (lto_input_scc): New function.
8824         (lto_input_tree_1): Split out from ...
8825         (lto_input_tree): ... this.  Handle LTO_tree_scc.
8826         (lto_data_in_create): Create the streamer cache without hashes.
8827         * lto-streamer-out.c (create_output_block): Create the streamer
8828         cache with hashes when not doing WPA.
8829         (lto_write_tree_1): Split out from ...
8830         (lto_write_tree): ... this.
8831         (get_symbol_initial_value): New function.
8832         (lto_output_tree_1): Split out from ...
8833         (lto_output_tree): ... this.  Write trees as series of SCCs
8834         using a DFS walk via DFS_write_tree.
8835         (struct sccs, struct scc_entry): New types.
8836         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
8837         (DFS_write_tree_body): New function.
8838         (DFS_write_tree): Likewise.
8839         (hash_tree): Likewise.
8840         (scc_entry_compare): Likewise.
8841         (hash_scc): Likewise.
8842         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
8843         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
8844         TREE_CHAIN as regular reference.
8845         (streamer_read_integer_cst): Remove.
8846         (streamer_get_pickled_tree): Adjust.
8847         * tree-streamer-out.c (streamer_write_chain): Disable streaming
8848         of DECL_EXTERNALs in BLOCK_VARS for now.
8849         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
8850         reference.
8851         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
8852         Add hash value argument and record that if hashes are recorded
8853         in the cache.
8854         (streamer_tree_cache_insert_1): Adjust.
8855         (streamer_tree_cache_insert): Likewise.
8856         (streamer_tree_cache_insert_at): Rename to ...
8857         (streamer_tree_cache_replace_tree): ... this and adjust.
8858         (streamer_tree_cache_append): Adjust.
8859         (record_common_node): Likewise.
8860         (streamer_tree_cache_create): Add argument whether to
8861         record hash values together with trees.
8862         (streamer_tree_cache_delete): Adjust.
8863         * tree-streamer.h (struct streamer_tree_cache_d): Add
8864         vector of hashes.
8865         (streamer_read_integer_cst): Remove.
8866         (streamer_tree_cache_insert): Adjust.
8867         (streamer_tree_cache_append): Likewise.
8868         (streamer_tree_cache_insert_at): Rename to ...
8869         (streamer_tree_cache_replace_tree): ... this and adjust.
8870         (streamer_tree_cache_create): Add argument whether to record hashes.
8871         (streamer_tree_cache_get): Rename to ...
8872         (streamer_tree_cache_get_tree): ... this.
8873         (streamer_tree_cache_get_hash): New function.
8874         * tree.c (cache_integer_cst): New function.
8875         * tree.h (cache_integer_cst): Declare.
8876         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
8877         * lto-symtab.c (lto_varpool_replace_node): Only release
8878         DECL_INITIAL of non-prevailing decls.
8879         * varpool.c (varpool_remove_initializer): Do not release
8880         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
8882 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8884         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
8885         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
8886         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
8887         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
8888         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
8889         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
8890         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
8891         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
8892         instead of TARGET_64BIT.
8893         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
8894         Require ISA_HAS_<D>DIV.
8896 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
8898         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
8899         (mips*-*-linux*): Move default with_llsc setting to where other
8900         defaults are set.
8901         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
8902         with_arch block.
8903         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
8904         Likewise.  Remove default with_tune setting.  Move default float
8905         setting to its own block.  Handle with_llsc in the same block as above.
8907 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
8909         PR rtl-optimization/57425
8910         PR rtl-optimization/57569
8911         * alias.c (write_dependence_p): Add new parameters mem_mode,
8912         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
8913         Changed all callers.
8914         (canon_anti_dependence): New function.
8915         * cse.c (check_dependence): Use canon_anti_dependence.
8916         * cselib.c (cselib_invalidate_mem): Likewise.
8917         * rtl.h (canon_anti_dependence): Declare.
8919 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8921         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
8922         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
8923         ".set mips3" for 64-bit targets.
8925 2013-06-15  Dehao Chen  <dehao@google.com>
8927         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
8928         * gimple-low.c (gimple_check_call_matching_types): Likewise.
8929         (gimple_check_call_args): Likewise.
8930         * value-prof.c (check_ic_target): Likewise.
8931         * ipa-inline.c (early_inliner): Likewise.
8932         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
8933         * cgraph.c (cgraph_create_edge_1): Likewise.
8934         (cgraph_make_edge_direct): Likewise.
8936 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
8938         PR target/57615
8939         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
8940         rs6000_output_move_128bit to handle emitting quad memory
8941         operations.  Set attribute length to 8 bytes.
8943 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
8945         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
8946         New pattern.
8947         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
8948         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
8949         (aarch64_<su>mlsl<mode>): Likewise.
8951 2013-06-14  Mike Stump  <mikestump@comcast.net>
8953         * Makefile.in (TARGET_H): Add insn-codes.h.
8955 2013-06-14  Alan Modra  <amodra@gmail.com>
8957         PR middle-end/57134
8958         PR middle-end/57586
8959         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
8960         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
8961         bitfield expansion when EXPAND_MEMORY.
8962         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
8964 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
8966         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
8967         test for clearing quad memory on 32-bit later.
8969 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8971         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
8972         (fold_negate_expr): Likewise.
8973         (fold_real_zero_addition_p): Handle vectors.
8974         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
8976 2013-06-14  Alan Modra  <amodra@gmail.com>
8978         * varasm.c (force_const_mem): Revert 2013-06-07 change.
8980 2013-06-13  Jan Hubicka  <jh@suse.cz>
8982         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
8983         Local comdats are not externally visible.
8984         * symtab.c (dump_symtab_base): Dump externally visible.
8985         (verify_symtab_base): Verify back links in the symtab hash.
8987 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8989         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
8990         CONVERT_EXPR as equal nodes.
8992 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8994         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
8996 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8998         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
8999         Generalize to complex and vector.
9000         * tree.c (build_all_ones_cst): New function.
9001         * tree.h (build_all_ones_cst): Declare it.
9003 2013-06-13  Alan Modra  <amodra@gmail.com>
9005         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9006         * config/rs6000/rs6000.md (signbittf2): New insn.
9007         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9008         (abstf2_internal, cmptf_internal2): Likewise.
9009         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9011 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
9012             Pat Haugen  <pthaugen@us.ibm.com>
9013             Peter Bergner  <bergner@vnet.ibm.com>
9015         * config/rs6000/rs6000.c (emit_load_locked): Add support for
9016         power8 byte, half-word, and quad-word atomic instructions.
9017         (emit_store_conditional): Likewise.
9018         (rs6000_expand_atomic_compare_and_swap): Likewise.
9019         (rs6000_expand_atomic_op): Likewise.
9021         * config/rs6000/sync.md (larx): Add new modes for power8.
9022         (stcx): Likewise.
9023         (AINT): New mode iterator to include TImode as well as normal
9024         integer modes on power8.
9025         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9026         that VSX registers are not considered.  Use AINT mode iterator
9027         instead of INT1 to allow inclusion of quad word atomic operations
9028         on power8.
9029         (load_locked<mode>): Likewise.
9030         (store_conditional<mode>): Likewise.
9031         (atomic_compare_and_swap<mode>): Likewise.
9032         (atomic_exchange<mode>): Likewise.
9033         (atomic_nand<mode>): Likewise.
9034         (atomic_fetch_<fetchop_name><mode>): Likewise.
9035         (atomic_nand_fetch<mode>): Likewise.
9036         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9037         each type.
9038         (ATOMIC): On power8, add QImode, HImode modes.
9039         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9040         modes that promote to SImode.
9041         (load_lockedti): Convert TImode arguments to PTImode, so that we
9042         get a guaranteed even/odd register pair.
9043         (load_lockedpti): Likewise.
9044         (store_conditionalti): Likewise.
9045         (store_conditionalpti): Likewise.
9047         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9048         atomic load/store instructions.
9049         (HSI): Likewise.
9051 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9053         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9054         loads.
9055         (insn_count): New attribute, with most cases extracted from...
9056         (length): ...here.  Redefine most cases in terms of insn_count.
9057         (single_insn): Delete.
9058         (can_delay): Use insn_count to check for single instructions.
9059         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9060         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9061         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9062         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9063         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9064         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9065         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9066         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9067         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9068         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9069         rather than "length".
9070         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9071         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9072         Use "insn_count" rather than "length".
9073         * config/mips/mips-dsp.md
9074         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9075         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9076         length attributes.
9078 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9080         PR tree-optimization/57361
9081         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9083 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9085         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9086         to split.
9087         (aarch64_simd_combine<mode>): New instruction expansion.
9088         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9089         function prototype.
9090         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9091         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9093 2013-06-12  Jan Hubicka  <jh@suse.cz>
9095         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9096         decl has when in streaming stage.
9097         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9098         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9100 2013-06-12  Roland Stigge  <stigge@antcom.de>
9102         PR target/57578
9103         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9105 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9107         PR tree-optimization/57537
9108         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9109         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9111 2013-06-12  Richard Biener  <rguenther@suse.de>
9113         * data-streamer.h (streamer_write_char_stream): CSE
9114         obs->current_pointer.
9115         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9116         streamer_write_char_stream manually and optimize the resulting loop.
9117         (streamer_write_hwi_stream): Likewise.
9119 2013-06-12  Jan Hubicka  <jh@suse.cz>
9121         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9122         * cgraph.h (varpool_create_empty_node): Declare.
9123         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9124         duplicated nodes.
9125         * symtab.c (symtab_unregister_node): Be lax about missin entries
9126         in node hash.
9127         (symtab_get_node): Update comment.
9128         * varpool.c (varpool_create_empty_node): Break out from ...
9129         (varpool_node_for_decl): ... here.
9130         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9132 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9134         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9135         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9136         part.  Use straight-line flow at the end.
9137         <COMPONENT_REF>: Remove superfluous else.
9138         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9140 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9142         PR target/56564
9143         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9144         target hook even for !TREE_PUBLIC decls.  If no resolution info
9145         is available, return false for common and external decls.
9147 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9149         * config/rl78/constraints.md (U): New constraint.
9150         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9151         valloc attribute.
9153 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9155         PR target/57589
9156         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9157         to allow returning address to AT_PLATFORM name.
9159 2013-06-11  Jan Hubicka  <jh@suse.cz>
9161         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9162         * cgraph.h (symtab_node_base): Add weakref flag.
9163         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9164         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9165         (output_weakrefs): Use weakref flag.
9166         * fold-const.c (simple_operand_p): Handle WEAK.
9167         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9168         * ipa.c (varpool_externally_visible_p): Drop weakref.
9169         (function_and_variable_visibility): Update comment; fix weakref
9170         sanity checks; do not clear DECL_WEAK on them.
9171         * lto-cgraph.c (lto_output_node): update.
9172         (lto_output_varpool_node): Update.
9173         (input_overwrite_node): Update.
9174         (input_node): Update.
9175         (input_varpool_node): Update.
9176         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9177         (lto_symtab_merge_symbols): Add sanity check.
9178         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9179         * passes.c (rest_of_decl_compilation): Set static flag, too.
9180         * symtab.c (dump_symtab_base): Dump weakref.
9181         (verify_symtab_base): Sanity check weakrefs.
9182         (symtab_make_decl_local): Remove duplicated code.
9183         (symtab_alias_ultimate_target): Simplify.
9184         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9186 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9188         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9189         sequence_vect == NULL.
9191 2013-06-11  DJ Delorie  <dj@redhat.com>
9193         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9194         (rl78_unwind_word_mode): New.
9196 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9198         * final.c (debug_prefix_maps): Make static.
9200 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9202         * function.c (initial_trampoline): Remove stray copy.
9204 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9206         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9208 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9210         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9211         within bounds at the beginning of the function.
9213 2013-06-11  Alan Modra  <amodra@gmail.com>
9215         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9216         reporting.
9217         (get_named_section): Don't NULL !DECL_P decl.
9219 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9221         * doc/invoke.texi (core-avx2): Document.
9222         (slm): Likewise.
9223         (atom): Updated with MOVBE.
9225 2013-06-11  Richard Biener  <rguenther@suse.de>
9227         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9229 2013-06-11  Anton Blanchard  <anton@samba.org>
9231         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9232         correct shift value in little-endian mode.
9234 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9236         PR target/56564
9237         * varasm.c (get_variable_align): Move #endif to the right place.
9239 2013-06-10  Cary Coutant  <ccoutant@google.com>
9241         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9242         for hash so that hash table traversal order is deterministic.
9244 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
9245             Pat Haugen  <pthaugen@us.ibm.com>
9246             Peter Bergner  <bergner@vnet.ibm.com>
9248         * config/rs6000/vector.md (GPR move splitter): Do not split moves
9249         of vectors in GPRS if they are direct moves or quad word load or
9250         store moves.
9252         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9253         declaration.
9254         (direct_move_p): Likewise.
9255         (quad_load_store_p): Likewise.
9257         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9258         classes into bins based on the physical register type.
9259         (reg_class_to_reg_type): Likewise.
9260         (IS_STD_REG_TYPE): Likewise.
9261         (IS_FP_VECT_REG_TYPE): Likewise.
9262         (reload_fpr_gpr): Arrays to determine what insn to use if we can
9263         use direct move instructions.
9264         (reload_gpr_vsx): Likewise.
9265         (reload_vsx_gpr): Likewise.
9266         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9267         information that is a simplification of register classes.  Also
9268         precalculate direct move reload helpers.
9269         (direct_move_p): New function to return true if the operation can
9270         be done as a direct move instruciton.
9271         (quad_load_store_p): New function to return true if the operation
9272         is a quad memory operation.
9273         (rs6000_legitimize_address): If quad memory, only allow register
9274         indirect for TImode addresses.
9275         (rs6000_legitimate_address_p): Likewise.
9276         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9277         (rs6000_reload_register_type): Likewise.
9278         (register_to_reg_type): Return register type.
9279         (rs6000_secondary_reload_simple_move): New helper function for
9280         secondary reload and secondary memory needed to identify anything
9281         that is a simple move, and does not need reloading.
9282         (rs6000_secondary_reload_direct_move): New helper function for
9283         secondary reload to identify cases that can be done with several
9284         instructions via the direct move instructions.
9285         (rs6000_secondary_reload_move): New helper function for secondary
9286         reload to identify moves between register types that can be done.
9287         (rs6000_secondary_reload): Add support for quad memory operations
9288         and for direct move.
9289         (rs6000_secondary_memory_needed): Likewise.
9290         (rs6000_debug_secondary_memory_needed): Change argument names.
9291         (rs6000_output_move_128bit): New function to return the move to
9292         use for 128-bit moves, including knowing about the various
9293         limitations of quad memory operations.
9295         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9296         memory operations.  call rs6000_output_move_128bit for the actual
9297         instruciton(s) to generate.
9298         (vsx_movti_64bit): Likewise.
9300         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9301         (UNSPEC_P8V_MTVSRWZ): Likewise.
9302         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9303         (UNSPEC_P8V_MTVSRD): Likewise.
9304         (UNSPEC_P8V_XXPERMDI): Likewise.
9305         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9306         (UNSPEC_FUSION_GPR): Likewise.
9307         (FMOVE128_GPR): New iterator for direct move.
9308         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9309         (f32_sv): Likewise.
9310         (f32_dm): Likewise.
9311         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9312         loads and direct move instructions.
9313         (zero_extendsidi2_lfiwzx): Likewise.
9314         (extendsidi2_lfiwax): Likewise.
9315         (extendsidi2_nocell): Likewise.
9316         (floatsi<mode>2_lfiwax): Likewise.
9317         (lfiwax): Likewise.
9318         (floatunssi<mode>2_lfiwzx): Likewise.
9319         (lfiwzx): Likewise.
9320         (fix_trunc<mode>_stfiwx): Likewise.
9321         (fixuns_trunc<mode>_stfiwx): Likewise.
9322         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9323         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9324         (parity<mode>2_cmpb): Set length/type attr.
9325         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9326         for 'mr.' to fast_compare.
9327         (bpermd_<mode>): Change type attr to popcnt.
9328         (p8_fmrgow_<mode>): New insns for power8 direct move support.
9329         (p8_mtvsrwz_1): Likewise.
9330         (p8_mtvsrwz_2): Likewise.
9331         (reload_fpr_from_gpr<mode>): Likewise.
9332         (p8_mtvsrd_1): Likewise.
9333         (p8_mtvsrd_2): Likewise.
9334         (p8_xxpermdi_<mode>): Likewise.
9335         (reload_vsx_from_gpr<mode>): Likewise.
9336         (reload_vsx_from_gprsf): Likewise.
9337         (p8_mfvsrd_3_<mode>): LIkewise.
9338         (reload_gpr_from_vsx<mode>): Likewise.
9339         (reload_gpr_from_vsxsf): Likewise.
9340         (p8_mfvsrd_4_disf): Likewise.
9341         (multi-word GPR splits): Do not split direct moves or quad memory
9342         operations.
9344 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
9346         * tree-into-ssa.c (interesting_blocks): Make static.
9348 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
9350         PR target/56564
9351         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9352         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9353         Use DATA_ABI_ALIGNMENT for that case instead if defined.
9354         (get_variable_align): New function.
9355         (get_variable_section, emit_bss, emit_common,
9356         assemble_variable_contents, place_block_symbol): Use
9357         get_variable_align instead of DECL_ALIGN.
9358         (assemble_noswitch_variable): Add align argument, use it
9359         instead of DECL_ALIGN.
9360         (assemble_variable): Adjust caller.  Use get_variable_align
9361         instead of DECL_ALIGN.
9362         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9363         caller.
9364         (DATA_ABI_ALIGNMENT): Define.
9365         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9366         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
9367         opt is false, only return the psABI mandated alignment increase.
9368         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9369         (DATA_ABI_ALIGNMENT): ... this.
9370         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9371         (DATA_ABI_ALIGNMENT): ... this.
9372         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9373         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9374         (DATA_ABI_ALIGNMENT): ... this.
9375         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9376         * doc/tm.texi: Regenerated.
9378 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
9380         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9381         cmp_code to construct REG_EQUAL note.
9383 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
9385         PR target/57568
9386         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9387         that operands[2] doesn't overlap with operands[0].
9389 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
9390             Jan Hubicka  <jh@suse.cz>
9392         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9393         hack to mark symbols as used.
9395 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
9397         PR rtl-optimization/57559
9398         * lra-constraints.c (process_alt_operands): Don't discourage
9399         memory with known offset for offsetable memory constraint.
9400         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9402 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
9404         * varasm.c (struct oc_local_state): Reorder fields.
9405         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9406         and adjust accordingly.
9407         (output_constructor): Reorder initialization code and adjust call to
9408         output_constructor_bitfield.
9410 2013-06-07  Jan Hubicka  <jh@suse.cz>
9412         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9414 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
9416         * tree-object-size.c (unknown): Make const.
9418 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9420         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9421         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9422         for last alternative in the cpu_facility attribute.
9424 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9426         PR target/56315
9427         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9428         (xordi3): Change operand 2 constraint to arm_xordi_operand.
9429         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9430         * config/arm/constraints.md (Dg): New constraint.
9431         * config/arm/neon.md (xordi3_neon): Remove.
9432         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9433         * config/arm/predicates.md (arm_xordi_operand): New predicate.
9435 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9437         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9438         Clean up alternatives.
9440 2013-06-07  Alan Modra  <amodra@gmail.com>
9442         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9443         va_list_gpr_size.
9445 2013-06-07  Alan Modra  <amodra@gmail.com>
9447         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9449 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9451         * config/arm/constraints.md (Df): New constraint.
9452         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9453         Correct length attribute for last two alternatives.
9455 2013-06-07  Alan Modra  <amodra@gmail.com>
9457         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9458         override user -mfp-in-toc.
9459         (offsettable_ok_by_alignment): Consider just the current access
9460         rather than the whole object, unless BLKmode.  Handle
9461         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9462         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9463         for -mcmodel=medium.
9464         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9465         override user -mfp-in-toc or -msum-in-toc.  Default to
9466         -mno-fp-in-toc for -mcmodel=medium.
9468 2013-06-06  DJ Delorie  <dj@redhat.com>
9470         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9471         TARGET_VALID_POINTER_MODE.
9473 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
9474             Pat Haugen  <pthaugen@us.ibm.com>
9475             Peter Bergner  <bergner@vnet.ibm.com>
9477         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9478         Document new power8 builtins.
9480         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9481         condition code register, to allow 128-bit logical operations to be
9482         done in the VSX or GPR registers.
9483         (nor<mode>3): Use the canonical form for nor.
9484         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9485         vclz*, and vpopcnt* vector instructions.
9486         (nand<mode>3): Likewise.
9487         (orc<mode>3): Likewise.
9488         (clz<mode>2): LIkewise.
9489         (popcount<mode>2): Likewise.
9491         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9492         that only the GPRs are recognized.
9494         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9495         support for new power8 builtins.
9497         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9498         builtin functions.
9499         (xscvdpspn): Likewise.
9500         (vclz): Likewise.
9501         (vclzb): Likewise.
9502         (vclzh): Likewise.
9503         (vclzw): Likewise.
9504         (vclzd): Likewise.
9505         (vpopcnt): Likewise.
9506         (vpopcntb): Likewise.
9507         (vpopcnth): Likewise.
9508         (vpopcntw): Likewise.
9509         (vpopcntd): Likewise.
9510         (vgbbd): Likewise.
9511         (vmrgew): Likewise.
9512         (vmrgow): Likewise.
9513         (eqv): Likewise.
9514         (eqv_v16qi3): Likewise.
9515         (eqv_v8hi3): Likewise.
9516         (eqv_v4si3): Likewise.
9517         (eqv_v2di3): Likewise.
9518         (eqv_v4sf3): Likewise.
9519         (eqv_v2df3): Likewise.
9520         (nand): Likewise.
9521         (nand_v16qi3): Likewise.
9522         (nand_v8hi3): Likewise.
9523         (nand_v4si3): Likewise.
9524         (nand_v2di3): Likewise.
9525         (nand_v4sf3): Likewise.
9526         (nand_v2df3): Likewise.
9527         (orc): Likewise.
9528         (orc_v16qi3): Likewise.
9529         (orc_v8hi3): Likewise.
9530         (orc_v4si3): Likewise.
9531         (orc_v2di3): Likewise.
9532         (orc_v4sf3): Likewise.
9533         (orc_v2df3): Likewise.
9535         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9536         allow power8 quad mode in 64-bit.
9537         (rs6000_builtin_vectorized_function): Add support to vectorize
9538         ISA 2.07 count leading zeros, population count builtins.
9539         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9540         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9541         (builtin_function_type): Add vgbbd builtin function which takes an
9542         unsigned argument.
9543         (altivec_expand_vec_perm_const): Add support for new power8 merge
9544         instructions.
9546         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9547         that does not include TImdoe for use with 32-bit.
9548         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9549         instructions.
9550         (UNSPEC_VSX_CVDPSPN): Likewise.
9551         (vsx_xscvdpspn): Likewise.
9552         (vsx_xscvspdpn): Likewise.
9553         (vsx_xscvdpspn_scalar): Likewise.
9554         (vsx_xscvspdpn_directmove): Likewise.
9555         (vsx_and<mode>3): Split logical operations into 32-bit and
9556         64-bit. Add support to do logical operations on TImode as well as
9557         VSX vector types.  Allow logical operations to be done in either
9558         VSX registers or in general purpose registers in 64-bit mode.  Add
9559         splitters if GPRs were used. For AND, add clobber of CCmode to
9560         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
9561         encoding.
9562         (vsx_and<mode>3_32bit): Likewise.
9563         (vsx_and<mode>3_64bit): Likewise.
9564         (vsx_ior<mode>3): Likewise.
9565         (vsx_ior<mode>3_32bit): Likewise.
9566         (vsx_ior<mode>3_64bit): Likewise.
9567         (vsx_xor<mode>3): Likewise.
9568         (vsx_xor<mode>3_32bit): Likewise.
9569         (vsx_xor<mode>3_64bit): Likewise.
9570         (vsx_one_cmpl<mode>2): Likewise.
9571         (vsx_one_cmpl<mode>2_32bit): Likewise.
9572         (vsx_one_cmpl<mode>2_64bit): Likewise.
9573         (vsx_nor<mode>3): Likewise.
9574         (vsx_nor<mode>3_32bit): Likewise.
9575         (vsx_nor<mode>3_64bit): Likewise.
9576         (vsx_andc<mode>3): Likewise.
9577         (vsx_andc<mode>3_32bit): Likewise.
9578         (vsx_andc<mode>3_64bit): Likewise.
9579         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
9580         and xxlorc instructions.
9581         (vsx_eqv<mode>3_64bit): Likewise.
9582         (vsx_nand<mode>3_32bit): Likewise.
9583         (vsx_nand<mode>3_64bit): Likewise.
9584         (vsx_orc<mode>3_32bit): Likewise.
9585         (vsx_orc<mode>3_64bit): Likewise.
9587         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
9589         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
9590         instruction.
9591         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
9592         (p8_vmrgow): Likewise.
9593         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
9594         GPRs to be split under VSX.
9595         (p8v_clz<mode>2): Add power8 count leading zero support.
9596         (p8v_popcount<mode>2): Add power8 population count support.
9597         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
9598         support.
9600         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
9601         instruction.
9603         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
9604         builtin functions.
9605         (vec_nand): Likewise.
9606         (vec_vclz): Likewise.
9607         (vec_vclzb): Likewise.
9608         (vec_vclzd): Likewise.
9609         (vec_vclzh): Likewise.
9610         (vec_vclzw): Likewise.
9611         (vec_vgbbd): Likewise.
9612         (vec_vmrgew): Likewise.
9613         (vec_vmrgow): Likewise.
9614         (vec_vpopcnt): Likewise.
9615         (vec_vpopcntb): Likewise.
9616         (vec_vpopcntd): Likewise.
9617         (vec_vpopcnth): Likewise.
9618         (vec_vpopcntw): Likewise.
9620 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9622         PR rtl-optimization/57468
9623         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
9624         spilled pseudos.
9626 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9628         PR rtl-optimization/57459
9629         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
9630         type when setting live regs.
9632 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9634         * config/s390/s390.opt (mlra): New option.
9635         * config/s390/s390.c (s390_decompose_address): Check displacement
9636         for all registers for LRA.
9637         (s390_secondary_reload): Don't used secondary reloads for LRA.
9638         (s390_lra_p): New function.
9639         (TARGET_LRA_P): Define.
9640         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
9641         of attribute cpu_facility to zarch for the last alternative.
9642         (*cmpmem_short): Ditto.
9644 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
9646         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
9647         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
9648         (arm_expand_prologue): Likewise.
9650 2013-06-06  Teresa Johnson  <tejohnson@google.com>
9652         PR c++/53743
9653         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
9654         as this is now done by redirect_edge_and_branch_force.
9655         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
9656         barriers, and fix interaction with splitting.
9657         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
9658         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
9659         reflect changes made in the routine.
9660         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
9661         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
9662         since this is called in cfglayout mode, and replace partition fixup
9663         with assert as that is now done by force_nonfallthru_and_redirect.
9664         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
9665         already be marked with region crossing note.
9666         (insert_section_boundary_note): Make non-static, gate on flag
9667         has_bb_partition, rewrite to also check for multiple partitions.
9668         (rest_of_handle_reorder_blocks): Remove call to
9669         insert_section_boundary_note, now done later during free_cfg.
9670         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
9671         * bb-reorder.h (insert_section_boundary_note): Declare.
9672         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
9673         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
9674         invoke insert_section_boundary_note.
9675         (try_redirect_by_replacing_jump): Remove unnecessary
9676         check for region crossing note.
9677         (fixup_partition_crossing): New function.
9678         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
9679         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
9680         in non-cfglayout mode.
9681         (force_nonfallthru_and_redirect): Fixup partition boundaries,
9682         remove old code that tried to do this. Emit barrier correctly
9683         when we are in cfglayout mode.
9684         (last_bb_in_partition): New function.
9685         (rtl_split_edge): Correctly fixup partition boundaries.
9686         (commit_one_edge_insertion): Remove old code that tried to
9687         fixup region crossing edge since this is now handled in
9688         split_block, and set up insertion point correctly since
9689         block may now end in a jump.
9690         (verify_hot_cold_block_grouping): Guard against checking when not in
9691         linearized RTL mode.
9692         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
9693         notes.
9694         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
9695         rtl_verify_flow_info, so not called in cfglayout mode.
9696         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
9697         (fixup_reorder_chain): Remove old code that attempted to fixup region
9698         crossing note as this is now handled in force_nonfallthru_and_redirect.
9699         (duplicate_insn_chain): Don't duplicate switch section notes.
9700         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
9701         note.
9702         * basic-block.h (emit_barrier_after_bb): Declare.
9704 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9706         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
9707         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
9708         arm_usatsihi): Adjust alternatives for arm_restrict_it.
9710 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9712         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
9713         where appropriate.
9714         * config/arm/ldmstm.md: Regenerate.
9716 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9718         * config/arm/sync.md (atomic_loaddi_1):
9719         Disable predication for arm_restrict_it.
9720         (arm_load_exclusive<mode>): Likewise.
9721         (arm_load_exclusivesi): Likewise.
9722         (arm_load_exclusivedi): Likewise.
9723         (arm_load_acquire_exclusive<mode>): Likewise.
9724         (arm_load_acquire_exclusivesi): Likewise.
9725         (arm_load_acquire_exclusivedi): Likewise.
9726         (arm_store_exclusive<mode>): Likewise.
9727         (arm_store_exclusive<mode>): Likewise.
9728         (arm_store_release_exclusivedi): Likewise.
9729         (arm_store_release_exclusive<mode>): Likewise.
9731 2013-06-06  Richard Biener  <rguenther@suse.de>
9733         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
9734         after LTO_null.
9735         (lto_tag_is_tree_code_p): Adjust.
9736         (lto_tag_is_gimple_code_p): Likewise.
9737         (lto_gimple_code_to_tag): Likewise.
9738         (lto_tag_to_gimple_code): Likewise.
9739         (lto_tree_code_to_tag): Likewise.
9740         (lto_tag_to_tree_code): Likewise.
9741         * data-streamer.h (streamer_write_hwi_in_range): Use
9742         uhwi streaming to stream the normalized range.
9743         (streamer_read_hwi_in_range): Likewise.
9745 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9747         * config/arm/arm.md (enabled_for_depr_it): New attribute.
9748         (predicable_short_it): Likewise.
9749         (predicated): Likewise.
9750         (enabled): Handle above.
9751         (define_cond_exec): Set predicated attribute to yes.
9753 2013-06-05  Mike Stump  <mikestump@comcast.net>
9755         * gdbinit.in (__FUNCTION__): Add.
9757 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
9759         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
9760         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
9762 2013-06-05  Jan Hubicka  <jh@suse.cz>
9764         * varasm.c (mark_decl_referenced): Revert the removal until targets
9765         are fixed.
9767 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
9769         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
9770         instead of mark_decl_referenced.
9772 2013-06-05  Jan Hubicka  <jh@suse.cz>
9774         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
9775         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
9776         and symtab_used_from_object_file_p.
9777         (cgraph_make_node_local_1): Clear forced_by_abi.
9778         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
9779         * cgraph.h (symtab_node_base): Add forced_by_abi.
9780         (decide_is_variable_needed): Remove.
9781         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
9782         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
9783         (decide_is_symbol_needed): ... this one; handle symbols in general;
9784         always analyze virtuals; honnor forced_by_abi.
9785         (cgraph_finalize_function): Update.
9786         (varpool_finalize_decl): Update.
9787         (symbol_defined_and_needed): Remove.
9788         (analyze_functions): Update.
9789         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9790         output_refs, input_overwrite_node): Handle forced_by_abi.
9791         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
9792         (address_taken_from_non_vtable_p): ... this one.
9793         (comdat_can_be_unshared_p_1): New function.
9794         (cgraph_comdat_can_be_unshared_p): Rename to ...
9795         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
9796         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
9797         (function_and_variable_visibility): Clear forced_by_abi as needed.
9798         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
9799         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
9800         * symtab.c (dump_symtab_base): Dump forced_by_abi.
9801         * varpool.c (decide_is_variable_needed): Remove.
9803 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9805         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
9806         (arm_option_override): Override arm_restrict_it where appropriate.
9807         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
9808         * config/arm/arm.opt (mrestrict-it): New command-line option.
9809         * doc/invoke.texi: Document -mrestrict-it.
9811 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9813         * tsan.c (tsan_atomic_table): Make const.
9815 2013-06-05  Richard Biener  <rguenther@suse.de>
9817         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
9818         index associated with the tree we are supposed to replace.
9819         * tree-streamer-out.c (pack_ts_base_value_fields): Output
9820         TREE_ASM_WRITTEN as zero for everything but SSA names.
9822 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9824         * tree-ssa-structalias.c (call_stmt_vars): Make static.
9826 2013-06-04  Jan Hubicka  <jh@suse.cz>
9828         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
9829         (input_node, input_varpool_node): Handle correctly external same
9830         body aliases.
9831         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
9832         nodes at ltrans stage.
9834 2013-06-04  Jan Hubicka  <jh@suse.cz>
9836         * ipa-inline.c (update_caller_keys): Fix availability test.
9837         (update_callee_keys): Likewise.
9838         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
9839         to follow ELF standard.
9841 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
9843         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
9844         (mips64r5900el-*-elf*): New configurations.
9845         * config/mips/mips-cpus.def (r5900): New processor.
9846         * config/mips/mips-tables.opt: Regenerate.
9847         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
9848         (mips_issue_rate): Handle PROCESSOR_R5900.
9849         (mips_reorg_process_insns): Force reorder mode for the R5900.
9850         * config/mips/mips.h (TARGET_MIPS5900): Define.
9851         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
9852         TARGET_MIPS5900.
9853         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
9854         TARGET_MIPS5900.
9855         * config/mips/mips.md (processor): Add r5900.
9856         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
9858 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9860         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
9861         into function to generate MOVI instruction.
9862         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
9863         (aarch64_preferred_simd_mode): Turn into wrapper.
9864         (aarch64_output_scalar_simd_mov_immediate): New function.
9865         * config/aarch64/aarch64-protos.h: Add prototype for above.
9867 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9869         * config/aarch64/aarch64.c (simd_immediate_info): Remove
9870         element_char member.
9871         (sizetochar): Return signed char.
9872         (aarch64_simd_valid_immediate): Remove elchar and other
9873         unnecessary variables.
9874         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
9875         Calculate element_char as required.
9876         * config/aarch64/aarch64-protos.h: Update and move prototype
9877         for aarch64_output_simd_mov_immediate.
9878         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9879         Update arguments.
9881 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9883         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
9884         information completed by aarch64_simd_valid_immediate.
9885         (aarch64_legitimate_constant_p): Update arguments.
9886         (aarch64_simd_valid_immediate): Work with struct rather than many
9887         pointers.
9888         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
9889         (aarch64_simd_make_constant): Update arguments.
9890         (aarch64_output_simd_mov_immediate): Work with struct rather than
9891         many pointers.  Output immediate directly rather than as operand.
9892         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9893         Update prototype.
9894         * config/aarch64/constraints.md (Dn): Update arguments.
9896 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9898         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
9899         longer static.
9900         (aarch64_simd_immediate_valid_for_move): Remove.
9901         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
9902         (aarch64_simd_make_constant): Update call.
9903         (aarch64_output_simd_mov_immediate): Update call.
9904         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9905         Add prototype.
9906         * config/aarch64/constraints.md (Dn): Update call.
9908 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9910         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
9911         return type to bool for prototype.
9912         (aarch64_legitimate_constant_p): Check for true instead of not -1.
9913         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
9914         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
9916 2013-06-04  Catherine Moore  <clm@codesourcery.com>
9918         * config/mips/mips.opt (meva): New.
9919         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
9920         (ASM_SPEC): Handle -meva.
9921         * doc/invoke.texi (meva):  Document.
9923 2013-06-04  Alan Modra  <amodra@gmail.com>
9925         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
9926         constant output.
9928 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9930         * rtl.def: Add extra fourth optional field to define_cond_exec.
9931         * gensupport.c (process_one_cond_exec): Process attributes from
9932         define_cond_exec.
9933         * doc/md.texi: Document fourth field in define_cond_exec.
9935 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
9937         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
9938         out the processing order as in store_bit_field_1.
9940 2013-06-04  Jan Hubicka  <jh@suse.cz>
9942         PR middle-end/57500
9943         * cgraphunit.c (cgraph_process_same_body_aliases): Create
9944         non-VAR_DECL node if it does not exist yet.
9946 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
9948         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
9949         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
9950         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
9951         target_cpu_default setting.
9953 2013-06-03  Teresa Johnson  <tejohnson@google.com>
9955         * dumpfile.c (opt_info_switch_p): Change -fopt-info
9956         default to -fopt-info=optimized instead of all.
9957         * doc/invoke.texi: Ditto.
9958         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
9959         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
9960         (execute_vect_slp): Emit BB vectorization success under
9961         MSG_OPTIMIZED_LOCATIONS.
9962         * tree-vect-slp.c (vect_slp_transform_bb): Change
9963         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
9964         * tree-vect-loop.c (vect_transform_loop): Ditto.
9966 2013-06-03  Jason Merrill  <jason@redhat.com>
9968         PR c++/57415
9969         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9970         Use TARGET_EXPR for C++.
9972 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
9974         PR rtl-optimization/57268
9975         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
9976         if DEBUG_INSN_P (insn).
9978         Reapply
9979         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
9981         PR rtl-optimization/57268
9982         * sched-deps.c (sched_analyze_2): Flush dependence lists if
9983         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
9985 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
9987         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
9988         (ix86_avoid_lea_for_addr): Likewise.
9989         (exact_dependency_1): Likewise.
9990         (ix86_adjust_cost): Likewise.
9991         (swap_top_of_ready_list): Fix formatting and !reload_completed check
9992         removed.
9993         (do_reorder_for_imul): Fix typo, formatting and
9994         !reload_completed check removed.
9995         (ix86_sched_reorder): Fix typo and formatting.
9996         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
9997         list.
9999 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
10001         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
10003 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
10005         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10006         <STRING_CST>: Likewise.
10007         <VECTOR_CST>: Likewise.
10009 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
10010             Mikael Morin  <mikael@gcc.gnu.org>
10012         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10013         * config.in: Regenerated.
10014         * configure: Regenerated.
10016 2013-06-01  Jan Hubicka  <jh@suse.cz>
10018         PR middle-end/57366
10019         * cgraphunit.c (compile): When weakref is not supported,
10020         set up transparent aliases before final output pass.
10021         * varasm.c (assemble_alias): Do not try to do it here.
10023 2013-06-01  Jan Hubicka  <jh@suse.cz>
10025         PR middle-end/57467
10026         * passes.c (for_per_function): Skip unanalyzed functions.
10028 2013-06-01  Jan Hubicka  <jh@suse.cz>
10030         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10031         (lto_symtab_merge_symbols_1): ... this one.
10032         (lto_symtab_merge_cgraph_nodes): Rename to ...
10033         (lto_symtab_merge_symbols): ... this one; simplify.
10034         * cgraph.c (same_body_aliases_done): Rename to ...
10035         (cpp_implicit_aliases_done): ... this one.
10036         (cgraph_create_function_alias): Update.
10037         (cgraph_same_body_alias): Update.
10038         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10039         (verify_edge_corresponds_to_fndecl): Simplify.
10040         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10041         (cgraph_node): Remove same_body_alias.
10042         (varpool_node): Remove alias_of and extra_name_alias.
10043         (same_body_aliases_done): Rename to ..
10044         (cpp_implicit_aliases_done): ... this one.
10045         (symtab_alias_ultimate_target): Add default parameter.
10046         (symtab_resolve_alias): New function.
10047         (fixup_same_cpp_alias_visibility): Declare.
10048         (cgraph_function_node): Add default parameter.
10049         (cgraph_node_asm_name): Likewise.
10050         (cgraph_function_or_thunk_node): Add default parameter; do
10051         not ICE when it is NULL.
10052         (varpool_variable_node): Likewise.
10053         * tree-emutls.c (create_emultls_var): Update.
10054         (ipa_lower_emutls): Update.
10055         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10056         (cgraph_reset_node): Reset alias info.
10057         (cgraph_finalize_function): Update.
10058         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10059         (analyze_function): Simplify.
10060         (cgraph_process_same_body_aliases): Simplify.
10061         (analyze_functions): Fixup same body aliases.
10062         (handle_alias_pairs): Simplify.
10063         (assemble_thunk): Update.
10064         (assemble_thunks_and_aliases): Update.
10065         (output_weakrefs): Rewrite.
10066         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10067         (lto_output_varpool_node): Likewise.
10068         (compute_ltrans_boundary): Remve assert.
10069         (get_alias_symbol): New functoin.
10070         (input_node): Rewrite alias handling.
10071         (input_varpool_node): Likewise.
10072         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10073         * ipa.c (process_references): Handle weakrefs correctly.
10074         (symtab_remove_unreachable_nodes): Likewise.
10075         * trans-mem.c (get_cg_data): Update.
10076         (ipa_tm_create_version_alias): Update.
10077         (ipa_tm_execute): Update.
10078         * symtab.c (dump_symtab_base): Dump aliases.
10079         (verify_symtab_base): Verify aliases.
10080         (symtab_node_availability): New function.
10081         (symtab_alias_ultimate_target): Simplify.
10082         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10083         handle all the fixup cases.
10084         (symtab_resolve_alias): New function.
10085         * passes.c (ipa_write_summaries): Handle weakrefs.
10086         * varpool.c (varpool_analyze_node): Simplify.
10087         (assemble_aliases): Update.
10088         (varpool_create_variable_alias): Simplify.
10089         (varpool_extra_name_alias): Simplify.
10090         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10091         (lto_symtab_merge_symbols): ... this one.
10093 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10095         Revert
10096         PR rtl-optimization/57268
10097         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10098         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10100 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10102         Partially reverted:
10103         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10105         PR middle-end/57073
10106         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10107         further up.
10109 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10111         PR rtl-optimization/57268
10112         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10113         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10115 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10117         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10118         unordered comparison operators when -fno-trapping-math is in effect
10119         on the e500.
10120         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10121         and implement unordered comparison operators properly on the e500.
10123 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10125         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10126         for constant scalar integers.
10127         (simplify_relational_operation_1): Likewise.
10129 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10131         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10132         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10133         Fix comment.
10135 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10136             Igor Zamyatin  <igor.zamyatin@intel.com>
10138         Silvermont (SLM) architecture performance tuning.
10139         * config/i386/i386.h (enum ix86_tune_indices): Add
10140         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10141         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10143         * config/i386/i386.c (initial_ix86_tune_features)
10144         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10145         (ix86_lea_outperforms): Handle Silvermont tuning.
10146         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10147         call.
10148         (ix86_use_lea_for_mov): Likewise.
10149         (ix86_avoid_lea_for_addr): Likewise.
10150         (ix86_lea_for_add_ok): Likewise.
10151         (exact_dependency_1): New function.
10152         (exact_store_load_dependency): Likewise.
10153         (ix86_adjust_cost): Handle Silvermont tuning.
10154         (do_reoder_for_imul): Likewise.
10155         (swap_top_of_ready_list): New function.
10156         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10158         * config/i386/i386.md (peepholes that split memory operand in fp
10159         converts): New.
10161 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10163         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10164         Remove un-necessary braces.
10166 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10168         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10169         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10171 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10173         PR middle-end/57073
10174         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10176 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10178         PR target/56315
10179         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10180         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10181         * config/arm/neon.md (iordi3_neon): Remove.
10182         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10183         * config/arm/predicates.md (imm_for_neon_logic_operand):
10184         Move to earlier in the file.
10185         (neon_logic_op2): Likewise.
10186         (arm_iordi_operand_neon): New predicate.
10188 2013-05-31  Richard Biener  <rguenther@suse.de>
10190         PR tree-optimization/57478
10191         PR tree-optimization/57453
10192         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10193         are life as well.
10195 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10197         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10198         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10200 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10201             Thomas Koenig  <tkoenig@gcc.gnu.org>
10203         PR middle-end/57073
10204         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10205         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10207 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10209         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10211 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10213         * target.def (register_usage_leveling_p): New hook.
10214         * targhooks.c (default_register_usage_leveling_p): New.
10215         * targhooks.h (default_register_usage_leveling_p): New prototype.
10216         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10217         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10218         * doc/tm.texi: Update.
10219         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10221 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10223         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10224         (*insv_reg<mode>): New define_insn.
10226 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10228         PR rtl-optimization/57439
10229         * postreload.c (move2add_valid_value_p): Check that we have
10230         a zero subreg_regno_offset when accessing the register in
10231         the requested mode.
10233 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10234             Igor Zamyatin  <igor.zamyatin@intel.com>
10236         Silvermont (SLM) architecture pipeline model, tuning and
10237         insn selection.
10238         * config.gcc: Add slm config options and target.
10240         * config/i386/slm.md: New.
10242         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10244         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
10245         PROCESSOR_SLM.
10246         (ix86_target_macros_internal): Likewise.
10248         * config/i386/i386.c (slm_cost): New cost.
10249         (m_SLM): New macro flag.
10250         (initial_ix86_tune_features): Set m_SLM.
10251         (x86_accumulate_outgoing_args): Likewise.
10252         (x86_arch_always_fancy_math_387): Likewise.
10253         (processor_target_table): Add slm cost.
10254         (cpu_names): Add slm cpu name.
10255         (x86_option_override_internal): Set SLM ISA.
10256         (ix86_issue_rate): New case PROCESSOR_SLM.
10257         (ia32_multipass_dfa_lookahead): Likewise.
10258         (fold_builtin_cpu): Add slm.
10260         * config/i386/i386.h (TARGET_SLM): New target macro.
10261         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10262         (processor_type): Add PROCESSOR_SLM.
10264         * config/i386/i386.md (cpu): Add new value "slm".
10265         (slm.md): Include slm.md.
10267 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
10268             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10270         * config/arm/arm-protos.h: Add and update function protos.
10271         * config/arm/arm.c (use_simple_return_p): New added.
10272         (thumb2_expand_return): Check simple_return flag.
10273         * config/arm/arm.md: Add simple_return and conditional simple_return.
10274         * config/arm/iterators.md: Add iterator for return and simple_return.
10276 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10278         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10279         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10280         (arm_emit_vfp_multi_reg_pop): Likewise.
10281         (thumb2_emit_ldrd_pop): Likewise.
10282         (arm_expand_epilogue): Add misc REG_CFA notes.
10283         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10285 2013-05-29  Lawrence Crowl  <crowl@google.com>
10287         * config/arm/t-arm: Update for below.
10289         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10290         Change type to hash_table.  Update dependent calls and types.
10292         * config/i386/t-cygming: Update for below.
10294         * config/i386/t-interix: Update for below.
10296         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10297         Change type to hash_table.  Update dependent calls and types.
10298         (i386_find_on_wrapper_list::wrappers): Likewise.
10300         * config/ia64/t-ia64: Update for below.
10302         * config/ia64/ia64.c (bundle_state_table):
10303         Change type to hash_table.  Update dependent calls and types.
10305         * config/mips/mips.c (mips_reorg_process_insns::htab):
10306         Change type to hash_table.  Update dependent calls and types.
10308         * config/sol2.c (solaris_comdat_htab):
10309         Change type to hash_table.  Update dependent calls and types.
10311         * config/t-sol2: Update for above.
10313 2013-05-29  Teresa Johnson  <tejohnson@google.com>
10315         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10316         functions are not yet marked as defined.
10318 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
10319             Pat Haugen  <pthaugen@us.ibm.com>
10320             Peter Bergner  <bergner@vnet.ibm.com>
10322         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10323         instructions.
10324         (VEC_A): Likewise.
10325         (VEC_C): Likewise.
10326         (vrotl<mode>3): Likewise.
10327         (vashl<mode>3): Likewise.
10328         (vlshr<mode>3): Likewise.
10329         (vashr<mode>3): Likewise.
10331         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10332         support for power8 V2DI builtins.
10334         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10335         power8 V2DI builtins.
10336         (vupkhsw): Likewise.
10337         (vupklsw): Likewise.
10338         (vaddudm): Likewise.
10339         (vminsd): Likewise.
10340         (vmaxsd): Likewise.
10341         (vminud): Likewise.
10342         (vmaxud): Likewise.
10343         (vpkudum): Likewise.
10344         (vpksdss): Likewise.
10345         (vpkudus): Likewise.
10346         (vpksdus): Likewise.
10347         (vrld): Likewise.
10348         (vsld): Likewise.
10349         (vsrd): Likewise.
10350         (vsrad): Likewise.
10351         (vsubudm): Likewise.
10352         (vcmpequd): Likewise.
10353         (vcmpgtsd): Likewise.
10354         (vcmpgtud): Likewise.
10355         (vcmpequd_p): Likewise.
10356         (vcmpgtsd_p): Likewise.
10357         (vcmpgtud_p): Likewise.
10358         (vupkhsw): Likewise.
10359         (vupklsw): Likewise.
10360         (vaddudm): Likewise.
10361         (vmaxsd): Likewise.
10362         (vmaxud): Likewise.
10363         (vminsd): Likewise.
10364         (vminud): Likewise.
10365         (vpksdss): Likewise.
10366         (vpksdus): Likewise.
10367         (vpkudum): Likewise.
10368         (vpkudus): Likewise.
10369         (vrld): Likewise.
10370         (vsld): Likewise.
10371         (vsrad): Likewise.
10372         (vsrd): Likewise.
10373         (vsubudm): Likewise.
10375         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10376         support for power8 V2DI instructions.
10378         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10379         power8 V2DI instructions.  Combine pack and unpack insns to use an
10380         iterator for each mode.  Check whether a particular mode supports
10381         Altivec instructions instead of just checking TARGET_ALTIVEC.
10382         (UNSPEC_VPKUWUM): Likewise.
10383         (UNSPEC_VPKSHSS): Likewise.
10384         (UNSPEC_VPKSWSS): Likewise.
10385         (UNSPEC_VPKUHUS): Likewise.
10386         (UNSPEC_VPKSHUS): Likewise.
10387         (UNSPEC_VPKUWUS): Likewise.
10388         (UNSPEC_VPKSWUS): Likewise.
10389         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10390         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10391         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10392         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10393         (UNSPEC_VUPKHSB): Likewise.
10394         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10395         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10396         (UNSPEC_VUPKHSH): Likewise.
10397         (UNSPEC_VUPKLSB): Likewise.
10398         (UNSPEC_VUPKLSH): Likewise.
10399         (VI2): Likewise.
10400         (VI_char): Likewise.
10401         (VI_scalar): Likewise.
10402         (VI_unit): Likewise.
10403         (VP): Likewise.
10404         (VP_small): Likewise.
10405         (VP_small_lc): Likewise.
10406         (VU_char): Likewise.
10407         (add<mode>3): Likewise.
10408         (altivec_vaddcuw): Likewise.
10409         (altivec_vaddu<VI_char>s): Likewise.
10410         (altivec_vadds<VI_char>s): Likewise.
10411         (sub<mode>3): Likewise.
10412         (altivec_vsubcuw): Likewise.
10413         (altivec_vsubu<VI_char>s): Likewise.
10414         (altivec_vsubs<VI_char>s): Likewise.
10415         (altivec_vavgs<VI_char>): Likewise.
10416         (altivec_vcmpbfp): Likewise.
10417         (altivec_eq<mode>): Likewise.
10418         (altivec_gt<mode>): Likewise.
10419         (altivec_gtu<mode>): Likewise.
10420         (umax<mode>3): Likewise.
10421         (smax<mode>3): Likewise.
10422         (umin<mode>3): Likewise.
10423         (smin<mode>3): Likewise.
10424         (altivec_vpkuhum): Likewise.
10425         (altivec_vpkuwum): Likewise.
10426         (altivec_vpkshss): Likewise.
10427         (altivec_vpkswss): Likewise.
10428         (altivec_vpkuhus): Likewise.
10429         (altivec_vpkshus): Likewise.
10430         (altivec_vpkuwus): Likewise.
10431         (altivec_vpkswus): Likewise.
10432         (altivec_vpks<VI_char>ss): Likewise.
10433         (altivec_vpks<VI_char>us): Likewise.
10434         (altivec_vpku<VI_char>us): Likewise.
10435         (altivec_vpku<VI_char>um): Likewise.
10436         (altivec_vrl<VI_char>): Likewise.
10437         (altivec_vsl<VI_char>): Likewise.
10438         (altivec_vsr<VI_char>): Likewise.
10439         (altivec_vsra<VI_char>): Likewise.
10440         (altivec_vsldoi_<mode>): Likewise.
10441         (altivec_vupkhsb): Likewise.
10442         (altivec_vupkhs<VU_char>): Likewise.
10443         (altivec_vupkls<VU_char>): Likewise.
10444         (altivec_vupkhsh): Likewise.
10445         (altivec_vupklsb): Likewise.
10446         (altivec_vupklsh): Likewise.
10447         (altivec_vcmpequ<VI_char>_p): Likewise.
10448         (altivec_vcmpgts<VI_char>_p): Likewise.
10449         (altivec_vcmpgtu<VI_char>_p): Likewise.
10450         (abs<mode>2): Likewise.
10451         (vec_unpacks_hi_v16qi): Likewise.
10452         (vec_unpacks_hi_v8hi): Likewise.
10453         (vec_unpacks_lo_v16qi): Likewise.
10454         (vec_unpacks_hi_<VP_small_lc>): Likewise.
10455         (vec_unpacks_lo_v8hi): Likewise.
10456         (vec_unpacks_lo_<VP_small_lc>): Likewise.
10457         (vec_pack_trunc_v8h): Likewise.
10458         (vec_pack_trunc_v4si): Likewise.
10459         (vec_pack_trunc_<mode>): Likewise.
10461         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10462         V2DI builtins.
10463         (vec_vmaxsd): Likewise.
10464         (vec_vmaxud): Likewise.
10465         (vec_vminsd): Likewise.
10466         (vec_vminud): Likewise.
10467         (vec_vpksdss): Likewise.
10468         (vec_vpksdus): Likewise.
10469         (vec_vpkudum): Likewise.
10470         (vec_vpkudus): Likewise.
10471         (vec_vrld): Likewise.
10472         (vec_vsld): Likewise.
10473         (vec_vsrad): Likewise.
10474         (vec_vsrd): Likewise.
10475         (vec_vsubudm): Likewise.
10476         (vec_vupkhsw): Likewise.
10477         (vec_vupklsw): Likewise.
10479 2013-05-29  Jan Hubicka  <jh@suse.cz>
10481         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10482         flags; reorder rest of fields in more consistent way.
10483         (varpool_node): Remove analyzed, finalized and alias.
10484         (cgraph_ndoe): Likewise.
10485         (symtab_alias_ultimate_target): New function.
10486         (cgraph_function_node): Move offline.
10487         (cgraph_reset_node): Declare.
10488         (cgraph_comdat_can_be_unshared_p): Remove.
10489         (varpool_remove_initializer): Declare.
10490         (varpool_first_defined_variable, varpool_next_defined_variable
10491         cgraph_first_defined_function, cgraph_next_defined_function): Update.
10492         (cgraph_function_with_gimple_body_p): Update.
10493         (varpool_all_refs_explicit_p): Update.
10494         (symtab_alias_target): New function.
10495         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10496         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10497         (cgraph_function_or_thunk_node): Simplify using
10498         symtab_alias_ultimate_target.
10499         (varpool_variable_node): Likewise.
10500         * cgraph.c (cgraph_create_function_alias): Update.
10501         (cgraph_add_thunk): Update.
10502         (cgraph_remove_node): Update.
10503         (dump_cgraph_node): Do not dump removed flags.
10504         (cgraph_function_body_availability): Update.
10505         (cgraph_propagate_frequency): Update.
10506         (verify_cgraph_node): Check sanity of local flag.
10507         (cgraph_function_node): Move here from cgraph.h; revamp for
10508         cgraph_function_or_thunk_node.
10509         * lto-symtab.c (lto_varpool_replace_node): Update.
10510         (lto_symtab_resolve_can_prevail_p): Update.
10511         (lto_symtab_merge_cgraph_nodes): Update.
10512         * ipa-cp.c (determine_versionability, initialize_node_lattices,
10513         propagate_constants_accross_call, devirtualization_time_bonus,
10514         ipcp_propagate_stage): Update.
10515         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10516         * ipa-inline-transform.c (clone_inlined_nodes,
10517         preserve_function_body_p): Update.
10518         * ipa-reference.c (propagate): Update.
10519         (write_node_summary_p): Update.
10520         * toplev.c (wrapup_global_declaration_2): Update.
10521         * cgraphunit.c (cgraph_analyze_function): Rename to ...
10522         (analyze_function) ... this one.
10523         (cgraph_process_new_functions): Update.
10524         (cgraph_reset_node): Export.
10525         (cgraph_finalize_function): Update.
10526         (cgraph_add_new_function): Update.
10527         (process_function_and_variable_attributes): Update.
10528         (varpool_finalize_decl): Update.
10529         (symbol_finalized): Remove.
10530         (symbol_finalized_and_needed): Rename to ...
10531         (symbol_defined_and_needed): ... update.
10532         (cgraph_analyze_functions): Update.
10533         (handle_alias_pairs): Update.
10534         (mark_functions_to_output): Update.
10535         (assemble_thunk): Update.
10536         (output_in_order): Update.
10537         (output_weakrefs): Update.
10538         (finalize_compilation_unit): Update.
10539         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10540         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10541         input_node, input_varpool_node): Update.
10542         * dbxout.c (dbxout_expand_expr): Update.
10543         * cgraphclones.c (cgraph_clone_node): Update.
10544         (cgraph_copy_node_for_versioning): Update.
10545         (cgraph_materialize_clone): Update.
10546         (cgraph_materialize_all_clones): Update.
10547         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10548         propagate_pure_const, propagate_nothrow): Update.
10549         * lto-streamer-out.c (lto_output, write_symbol): Update.
10550         * ipa-utils.c (ipa_reverse_postorder): Update.
10551         * ipa-inline.c (can_inline_edge_p): Update.
10552         (update_caller_keys, ipa_inline): Update.
10553         * dwarf2out.c (reference_to_unused,
10554         premark_types_used_by_global_vars_helper): Update.
10555         * tree-eh.c (tree_could_trap_p): Update.
10556         * ipa-split.c (consider_split, execute_split_functions): Update.
10557         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10558         has_addr_references_p): Update; move ahead in file for better
10559         readability.
10560         (process_references): Simplify.
10561         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10562         bodies are removed.
10563         (cgraph_comdat_can_be_unshared_p): Make static.
10564         (cgraph_externally_visible_p): Update.
10565         (varpool_externally_visible_p): Update.
10566         (function_and_variable_visibility): Update.
10567         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10568         ipa_tm_mark_force_output_node): Update.
10569         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10570         estimate_edge_devirt_benefit, inline_generate_summary,
10571         inline_write_summary): Update.
10572         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10573         * ipa-prop.c (ipa_compute_jump_functions): Update.
10574         (ipa_print_node_params, ipa_prop_read_section,
10575         ipa_update_after_lto_read, read_replacements_section): Update.
10576         * varasm.c (mark_decl_referenced): Update.
10577         (assemble_alias, dump_tm_clone_pairs): Update.
10578         * tree-inline.c (copy_bb): Update.
10579         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
10580         Update.
10581         * symtab.c (dump_symtab_base): Print new flags.
10582         (verify_symtab_base): Verify new flags.
10583         (symtab_alias_ultimate_target): New function.
10584         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
10585         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
10586         Update.
10587         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
10588         Update.
10589         * i386.c (ix86_get_function_versions_dispatcher,
10590         ix86_generate_version_dispatcher_body): Update.
10591         (fold_builtin_cpu): Use varpool_add_new_variable.
10592         * varpool.c (varpool_remove_initializer): Break out from ...
10593         (varpool_remove_node): ... this one.
10594         (dump_varpool_node, varpool_node_for_asm,
10595         cgraph_variable_initializer_availability, varpool_analyze_node,
10596         varpool_assemble_decl, varpool_remove_unreferenced_decls,
10597         varpool_finalize_named_section_flags, varpool_create_variable_alias):
10598         Update.
10600 2013-05-29  Jan Hubicka  <jh@suse.cz>
10602         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
10604 2013-05-29  Easwaran Raman  <eraman@google.com>
10606         PR tree-optimization/57442
10607         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
10608         when control exits the main loop.
10610 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
10612         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
10613         and RX600.
10614         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
10615         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
10616         * rx/t-rx: Add rx100 under multi library matches option for nofpu
10617         option.
10619 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10621         PR tree-optimization/57441
10622         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
10623         Don't limit size of incr_vec to number of candidates.
10625 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
10627         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
10628         and mips16 directories.
10629         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
10630         (MULTILIB_DIRNAMES): Ditto.
10631         (MULTILIB_EXCEPTIONS): Add new exceptions.
10632         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
10633         (MULTILIB_DIRNAMES): Ditto.
10634         (MULTILIB_EXCEPTIONS): Add new exceptions.
10636 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10637             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10639         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
10640         SYMBOL_TINY_ABSOLUTE.
10641         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
10642         SYMBOL_TINY_ABSOLUTE.
10643         (aarch64_expand_mov_immediate): Likewise.
10644         (aarch64_classify_symbol): Likewise.
10645         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
10646         Permit SYMBOL_TINY_ABSOLUTE.
10647         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
10649 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10650             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10652         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
10653         Refactor if/switch.  Replace gcc_assert with if.
10655 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10657         * config/i386/i386.c (initial_ix86_tune_features): Enable
10658         FP Reassociation for AMD bdver1 and bdver2.
10660 2013-05-29  Martin Jambor  <mjambor@suse.cz>
10662         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
10663         and IMAGPART_EXPR do not occur within other handled_components.
10665 2013-05-29  Richard Biener  <rguenther@suse.de>
10667         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
10668         access on whether the use is in the BB we currently try to
10669         vectorize.
10670         (vect_bb_vectorization_profitable_p): Pass the BB we currently
10671         vectorize to vect_bb_slp_scalar_cost.
10673 2013-05-29  Richard Biener  <rguenther@suse.de>
10675         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
10676         computing scalar cost offsetted by stmts that are kept live
10677         by scalar uses.
10678         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
10679         for computation of scalar cost.
10681 2013-05-28  Steve Ellcey  <sellcey@mips.com>
10683         * config/mips/mips-cpus.def (mips32r2): Change processor type.
10685 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
10687         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
10688         array notation built-in reduction functions.
10689         * doc/passes.texi (Passes): Added documentation about changes done
10690         for Cilk Plus.
10691         * doc/invoke.texi (C Dialect Options): Added documentation about
10692         the -fcilkplus flag.
10693         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
10694         (BUILTINS_DEF): Depend on cilkplus.def.
10695         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
10696         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
10697         * cilkplus.def: New file.
10699 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
10701         PR rtl-optimization/57439
10702         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
10704 2013-05-28  Easwaran Raman  <eraman@google.com>
10706         PR tree-optimization/57337
10707         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
10708         (find_insert_point): Correctly identify the insertion point
10709         when two statements with the same UID is compared.
10711 2013-05-28  Richard Biener  <rguenther@suse.de>
10713         PR tree-optimization/56787
10714         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
10715         from the list of data references.
10716         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
10717         clobbers.
10718         (vect_analyze_loop_operations): Likewise.
10719         (vect_transform_loop): Remove clobbers.
10721 2013-05-28  Martin Jambor  <mjambor@suse.cz>
10723         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
10724         and REALPART_EXPRs have scalar type.
10726 2013-05-28  Richard Biener  <rguenther@suse.de>
10728         PR tree-optimization/57411
10729         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
10730         virtual operands.
10731         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
10732         virtual operand propagation.
10734 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10736         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
10737         destination register for bmasksi_vis.
10738         (vector_init_bshuffle): Likewise.
10739         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
10741 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10743         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
10744         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
10745         mode if the instruction isn't available in the original mode.
10746         * config/sparc/sparc.opt (mfix-ut699): New option.
10747         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
10748         (divdf3): Turn into expander.
10749         (divdf3_nofix): New insn.
10750         (divdf3_fix): Likewise.
10751         (divsf3): Disable if -mfix-ut699.
10752         (sqrtdf2): Turn into expander.
10753         (sqrtdf2_nofix): New insn.
10754         (sqrtdf2_fix): Likewise.
10755         (sqrtsf2): Disable if -mfix-ut699.
10757 2013-05-27  Richard Biener  <rguenther@suse.de>
10759         PR middle-end/57412
10760         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
10761         block for the new loop.
10763 2013-05-27  Richard Biener  <rguenther@suse.de>
10765         PR tree-optimization/57343
10766         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
10767         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
10768         (number_of_iterations_cond): Do not build the folded tree.
10770 2013-05-27  Richard Biener  <rguenther@suse.de>
10772         Revert
10773         PR middle-end/57381
10774         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10775         OEP_CONSTANT_ADDRESS_OF retained.
10777         PR tree-optimization/57417
10778         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
10779         for unchanged base.
10780         (set_ssa_val_to): Compare addresses using
10781         get_addr_base_and_unit_offset.
10783 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
10785         PR rtl-optimization/56833
10786         * postreload.c (move2add_record_mode): New function.
10787         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
10788         (move2add_use_add2_insn): Use move2add_record_sym_value.
10789         (move2add_use_add3_insn): Likewise.
10790         (reload_cse_move2add): Use move2add_valid_value_p and
10791         move2add_record_mode.  Invalidate call-clobbered and REG_INC
10792         affected regs by setting reg_mode to VOIDmode.
10793         (move2add_note_store): Don't pretend the inside of a SUBREG is
10794         the actual destination.  Invalidate single/leading registers by
10795         setting reg_mode to VOIDmode.
10796         Use move2add_record_sym_value, move2add_valid_value_p and
10797         move2add_record_mode.
10799 2013-05-27  Richard Biener  <rguenther@suse.de>
10801         PR tree-optimization/57396
10802         * tree-affine.c (double_int_constant_multiple_p): Properly
10803         return false for val == 0 and div != 0.
10805 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10807         * config/mips/mips.h: Use #elif in preprocessor conditions.
10809 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10811         PR target/53916
10812         * config/mips/constraints.md (kl): New constraint.
10813         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
10814         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
10815         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
10816         from LO for MIPS16.
10817         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
10819 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10821         PR target/55777
10822         * config/mips/mips.c (mips_can_inline_p): New function.
10823         (TARGET_CAN_INLINE_P): Define.
10825 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
10827         * sched-int.h (ds_t, dw_t): Make unsigned int.
10828         Fix documentation that describes how all the ds_t bits are used.
10829         Reserve the last bit for delayed-branch scheduling.
10830         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
10831         (BITS_PER_DEP_WEAK): Fix definition and documentation.
10832         (gen_dep_weak_1): Remove prototype.
10833         * sched-deps.c (get_dep_weak_1): Make static.
10834         * target.def (speculate_insn, needs_block_p, gen_spec_check,
10835         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
10836         * doc/tm.texi: Regenerate.
10837         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
10839 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
10841         PR debug/56950
10842         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
10844 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
10845             Sandra Loosemore  <sandra@codesourcery.com>
10847         * config.gcc (powerpc-*): Allow native for with-cpu.
10849 2013-05-24  Jeff Law  <law@redhat.com>
10851         PR tree-optimization/57124
10852         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
10853         conversion feeding a condition if the range has an overflow
10854         if -fstrict-overflow.  Add warnings for when we do make the
10855         transformation.
10857 2013-05-24  Dehao Chen  <dehao@google.com>
10859         * tree-cfg.c (locus_discrim_map): Fix the typo.
10860         (locus_discrim_hasher): Likewise.
10861         (locus_discrim_hasher::hash): Likewise.
10862         (locus_discrim_hasher::equal): Likewise.
10864 2013-05-24  Martin Jambor  <mjambor@suse.cz>
10866         PR tree-optimization/57294
10867         * cgraph.h (ipa_record_stmt_references): Declare.
10868         * cgraphbuild.c (ipa_record_stmt_references): New function.
10869         (build_cgraph_edges): Use ipa_record_stmt_references.
10870         (rebuild_cgraph_edges): Likewise.
10871         (cgraph_rebuild_references): Likewise.
10872         * ipa-prop.c (ipa_modify_call_arguments): Discard references
10873         associated with the old statement and build references from the
10874         newly built statements.
10875         * ipa-ref.c (ipa_remove_stmt_references): New function.
10876         * ipa-ref.h (ipa_remove_stmt_references): Declare.
10878 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
10880         * lra-constraints.c (emit_spill_move): Use smaller mode for
10881         mem-mem moves.
10882         (check_and_process_move): Consider mem-reg moves for secondary
10883         too.
10884         (curr_insn_transform): Don't lose insns emitted before for
10885         secondary memory moves.
10886         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
10887         reg set up in the current insn.
10889 2013-05-24  Dehao Chen  <dehao@google.com>
10891         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
10892         hash function.
10893         (locus_descrim_hasher::equal): Likewise.
10894         (build_gimple_cfg): New discriminator assignment algorithm.
10895         (make_edges): Likewise.
10896         (next_discriminator_for_locus): Likewise.
10897         (same_line_p): Likewise.
10898         (assign_discriminators): Likewise.
10899         (make_cond_expr_edges): Likewise.
10900         (make_gimple_switch_edges): Likewise.
10901         (make_goto_expr_edges): Likewise.
10902         (make_gimple_asm_edges): Likewise.
10904 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
10906         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
10907         X format specifier to only display bottom 16 bits.
10908         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
10909         immediate to match for operand 2, since it will be masked.
10911 2013-05-24  Richard Biener  <rguenther@suse.de>
10913         PR tree-optimization/57287
10914         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
10915         all SSA names that occur in abnormal PHIs.
10917 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10919         PR tree-ssa/57385
10920         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
10921         that index is not negative.
10923 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
10925         PR rtl-optimization/55177
10926         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
10927         (simplify_byte_swapping_operation): New.
10928         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
10929         (simplify_relational_operation_1): Deal with BSWAP.
10931 2013-05-23  Richard Henderson  <rth@redhat.com>
10933         PR target/56742
10934         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
10935         (ix86_reorg): Call it.
10937 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
10939         PR target/57379
10940         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
10941         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
10942         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
10944 2013-05-23  Christian Bruel  <christian.bruel@st.com>
10946         PR debug/57351
10947         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
10949 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10950             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10952         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
10953         * config/aarch64/constraints.md (Usa): Remove.
10954         * doc/md.texi (AArch64 Usa): Remove.
10956 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10957             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10959         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
10960         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
10961         * config/aarch64/predicates.md (aarch64_const_address): Remove.
10962         (aarch64_mov_operand): Use aarch64_mov_operand_p.
10964 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
10966         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
10967         instruction (AdvSIMD).
10968         * config/aarch64/aarch64-builtins.c
10969         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
10970         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
10972 2013-05-23  Martin Jambor  <mjambor@suse.cz>
10974         PR middle-end/57347
10975         * tree.h (contains_bitfld_component_ref_p): Declare.
10976         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
10977         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
10978         caller.
10979         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
10980         not access a bit-field.  Assert all final offsets are byte-aligned.
10982 2013-05-23  Richard Biener  <rguenther@suse.de>
10984         PR tree-optimization/57380
10985         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
10986         least one invariant or re-used load.
10987         * passes.c (init_optimization_passes): Move pass_phiprop before
10988         pass_forwprop.
10990 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
10992         * config/aarch64/aarch64-simd.md
10993         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
10995 2013-05-23  Richard Biener  <rguenther@suse.de>
10997         PR middle-end/57381
10998         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10999         OEP_CONSTANT_ADDRESS_OF retained.
11001 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
11003         PR middle-end/57344
11004         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
11005         don't lower unit.  Handle unit not being always BITS_PER_WORD.
11007 2013-05-23  Richard Biener  <rguenther@suse.de>
11009         PR rtl-optimization/57341
11010         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11011         instead of true_dependence.
11013 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
11015         * bb-reorder.c (branch_threshold): Make const.
11016         (exec_threshold): Ditto.
11018 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11019             Pat Haugen  <pthaugen@us.ibm.com>
11020             Peter Bergner  <bergner@vnet.ibm.com>
11022         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11023         documentation for the power8 crypto builtins.
11025         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11027         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11028         macros for defining power8 builtin functions.
11029         (BU_P8V_AV_2): Likewise.
11030         (BU_P8V_AV_P): Likewise.
11031         (BU_P8V_VSX_1): Likewise.
11032         (BU_P8V_OVERLOAD_1): Likewise.
11033         (BU_P8V_OVERLOAD_2): Likewise.
11034         (BU_CRYPTO_1): Likewise.
11035         (BU_CRYPTO_2): Likewise.
11036         (BU_CRYPTO_3): Likewise.
11037         (BU_CRYPTO_OVERLOAD_1): Likewise.
11038         (BU_CRYPTO_OVERLOAD_2): Likewise.
11039         (XSCVSPDP): Fix typo, point to the correct instruction.
11040         (VCIPHER): Add power8 crypto builtins.
11041         (VCIPHERLAST): Likewise.
11042         (VNCIPHER): Likewise.
11043         (VNCIPHERLAST): Likewise.
11044         (VPMSUMB): Likewise.
11045         (VPMSUMH): Likewise.
11046         (VPMSUMW): Likewise.
11047         (VPERMXOR_V2DI): Likewise.
11048         (VPERMXOR_V4SI: Likewise.
11049         (VPERMXOR_V8HI: Likewise.
11050         (VPERMXOR_V16QI: Likewise.
11051         (VSHASIGMAW): Likewise.
11052         (VSHASIGMAD): Likewise.
11053         (VPMSUM): Likewise.
11054         (VPERMXOR): Likewise.
11055         (VSHASIGMA): Likewise.
11057         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11058         __CRYPTO__ if the crypto instructions are available.
11059         (altivec_overloaded_builtins): Add support for overloaded power8
11060         builtins.
11062         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11063         support for power8 crypto builtins.
11064         (builtin_function_type): Likewise.
11065         (altivec_init_builtins): Add support for builtins that take vector
11066         long long (V2DI) arguments.
11068         * config/rs6000/crypto.md: New file, define power8 crypto
11069         instructions.
11071 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11072             Pat Haugen  <pthaugen@us.ibm.com>
11073             Peter Bergner  <bergner@vnet.ibm.com>
11075         * doc/invoke.texi (Option Summary): Add power8 options.
11076         (RS/6000 and PowerPC Options): Likewise.
11078         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11079         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11080         wm, wn, wr documentation.
11082         * config/rs6000/constraints.md (wm): New constraint for VSX
11083         registers if direct move instructions are enabled.
11084         (wn): New constraint for no registers.
11085         (wq): New constraint for quad word even GPR registers.
11086         (wr): New constraint if 64-bit instructions are enabled.
11087         (wv): New constraint if power8 vector instructions are enabled.
11088         (wQ): New constraint for quad word memory locations.
11090         * config/rs6000/predicates.md (const_0_to_15_operand): New
11091         constraint for 0..15 for crypto instructions.
11092         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11093         as GPR and floating point registers.
11094         (int_reg_operand): New predicate to match only GPR registers.
11095         (base_reg_operand): New predicate to match base registers.
11096         (quad_int_reg_operand): New predicate to match even GPR registers
11097         for quad memory operations.
11098         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11099         operations in both GPR and VSX registers.
11100         (quad_memory_operand): New predicate for quad memory operations.
11101         (reg_or_indexed_operand): New predicate for direct move support.
11103         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11104         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11105         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11106         (POWERPC_MASKS): Add power8 options.
11107         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11108         various options.
11110         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11111         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11113         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11114         (-mpower8-fusion): New power8 options.
11115         (-mpower8-fusion-sign): Likewise.
11116         (-mpower8-vector): Likewise.
11117         (-mcrypto): Likewise.
11118         (-mdirect-move): Likewise.
11119         (-mquad-memory): Likewise.
11121         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11122         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11123         registers.
11124         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11125         (rs6000_debug_vector_unit): Add p8_vector.
11126         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11127         definitions.  Also print fusion state.
11128         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11129         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11130         (rs6000_option_override_internal): Add support for power8.
11131         (rs6000_common_init_builtins): Add debugging for skipped builtins
11132         if -mdebug=builtin.
11133         (rs6000_adjust_cost): Add power8 support.
11134         (rs6000_issue_rate): Likewise.
11135         (insn_must_be_first_in_group): Likewise.
11136         (insn_must_be_last_in_group): Likewise.
11137         (force_new_group): Likewise.
11138         (rs6000_register_move_cost): Likewise.
11139         (rs6000_opt_masks): Likewise.
11141         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11142         power8 capable assembler, default to power7 options.
11143         (TARGET_DIRECT_MOVE): Likewise.
11144         (TARGET_CRYPTO): Likewise.
11145         (TARGET_P8_VECTOR): Likewise.
11146         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11147         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11148         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11149         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11150         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11151         (TARGET_XSCVDPSPN): Likewise.
11152         (TARGET_XSCVSPDPN): Likewsie.
11153         (TARGET_SYNC_HI_QI): Likewise.
11154         (TARGET_SYNC_TI): Likewise.
11155         (MASK_CRYPTO): Likewise.
11156         (MASK_DIRECT_MOVE): Likewise.
11157         (MASK_P8_FUSION): Likewise.
11158         (MASK_P8_VECTOR): Likewise.
11159         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11160         temporary used by some of the direct move instructions to get two FP
11161         temporary registers does not force creation of a stack frame.
11162         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11163         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11164         that any VSX registers are tieable, even if they are also an
11165         Altivec vector mode.
11166         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11167         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11168         (RS6000_BTM_CRYPTO): Likewise.
11169         (RS6000_BTM_COMMON): Likewise.
11171         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11172         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11173         (enum rs6000_vector): Add power8 vector support.
11175 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11177         PR target/19599
11178         PR target/57340
11179         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11180         (any_sibcall_could_use_r3): this and handle indirect calls.
11181         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11183 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11185         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11187 2013-05-22  Richard Biener  <rguenther@suse.de>
11189         PR middle-end/57349
11190         * profile.c (branch_prob): Do not split blocks that are
11191         abnormally receiving from ECF_RETURNS_TWICE functions.
11193 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11195         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11196         address mode.  Move pointer mode initialization to the same place.
11198 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11200         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11201         while it has any effect.
11203 2013-05-21  Easwaran Raman  <eraman@google.com>
11205         PR tree-optimization/57322
11206         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11207         UID of the statement added to the BB to be 1.
11209 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11211         PR tree-optimization/57331
11212         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11213         of conversion from pointer type to integral type with integer.
11215 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11217         PR lto/57289
11218         * ipa-prop.c (ipa_read_node_info): Process param_used and
11219         controlled_uses in the same order as when writing.
11221 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11223         PR plugins/56754
11224         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11226 2013-05-21  Richard Biener  <rguenther@suse.de>
11228         PR tree-optimization/57318
11229         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11230         estimate stmts with side-effects as likely eliminated.
11232 2013-05-21  Richard Biener  <rguenther@suse.de>
11234         PR tree-optimization/57330
11235         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11236         preserve the call stmts fntype.
11238 2013-05-21  Richard Biener  <rguenther@suse.de>
11240         PR tree-optimization/57303
11241         * tree-ssa-sink.c (statement_sink_location): Improve killing
11242         stmt detection and properly handle self-assignments.
11244 2013-05-21  Christian Bruel  <christian.bruel@st.com>
11246         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11247         spanning registers. LEAF_REG_REMAP is supported only for contiguous
11248         registers. Set register size out of the PARALLEL loop.
11250 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
11252         PR target/56547
11253         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11254         (*fmasf4, *fmasf4_media): New insns.
11256 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11258         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11259         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11260         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11261         (mips_idiv_insns): Update the comments to say that the returned
11262         instruction counts are in units of BASE_INSN_LENGTH.
11263         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11264         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
11265         using 2 rather than 4 as the length of indirect MIPS16 and
11266         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
11267         length of a NOP.  Don't divide MIPS16 lengths by 2.
11268         (mips16_split_long_branches): Assume a branch is long if the
11269         length is greater than 4 rather than 8.
11270         * config/mips/mips.md (length): Give MIPS16 lengths directly,
11271         rather than multiplying them by 2.  Multiply instruction counts
11272         by BASE_INSN_LENGTH rather than 4.
11273         (*jump_mips16, tls_get_tp_mips16_<mode>)
11274         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11276 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11278         * config/mips/mips.md (extended_mips16): Remove branch case.
11279         (length): Remove duplicated extended_mips16 test.
11281 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11283         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11285 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
11287         * recog.h (Recog_data): Rename to...
11288         (recog_data_d): ...this.
11289         (recog_data): Update accordingly.
11290         * recog.c (recog_data): Likewise.
11291         * reload.c (save_recog_data): Likewise.
11292         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11293         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11295 2013-05-17  Julian Brown  <julian@codesourcery.com>
11297         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11298         found in a REG_EQUAL note, invalidate it.
11300 2013-05-17  Easwaran Raman  <eraman@google.com>
11302         * tree-ssa-reassoc.c (find_insert_point): New function.
11303         (insert_stmt_after): Likewise.
11304         (get_def_stmt): Likewise.
11305         (ensure_ops_are_available): Likewise.
11306         (not_dominated_by): Likewise.
11307         (rewrite_expr_tree): Do not move statements beyond what is
11308         necessary. Remove call to swap_ops_for_binary_stmt...
11309         (reassociate_bb): ... and move it here.
11310         (build_and_add_sum): Assign UIDs for new statements.
11311         (linearize_expr): Likewise.
11312         (do_reassoc): Renumber gimple statement UIDs.
11314 2013-05-17  Jan Hubicka  <jh@suse.cz>
11316         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11317         weakrefs.
11318         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11319         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11320         weakrefs.
11321         (output_weakrefs): Update.
11323 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
11324             Martin Jambor  <mjambor@suse.cz>
11326         PR middle-end/57276
11327         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11328         value that corresponds to the given aggval is found in values vector.
11330 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
11332         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11333         sse, sse2, sse3, ssse3 and sse4a flags to options.
11335 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
11337         * gengtype-state.c: (s_expr_writer): New class, to handle
11338         prettifying of output layout of s-expressions.
11339         (state_writer): New class, to write out gtype.state.
11340         (state_written_type_count): Move this variable into member data of
11341         state_writer.
11342         (s_expr_writer::s_expr_writer): New code: constructor for new class
11343         (state_writer::state_writer(): ditto
11344         (s_expr_writer::write_new_line): New function
11345         (s_expr_writer::write_any_indent): ditto
11346         (s_expr_writer::begin_s_expr): ditto
11347         (s_expr_writer::end_s_expr): ditto
11348         (write_state_fileloc): convert to method of state_writer...
11349         (state_writer:: write_state_fileloc): ...and use methods of
11350         s_expr_writer to write indentation into the gtype.state output file
11351         to visually represent the hierarchical structure of the list
11352         structures
11353         (write_state_fields): ditto, renaming to...
11354         (state_writer::write_state_fields)
11355         (write_state_a_string): ditto, renaming to...
11356         (state_writer::write_state_a_string)
11357         (write_state_string_option): ditto, renaming to...
11358         (state_writer::write_state_string_option)
11359         (write_state_type_option): ditto, renaming to...
11360         (state_writer::write_state_type_option)
11361         (write_state_nested_option): ditto, renaming to...
11362         (state_writer::write_state_nested_option)
11363         (write_state_option): ditto, renaming to...
11364         (state_writer::write_state_option)
11365         (write_state_options): ditto, renaming to...
11366         (state_writer::write_state_options)
11367         (write_state_lang_bitmap): ditto, renaming to...
11368         (state_writer::write_state_lang_bitmap)
11369         (write_state_version): ditto, renaming to...
11370         (state_writer::write_state_version)
11371         (write_state_scalar_type): ditto, renaming to...
11372         (state_writer::write_state_scalar_type)
11373         (write_state_string_type): ditto, renaming to...
11374         (state_writer::write_state_string_type)
11375         (write_state_undefined_type): ditto, renaming to...
11376         (state_writer::write_state_undefined_type)
11377         (write_state_struct_union_type): ditto, renaming to...
11378         (state_writer::write_state_struct_union_type)
11379         (write_state_struct_type): ditto, renaming to...
11380         (state_writer::write_state_struct_type)
11381         (write_state_user_struct_type): ditto, renaming to...
11382         (state_writer::write_state_user_struct_type)
11383         (write_state_lang_struct_type): ditto, renaming to...
11384         (state_writer::write_state_lang_struct_type)
11385         (write_state_param_struct_type): ditto, renaming to...
11386         (state_writer::write_state_param_struct_type)
11387         (write_state_pointer_type): ditto, renaming to...
11388         (state_writer::write_state_pointer_type)
11389         (write_state_array_type): ditto, renaming to...
11390         (state_writer::write_state_array_type)
11391         (write_state_gc_used): ditto, renaming to...
11392         (state_writer::write_state_gc_used)
11393         (write_state_common_type_content): ditto, renaming to...
11394         (state_writer::write_state_common_type_content)
11395         (write_state_type): ditto, renaming to...
11396         (state_writer::write_state_type)
11397         (write_state_pair_list): ditto, renaming to...
11398         (state_writer::write_state_pair_list)
11399         (write_state_pair): ditto, renaming to...
11400         (state_writer::write_state_pair)
11401         (write_state_typedefs): ditto, renaming to...
11402         (state_writer::write_state_typedefs)
11403         (write_state_structures): ditto, renaming to...
11404         (state_writer::write_state_structures)
11405         (write_state_param_structs): ditto, renaming to...
11406         (state_writer::write_state_param_structs)
11407         (write_state_variables): ditto, renaming to...
11408         (state_writer::write_state_variables)
11409         (write_state_srcdir): ditto, renaming to...
11410         (state_writer::write_state_srcdir)
11411         (write_state_files_list): ditto, renaming to...
11412         (state_writer::write_state_files_list)
11413         (write_state_languages): ditto, renaming to...
11414         (state_writer::write_state_languages)
11415         (write_state): create a state_writer instance and use it when
11416         writing out the state file
11418 2013-05-17  Mike Stump  <mikestump@comcast.net>
11420         PR rtl-optimization/57304
11421         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11422         accessing DF_REF_REAL_LOC.
11424 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11426         PR rtl-optimization/57281
11427         PR rtl-optimization/57300
11428         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11429         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11430         what the other splitter did if the registers are dead.
11432 2013-05-17  Richard Biener  <rguenther@suse.de>
11434         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11435         MEM_REF offsets.
11437 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11439         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11440         linking.
11442 2013-05-17  Marek Polacek  <polacek@redhat.com>
11444         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11445         length when doing non-zero store of storing '\0' to '\0'.
11447 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11449         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11450         vect_external_def oprnd1 with loop_vinfo, try to emit
11451         optional cast, negation and and stmts on the loop preheader
11452         edge instead of into the pattern def seq.
11454         PR tree-optimization/57051
11455         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11456         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11458 2013-05-16  Nick Clifton  <nickc@redhat.com>
11460         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11461         (rl78_is_naked_func): New function.
11462         (rl78_expand_prologue): Skip prologue generation for naked functions.
11463         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11464         * doc/extend.texi (naked): Add RL78 to the list of processors
11465         that supports this attribute.
11467 2013-05-16  Jeff Law  <law@redhat.com>
11469         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11471 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11473         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11474         cache parameters using detect_caches_amd also for CYRIX,
11475         NSC and TM2 signatures.
11477 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11478             Dzianis Kahanovich  <mahatma@eu.by>
11480         PR target/45359
11481         PR target/46396
11482         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11483         VIA/Centaur processors and determine their cache parameters
11484         using detect_caches_amd.
11486 2013-05-16  Teresa Johnson  <tejohnson@google.com>
11488         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11489         (rtl_verify_edges): New function.
11490         (rtl_verify_bb_insns): Ditto.
11491         (rtl_verify_bb_pointers): Ditto.
11492         (rtl_verify_bb_insn_chain): Ditto.
11493         (rtl_verify_fallthru): Ditto.
11494         (rtl_verify_bb_layout): Ditto.
11495         (rtl_verify_flow_info_1): Outline checks into new functions.
11496         (rtl_verify_flow_info): Ditto.
11498 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
11500         * cfghooks.c (copy_bbs): Add update_dominance argument.
11501         * cfghooks.h (copy_bbs): Update prototype.
11502         * tree-cfg.c (gimple_duplicate_sese_region):
11503         Add update_dominance argument.
11504         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11505         * tree-ssa-loop-ch.c (copy_loop_headers): Update
11506         gimple_duplicate_sese_region call.
11507         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11508         Update copy_bbs call.
11509         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11510         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11512 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11514         * tree-vectorizer.h (NUM_PATTERNS): Increment.
11515         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11516         vect_recog_rotate_pattern.
11517         (vect_recog_rotate_pattern): New function.
11519 2013-05-16  Jason Merrill  <jason@redhat.com>
11521         * Makefile.in (LLINKER): New variable.
11522         (mostlyclean): Remove link mutex.
11523         * configure.ac: Handle --enable-link-mutex.
11524         * lock-and-run.sh: New script.
11526 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11528         PR target/19599
11529         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11530         for NULL decl.
11532 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11534         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11536 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11538         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11539         * config/arm/arm.c (next_consecutive_mem): New function.
11540         (gen_movmem_ldrd_strd): Likewise.
11541         * config/arm/arm.md (movmemqi): Update condition and code.
11542         (unaligned_loaddi, unaligned_storedi): New patterns.
11544 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11546         * config.gcc: Obsolete *-*-solaris2.9*.
11547         * doc/install.texi (Specific, *-*-solaris2*): Document it.
11549 2013-05-16  Richard Biener  <rguenther@suse.de>
11551         * passes.c (init_optimization_passes): Move pass_parallelize_loops
11552         earlier, after GRAPHITE transforms and IV canonicalization.
11554 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11556         * omp-low.c (extract_omp_for_data): For collapsed loops,
11557         if at least one of the loops is known at compile time to
11558         iterate zero times, set count to 0.
11559         (expand_omp_regimplify_p): New function.
11560         (expand_omp_for_generic): For collapsed loops, if at least
11561         one of the loops isn't known to iterate at least once,
11562         add runtime check with setting count to 0.
11563         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11564         For unsigned types if it isn't known at compile time that
11565         the loop will iterate at least once, add runtime check to bypass
11566         the whole loop if initial condition isn't true.
11568 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
11570         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11572 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
11574         PR middle-end/57286
11575         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
11576         transformations to avoid an infinite loop.
11578 2013-05-16  Marek Polacek  <polacek@redhat.com>
11580         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
11582 2013-05-15  Leif Ekblad  <leif@rdos.net>
11584         * config/i386/i386.c (ix86_decompose_address): Use
11585         DEFAULT_TLS_SEG_REG to access TLS segment register.
11586         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
11587         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
11588         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
11590 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
11592         PR target/57260
11593         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
11594         sibling calls to functions that would normally be lazily bound,
11595         unless $gp is call-clobbered.
11597 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
11599         * config/i386/i386.c (ix86_option_override_internal): Update
11600         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
11601         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
11602         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
11603         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
11604         of TARGET_3DNOW.
11605         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
11607 2013-05-15  Andreas Schwab  <schwab@suse.de>
11609         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
11610         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
11611         third operand.
11613 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11615         * loop-unroll.c (report_unroll_peel): Check decision before
11616         emitting unroll/peel message.
11618 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11620         * function.h (has_bb_partition): New rtl_data flag.
11621         (bb_reorder_complete): Ditto.
11622         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
11623         instead of flag_reorder_blocks_and_partition.
11624         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
11625         with some enhancements.
11626         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
11627         * bb-reorder.c (connect_traces): Check for has_bb_partition
11628         instead of flag_reorder_blocks_and_partition.
11629         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
11630         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
11631         verify_hot_cold_block_grouping.
11632         (partition_hot_cold_basic_blocks): Set has_bb_partition.
11634 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11636         PR target/19599
11637         * config/arm/predicates.md (call_insn_operand): New predicate.
11638         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
11639         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
11640         if insn is not a tail call.
11641         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
11642         registers.
11643         * config/arm/arm.h (enum reg_class): New caller save register class.
11644         (REG_CLASS_NAMES): Likewise.
11645         (REG_CLASS_CONTENTS): Likewise.
11646         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
11647         without decls.
11649 2013-05-15  Richard Biener  <rguenther@suse.de>
11651         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
11652         of MSG_OPTIMIZED_LOCATIONS.
11653         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
11654         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
11655         message.
11656         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
11657         of MSG_OPTIMIZED_LOCATIONS.
11658         (execute_vect_slp): Likewise.
11659         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
11660         (vect_create_cond_for_alias_checks): Likewise.
11661         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11662         (vect_recog_widen_mult_pattern): Likewise.
11663         (vect_recog_widen_sum_pattern): Likewise.
11664         (vect_recog_over_widening_pattern): Likewise.
11665         (vect_recog_widen_shift_pattern): Likewise.
11666         (vect_recog_vector_vector_shift_pattern): Likewise.
11667         (vect_recog_divmod_pattern): Likewise.
11668         (vect_recog_mixed_size_cond_pattern): Likewise.
11669         (vect_recog_bool_pattern): Likewise.
11670         (vect_pattern_recog_1): Likewise.
11672 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11674         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
11675         non-functions to builtin_unreachable.
11676         * ipa-inline-transform.c (inline_call): Do not assert estimates were
11677         correct when new direct edges were discovered.
11679 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11681         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
11682         header, print symbol order instead of node uid, print more information
11683         about indirect edge targets.
11684         (ipa_make_edge_direct_to_target): Print symbol order instead of node
11685         uids.
11686         (ipa_make_edge_direct_to_target): Likewise.
11687         (remove_described_reference): Likewise.
11688         (propagate_controlled_uses): Likewise.
11689         (ipa_print_node_params): Also print symbol order.
11690         (ipcp_transform_function): Print symbol order instead of node uids.
11691         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
11692         (cgraph_get_create_real_symbol_node): Likewise.
11693         * ipa-cp.c (print_lattice): Likewise.
11694         (print_all_lattices): Likewise.
11695         (determine_versionability): Likewise.
11696         (initialize_node_lattices): Likewise.
11697         (estimate_local_effects): Likewise.
11698         (update_profiling_info): Likewise.
11699         (create_specialized_node): Likewise.
11700         (perhaps_add_new_callers): Likewise.
11701         (decide_about_value): Likewise.
11702         (decide_whether_version_node): Likewise.
11703         (identify_dead_nodes): Likewise.
11704         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
11705         (dump_inline_summary): Likewise.
11706         (estimate_node_size_and_time): Likewise.
11707         (inline_analyze_function): Likewise.
11708         * ipa-inline.c (report_inline_failed_reason): Likewise.
11709         (want_early_inline_function_p): Likewise.
11710         (edge_badness): Likewise.
11711         (update_edge_key): Likewise.
11712         (inline_small_functions): Likewise.  Add dumping of order to two other
11713         dumps.
11714         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
11715         instead of node uids.
11716         (propagate_pure_const): Likewise.
11717         (propagate_pure_const): Likewise.
11718         * ipa-utils.c (dump_cgraph_node_set): Likewise.
11719         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
11720         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
11721         of node uids.
11722         * tree-pretty-print.c (dump_function_header): Likewise.
11723         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
11724         Print symbol order instead of node uids.
11726 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11728         * config/s390/s390.c (s390_register_move_cost): Don't impose the
11729         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
11731 2013-05-15  Richard Biener  <rguenther@suse.de>
11733         PR tree-optimization/57275
11734         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
11735         return value for fail to do runtime alias checks for gather loads.
11737 2013-05-15  Jan Hubicka  <jh@suse.cz>
11739         PR lto/57038
11740         PR lto/47375
11741         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
11742         weakrefs are not external.
11743         (lto_symtab_merge_decls): Fix thinko when dealing with
11744         non-lto_symtab decls.
11745         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
11746         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
11747         * varpool.c (dump_varpool_node): Dump more flags.
11749 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11751         * config/i386/i386.c (processor_alias_table): Add instruction
11752         FSGSBASE for AMD bdver3 architecture.
11754 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11756         * tree.c (warn_deprecated_use): Print file:line using locus color.
11757         * diagnostic.c (diagnostic_report_current_module): Print file:line
11758         and file:line:column using locus color.
11760 2013-05-14  Mike Stump  <mikestump@comcast.net>
11762         * gdbinit.in: Add __null.
11764 2013-05-14  Mike Stump  <mikestump@comcast.net>
11766         * recog.h: Rename struct recog_data to Recog_data.
11767         * recog.c: Likewise.
11768         * reload.c (can_reload_into): Likewise.
11769         * config/picochip/picochip.c: Likewise.
11771 2013-05-14  Mike Stump  <mikestump@comcast.net>
11773         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
11775 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
11777         * resource.h (struct resources): Remove unch_memory member.
11778         (CLEAR_RESOURCE): Don't clear unch_memory.
11779         * resource.c (mark_referenced_resources): Don't set it.
11780         (mark_set_resources): Likewise.
11781         (mark_target_live_regs): Don't clear it.
11782         (init_resource_info): Likewise.
11783         * reorg.c (resource_conflicts_p): Don't compare it.
11784         (redundant_insn): Don't set it.
11786         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
11787         Remove prototypes.
11788         * emit-rtl.c (next_label): Remove unused function.
11789         (skip_consecutive_labels, link_cc0_insns): Move to ...
11790         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
11791         only place where these functions are used, and make them static.
11793 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11795         * fold-const.c (fold_negate_expr): Handle vectors.
11796         (fold_truth_not_expr): Make it static.
11797         (fold_invert_truthvalue): New static function.
11798         (invert_truthvalue_loc): Handle vectors. Do not call
11799         fold_truth_not_expr directly.
11800         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
11801         <TRUTH_NOT_EXPR>: Do not cast to boolean.
11802         (fold_comparison): Handle vector constants.
11803         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
11804         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
11805         * tree.h (fold_truth_not_expr): Remove declaration.
11807 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
11809         * config/aarch64/aarch64-simd.md
11810         (aarch64_vcond_internal<mode>): Rename to...
11811         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
11812         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
11813         float modes. Clarify all iterator modes.
11814         (vcond<mode><mode>): Use new name for vcond expanders.
11815         (vcond<v_cmp_result><mode>): Likewise.
11816         (vcondu<mode><mode>: Likewise.
11817         * config/aarch64/iterators.md (VDQF_COND): New.
11819 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11821         PR bootstrap/57266
11822         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
11823         variable for the shift amount. Check that we shift by non-negative
11824         amounts.
11826 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
11828         PR target/42017
11829         * config/arm/arm.h (EPILOGUE_USES): Only return true
11830         for LR_REGNUM after epilogue_completed.
11832 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
11834         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
11835         is error_mark_node.
11837 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11839         PR target/57261
11840         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
11841         and Solaris 11+/x86 with gld.
11842         * configure: Regenerate.
11844 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11846         * expmed.c (expand_shift_1): Canonicalize rotates by
11847         constant bitsize / 2 to bitsize - 1.
11848         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
11849         case ROTATERT>: Likewise.
11851         Revert:
11852         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
11854         * config/i386/i386.md (rotateinv): New code attr.
11855         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
11856         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
11857         roll $31, %eax, etc.
11859 2013-05-14  Richard Biener  <rguenther@suse.de>
11861         PR middle-end/57235
11862         * tree-eh.c (sink_clobbers): Give up for successors with
11863         multiple predecessors and no virtual uses.
11865 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
11867         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
11868         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
11870 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11872         PR middle-end/57251
11873         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
11874         the case when both op0 and op1 have VOIDmode.
11876 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11878         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
11879         in multiply-accumulate mode.
11881 2013-05-13  Guozhi Wei  <carrot@google.com>
11883         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
11885 2013-05-13  Kai Tietz  <ktietz@redhat.com>
11887         PR target/56975
11888         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
11889         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
11890         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
11891         * config/i386/i386.c (ix86_option_override_internal): Likewise.
11892         (ix86_expand_prologue): Likewise.
11893         (ix86_expand_split_stack_prologue): Likewise.
11894         (legitimate_pic_address_disp_p): Likewise.
11895         (legitimize_pic_address): Likewise.
11896         (legitimize_tls_address): Likewise.
11897         (legitimize_pe_coff_symbol): Likewise.
11898         (output_pic_addr_const): Likewise.
11899         (construct_plt_address): Likewise.
11900         (ix86_expand_call): Likewise.
11901         (x86_output_mi_thunk): Likewise.
11902         (x86_function_profiler): Likewise.
11904 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
11906         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
11907         similar switch cases.
11908         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
11909         (aarch64_simd_mov_to_<mode>low): Delete.
11910         (aarch64_simd_mov_to_<mode>high): Delete.
11911         (move_lo_quad_<mode>): Add w<-r alternative.
11912         (aarch64_simd_move_hi_quad_<mode>): Likewise.
11913         (aarch64_simd_mov_from_*): Update type attribute.
11914         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
11915         statement.
11917 2013-05-13  Jan Hubicka  <jh@suse.cz>
11919         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
11920         * config/i386/i386.c (ix86_compute_frame_layout,
11921         ix86_expand_epilogue, emit_i387_cw_initialization,
11922         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
11923         ix86_local_alignment): Fix use of size/speed predicates.
11925 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11927         PR tree-optimization/45216
11928         PR tree-optimization/57157
11929         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
11930         the (-Y) & (B - 1) variant if OP is |.
11931         * expmed.c (expand_shift_1): For rotations by const0_rtx just
11932         return shifted.  Use (-op1) & (prec - 1) as other_amount
11933         instead of prec - op1.
11935 2013-05-13  Martin Jambor  <mjambor@suse.cz>
11937         PR middle-end/42371
11938         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
11939         (ipa_constant_data): New type.
11940         (ipa_jump_func): Use ipa_constant_data to hold information about
11941         constant jump functions.
11942         (ipa_get_jf_constant): Adjust to jump function type changes.
11943         (ipa_get_jf_constant_rdesc): New function.
11944         (ipa_param_descriptor): New field controlled_uses.
11945         (ipa_get_controlled_uses): New function.
11946         (ipa_set_controlled_uses): Likewise.
11947         * ipa-ref.h (ipa_find_reference): Declare.
11948         * ipa-prop.c (ipa_cst_ref_desc): New type.
11949         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
11950         changes.
11951         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
11952         New parameter cs.  Adjust all callers.
11953         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
11954         (remove_described_reference): New function.
11955         (jfunc_rdesc_usable): Likewise.
11956         (try_make_edge_direct_simple_call): Decrement controlled use count,
11957         attempt to remove reference if it hits zero.
11958         (combine_controlled_uses_counters): New function.
11959         (propagate_controlled_uses): Likewise.
11960         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
11961         (ipa_edge_duplication_hook): Duplicate reference descriptions.
11962         (ipa_print_node_params): Print described use counter.
11963         (ipa_write_jump_function): Adjust to jump function type changes.
11964         (ipa_read_jump_function): New parameter CS, pass it to
11965         ipa_set_jf_constant.  Adjust caller.
11966         (ipa_write_node_info): Stream controlled use count
11967         (ipa_read_node_info): Likewise.
11968         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
11969         asserting.
11970         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
11971         count.  Remove cloning-added reference if it reaches zero.
11972         * ipa-ref.c (ipa_find_reference): New function.
11974 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11976         * config/i386/i386.c (processor_target_table): Modified default
11977         alignment values for AMD BD and BT architectures.
11979 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
11981         * tree-vect-generic.c (uniform_vector_p): Move ...
11982         * tree.c (uniform_vector_p): ... here.
11983         * tree.h (uniform_vector_p): Declare it.
11984         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
11985         into a scalar.
11987 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11989         PR tree-optimization/57230
11990         * tree-ssa-strlen.c (handle_char_store): Record length for
11991         array store from STRING_CST.
11993         PR tree-optimization/57230
11994         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
11995         check.
11997 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
11999         * config/epiphany/epiphany.c (epiphany_init): Check size of
12000         NUM_MODES_FOR_MODE_SWITCHING.
12001         (epiphany_expand_prologue):
12002         Remove CONFIG_REGNUM initial value handling code.
12003         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
12004         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
12005         (emit_set_fp_mode, epiphany_mode_after): Likewise.
12006         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12007         Don't return 1 for FP_MODE_NONE.
12008         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12009         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12010         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12011         * config/epiphany/epiphany.md (save_config): New pattern.
12013 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
12015         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12017 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
12019         * config/i386/i386.md (memory): Handle sseishft1.
12020         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12021         (*vec_extractv2di_1): Ditto.
12023 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12025         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12026         saved registers.
12028 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12030         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12031         Add mthumb/march=armv7-a multilib.
12032         Add mthumb/march=armv7-r multilib.
12033         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12035 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12037         * config/v850/t-rtems: Add more multilibs.
12039 2013-05-10  Richard Biener  <rguenther@suse.de>
12041         PR tree-optimization/57214
12042         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12043         not propagate from SSA names that occur in abnormal PHI nodes.
12045 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12047         * stor-layout.c (element_precision): New function.
12048         * machmode.h (element_precision): Declare it.
12049         * tree.c (build_minus_one_cst): New function.
12050         (element_precision): Likewise.
12051         * tree.h (build_minus_one_cst): Declare new function.
12052         (element_precision): Likewise.
12053         * fold-const.c (operand_equal_p): Use element_precision.
12054         (fold_binary_loc): Handle vector types.
12055         * convert.c (convert_to_integer): Use element_precision.
12056         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12057         separately.
12059 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12061         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12062         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12063         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12064         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12065         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12066         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12067         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12068         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12069         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12070         (Uuw8): New constraints.
12071         (Usb4): Move into alphabetical order.
12072         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12073         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12074         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12075         previously unnamed patterns.
12076         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12077         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12078         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12079         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12080         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12081         of specific lengths.
12083 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12085         * config/i386/i386.md (rotateinv): New code attr.
12086         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12087         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12088         roll $31, %eax, etc.
12090         PR tree-optimization/45216
12091         PR tree-optimization/57157
12092         * tree-ssa-forwprop.c (simplify_rotate): New function.
12093         (ssa_forward_propagate_and_combine): Call it.
12095 2013-05-10  Richard Biener  <rguenther@suse.de>
12097         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12098         disable peeling when we version for aliasing.
12099         (vector_alignment_reachable_p): Honor explicit user alignment.
12100         (vect_supportable_dr_alignment): Likewise.
12101         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12102         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12103         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12104         then peeling to arrange for the cost-model check to come first.
12106 2013-05-10  Alan Modra  <amodra@gmail.com>
12108         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12109         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12110         * configure: Regenerate.
12112 2013-05-10  Alan Modra  <amodra@gmail.com>
12114         PR target/55033
12115         * varasm.c (default_elf_select_section): Move !DECL_P check..
12116         (get_named_section): ..to here before calling get_section_name.
12117         Adjust assertion.
12118         (default_section_type_flags): Add DECL_P check.
12119         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12120         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12122 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12124         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12125         When using gen_stack_adjust_str with a register offset, add a
12126         REG_FRAME_RELATED_EXPR note.
12128 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12130         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12131         (*vec_extractv4si_zext_mem): Ditto.
12132         (*vec_extractv2di): Add 0->x and x->x alternatives.
12133         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12134         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12136 2013-05-09  Jason Merrill  <jason@redhat.com>
12138         N3639 C++1y VLA support
12139         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12140         DECL_VALUE_EXPR.
12142         * tree.c (build_constructor_va): New.
12143         * tree.h: Declare it.
12145 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12147         PR lto/57084
12148         * gimple-fold.c (canonicalize_constructor_val): Call
12149         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12151 2013-05-09  Jan Hubicka  <jh@suse.cz>
12152             Richard Biener  <rguenther@suse.de>
12154         PR lto/54095
12155         * symtab.c (symtab_make_decl_local): Do not add private names.
12157 2013-05-09  Jan Hubicka  <jh@suse.cz>
12159         PR lto/54095
12160         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12161         (unlink_from_assembler_name_hash): Likewise.
12162         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12163         symtab_unregister_node, symtab_initialize_asm_name_hash,
12164         change_decl_assembler_name): Update.
12166 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12168         * config/aarch64/aarch64.md: New movtf split.
12169         (*movtf_aarch64): Update.
12170         (aarch64_movdi_tilow): Handle TF modes and rename to
12171         aarch64_movdi_<mode>low.
12172         (aarch64_movdi_tihigh): Handle TF modes and rename to
12173         aarch64_movdi_<mode>high
12174         (aarch64_movtihigh_di): Handle TF modes and rename to
12175         aarch64_mov<mode>high_di
12176         (aarch64_movtilow_di): Handle TF modes and rename to
12177         aarch64_mov<mode>low_di
12178         (aarch64_movtilow_tilow): Remove spurious whitespace.
12179         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12180         splits.
12181         (aarch64_print_operand): Update.
12183 2013-05-09  Alan Modra  <amodra@gmail.com>
12185         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12186         powerpc64le.
12187         * configure: Regenerate.
12189 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12191         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12192         splitter preparation statements.
12193         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12194         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12195         adjust_address_nv.
12197 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12199         * gimple-ssa-strength-reduction.c (count_candidates): Change
12200         return value to int.
12201         (analyze_candidates_and_replace): Change type of length to int.
12203 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12205         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12206         (*vec_extract<mode>): Use VI12_128 mode iterator.
12207         (*vec_extract<mode>_mem): Ditto.
12208         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12209         attribute.
12211 2013-05-08  Diego Novillo  <dnovillo@google.com>
12213         PR bootstrap/54659
12215         Revert:
12216         2012-08-17  Diego Novillo  <dnovillo@google.com>
12218         PR bootstrap/54281
12219         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12220         * config.in: Regenerate.
12221         * configure: Regenerate.
12222         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12224 2013-05-08  Jan Hubicka  <jh@suse.cz>
12226         PR lto/54095
12227         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12228         * cgraph.h (symtab_node_base): Add unique_name.
12229         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12230         input_overwrite_node, input_varpool_node): Stream unique_name.
12231         * cgraphclones.c (cgraph_create_virtual_clone,
12232         cgraph_function_versioning): Set unique_name.
12233         * ipa.c (function_and_variable_visibility): Set unique_name.
12235 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12237         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12238         (alloc_cand_and_find_basis): Restrict conditional candidate
12239         processing to CAND_MULTs.
12241 2013-05-08  Jan Hubicka  <jh@suse.cz>
12243         PR lto/54095
12244         lto-symtab.c (lto_symtab_symbol_p): New function.
12245         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12246         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12247         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12248         Skip static symbols.
12250 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
12252         PR tree-optimization/57200
12253         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12254         Only call inform if the preceding warning_at returns true.
12256 2013-05-07  Han Shen  <shenhan@google.com>
12258         * cfgexpand.c (record_or_union_type_has_array_p): New function.
12259         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12260         * common.opt (fstack-protector-strong): New option.
12261         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12262         * doc/invoke.texi (Optimization Options): Document
12263         "-fstack-protector-strong".
12264         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12266 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12268         * config/mips/mips.c (mips_machine_reorg2): Return 0.
12270 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
12272         * ira.c (update_equiv_regs): Add insn having equiv memory even if
12273         it is not lhs of the insn.
12274         (setup_reg_equiv): Remove insn having equiv memory which it is not
12275         lhs of the insn.
12276         * lra-constraints.c (process_address): Try to improve generation
12277         code for address base + disp.
12278         (lra_constraints): Make correct the code for checking insn setting
12279         up backward equivalence.  Remove insn only if it is in the init
12280         insn list.
12281         * lra-eliminations.c (update_reg_eliminate): Change return value.
12282         (lra_eliminate): Use the result.
12284 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
12286         * config/i386/sse.md (ssescalarnummask): New mode attribute.
12287         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12288         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12289         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12290         register target operands.
12291         (*vec_extractv8hi_sse2): New pattern.
12292         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12293         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12294         (*vec_extract<mode>_mem): New insn and split pattern.
12296 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
12298         * config/arm/arm.c (arm_asan_shadow_offset): New function.
12299         (TARGET_ASAN_SHADOW_OFFSET): Define.
12300         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12301         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12303 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12305         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12306         (incr_vec_index): Return -1 if increment not found.
12307         (create_add_on_incoming_edge): Assert if increment not found.
12308         (record_increment): Limit number of increments recorded.
12309         (all_phi_incrs_profitable): Return false if an increment not found.
12310         (replace_profitable_candidates): Don't process increments that were
12311         not recorded.
12312         (analyze_candidates_and_replace): Limit size of incr_vec.
12314 2013-05-07  Richard Biener  <rguenther@suse.de>
12316         * calls.c (special_function_p): setjmp-like functions are leaf.
12317         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12318         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12320 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
12322         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12323         (aarch64_simd_mov<mode>): New expander.
12324         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12325         (aarch64_simd_mov_to_<mode>high): Likewise.
12326         (aarch64_simd_mov_from_<mode>low): Likewise.
12327         (aarch64_simd_mov_from_<mode>high): Likewise.
12328         (aarch64_dup_lane<mode>): Update.
12329         (aarch64_dup_lanedi): New instruction pattern.
12330         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12331         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12333 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12335         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12336         (replace_mult_candidate): Remove unnecessary argument; remove
12337         unnecessary parameter from call to introduce_cast_before_cand.
12338         (replace_unconditional_candidate): Remove unnecessary parameter
12339         from call to replace_mult_candidate.
12340         (replace_conditional_candidate): Likewise.
12341         (insert_initializers): Use make_temp_ssa_name.
12342         (introduce_cast_before_cand): Remove unnecessary argument; use
12343         make_temp_ssa_name.
12344         (replace_one_candidate): Remove unnecessary argument; remove
12345         unnecessary parameter from calls to introduce_cast_before_cand.
12346         (replace_profitable_candidates): Remove unnecessary parameters
12347         from calls to replace_one_candidate.
12349 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12351         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12352         phi def as possibly hiding a basis for a CAND_ADD whose operands
12353         have been commuted in the analysis.
12354         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12356 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12358         * config/aarch64/aarch64.md
12359         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12360         shift value between 0-4.
12362 2013-05-07  Richard Biener  <rguenther@suse.de>
12364         * double-int.h (rshift): New overload.
12365         * double-int.c (rshift): New function.
12366         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12367         (create_reference_ops_from_ref): Remove.
12368         (vn_reference_insert): Use shared ops for constructing the
12369         reference and copy it.
12371 2013-05-07  Richard Biener  <rguenther@suse.de>
12373         PR middle-end/57190
12374         * tree-eh.c (sink_clobbers): Properly propagate
12375         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12377 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
12379         PR tree-optimization/57149
12380         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12381         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12382         collect_phi_def_edges, execute_late_warn_uninitialized): Use
12383         uninit_undefined_value_p instead of ssa_undefined_value_p.
12385         PR debug/57184
12386         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12387         for modifier == EXPAND_INITIALIZER.
12389 2013-05-07  Anton Blanchard  <anton@samba.org>
12391         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12392         for powerpc64 little endian.
12393         * configure: Regenerate.
12395 2013-05-06  Graham Stott  <grahams@btinternet.com>
12397         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12398         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12399         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12400         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12402 2013-05-06  Graham Stott  <grahams@btinternet.com>
12404         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12405         codes which allow non-lvalues.
12407 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
12409         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12410         components are all 1s.
12411         (integer_minus_onep): New function.
12412         * tree.h (integer_minus_onep): Declare it.
12413         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12414         integer_minus_onep instead of integer_all_onesp.
12416 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12418         PR target/52933
12419         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12420         variations of these patterns.
12422 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12424         * config/i386/i386.md (isa): Add x64_sse4 member.
12425         (enabled): Handle x64_sse4.
12426         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12427         instruction for 64bit SSE4_1 targets.  Update insn attributes.
12428         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12429         instruction for SSE4_1 targets.  Update insn attributes.
12430         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12431         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12432         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12433         const_1 selector.
12434         (*vec_extractv4si): Rename from *sse4_1_pextrd.
12435         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12436         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12438 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12440         PR target/57108
12441         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12443 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
12445         * final.c (do_assembler_dialects): Don't handle curly braces and
12446         vertical bar escaped by % as dialect delimiters.
12447         (output_asm_insn): Print curly braces and vertical bar if escaped
12448         by % and ASSEMBLER_DIALECT defined.
12449         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12450         * doc/tm.texi: Regenerated.
12452 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12454         * config/mips/mips.c: Include tree-pass.h.
12455         (mips_reorg): Split in pre- and post-dbr_schedule parts.
12456         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12457         (pass_mips_machine_reorg2): New machine specific pass.
12458         (insert_pass_mips_machine_reorg2): New pass plugin definition.
12459         (mips_option_override): Register the new pass.
12460         * rtl.h (cleanup_barriers): Remove prototype.
12461         (dbr_schedule): Likewise.
12462         * jump.c (cleanup_barriers): Make static.
12463         * reorg.c (dbr_schedule): Likewise.
12465 2013-05-06  Richard Biener  <rguenther@suse.de>
12467         PR tree-optimization/57185
12468         * tree-parloops.c (add_field_for_reduction): Handle anonymous
12469         SSA names properly.
12471 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12473         PR target/57106
12474         * config/i386/i386.c (add_parameter_dependencies): Add dependence
12475         between "first_arg" and "insn", not "last" and "insn".
12477 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12479         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12480         (find_candidates_in_block): Re-enable slsr_process_phi.
12481         (create_phi_basis): Fix double counting of candidate adjustment.
12483 2013-05-06  Richard Biener  <rguenther@suse.de>
12485         PR middle-end/57147
12486         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12487         the edge is also fallthru, preserve it and just clear the
12488         abnormal flag.
12489         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12490         also complex, preserve that and just clear the fallthru flag.
12491         * tree-inline.c (update_ssa_across_abnormal_edges): Also
12492         update virtual operands.
12494 2013-05-06  Alan Modra  <amodra@gmail.com>
12496         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12497         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12498         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12499         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12500         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12501         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12503 2013-05-06  Alan Modra  <amodra@gmail.com>
12505         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12506         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
12507         (DEFAULT_ASM_ENDIAN): Define.
12508         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12509         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12510         Update -K PIC clause from sysv4.h.
12511         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12512         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12514 2013-05-06  Alan Modra  <amodra@gmail.com>
12516         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12517         twice for little-endian.
12518         (ashrdi3_no_power, ashrdi3): Support little-endian.
12520 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12522         PR target/55303
12523         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12524         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12525         related expanders.
12526         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12527         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12528         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12529         New predicates.
12531 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
12532             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12534         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12535         * config/pa/pa.opt: Make mbig-switch a no-op.
12536         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12537         (CASE_VECTOR_MODE): Always return SImode.
12538         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12539         for the !TARGET_BIG_SWITCH case.
12540         * config/pa/pa-linux.h: Likewise.
12541         * config/pa/pa-openbsd.h: Likewise.
12542         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12543         * config/pa/pa.md (short_jump): Remove define_insn.
12544         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12545         (casesi0): Remove define_insn.
12546         (type): Remove btable_branch.
12547         (pa_combine_type): Likewise.
12548         (in_nullified_branch_delay): Likewise.
12549         (in_call_delay): Likewise.
12550         (define_delay): Likewise.
12551         (define_insn_reservation "Z3"): Likewise.
12552         (define_insn_reservation "Z4"): Likewise.
12553         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12554         (pa_adjust_insn_length): Remove adjustment for btable branches.
12555         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12556         and mno-big-switch
12558 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
12560         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12561         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12562         Add m->r,x alternatives.
12563         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12564         splitters using SWI48x mode iterator.
12565         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
12566         TARGET_64BIT.  Add m->x alternative.
12567         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12568         Add o->x alternative.  Enable for TARGET_SSE.
12569         (sse_storeq): Remove expander.
12570         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12571         with memory input operand.
12572         (*vec_extractv2di_1 splitter): New.
12573         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12574         * config/i386/i386.md (ssevecmodelower): New mode attribute.
12576 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12578         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
12579         (INT_LOWPART): Delete.
12580         (extract_MB): Adjust.
12581         (extract_ME): Adjust.
12582         (print_operand): Adjust.
12584 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12586         * config/rs6000/predicates.md (reg_or_add_cint_operand,
12587         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12588         (reg_or_logical_cint_operand, easy_fp_constant,
12589         logical_const_operand): Delete "CONST_DOUBLE" case.
12590         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
12591         "HOST_BITS_PER_WIDE_INT == 64" test.
12592         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
12593         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
12594         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
12595         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
12596         test.
12597         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
12598         CONST_DOUBLE DImode/VOIDmode case.
12599         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
12600         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12601         CONST_DOUBLE VOIDmode case.
12602         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12603         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
12604         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
12605         Delete CONST_DOUBLE case.
12606         (splitters for mov FMOVE64 const_double): Delete
12607         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12608         "HOST_BITS_PER_WIDE_INT >= 64" test.
12609         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
12610         case.
12611         (mov DI const_double): Delete.
12613 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
12615         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
12616         on op shows all bits zero in mode of a lowpart subreg, return zero.
12618 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
12620         PR target/57150
12621         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
12622         to save TFmode registers and DImode to save TImode registers for
12623         caller save operations.
12624         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
12625         mark being partially clobbered since they only use the first
12626         double word.
12628         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
12629         and TDmode only use the upper 64-bits of each VSX register.
12631 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12633         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
12634         (find_candidates_in_block): Disable slsr_process_phi.
12636 2013-05-03  Guozhi Wei  <carrot@google.com>
12638         * coverage.c (coverage_obj_init): Move the construction of gcov
12639         constructor to ...
12640         (build_init_ctor): ... here.
12642 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12644         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
12645         (slsr_cand_d): Redefine def_phi.
12646         (stride_status, phi_adjust_status, count_phis_status): New enums.
12647         (find_phi_def): New.
12648         (find_basis_for_base_expr): New.
12649         (find_basis_for_candidate): Handle hidden bases.
12650         (alloc_cand_and_find_basis): Handle phi candidates.
12651         (slsr_process_phi): New.
12652         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
12653         (create_mul_imm_cand): Likewise.
12654         (create_add_ssa_cand): Exclude phi base candidates.
12655         (create_add_imm_cand): Likewise.
12656         (slsr_process_cast): Likewise.
12657         (slsr_process_copy): Likewise.
12658         (find_candidates_in_block): Handle phi candidates.
12659         (dump_candidate): Likewise.
12660         (unconditional_cands): Delete.
12661         (unconditional_cands_with_known_stride_p): Delete.
12662         (phi_dependent_cand_p): New.
12663         (cand_increment): Handle phi-dependent candidates.
12664         (replace_dependent): Delete.
12665         (replace_mult_candidate): New.
12666         (replace_unconditional_candidate): New.
12667         (incr_vec_index): Move to avoid forward reference.
12668         (create_add_on_incoming_edge): New.
12669         (create_phi_basis): New.
12670         (replace_dependents): Delete.
12671         (replace_conditional_candidate): New.
12672         (phi_add_costs): New.
12673         (replace_uncond_cands_and_profitable_phis): New.
12674         (record_increment): Handle phi adjustments.
12675         (record_phi_increments): New.
12676         (record_increments): Handle phi adjustments.
12677         (phi_incr_cost): New.
12678         (lowest_cost_path): Handle phis.
12679         (total_savings): Likewise.
12680         (analyze_increments): Likewise.
12681         (ncd_with_phi): New.
12682         (ncd_of_cand_and_phis): New.
12683         (nearest_common_dominator_for_cands): Handle phi increments.
12684         (all_phi_incrs_profitable): New.
12685         (replace_profitable_candidates): Handle phi-dependent candidates.
12686         (analyze_candidates_and_replace): Likewise.
12688 2013-05-03  Teresa Johnson  <tejohnson@google.com>
12690         PR bootstrap/57154
12691         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
12692         do not exceed REG_BR_PROB_BASE.
12694 2013-05-03  Jeff Law  <law@redhat.com>
12696         PR tree-optimization/57144
12697         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
12698         operand of the condition will bit into the new type when eliminating
12699         a cast feeding a condition.
12701 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
12703         PR rtl-optimization/57130
12704         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
12705         of COMPARE as in_code to the recursive call if needed.
12707 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
12709         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
12710         (enabled): Handle new members.
12711         * config/i386/sse.md (*vec_concatv2si): Merge from
12712         *vec_concatv2si_sse2 and vec_concatv2si_sse.
12713         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
12715 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
12717         PR tree-optimization/57027
12718         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
12719         for fnms opportunity, check we got the prerequisite kind
12720         of tree / gimple before using accessor functions.
12722 2013-05-03  Richard Biener  <rguenther@suse.de>
12724         * double-int.h (lshift): New overload without precision
12725         and arith argument.
12726         (operator *=, operator +=, operator -=): Move ...
12727         * double-int.c (operator *=, operator +=, operator -=): ... here
12728         and implement more efficiently.
12729         (mul_double_with_sign): Remove.
12730         (lshift_double): Adjust to take unsinged shift argument, push
12731         dispatching code to callers.
12732         (mul_double_wide_with_sign): Add early out for callers that
12733         are not interested in high parts or overflow.
12734         (lshift): New function.
12735         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
12736         dispatch code here.
12737         (lrotate, rrotate): Use logical shifts.
12738         * expr.c (get_inner_reference): Use lshift.
12739         * fixed-value.c (do_fixed_divide): Likewise.
12740         * tree-dfa.c (get_ref_base_and_extent): Likewise.
12741         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
12742         (indirect_refs_may_alias_p): Likewise.
12743         (stmt_kills_ref_p_1): Likewise.
12745 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12747         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
12749 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12751         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
12752         scalar form of FABD instruction.
12754 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
12756         * lra-constraints.c (process_alt_operands): Add checking alt
12757         number to choose the best alternative.
12759 2013-05-02  Richard Biener  <rguenther@suse.de>
12761         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
12762         bitmap and its handling.
12763         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
12765 2013-05-02  Richard Biener  <rguenther@suse.de>
12767         PR middle-end/57140
12768         * tree-inline.c (copy_loops): Properly handle removed loops.
12769         (copy_cfg_body): Mark destination loops for fixup if source
12770         loops needed fixup.
12772 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
12774         PR target/56732
12775         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
12776         generating simple_return for naked functions.
12778 2013-05-02  Martin Jambor  <mjambor@suse.cz>
12780         PR middle-end/56988
12781         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
12782         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
12783         flags match.
12784         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
12785         ipa_agg_replacement_value structures.
12786         (known_aggs_to_agg_replacement_list): Likewise.
12787         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
12788         (read_agg_replacement_chain): Likewise.
12789         (ipcp_transform_function): Also check that by_ref flags match.
12791 2013-05-02  Richard Biener  <rguenther@suse.de>
12793         * graphds.h (struct graph): Add obstack member.
12794         * graphds.c (new_graph): Initialize obstack and allocate
12795         vertices from it.
12796         (add_edge): Allocate edge from the obstack.
12797         (free_graph): Free the obstack instead of all edges and vertices.
12799 2013-05-02  Teresa Johnson  <tejohnson@google.com>
12801         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
12802         divides.
12803         * cfg.c (update_bb_profile_for_threading): Ditto.
12804         * tree-inline.c (copy_bb): Ditto.
12805         (copy_edges_for_bb): Ditto.
12806         (initialize_cfun): Ditto.
12807         (copy_cfg_body): Ditto.
12808         (expand_call_inline): Ditto.
12809         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
12810         (estimate_node_size_and_time): Ditto.
12811         (inline_merge_summary): Ditto.
12812         * cgraphclones.c (cgraph_clone_edge): Ditto.
12813         (cgraph_clone_node): Ditto.
12814         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12815         (compute_trg_info): Ditto.
12817 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12819         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
12820         S reg when fp attribute set.
12821         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
12823 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12825         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
12826         New pattern.
12827         (*and_one_cmplsi3_compare0_uxtw): Likewise.
12828         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
12829         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
12831 2013-05-02  Richard Biener  <rguenther@suse.de>
12833         * tree-scalar-evolution.c (scev_info_hasher): Remove.
12834         (struct instantiate_cache_entry): New type.
12835         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
12836         (struct instantiate_cache_type): New type.
12837         (set_instantiated_value, get_instantiated_value): Remove.
12838         (get_instantiated_value_entry): New function.
12839         (instantiate_scev_name): Use the new cache and adjust.
12840         (instantiate_scev_poly): Adjust.
12841         (instantiate_scev_binary): Likewise.
12842         (instantiate_array_ref): Likewise.
12843         (instantiate_scev_convert): Likewise.
12844         (instantiate_scev_not): Likewise.
12845         (instantiate_scev_3): Likewise.
12846         (instantiate_scev_2): Likewise.
12847         (instantiate_scev_r): Likewise.
12848         (instantiate_scev): Likewise.
12849         (resolve_mixers): Likewise.
12851 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
12853         PR target/57091
12854         * lra-constraints.c (best_small_class_operands_num): Remove.
12855         (process_alt_operands): Remove small_class_operands_num.  Take
12856         small classes operands into losers and only if the operand is not
12857         matched.  Modify debugging output.
12858         (curr_insn_transform): Remove best_small_class_operands_num.
12859         Print insn name.
12861 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12863         * config/aarch64/aarch64-builtins.c
12864         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
12865         * config/aarch64/aarch64-simd-builtins.def
12866         (reduc_splus_): Add new modes.
12867         (reduc_uplus_): New.
12868         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
12869         (reduc_uplus_v4sf): Likewise.
12870         (reduc_splus_v4sf): Likewise.
12871         (aarch64_addv<mode>): Likewise.
12872         (reduc_uplus_<mode>): Likewise.
12873         (reduc_splus_<mode>): Likewise.
12874         (aarch64_addvv2di): Likewise.
12875         (reduc_uplus_v2di): Likewise.
12876         (reduc_splus_v2di): Likewise.
12877         (aarch64_addvv2si): Likewise.
12878         (reduc_uplus_v2si): Likewise.
12879         (reduc_splus_v2si): Likewise.
12880         (reduc_<sur>plus_<mode>): New.
12881         (reduc_<sur>plus_v2di): Likewise.
12882         (reduc_<sur>plus_v2si): Likewise.
12883         (reduc_<sur>plus_v4sf): Likewise.
12884         (aarch64_addpv4sf): Likewise.
12885         * config/aarch64/arm_neon.h
12886         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
12887         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
12888         add UNSPEC_SADDV, UNSPEC_UADDV.
12889         (SUADDV): New.
12890         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
12892 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12894         * config/aarch64/arm_neon.h
12895         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
12897 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12899         * config/aarch64/aarch64-builtins
12900         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
12902 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12904         * config/aarch64/aarch64-simd-builtins.def
12905         (reduc_smax_): New.
12906         (reduc_smin_): Likewise.
12907         (reduc_umax_): Likewise.
12908         (reduc_umin_): Likewise.
12909         (reduc_smax_nan_): Likewise.
12910         (reduc_smin_nan_): Likewise.
12911         (fmax): Remove.
12912         (fmin): Likewise.
12913         (smax): Update for V2SF, V4SF and V2DF modes.
12914         (smin): Likewise.
12915         (smax_nan): New.
12916         (smin_nan): Likewise.
12917         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
12918         (<su><maxmin><mode>3): ...This, refactor.
12919         (s<maxmin><mode>3): New.
12920         (<maxmin_uns><mode>3): Likewise.
12921         (reduc_<maxmin_uns>_<mode>): Refactor.
12922         (reduc_<maxmin_uns>_v4sf): Likewise.
12923         (reduc_<maxmin_uns>_v2si): Likewise.
12924         (aarch64_<fmaxmin><mode>: Remove.
12925         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
12926         new builtin names.
12927         (vmin<q>_f<32,64>): Likewise.
12928         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12929         (FMAXMIN): New.
12930         (su): Add mappings for smax, smin, umax, umin.
12931         (maxmin): New.
12932         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12933         (FMAXMIN): Rename as...
12934         (FMAXMIN_UNS): ...This.
12935         (maxminv): Remove.
12936         (fmaxminv): Likewise.
12937         (fmaxmin): Likewise.
12938         (maxmin_uns): New.
12939         (maxmin_uns_op): Likewise.
12941 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12943         * config/aarch64/arm_neon.h
12944         (vac<ge, gt><sd>_f<32, 64>): Rename to...
12945         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
12946         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
12948 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12950         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
12951         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
12953 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12955         * config/aarch64/aarch64-simd.md
12956         (vcond<mode>_internal): Handle special cases for constant masks.
12957         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
12958         (vcondu<mode><mode>): Likewise.
12959         (vcond<v_cmp_result><mode>): New.
12961 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12963         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
12964         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
12965         * config/aarch64/aarch64-simd-builtins.def
12966         (cmeq): Update to BUILTIN_VALLDI.
12967         (cmgt): Likewise.
12968         (cmge): Likewise.
12969         (cmle): Likewise.
12970         (cmlt): Likewise.
12971         * config/aarch64/arm_neon.h
12972         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
12973         to builtins or C as appropriate.
12975 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12977         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
12978         (cmgeu): ...This.
12979         (cmhi): Rename to...
12980         (cmgtu): ...This.
12981         * config/aarch64/aarch64-simd.md
12982         (simd_mode): Add SF.
12983         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
12984         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
12985         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
12986         (cstore<mode>_neg): ...This.
12987         * config/aarch64/iterators.md
12988         (VALLF): new.
12989         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
12990         (COMPARISONS): New.
12991         (UCOMPARISONS): Likewise.
12992         (optab): Add missing comparisons.
12993         (n_optab): New.
12994         (cmp_1): Likewise.
12995         (cmp_2): Likewise.
12996         (CMP): Likewise.
12997         (cmp): Remove.
12998         (VCMP_S): Likewise.
12999         (VCMP_U): Likewise.
13000         (V_cmp_result): Add DF, SF modes.
13001         (v_cmp_result): Likewise.
13002         (v): Likewise.
13003         (vmtype): Likewise.
13004         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13006 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
13008         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13009         define_insn to define_insn_and_split.
13010         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13011         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13012         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13013         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13014         (thumb2_negscc): Likewise.
13016 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13018         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13020 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13022         * config/arm/thumb2.md: Remove trailing whitespaces.
13024 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13026         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13027         Use gen_int_mode rather than GEN_INT.
13029 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13031         * value-prof.c (stream_in_histogram_value): Remove the strayed
13032         debug_gimple_stmt.
13034 2013-04-30  Richard Biener  <rguenther@suse.de>
13036         PR middle-end/57122
13037         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13039 2013-04-30  Richard Biener  <rguenther@suse.de>
13041         PR middle-end/57107
13042         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13044 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13046         PR rtl-optimization/56957
13047         PR rtl-optimization/57105
13048         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13049         variable.  Use just INSN_UID for determining whether an insn
13050         should be only disconnected from the insn stream.
13051         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13053 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13055         PR tree-optimization/57104
13056         * tsan.c (instrument_expr): Don't instrument accesses to
13057         DECL_HARD_REGISTER VAR_DECLs.
13059 2013-04-30  Richard Biener  <rguenther@suse.de>
13061         * function.h (loops_for_fn): New inline function.
13062         (set_loops_for_fn): Likewise.
13063         * cfgloop.h (place_new_loop): Add struct function parameter.
13064         (get_loop): Likewise.
13065         (get_loops): Likewise.
13066         (number_of_loops): Likewise.
13067         (fel_next): Adjust.
13068         (fel_init): Likewise.
13069         * cfg.c (get_loop_copy): Adjust.
13070         * cfgloop.c (flow_loops_dump): Likewise.
13071         (record_loop_exits): Likewise.
13072         (verify_loop_structure): Likewise.
13073         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13074         (estimate_reg_pressure_cost): Likewise.
13075         (mark_loop_exit_edges): Likewise.
13076         * cfgloopmanip.c (place_new_loop): Likewise.
13077         (add_loop): Likewise.
13078         (duplicate_loop): Likewise.
13079         * graph.c (draw_cfg_nodes): Likewise.
13080         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13081         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13082         (extract_affine_chrec): Likewise.
13083         (build_scop_iteration_domain): Likewise.
13084         * graphite.c (graphite_initialize): Likewise.
13085         * ira-build.c (create_loop_tree_nodes): Likewise.
13086         (more_one_region_p): Likewise.
13087         (rebuild_regno_allocno_maps): Likewise.
13088         (mark_loops_for_removal): Likewise.
13089         (mark_all_loops_for_removal): Likewise.
13090         (remove_unnecessary_regions): Likewise.
13091         (ira_build): Likewise.
13092         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13093         * loop-init.c (fix_loop_structure): Likewise.
13094         (gate_rtl_move_loop_invariants): Likewise.
13095         (gate_rtl_unswitch): Likewise.
13096         (gate_rtl_unroll_and_peel_loops): Likewise.
13097         (rtl_doloop): Likewise.
13098         * lto-streamer-in.c (input_cfg): Likewise.
13099         * lto-streamer-out.c (output_cfg): Likewise.
13100         * modulo-sched.c (sms_schedule): Likewise.
13101         * predict.c (tree_estimate_probability): Likewise.
13102         (tree_estimate_probability_driver): Likewise.
13103         (estimate_loops): Likewise.
13104         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13105         (move_sese_region_to_fn): Likewise.
13106         (debug_loop_num): Likewise.
13107         * tree-chrec.c (chrec_evaluate): Likewise.
13108         (hide_evolution_in_other_loops_than_loop): Likewise.
13109         (chrec_component_in_loop_num): Likewise.
13110         (reset_evolution_in_loop): Likewise.
13111         (evolution_function_is_invariant_rec_p): Likewise.
13112         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13113         * tree-inline.c (copy_loops): Likewise.
13114         (copy_cfg_body): Likewise.
13115         (tree_function_versioning): Likewise.
13116         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13117         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13118         Likewise.
13119         (add_to_evolution_1): Likewise.
13120         (scev_const_prop): Likewise.
13121         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13122         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13123         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13124         (tree_ssa_lim_initialize): Likewise.
13125         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13126         (verify_loop_closed_ssa): Likewise.
13127         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13128         (tree_ssa_loop_im): Likewise.
13129         (tree_ssa_loop_unswitch): Likewise.
13130         (tree_vectorize): Likewise.
13131         (check_data_deps): Likewise.
13132         (tree_ssa_loop_ivcanon): Likewise.
13133         (tree_ssa_loop_bounds): Likewise.
13134         (tree_complete_unroll): Likewise.
13135         (tree_complete_unroll_inner): Likewise.
13136         (tree_parallelize_loops): Likewise.
13137         (tree_ssa_loop_prefetch): Likewise.
13138         (tree_ssa_loop_ivopts): Likewise.
13139         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13140         * tree-vectorizer.c (vectorize_loops): Likewise.
13142 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13144         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13145         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13146         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13147         with EABI_LINK_SPEC.
13149 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13151         PR target/44578
13152         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13153         alternative.
13155 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13157         PR target/57097
13158         * lra-constraints.c (process_alt_operands): Discourage a bit more
13159         using memory for pseudos.  Print cost dump for alternatives.
13160         Modify cost values for conflicts with early clobbers.
13161         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13163 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13165         PR target/57098
13166         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13168 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13170         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13171         from/to S register.
13172         (movdi_aarch64): Support LDR/STR from/to D register.
13174 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13176         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13177         or higher by default.
13179 2013-04-29  Richard Biener  <rguenther@suse.de>
13181         PR middle-end/57075
13182         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13183         even if not adding abnormal edges for calls that can make
13184         abnormal gotos.
13186 2013-04-29  Richard Biener  <rguenther@suse.de>
13188         PR middle-end/57103
13189         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13190         TREE_BLOCK.
13191         (move_stmt_r): Remove redundant checking.
13193 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13195         PR bootstrap/57077
13196         * basic-block.h (apply_scale): New function.
13197         (apply_probability): Use apply_scale.
13198         * gimple-streamer-in.c (input_bb): Ditto.
13199         * lto-streamer-in.c (input_cfg): Ditto.
13200         * lto-cgraph.c (merge_profile_summaries): Ditto.
13201         * tree-optimize.c (execute_fixup_cfg): Ditto.
13202         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13203         (copy_edges_for_bb): Ditto.
13204         (copy_cfg_body): Ditto.
13206 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13208         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13209         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13210         (tail_merge_optimize): Handle current_loops == NULL.
13212 2013-04-26  Jeff Law  <law@redhat.com>
13214         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13215         (simplify_cond_using_ranges): Generalize code to simplify
13216         COND_EXPRs where one argument is a constant and the other
13217         is an SSA_NAME created by an integral type conversion.
13219 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13221         * config/arm/arm.md (store_minmaxsi): Use only when
13222         optimize_insn_for_size_p.
13224 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13226         PR target/57108
13227         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13229 2013-04-29  Richard Biener  <rguenther@suse.de>
13231         PR middle-end/57089
13232         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13233         loop tree make sure to schedule a fixup for the child as well.
13234         (expand_omp_for_generic): Properly add loops.
13235         (expand_omp_for_static_nochunk): Likewise.
13236         (expand_omp_for_static_chunk): Likewise.
13237         (expand_omp_for): For the degenerate case fixup loops.
13238         (expand_omp_sections): Fix default bb placement in loops.
13239         (expand_omp_atomic_pipeline): Properly add loops.
13241 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13243         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13245 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13247         * tree-ssa-tail-merge.c: Update header comment.
13249 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13251         * config/aarch64/arm_neon.h
13252         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13253         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13254         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13255         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13256         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13257         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13258         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13259         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13261 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13263         * config/aarch64/aarch64-simd.md
13264         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13265         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13266         fix_trunc, fixuns_trunc.
13267         (ftrunc<VDQF:mode>2): New.
13268         * config/aarch64/iterators.md (optab): Add fix, fixuns.
13269         (fix_trunc_optab): New.
13271 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13273         * config/aarch64/aarch64-builtins.c
13274         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13275         iceilf, lround, iroundf.
13277 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13279         PR target/54349
13280         * config/i386/i386.h (enum ix86_tune_indices)
13281         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13282         New, split from X86_TUNE_INTER_UNIT_MOVES.
13283         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13284         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13285         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13286         (TARGET_INTER_UNIT_MOVES): Remove.
13287         * config/i386/i386.c (initial_ix86_tune_features): Update.
13288         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13289         (ix86_expand_convert_uns_didf_sse): Use
13290         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13291         (ix86_expand_vector_init_one_nonzero): Ditto.
13292         (ix86_expand_vector_init_interleave): Ditto.
13293         (inline_secondary_memory_needed): Return true for moves from SSE class
13294         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13295         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13296         * config/i386/constraints.md (Yi, Ym): Depend on
13297         TARGET_INTER_UNIT_MOVES_TO_VEC.
13298         (Yj, Yn): New constraints.
13299         * config/i386/i386.md (*movdi_internal): Change constraints of
13300         operand 1 from Yi to Yj and from Ym to Yn.
13301         (*movsi_internal): Ditto.
13302         (*movdf_internal): Ditto.
13303         (*movsf_internal): Ditto.
13304         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13305         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13306         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13307         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13308         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13309         * config/i386/sse.md (movdi_to_sse): Ditto.
13310         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13311         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13312         TARGET_INTER_UNIT_MOVES.
13313         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13314         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13315         instead of TARGET_INTER_UNIT_MOVES.
13316         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13317         operand 1 from Yi to Yj and from Ym to Yn.
13319 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13321         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13322         (float_truncate_hi_): Likewise.
13323         (float_extend_lo_): Likewise.
13324         (float_truncate_lo_): Likewise.
13325         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13326         (aarch64_float_extend_lo_v2df): Likewise.
13327         (vec_unpacks_hi_v4sf): Likewise.
13328         (aarch64_float_truncate_lo_v2sf): Likewise.
13329         (aarch64_float_truncate_hi_v4sf): Likewise.
13330         (vec_pack_trunc_v2df): Likewise.
13331         (vec_pack_trunc_df): Likewise.
13333 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13335         * config/aarch64/aarch64-builtins.c
13336         (aarch64_fold_builtin): Fold float conversions.
13337         * config/aarch64/aarch64-simd-builtins.def
13338         (floatv2si, floatv4si, floatv2di): New.
13339         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13340         * config/aarch64/aarch64-simd.md
13341         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13342         * config/aarch64/iterators.md (FLOATUORS): New.
13343         (optab): Add float, floatuns.
13344         (su_optab): Likewise.
13346 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13348         * config/aarch64/aarch64-builtins.c
13349         (aarch64_builtin_vectorized_function): Use new names for
13350         fcvt builtins.
13351         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13352         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13353         (fcvtzu): Split as...
13354         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13355         (fcvtas): Split as...
13356         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13357         (fcvtau): Split as...
13358         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13359         (fcvtps): Split as...
13360         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13361         (fcvtpu): Split as...
13362         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13363         (fcvtms): Split as...
13364         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13365         (fcvtmu): Split as...
13366         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13367         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13368         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13369         (lfrintnusf, lfrintnudf): Likewise.
13370         * config/aarch64/aarch64-simd.md
13371         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13372         define_insn.
13373         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13374         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13375         (fcvt_pattern): Likewise.
13377 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13379         * config/aarch64/aarch64-simd.md
13380         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13381         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13383 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13385         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13386         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13387         (vrnd<a,m,n,p>_f32): Implement using builtins.
13388         (vrnd<i,x><q>_f<32, 64>): New.
13390 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13392         * config/aarch64/aarch64-builtins.c
13393         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13394         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13395         (frintz): Rename to...
13396         (btrunc): ...this.
13397         (frintp): Rename to...
13398         (ceil): ...this.
13399         (frintm): Rename to...
13400         (floor): ...this.
13401         (frinti): Rename to...
13402         (nearbyint): ...this.
13403         (frintx): Rename to...
13404         (rint): ...this.
13405         (frinta): Rename to...
13406         (round): ...this.
13407         * config/aarch64/aarch64-simd.md
13408         (aarch64_frint<frint_suffix><mode>): Delete.
13409         (<frint_pattern><mode>2): Convert to insn.
13410         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13411         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13412         (frint_pattern): Likewise.
13413         (frint_suffix): Likewise.
13415 2013-04-29  Richard Biener  <rguenther@suse.de>
13417         PR tree-optimization/57081
13418         * loop-init.c: Include tree-flow.h.
13419         (loop_optimizer_finalize): Free number of iteration estimates.
13420         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13422 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
13424         PR tree-optimization/57083
13425         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13426         non-singleton shift count range, zero extend low_bound for uns case.
13428         * config/i386/predicates.md (general_vector_operand): New predicate.
13429         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13430         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13431         if they aren't nonimmediate operands.  If their original values
13432         satisfy const_vector_equal_evenodd_p, don't shift them.
13433         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13434         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
13435         if not nonimmediate_operand.
13436         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13437         instead of register_operand.
13438         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13440 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
13442         * stor-layout.c (finalize_size_functions): Allocate a structure and
13443         reset cfun before dumping the functions.
13445 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
13447         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13449         PR target/56866
13450         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13451         use xop_pmacsdqh if uns_p.
13452         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13453         the immediate rotate count.
13455 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
13457         * rtl.h (struct rtx_def): Add comment for field jump.
13458         (LRA_SUBREG_P): New macro.
13459         * recog.c (register_operand): Check LRA_SUBREG_P.
13460         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13461         stack frame.
13462         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13463         (lra_final_code_change): Skip subreg change for operators.
13464         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13465         if there are no operand changes.
13466         * lra-constraints.c (curr_insn_set): New.
13467         (match_reload): Set LRA_SUBREG_P.
13468         (emit_spill_move): Ditto.
13469         (check_and_process_move): Use curr_insn_set. Process only single
13470         set insns.  Don't initialize sec_mem_p and change_p.
13471         (simplify_operand_subreg): Use LRA_SUBREG_P.
13472         (reg_in_class_p): New function.
13473         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
13474         of #ifdef.  Add code to remove cycling.
13475         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13476         non-null disp.  Reload inner instead of disp when base and index
13477         are null.  Try to put lo_sum into register.
13478         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13479         (check_and_process_move): Move code for move cost check to
13480         simple_move_p.  Remove equiv_substitution.
13481         (simple_move_p): New function.
13482         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
13483         curr_insn_set.  Call check_and_process_move only for single set
13484         insns.  Use the new function.  Move call of check_and_process_move
13485         after operand equiv substitution and address process.
13487 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13489         PR go/57045
13490         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13491         with nonlocal goto receivers or returns twice calls, ignore
13492         unininitialized values from abnormal edges to nl goto receiver
13493         or returns twice call.
13495 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13497         PR tree-optimization/57051
13498         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13499         and VEC_RSHIFT_EXPR if shift count is a multiple of element
13500         bitsize.
13502 2013-04-26  Richard Biener  <rguenther@suse.de>
13504         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13505         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
13506         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13507         (fixup_loop_arrays_after_move): New function.
13508         (move_sese_region_to_fn): Properly outline the loop tree parts
13509         of the SESE region.
13511 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
13513         * config/i386/i386.md (type, unit): Fix long lines.
13515 2013-04-26  Richard Biener  <rguenther@suse.de>
13517         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13518         (lto-streamer-out.o): Likewise.
13519         * cfgloop.c (init_loops_structure): Export, add struct function
13520         argument and adjust.
13521         (flow_loops_find): Adjust.
13522         * cfgloop.h (enum loop_estimation): Add EST_LAST.
13523         (init_loops_structure): Declare.
13524         * lto-streamer-in.c: Include cfgloop.h.
13525         (input_cfg): Input the loop tree.
13526         * lto-streamer-out.c: Include cfgloop.h.
13527         (output_cfg): Output the loop tree.
13528         (output_struct_function_base): Do not drop PROP_loops.
13530 2013-03-26  Richard Biener  <rguenther@suse.de>
13532         * tree-cfg.c (execute_build_cfg): Build the loop tree.
13533         (pass_build_cfg): Provide PROP_loops.
13534         (move_sese_region_to_fn): Remove loops that are outlined into fn
13535         for now.
13536         * tree-inline.c: Include cfgloop.h.
13537         (initialize_cfun): Do not drop PROP_loops.
13538         (copy_loops): New function.
13539         (copy_cfg_body): Copy loop structure.
13540         (tree_function_versioning): Initialize destination loop tree.
13541         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13542         (pass_parallelize_loops): Do IL verification.
13543         * loop-init.c (loop_optimizer_init): Fixup loops if required.
13544         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13545         the CFG make sure we fixup loops as well.
13546         * tree-ssa-tail-merge.c: Include cfgloop.h.
13547         (replace_block_by): When merging loop latches mark loops for fixup.
13548         * lto-streamer-out.c (output_struct_function_base): Drop
13549         PROP_loops for now.
13550         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13551         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13552         * ipa-split.c: Include cfgloop.h.
13553         (split_function): Add the new return block to the loop tree root.
13554         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13555         whether we have removed the forwarder block.
13556         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13557         * cfgloop.h (place_new_loop): Declare.
13558         * cfgloopmanip.c (place_new_loop): Export.
13559         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13560         (tree-switch-conversion.o): Likewise.
13561         (tree-complex.o): Likewise.
13562         (tree-inline.o): Likewise.
13563         (tree-ssa-tailmerge.o): Likewise.
13564         (ipa-split.o): Likewise.
13565         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13566         (tree-ssa-copy.o): Likewise.
13567         * tree-switch-conversion.c: Include cfgloop.h
13568         (process_switch): If we emit a bit-test cascade, schedule loops
13569         for fixup.
13570         * tree-complex.c: Include cfgloop.h.
13571         (expand_complex_div_wide): Properly add new basic-blocks to loops.
13572         * asan.c: Include cfgloop.h.
13573         (create_cond_insert_point): Properly add new basic-blocks to
13574         loops, schedule loop fixup.
13575         * cfgloop.c (verify_loop_structure): Check that looks are not
13576         marked for fixup.
13577         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
13578         to loops.
13579         (expand_omp_for_generic): Likewise.
13580         (expand_omp_sections): Likewise.
13581         (expand_omp_atomic_pipeline): Schedule loops for fixup.
13582         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
13583         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
13584         is initialized, not when loops are present.
13585         * tree-parloops.c (parallelize_loops): Remove checking here.
13586         * passes.c (init_optimization_passes): Schedule a copy-propagation
13587         pass before complete unrolling of inner loops.
13589 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13591         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
13592         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
13593         (colorize_init): Add argument to _WIN32 version.
13594         * toplev.c: Include diagnostic-color.h.
13595         (process_options): Default to -fdiagnostics-color=auto if
13596         GCC_COLORS env var is in the environment.
13597         * common.opt (fdiagnostics-color=): Add Var and Init.
13598         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
13599         env var is in the environment, the default is auto rather than never.
13601         * diagnostic.h (file_name_as_prefix): Add context argument.
13602         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
13603         the string as locus.
13604         * langhooks.c (lhd_print_error_function): Adjust caller.
13606 2013-04-25  Lawrence Crowl  <crowl@google.com>
13608         * var-tracking.c (shared_hash_def::htab):
13609         Change type to hash_table.  Update dependent calls and types.
13611 2013-04-25  Lawrence Crowl  <crowl@google.com>
13613         * Makefile.in: Update as needed below.
13615         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
13616         Move declaration to after the type's method definitons.
13618         * attribs.c (htab_t scoped_attributes::attribute_hash):
13619         Change type to hash_table.  Update dependent calls and types.
13621         * bitmap.c (htab_t bitmap_desc_hash):
13622         Change type to hash_table.  Update dependent calls and types.
13624         * cselib.c (htab_t cselib_hash_table):
13625         Change type to hash_table.  Update dependent calls and types.
13627         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
13628         (hash_string_slot_node): Move implementation into lto-streamer.h
13629         struct string_slot_hasher.
13630         (eq_string_slot_node): Likewise.
13632         * data-streamer-out.c: Update output_block::string_hash_table
13633         dependent calls and types.
13635         * dwarf2cfi.c (htab_t trace_index):
13636         Change type to hash_table.  Update dependent calls and types.
13638         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
13639         Change type to hash_table.  Update dependent calls and types.
13640         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
13641         (htab_t optimize_external_refs::map): Likewise.
13642         (htab_t output_comp_unit::extern_map): Likewise.
13643         (htab_t output_comdat_type_unit::extern_map): Likewise.
13644         (htab_t output_macinfo::macinfo_htab): Likewise.
13645         (htab_t optimize_location_lists::htab): Likewise.
13646         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
13648         * except.c (htab_t ehspec_hash_type):
13649         Change type to hash_table.  Update dependent calls and types.
13650         (assign_filter_values::ttypes): Likewise.
13651         (assign_filter_values::ehspec): Likewise.
13652         (sjlj_assign_call_site_values::ar_hash): Likewise.
13653         (convert_to_eh_region_ranges::ar_hash): Likewise.
13655         * gcse.c (htab_t pre_ldst_table):
13656         Change type to hash_table.  Update dependent calls and types.
13658         * ggc-common.c (htab_t saving_htab):
13659         Change type to hash_table.  Update dependent calls and types.
13660         (htab_t loc_hash): Likewise.
13661         (htab_t ptr_hash): Likewise.
13662         (call_count): Rename ggc_call_count.
13663         (call_alloc): Rename ggc_call_alloc.
13664         (loc_descriptor): Rename make_loc_descriptor.
13665         (add_statistics): Rename ggc_add_statistics.
13667         * ggc-common.c (saving_htab):
13668         Change type to hash_table.  Update dependent calls and types.
13670         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
13671         (push_gimplify_context): Likewise.
13672         (pop_gimplify_context): Likewise.
13673         (struct gimple_temp_hash_elt): Added.
13674         (struct gimplify_hasher): Likewise.
13675         (struct gimplify_ctx.temp_htab):
13676         Change type to hash_table.  Update dependent calls and types.
13678         * gimple-fold.c: Include gimplify-ctx.h.
13680         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
13681         Change type to hash_table.  Update dependent calls and types.
13682         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
13683         avoid potential global name collision.
13685         * gimplify.c: Include gimplify-ctx.h.
13686         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
13687         (htab_t gimplify_ctx::temp_htab):
13688         Update dependent calls and types for new type hash_table.
13689         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
13690         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
13692         * gimplify-ctx.h: New.
13693         (struct gimple_temp_hash_elt): Move from gimplify.c.
13694         (class gimplify_hasher): New.
13695         (struct gimplify_ctx): Move from gimple.h.
13696         (htab_t gimplify_ctx::temp_htab):
13697         Change type to hash_table.  Update dependent calls and types.
13699         * graphite-clast-to-gimple.c: Include graphite-htab.h.
13700         (htab_t ivs_params::newivs_index):
13701         Change type to hash_table.  Update dependent calls and types.
13702         (htab_t ivs_params::params_index): Likewise.
13703         (htab_t print_generated_program::params_index): Likewise.
13704         (htab_t gloog::newivs_index): Likewise.
13705         (htab_t gloog::params_index): Likewise.
13707         * graphite.c: Include graphite-htab.h.
13708         4htab_t graphite_transform_loops::bb_pbb_mapping):
13709         Change type to hash_table.  Update dependent calls and types.
13711         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
13712         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
13713         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
13715         * graphite-dependences.c: Include graphite-htab.h.
13716         (loop_is_parallel_p): Change hash table type of parameter.
13718         * graphite-htab.h: New.
13719         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
13720         (extern find_pbb_via_hash): Move from graphite-poly.h.
13721         (extern loop_is_parallel_p): Move from graphite-poly.h.
13722         (extern get_loop_body_pbbs): Move from graphite-poly.h.
13724         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
13725         (extern loop_is_parallel_p): Move to graphite-htab.h.
13726         (extern get_loop_body_pbbs): Move to graphite-htab.h.
13728         * haifa-sched.c (htab_t delay_htab):
13729         Change type to hash_table.  Update dependent calls and types.
13730         (htab_t delay_htab_i2): Likewise.
13732         * ira-color.c (htab_t allocno_hard_regs_htab):
13733         Change type to hash_table.  Update dependent calls and types.
13735         * ira-costs.c (htab_t cost_classes_htab):
13736         Change type to hash_table.  Update dependent calls and types.
13738         * loop-invariant.c (htab_t merge_identical_invariants::eq):
13739         Change type to hash_table.  Update dependent calls and types.
13741         * loop-iv.c (htab_t bivs):
13742         Change type to hash_table.  Update dependent calls and types.
13744         * loop-unroll.c (htab_t opt_info::insns_to_split):
13745         Change type to hash_table.  Update dependent calls and types.
13746         (htab_t opt_info::insns_with_var_to_expand): Likewise.
13748         * lto-streamer.h (struct string_slot): Move from data-streamer.h
13749         (struct string_slot_hasher): New.
13750         (htab_t output_block::string_hash_table):
13751         Change type to hash_table.  Update dependent calls and types.
13753         * lto-streamer-in.c (freeing_string_slot_hasher): New.
13754         (htab_t file_name_hash_table):
13755         Change type to hash_table.  Update dependent calls and types.
13757         * lto-streamer-out.c: Update output_block::string_hash_table dependent
13758         calls and types.
13760         * lto-streamer.c (htab_t tree_htab):
13761         Change type to hash_table.  Update dependent calls and types.
13763         * omp-low.c: Include gimplify-ctx.h.
13765         * passes.c (htab_t name_to_pass_map):
13766         Change type to hash_table.  Update dependent calls and types.
13767         (pass_traverse): Rename to passes_pass_traverse.
13769         * plugin.c (htab_t event_tab):
13770         Change type to hash_table.  Update dependent calls and types.
13772         * postreload-gcse.c (htab_t expr_table):
13773         Change type to hash_table.  Update dependent calls and types.
13774         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
13776         * sese.c (debug_rename_map_1): Make extern.
13777         (htab_t copy_bb_and_scalar_dependences::rename_map):
13778         Change type to hash_table.  Update dependent calls and types.
13780         * sese.h (extern debug_rename_map): Move to .c file.
13782         * store-motion.c (htab_t store_motion_mems_table):
13783         Change type to hash_table.  Update dependent calls and types.
13785         * trans-mem.c (htab_t tm_new_mem_hash):
13786         Change type to hash_table.  Update dependent calls and types.
13788         * tree-browser.c (htab_t TB_up_ht):
13789         Change type to hash_table.  Update dependent calls and types.
13791         * tree-cfg.c (htab_t discriminator_per_locus):
13792         Change type to hash_table.  Update dependent calls and types.
13794         * tree-complex.c: Include tree-hasher.h
13795         (htab_t complex_variable_components):
13796         Change type to hash_table.  Update dependent calls and types.
13798         * tree-eh.c (htab_t finally_tree):
13799         Change type to hash_table.  Update dependent calls and types.
13801         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
13802         struct int_tree_hasher.
13803         (extern int_tree_map_eq): Likewise.
13804         (uid_decl_map_hash): Removed.
13805         (extern decl_tree_map_eq): Likewise.
13807         * tree-hasher.h: New.
13808         (struct int_tree_hasher): New.
13809         (typedef int_tree_htab_type): New.
13811         * tree-inline.c: Include gimplify-ctx.h.
13813         * tree-mudflap.c: Include gimplify-ctx.h.
13815         * tree-parloops.c: Include tree-hasher.h.
13816         (htab_t eliminate_local_variables_stmt::decl_address):
13817         Change type to hash_table.  Update dependent calls and types.
13818         (htab_t separate_decls_in_region::decl_copies): Likewise.
13820         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
13821         Change type to hash_table.  Update dependent calls and types.
13823         * tree-sra.c (candidates):
13824         Change type to hash_table.  Update dependent calls and types.
13826         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
13827         in tree-flow.h.
13828         (int_tree_map_hash): Likewise.
13830         * tree-ssa-dom.c (htab_t avail_exprs):
13831         Change type to hash_table.  Update dependent calls and types.
13833         * tree-ssa-live.c (var_map_base_init::tree_to_index):
13834         Change type to hash_table.  Update dependent calls and types.
13836         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
13837         Change type to hash_table.  Update dependent calls and types.
13839         * tree-ssa-phiopt.c (seen_ssa_names):
13840         Change type to hash_table.  Update dependent calls and types.
13842         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
13843         Change type to hash_table.  Update dependent calls and types.
13845         * tree-ssa-uncprop.c (equiv):
13846         Change type to hash_table.  Update dependent calls and types.
13848 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
13850         PR rtl-optimization/57003
13851         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
13852         call note_stores with kill_clobbered_value callback again after
13853         killing regs_invalidated_by_call.
13855 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13857         * config/aarch64/aarch64-simd.md
13858         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
13859         (aarch64_simd_bsl<mode>): Likewise.
13860         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
13862 2013-04-25  Marek Polacek  <polacek@redhat.com>
13864         PR tree-optimization/57066
13865         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
13867 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
13869         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
13871 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13873         * config/aarch64/aarch64-builtins.c
13874         (aarch64_fold_builtin): New.
13875         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
13876         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
13877         * config/aarch64/aarch64-simd-builtins.def (abs): New.
13878         * config/aarch64/arm_neon.h
13879         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
13881 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13882             Tejas Belagod  <tejas.belagod@arm.com>
13884         * config/aarch64/aarch64-builtins.c
13885         (aarch64_gimple_fold_builtin): New.
13886         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
13887         * config/aarch64/aarch64-simd-builtins.def (addv): New.
13888         * config/aarch64/aarch64-simd.md (addpv4sf): New.
13889         (addvv4sf): Update.
13890         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
13892 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13894         * config/aarch64/aarch64.md
13895         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
13897 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13899         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
13900         (*ngcsi_uxtw): New pattern.
13902 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13903             Julian Brown  <julian@codesourcery.com>
13905         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
13906         (TB_DREG): Add T_V4HF.
13907         (v4hf_UP): New macro.
13908         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13909         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13910         Handle initialisation of V4HF. Adjust initialisation of reinterpret
13911         built-ins.
13912         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13913         (arm_vector_mode_supported_p): Handle V4HF.
13914         (arm_mangle_map): Handle V4HFmode.
13915         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
13916         * config/arm/arm_neon_builtins.def: Add entries for
13917         vcvtv4hfv4sf, vcvtv4sfv4hf.
13918         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
13919         (neon_vcvtv4hfv4sf): Likewise.
13920         * config/arm/neon-gen.ml: Handle half-precision floating point
13921         features.
13922         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
13923         * config/arm/arm_neon.h: Regenerate.
13924         * config/arm/neon.ml (type elts): Add F16.
13925         (type vectype): Add T_float16x4, T_floatHF.
13926         (type vecmode): Add V4HF.
13927         (type features): Add Requires_FP_bit feature.
13928         (elt_width): Handle F16.
13929         (elt_class): Likewise.
13930         (elt_of_class_width): Likewise.
13931         (mode_of_elt): Refactor.
13932         (type_for_elt): Handle F16, fix error messages.
13933         (vectype_size): Handle T_float16x4.
13934         (vcvt_sh): New function.
13935         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
13936         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
13937         (string_of_mode): Handle V4HF.
13938         * doc/arm-neon-intrinsics.texi: Regenerate.
13940 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13942         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
13943         format specifier in 'X' case.
13945 2013-04-25  Alan Modra  <amodra@gmail.com>
13947         PR target/57052
13948         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
13949         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
13950         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
13951         Repeat for many other rotate/shift and mask patterns using subregs.
13952         Name lshiftrt insns.
13953         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
13954         on WORDS_BIG_ENDIAN.
13956 2013-04-25  Alan Modra  <amodra@gmail.com>
13958         * config.gcc: Support little-endian powerpc-linux targets.
13959         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
13960         (LINK_OS_LINUX_SPEC): Define.
13961         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
13962         Preserve MASK_LITTLE_ENDIAN.
13963         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
13964         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
13965         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
13966         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
13967         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
13968         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
13969         Correct fp word order for little-endian.  Don't shift toc entries
13970         smaller than a word for little-endian.
13971         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
13972         (bswapdi2 splits): Correct low-part subreg for little-endian.
13973         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
13974         low/high where such is correct only for be.
13975         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
13976         little-endian for -mcall-aixdesc.
13978 2013-04-25  Alan Modra  <amodra@gmail.com>
13980         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
13981         replace_equiv_address_nv.
13983 2013-04-25  Alan Modra  <amodra@gmail.com>
13985         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
13987 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13989         Revert:
13990         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13991         * rtl.h (struct rtx_def): ...
13993 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13995         PR rtl-optimizations/57046
13996         * lra-constraints (split_reg): Set up lra_risky_transformations_p
13997         for multi-reg splits.
13999 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
14001         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
14003 2013-04-24  Sterling Augustine  <saugustine@google.com>
14005         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14006         (comp_dir_string, debug_str_dwo_section): New.
14007         (DEBUG_STR_DWO_SECTION): Rename to ...
14008         (DEBUG_DWO_STR_SECTION): ... this.
14009         (DEBUG_NORM_STR_SECTION): Delete.
14010         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14011         (DEBUG_STR_DWO_SECTION_FLAGS): New.
14012         (find_AT_string): Move most logic to ...
14013         (find_AT_string_in_table): ... here.  New.
14014         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14015         add_skeleton_AT_string.  Delete logic.
14016         (output_skeleton_debug_sections): Remove call to
14017         add_top_level_skeleton_die_attrs.
14018         (add_comp_dir_attribute): Move logic to comp_dir_string.
14019         (dwarf2out_init): Initialize debug_str_dwo_section.
14020         (output_indirect_string): Call find_string_form.
14021         (output_indirect_strings): Rewrite.
14022         (prune_unused_types): Empty skeleton_debug_str_hash.
14023         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14024         (dwarf2out_finish):  Call output_indirect_strings.
14026 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14028         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14030 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14032         * rtl.h (struct rtx_def): Add comment for field jump.
14033         (LRA_SUBREG_P): New macro.
14034         * recog.c (register_operand): Check LRA_SUBREG_P.
14035         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14036         stack frame.
14037         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14038         (lra_final_code_change): Skip subreg change for operators.
14039         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14040         if there are no operand changes.
14041         * lra-constraints.c (curr_insn_set): New.
14042         (match_reload): Set LRA_SUBREG_P.
14043         (emit_spill_move): Ditto.
14044         (check_and_process_move): Use curr_insn_set. Process only single
14045         set insns.  Don't initialize sec_mem_p and change_p.
14046         (simplify_operand_subreg): Use LRA_SUBREG_P.
14047         (reg_in_class_p): New function.
14048         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14049         of #ifdef.  Add code to remove cycling.
14050         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14051         non-null disp.  Reload inner instead of disp when base and index
14052         are null.  Try to put lo_sum into register.
14053         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14054         (check_and_process_move): Move code for move cost check to
14055         simple_move_p.  Remove equiv_substitution.
14056         (simple_move_p): New function.
14057         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14058         curr_insn_set.  Call check_and_process_move only for single set
14059         insns.  Use the new function.  Move call of check_and_process_move
14060         after operand equiv substitution and address process.
14062 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14064         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14065         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14066         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14068 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14070         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14072 2013-04-24  Marek Polacek  <polacek@redhat.com>
14074         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14075         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14076         (select_loops_exit_conditions): Likewise.
14077         (number_of_iterations_for_all_loops): Likewise.
14078         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14079         (scev_analysis): Likewise.
14081 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14082             Chao-ying Fu  <fu@mips.com>
14084         * config/mips/micromips.md (jraddiusp): New pattern.
14085         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14086         instruction if possible.
14088 2013-04-24  Alan Modra  <amodra@gmail.com>
14090         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14092 2013-04-24  Julian Brown  <julian@codesourcery.com>
14093             Chung-Lin Tang  <cltang@codesourcery.com>
14095         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14096         dependency behavior in enumeration type DIE generation. Add TODO note
14097         to comments about future DW_FORM_sdata/udata re-work of related code.
14099 2013-04-23  Lawrence Crowl  <crowl@google.com>
14101         * Makefile.in: Update as needed below.
14103         * hash-table.h (class hash_table):
14104         Correct many methods with parameter types compare_type to the correct
14105         value_type.  (Correct code was unlikely to notice the change.)
14106         (hash_table::elements_with_deleted) New.
14107         (class hashtable::iterator): New.
14108         (hashtable::begin()): New.
14109         (hashtable::end()): New.
14110         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14112         * statistics.c (statistics_hashes):
14113         Change type to hash_table.  Update dependent calls and types.
14115         * tree-into-ssa.c (var_infos):
14116         Change type to hash_table.  Update dependent calls and types.
14118         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14119         Change type to hash_table.  Update dependent calls and types.
14121         * tree-ssa-loop-im.c (struct mem_ref.refs):
14122         Change type to hash_table.  Update dependent calls and types.
14124         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14125         Change type to hash_table.  Update dependent calls and types.
14127         * tree-ssa-sccvn.c (vn_tables_s::nary):
14128         Change type to hash_table.  Update dependent calls and types.
14129         (vn_tables_s::phis): Likewise.
14130         (vn_tables_s::references): Likewise.
14132         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14133         (vn_reference_eq): Update parameter and return types.
14135         * tree-ssa-structalias.c (pointer_equiv_class_table):
14136         Change type to hash_table.  Update dependent calls and types.
14137         (location_equiv_class_table): Likewise.
14139         * tree-vect-data-refs.c: Consequential changes for making
14140         peeling a hash_table.
14142         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14143         (destroy_loop_vec_info): Dependent hash_table update.
14145         * tree-vectorizer.h (peeling_htab):
14146         Change type to hash_table.  Update dependent calls and types.
14148 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14150         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14151         to check the register content is equal or not.
14152         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14153         to assign register content record.
14154         * lra-eliminations.c (update_reg_eliminate): Use
14155         lra_update_reg_val_offset to update register content offset.
14156         * lra-int.h (struct lra_reg): Add offset member.
14157         (lra_reg_val_equal_p): New static inline function.
14158         (lra_update_reg_val_offset): New static inline function.
14159         (lra_assign_reg_val): New static inline function.
14160         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14161         to assign register content record.
14162         (initialize_lra_reg_info_element): Initial offset to zero.
14164 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14166         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14167         operands.  Record compression.
14169 2013-04-23  Xinliang David Li  <davidxl@google.com>
14171         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14173 2013-04-23  Richard Biener  <rguenther@suse.de>
14175         PR middle-end/57036
14176         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14177         parameter, only add abnormal goto edges from the copied body
14178         if the call could perform abnormal gotos.
14179         (copy_cfg_body): Adjust.
14181 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14183         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14185 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14187         * coretypes.h (gimple_stmt_iterator): Add struct to make
14188         compatible with C.
14190 2013-04-23  Richard Biener  <rguenther@suse.de>
14192         PR tree-optimization/57026
14193         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14194         from SSA names occuring in abnormal PHI nodes.
14196 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14198         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14200 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14202         * lto/lto.c (print_lto_report_1): Declare early.
14203         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14205 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14207         * common.opt (-flto-report-wpa): Add.
14208         * doc/invoke.texi (-flto-report-wpa): Add.
14209         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14210         (lto_main): dito.
14212 2013-04-22  Xinliang David Li  <davidxl@google.com>
14214         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14215         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14216         * Makefile.in: New dependency
14218             David Daney <ddaney.cavm@gmail.com>
14220         * configure.ac (gcc_cv_as_micromips_support):  Use the
14221         --fatal-warnings option.
14222         * configure: Regenerate.
14224 2013-04-22  Marek Polacek  <polacek@redhat.com>
14226         PR sanitizer/56990
14227         * tsan.c (instrument_expr): Don't instrument expression
14228         in case its size is zero.
14230 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14232         PR target/57032
14233         Revert:
14234         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14236         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14238 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14240         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14241         (gimple_stmt_iterator): New typedef.
14242         * gimple.h (gimple_stmt_iterator): Rename to...
14243         (gimple_stmt_iterator_d): ... This.
14244         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14245         trees be valid for GIMPLE and GENERIC.
14246         (TARGET_GIMPLE_FOLD_BUILTIN): New.
14247         * gimple-fold.c (gimple_fold_call): Call target hook
14248         gimple_fold_builtin.
14249         * hooks.c (hook_bool_gsiptr_false): New.
14250         * hooks.h (hook_bool_gsiptr_false): New.
14251         * target.def (fold_stmt): New.
14252         * doc/tm.texi: Regenerate.
14254 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
14256         PR target/57018
14257         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14258         a set sp if no stack realignment.
14260 2013-04-22  Nick Clifton  <nickc@redhat.com>
14262         * config.gcc (tilegx-linux): Extend extra_objs rather than
14263         overwriting it.
14264         (tilepro-linux): Likewise.
14266 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14268         * config/aarch64/aarch64-builtins.c
14269         (CF): Remove.
14270         (CF0, CF1, CF2, CF3, CF4, CF10): New.
14271         (VAR<1-12>): Add MAP parameter.
14272         (BUILTIN_*): Likewise.
14273         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14274         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14275         (aarch64_ushl_n<mode>): Likewise.
14276         (aarch64_sshr_n<mode>): Likewise.
14277         (aarch64_ushr_n<mode>): Likewise.
14278         (aarch64_<maxmin><mode>): Likewise.
14279         (aarch64_sqrt<mode>): Likewise.
14280         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14281         (vshr<q>_n_*): Likewise.
14283 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14285         * config/aarch64/aarch64-builtins.c
14286         (aarch64_simd_builtin_type_mode): Handle SF types.
14287         (sf_UP): Define.
14288         (BUILTIN_GPF): Define.
14289         (aarch64_init_simd_builtins): Handle SF types.
14290         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14291         (frecps): Likewise.
14292         (frecpx): Likewise.
14293         * config/aarch64/aarch64-simd.md
14294         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14295         (aarch64_frecpe<mode>): New.
14296         (aarch64_frecps<mode>): Likewise.
14297         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14298         (v8type): Add frecp<esx>.
14299         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14300         (aarch64_frecps<mode>): Likewise.
14301         * config/aarch64/iterators.md (FRECP): New.
14302         (frecp_suffix): Likewise.
14303         * config/aarch64/arm_neon.h
14304         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14306 2013-04-22  Christian Bruel  <christian.bruel@st.com>
14308         PR target/56995
14309         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14310         (REG_CLASS_NAMES): Idem.
14311         (REG_CLASS_CONTENTS): Idem.
14312         (REGCLASS_HAS_FP_REG): Idem.
14313         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14314         (sh_conditional_register_usage): Idem.
14316 2013-04-21  Jeff Law  <law@redhat.com>
14318         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14319         (ssa_forward_propagate_and_combine): Use it.
14321 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14323         * lra.c: Update the flow chart diagram.
14325 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14327         PR rtl-optimization/56847
14328         * lra-constraints.c (process_alt_operands): Discourage alternative
14329         with non-matche doffsettable memory constraint fro memory with
14330         known offset.
14332 2013-04-19  Richard Biener  <rguenther@suse.de>
14334         PR tree-optimization/56982
14335         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14336         function.
14337         * gimplify.c (gimplify_call_expr): Notice special calls.
14338         (gimplify_modify_expr): Likewise.
14339         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14340         abnormal control flow receivers.
14341         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14342         in the same way as cfun->has_nonlocal_labels.
14343         (gimple_purge_dead_abnormal_call_edges): Likewise.
14344         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14345         receivers start a basic-block.
14347 2013-04-19  Richard Biener  <rguenther@suse.de>
14349         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14350         member ...
14351         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
14352         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14353         (SLP_TREE_LOAD_PERMUTATION): Add.
14354         (vect_transform_slp_perm_load): Adjust prototype.
14355         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14356         (vect_free_slp_instance): Likewise.
14357         (vect_create_new_slp_node): Likewise.
14358         (vect_supported_slp_permutation_p): Remove.
14359         (vect_slp_rearrange_stmts): Adjust.
14360         (vect_supported_load_permutation_p): Likewise.  Inline
14361         vect_supported_slp_permutation_p here.
14362         (vect_analyze_slp_instance): Compute load permutations per
14363         slp node instead of per instance.
14364         (vect_get_slp_defs): Adjust.
14365         (vect_transform_slp_perm_load): Likewise.
14366         (vect_schedule_slp_instance): Remove redundant code.
14367         (vect_schedule_slp): Remove hack for PR56270, add it ...
14368         * tree-vect-stmts.c (vectorizable_load): ... here, do not
14369         CSE loads for SLP.  Adjust.
14371 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14373         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14374         spelling in two comments.
14376 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14378         PR target/56797
14379         * config/arm/arm.c (load_multiple_sequence): Require SP
14380         as base register for loads if SP is in the register list.
14382 2013-04-19  Martin Jambor  <mjambor@suse.cz>
14384         PR tree-optimization/56718
14385         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14386         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14387         and made public.  Adjusted all callers.
14388         (ipa_intraprocedural_devirtualization): New function.
14389         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14390         (ipa_intraprocedural_devirtualization): Likewise.
14391         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14393 2013-04-19  Richard Biener  <rguenther@suse.de>
14395         PR tree-optimization/57000
14396         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14398 2013-04-19  Terry Guo  <terry.guo@arm.com>
14400         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14401         Replace with ...
14402         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
14403         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14404         (cortex_m4_fmacs): Use new reservations.
14405         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14407 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
14409         PR rtl-optimization/56999
14410         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14411         related code.
14412         (lra_coalesce): Remove split_origin_bitmap and related code.
14413         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
14414         ranges if necessary.
14416 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
14418         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14419         New array.
14420         (ix86_expand_call): Remove clobbered_registers array and use
14421         x86_64_ms_sysv_extra_clobbered_registers instead.
14422         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14423         Declare here.
14424         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14425         predicate.
14426         * config/i386/i386.md (*call_rex64_ms_sysv): Use
14427         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
14428         (*call_value_rex64_ms_sysv): Ditto.
14430 2013-04-18  Cary Coutant  <ccoutant@google.com>
14432         * dwarf2out.c (output_pubnames): Check die_perennial_p of
14433         parent instead of die_mark.
14435 2013-04-18  Diego Novillo  <dnovillo@google.com>
14437         * gimple.c (create_gimple_tmp): New.
14438         (get_expr_type): New.
14439         (build_assign): New.
14440         (build_type_cast): New.
14441         * gimple.h (enum ssa_mode): Define.
14442         (gimple_seq_set_location): New.
14443         * asan.c (build_check_stmt): Change some gimple_build_* calls
14444         to use build_assign and build_type_cast.
14446 2013-04-18  Richard Biener  <rguenther@suse.de>
14448         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14449         handle negative step.  Remove redundant checks.
14450         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14451         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14452         for negative step and grouped loads fail to vectorize.
14454 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14456         * emit-rtl.c (reset_insn_used_flags): New function.
14457         (reset_all_used_flags): Use it.
14458         (verify_insn_sharing): New function.
14459         (verify_rtl_sharing): Fix verification for SEQUENCEs.
14461 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
14463         PR tree-optimization/56984
14464         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14465         and (x >> M) >= N don't register any assertion if N << M is the
14466         minimum value.
14468 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14470         * lower-subreg.c (resolve_simple_move): If called self-recursive,
14471         do not delete_insn insns that have not yet been emitted, only
14472         unlink them with remove_insn.
14473         * df-scan.c (df_insn_delete): Revert r197492.
14475 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
14477         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14478         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14480 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14482         * config/arm/arm.md (movsicc_insn): Convert define_insn into
14483         define_insn_and_split.
14484         (and_scc,ior_scc,negscc): Likewise.
14485         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14487 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14489         * config/arm/arm.c (use_return_insn): Return 0 for targets that
14490         can benefit from using a sequence of LDRD instructions in epilogue
14491         instead of a single LDM instruction.
14493 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14495         PR 45688
14496         * doc/extend.texi: Fix typo.
14498 2013-04-17  Richard Biener  <rguenther@suse.de>
14500         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14501         (vect_build_slp_tree): ... here.
14502         (vect_build_slp_tree_1): Compute which stmts of the SLP group
14503         match.  Remove special-casing of mismatched complex loads.
14504         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14505         re-try the match with swapped commutative operands.
14506         (vect_supported_load_permutation_p): Remove special-casing of
14507         mismatched complex loads.
14508         (vect_analyze_slp_instance): Adjust.
14510 2013-04-17  Richard Biener  <rguenther@suse.de>
14512         PR rtl-optimization/56921
14513         * cfgloop.h (struct loop): Add simple_loop_desc member.
14514         (struct niter_desc): Mark with GTY(()).
14515         (simple_loop_desc): Do not use aux field but simple_loop_desc.
14516         * loop-iv.c (get_simple_loop_desc): Likewise.
14517         (free_simple_loop_desc): Likewise.
14519         Revert
14520         2013-04-16  Richard Biener  <rguenther@suse.de>
14522         PR rtl-optimization/56921
14523         * loop-init.c (pass_rtl_move_loop_invariants): Add
14524         TODO_do_not_ggc_collect to todo_flags_finish.
14525         (pass_rtl_unswitch): Same.
14526         (pass_rtl_unroll_and_peel_loops): Same.
14527         (pass_rtl_doloop): Same.
14529 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
14531         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14532         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14533         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14534         references.
14535         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14536         * tree-streamer.c (record_common_node): Adjust reference in comment.
14538 2013-04-17  Terry Guo  <terry.guo@arm.com>
14540         * config/arm/cortex-m4.md: Add a new bypass.
14542 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14544         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14545         New pattern.
14546         (*subs_<optab><mode>_multp2): New pattern.
14547         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14548         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14550 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14552         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14553         (*subs_mul_imm_<mode>): New pattern.
14555 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
14557         PR target/56948
14558         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14559         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
14560         (vsx_movti_32bit): Change j->wa to O->wa.
14562 2013-04-16  Richard Biener  <rguenther@suse.de>
14564         PR rtl-optimization/56921
14565         * loop-init.c (pass_rtl_move_loop_invariants): Add
14566         TODO_do_not_ggc_collect to todo_flags_finish.
14567         (pass_rtl_unswitch): Same.
14568         (pass_rtl_unroll_and_peel_loops): Same.
14569         (pass_rtl_doloop): Same.
14571 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
14573         * config/arm/arm.c (emit_multi_reg_push): New declaration
14574         for an existing function.
14575         (arm_emit_strd_push): New function.
14576         (arm_expand_prologue): Used here.
14577         (arm_emit_ldrd_pop): New function.
14578         (arm_expand_epilogue): Used here.
14579         (arm_get_frame_offsets): Update condition.
14580         (arm_emit_multi_reg_pop): Add a special case for load of a single
14581         register with writeback.
14583 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
14585         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
14586         description.
14588 2013-04-16  Richard Biener  <rguenther@suse.de>
14590         PR tree-optimization/56756
14591         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
14592         (first_mem_ref_loc): New.
14593         (execute_sm): Place the load temporarily before a previous
14594         access instead of in the latch edge to ensure its SSA dependencies
14595         are defined at points dominating the load.
14597 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
14599         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
14600         correct fix by moving header and footer insn to the footer of
14601         the merged basic block.  Clear BB_END of the merged-away block.
14603         PR middle-end/43631
14604         * emit-rtl.c (make_note_raw): New function.
14605         (link_insn_into_chain): New static inline function.
14606         (add_insn): Use it.
14607         (add_insn_before, add_insn_after): Factor insn chain linking code...
14608         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
14609         using link_insn_into_chain.
14610         (note_outside_basic_block_p): New helper function for emit_note_after
14611         and emit_note_before.
14612         (emit_note_after): Use nobb variant of add_insn_after if the note
14613         should not be contained in a basic block.
14614         (emit_note_before): Use nobb variant of add_insn_before if the note
14615         should not be contained in a basic block.
14616         (emit_note_copy): Use make_note_raw.
14617         (emit_note): Likewise.
14618         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
14619         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
14620         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
14621         the moved barrier the tail of the basic block it follows.
14622         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
14624 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
14626         PR tree-optimization/56962
14627         * gimple-ssa-strength-reduction.c (record_increment): Only set
14628         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
14629         either rhs1 or rhs2 is equal to c->base_expr.
14631 2013-04-15  Richard Biener  <rguenther@suse.de>
14633         PR tree-optimization/56933
14634         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
14635         member.
14636         (GROUP_READ_WRITE_DEPENDENCE): Remove.
14637         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
14638         * tree-vect-data-refs.c (vect_analyze_group_access): Move
14639         dependence check ...
14640         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
14641         ... here.
14642         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
14643         GROUP_READ_WRITE_DEPENDENCE.
14645 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14647         * emit-rtl.c (reset_all_used_flags): New function.
14648         (verify_rtl_sharing): Call reset_all_used_flags before and after
14649         performing the checks.
14651 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14653         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
14654         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
14655         * config/arm/constraints.md (De): New constraint.
14656         * config/arm/neon.md (anddi3_neon): Delete.
14657         (neon_vand<mode>): Expand to standard anddi3 pattern.
14658         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
14659         Move earlier in the file.
14660         (neon_inv_logic_op2): Likewise.
14661         (arm_anddi_operand_neon): New predicate.
14663 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14665         * configure.ac (gcc_cv_ld_as_needed): Set
14666         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
14667         Use -z ignore, -z record on *-*-solaris2*.
14668         (HAVE_LD_AS_NEEDED): Update comment.
14669         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
14670         * configure: Regenerate.
14671         * config.in: Regenerate.
14672         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
14673         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
14674         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
14675         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
14676         equivalents.  Fix markup.
14677         * doc/tm.texi: Regenerate.
14679 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
14681         * config/i386/i386.opt: New option mstack-protector-guard=.
14682         * config/i386/i386-opts.h: Add enum stack_protector_guard.
14683         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
14684         TARGET_SSP_TLS_GUARD.
14685         * config/i386/i386.c (ix86_option_override_internal): Set
14686         ix86_stack_protector_guard.
14687         * config/i386/i386.md (stack_protect_set): Enable for
14688         TARGET_SSP_TLS_GUARD only.
14689         (stack_protect_set_<mode>): Ditto.
14690         (stack_protect_test): Ditto.
14691         (stack_protect_test_<mode>): Ditto.
14692         * doc/invoke.texi (i386 Option): Document.
14694 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
14696         PR target/56890
14697         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
14698         (S_MODES): Set H_MODE bit.
14699         (SF_MODES): Set only S_MODE and SF_MODE bits.
14700         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
14701         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
14702         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
14703         <MODE_FLOAT>: Likewise.
14705 2013-04-15  Joey Ye  <joey.ye@arm.com>
14707         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
14709 2013-04-15  Joey Ye  <joey.ye@arm.com>
14711         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
14712         for real far jump.
14713         (thumb_far_jump_used_p): Count instruction size and set
14714         far_jump_used.
14716 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
14718         * reorg.c (fill_simple_delay_slots): Reindent block of code.
14719         * resource.c (mark_target_live_regs): Reformat conditional block.
14721 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14723         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
14724         notes, they are emitted only just before final.
14725         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
14727 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14729         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
14730         * cfgrtl.c (delete_insn): Call it here instead.
14731         * lra-spills.c (lra_final_code_change): Use delete_insn.
14732         * haifa-sched.c (sched_remove_insn): Likewise.
14733         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
14734         returning to the nop pool.
14735         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
14736         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
14738 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14740         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
14741         * doc/tm.texi: Regenerated.
14743 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
14745         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
14746         QImode checks.
14748 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14750         * df-core.c (df_find_def): Compare register numbers.
14751         (df_find_use): Likewise.
14753 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
14755         PR target/56903
14756         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
14757         lra_in_progress for return.
14759 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14761         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
14762         define_insn into define_insn_and_split and emit movsicc patterns.
14764 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14766         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
14768 2013-04-12  Richard Biener  <rguenther@suse.de>
14770         * tree-pass.h (TODO_do_not_ggc_collect): New.
14771         * passes.c (execute_one_ipa_transform_pass): Honor
14772         TODO_do_not_ggc_collect.
14773         (execute_one_pass): Likewise.
14775         Revert
14776         2013-04-10  Richard Biener  <rguenther@suse.de>
14778         * passes.c (init_optimization_passes): Remove reload pass.
14779         * ira.c (do_reload): Merge into ...
14780         (ira): ... this.
14781         (rest_of_handle_reload): Remove.
14782         (pass_reload): Likewise.
14783         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14784         of reload for vzeroupper pass placement.
14786 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
14788         PR tree-optimization/56918
14789         PR tree-optimization/56920
14790         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
14791         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
14792         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
14793         use wide_mul_with_sign method.
14795 2013-04-12  Richard Biener  <rguenther@suse.de>
14797         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
14798         not be considered a gimple constant.
14800 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
14802         * fold-const.c (const_binop): Handle vector shifts by a scalar.
14803         (fold_binary_loc): Call const_binop also for mixed vector-scalar
14804         operations.
14806 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14807             Jakub Jelinek  <jakub@redhat.com>
14809         * opts.c: Include diagnostic-color.h.
14810         (common_handle_option): Handle OPT_fdiagnostics_color_.
14811         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
14812         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
14813         (diagnostic-color.o): New.
14814         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
14815         (diagnostic_color_rule): New enum.
14816         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
14817         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
14818         the location string.
14819         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
14820         either NULL, or color kind.
14821         * diagnostic-color.c: New file.
14822         * diagnostic-color.h: New file.
14823         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
14824         arguments.
14825         * doc/invoke.texi (-fdiagnostics-color): Document.
14826         * pretty-print.h (pp_show_color): Define.
14827         (struct pretty_print_info): Add show_color field.
14828         * diagnostic.c: Include diagnostic-color.h.
14829         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
14830         macros.  Colorize error:, warning: etc. strings and also the location
14831         string.
14832         (diagnostic_show_locus): Colorize the caret line.
14833         * pretty-print.c: Include diagnostic-color.h.
14834         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
14835         inside of %< %> quotes or quoted through q format modifier.
14837 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14839         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
14841 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14843         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
14844         code in CC_NZ mode.
14845         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
14846         pattern.
14848 2013-04-11  Marek Polacek  <polacek@redhat.com>
14850         PR tree-optimization/48184
14851         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
14853 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
14855         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
14856         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
14857         (skip_simple_arithmetic): Tidy up.
14858         * tree.h (skip_simple_constant_arithmetic): Declare.
14860 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14862         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
14864 2013-04-11  Richard Biener  <rguenther@suse.de>
14866         * tree-vect-loop.c (get_initial_def_for_induction): Properly
14867         generate vector constants.
14869 2013-04-11  Richard Biener  <rguenther@suse.de>
14871         PR tree-optimization/56878
14872         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
14873         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
14874         New function.
14875         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14876         Prefer to align the DR with the most invariant base address.
14878 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14880         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
14881         comment.
14883 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
14885         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
14886         floating-point vector comparisons against 0.
14888 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
14890         PR tree-optimization/56899
14891         * fold-const.c (extract_muldiv_1): Apply distributive law
14892         only if TYPE_OVERFLOW_WRAPS (ctype).
14894 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
14896         PR target/56124
14897         * ira-costs.c (scan_one_insn): Check whether the source rtx of
14898         loading has side effect.
14900 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
14902         * config/sparc/sparc.c: Include tree-pass.h.
14903         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
14904         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
14905         head of file.  Change return type.  Split off gate function.
14906         (sparc_gate_work_around_errata): New function.
14907         (pass_work_around_errata): New pass definition.
14908         (insert_pass_work_around_errata) New pass insert definition to
14909         insert pass_work_around_errata just after delayed-branch scheduling.
14910         (sparc_option_override): Insert the pass.
14911         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
14913 2013-04-10  David S. Miller  <davem@davemloft.net>
14915         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
14916         or -mcpu=hypersparc.
14918         * target.def (cstore_mode): New hook.
14919         * target.h: Include insn-codes.h
14920         * targhooks.c: Likewise.
14921         (default_cstore_mode): New function.
14922         * targhooks.h: Declare it.
14923         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
14924         * doc/tm.texi: Rebuild.
14925         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
14926         target hook, rather than inspecting the insn_data.
14927         * config/sparc/sparc.c (sparc_cstore_mode): New function.
14928         (TARGET_CSTORE_MODE): Redefine.
14929         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
14930         result patterns.
14931         * config/sparc/predicates.md (cstore_result_operand): New special
14932         predicate.
14933         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
14934         Use it for operand 0.
14935         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
14936         (*snesi_special): Likewise.
14937         (*snesi_zero): Likewise.
14938         (*seqsi_zero): Likewise.
14939         (*sltu_insn): Likewise.
14940         (*sgeu_insn): Likewise.
14941         (*seqdi_special): Make operand 0 and comparison operation be of
14942         DImode.
14943         (*snedi_special): Likewise.
14944         (*snedi_special_vis3): Likewise.
14945         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
14946         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
14947         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
14948         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
14949         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
14950         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
14951         (*sltu_extend_sp64): Likewise.
14952         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
14953         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
14954         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
14955         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
14956         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
14958 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
14960         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
14961         (aarch64_start_file): Use the new function.
14963 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14964             Jason Merrill  <jason@redhat.com>
14966         * common.opt: Add -gdwarf.
14967         * opts.c (common_handle_option): Handle it.
14968         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
14970 2013-04-10  Richard Biener  <rguenther@suse.de>
14972         * passes.c (execute_todo): Do not call ggc_collect conditional here.
14973         (execute_one_ipa_transform_pass): But unconditionally here.
14974         (execute_one_pass): And here.
14975         (init_optimization_passes): Remove reload pass.
14976         * tree-pass.h (TODO_ggc_collect): Remove.
14977         (pass_reload): Likewise.
14978         * ira.c (do_reload): Merge into ...
14979         (ira): ... this.
14980         (rest_of_handle_reload): Remove.
14981         (pass_reload): Likewise.
14982         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14983         of reload for vzeroupper pass placement.
14984         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
14985         and todo_flags_finish of all passes.
14987 2013-04-10  Richard Biener  <rguenther@suse.de>
14989         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
14990         first_const_oprnd field, rename first_def_type to first_op_type.
14991         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
14992         (vect_get_and_check_slp_defs): Always use the type of the
14993         operand.  Allow mixed vect_external_def, vect_constant_def types.
14994         (vect_get_constant_vectors): Handle mixed vect_external_def,
14995         vect_constant_def types.
14997 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
14999         PR tree-optimization/55524
15000         * tree-ssa-math-opts.c
15001         (convert_mult_to_fma): Don't use an fms construct
15002         when we don't have an fms operation, but fnma, and it looks
15003         likely that we'll be able to use the latter.
15005 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
15007         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15008         function.
15009         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15010         inline fail caused by overwritable functions.
15012 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
15014         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15015         unnecessary bits in the constant power of two case.
15017 2013-04-10  Richard Biener  <rguenther@suse.de>
15019         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15020         broken code swapping operands.
15021         (vect_build_slp_tree): Do not compute load permutations here.
15022         (vect_analyze_slp_instance): Compute load permutations here,
15023         after building the SLP tree.
15025 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15027         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15028         of next/prev_real_insn.
15030 2013-04-09  Jan Hubicka  <jh@suse.cz>
15032         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15033         Drop aliased parameter.
15034         (function_and_variable_visibility): Do not handle alias pairs.
15035         * cgraph.c (varpool_externally_visible_p): Update prototype.
15036         * varpool.c (varpool_add_new_variable): Update.
15038 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15040         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15042 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15044         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15046         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15048 2013-04-09  Marek Polacek  <polacek@redhat.com>
15050         PR tree-optimization/48762
15051         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15053 2013-04-09  Richard Biener  <rguenther@suse.de>
15055         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15056         dealing with cost.
15057         (vect_build_slp_tree): Likewise.
15058         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15059         calculating the cost of a SLP instance.
15060         (vect_analyze_slp_instance): Use it from here, after building
15061         the SLP tree.
15063 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15065         PR middle-end/56883
15066         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15067         expand_omp_for_static_chunk): Use simple_p = true in
15068         force_gimple_operand_gsi calls when assigning to addressable decls.
15070 2013-04-09  Jeff Law  <law@redhat.com>
15072         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15073         when the boolean was created by converting a wider object which
15074         had a boolean range.
15076 2013-04-09  Richard Biener  <rguenther@suse.de>
15078         * tree-vectorizer.h (slp_void_p): Remove.
15079         (slp_tree): Typedef before _slp_tree declaration.
15080         (struct _slp_tree): Use a vector of slp_tree as children.
15081         (vect_get_place_in_interleaving_chain): Remove.
15082         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15083         Move ...
15084         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15085         and make static.
15086         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15087         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15088         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15089         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15090         Use slp_node instead of slp_void_p and adjust.
15092 2013-04-09  Richard Biener  <rguenther@suse.de>
15094         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15095         work that is not necessary.
15097 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15099         PR tree-optimization/56854
15100         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15101         forward into clobber stmts if it would change MEM_REF lhs into
15102         non-MEM_REF.
15104 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15106         * tree.c (type_hash_lookup, type_hash_add): Make static.
15107         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15109 2013-04-09  Richard Biener  <rguenther@suse.de>
15111         * tree.h (unsave_expr_now): Remove.
15112         * tree-inline.c (mark_local_for_remap_r): Remove.
15113         (unsave_expr_1): Likewise.
15114         (unsave_r): Likewise.
15115         (unsave_expr_now): Likewise.
15116         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15117         (propagate_tree_value): Likewise.
15119 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15121         * doc/rtl.texi (sequence): Rewrite documentation to match the
15122         current use of SEQUENCE rtl objects.
15123         * rtl.def (SEQUENCE): Likewise.
15125         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15126         Update documentation.
15127         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15128         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15130         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15132 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15134         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15135         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15136         methods.
15137         (estimate_edge_size_and_time): Add comment to suggest using rounding
15138         methods.
15139         (estimate_node_size_and_time): Ditto.
15140         (remap_edge_change_prob): Use helper rounding divide methods.
15141         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15142         (gimple_mod_pow2_value_transform): Ditto.
15143         (gimple_mod_subtract_transform): Ditto.
15144         (gimple_ic_transform): Ditto.
15145         (gimple_stringops_transform): Ditto.
15146         * stmt.c (conditional_probability): Ditto.
15147         (emit_case_dispatch_table): Ditto.
15148         * lto-cgraph.c (merge_profile_summaries): Ditto.
15149         * tree-optimize.c (execute_fixup_cfg): Ditto.
15150         * cfgcleanup.c (try_forward_edges): Ditto.
15151         * cfgloopmanip.c (scale_loop_profile): Ditto.
15152         (loopify): Ditto.
15153         (duplicate_loop_to_header_edge): Ditto.
15154         (lv_adjust_loop_entry_edge): Ditto.
15155         * tree-vect-loop.c (vect_transform_loop): Ditto.
15156         * profile.c (compute_branch_probabilities): Ditto.
15157         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15158         * lto-streamer-in.c (input_cfg): Ditto.
15159         * gimple-streamer-in.c (input_bb): Ditto.
15160         * ipa-cp.c (update_profiling_info): Ditto.
15161         (update_specialized_profile): Ditto.
15162         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15163         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15164         rounding methods.
15165         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15166         (compute_trg_info): Ditto.
15167         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15168         (purge_dead_edges): Ditto.
15169         * loop-unswitch.c (unswitch_loop): Ditto.
15170         * cgraphclones.c (cgraph_clone_edge): Ditto.
15171         (cgraph_clone_node): Ditto.
15172         * tree-inline.c (copy_bb): Ditto.
15173         (copy_edges_for_bb): Ditto.
15174         (initialize_cfun): Ditto.
15175         (copy_cfg_body): Ditto.
15176         (expand_call_inline): Ditto.
15178 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15180         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15181         TARGET_CYGWIN64 by TARGET_64BIT.
15183 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15185         * config/epiphany/epiphany.md (GPR_1): New constant.
15186         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15187         * config/epiphany/epiphany.c (gen_compare_reg):
15188         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15189         is already in place.
15190         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15191         Don't require being called during rtl expansion; If y operlaps r0,
15192         return 0.
15193         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15194         (epiphany_expand_epilogue): Likewise.
15196         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15197         Don't use CC_FPmode for ORDERED / UNORDERED.
15198         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15200         * config/epiphany/constraints.md (CnL): New constraint.
15201         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15202         * config/epiphany/predicates.md (add_operand): Allow 1024.
15204         * config/epiphany/epiphany.md (logical_op): New code iterator.
15205         (op_mnc): New code attribute.
15206         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15207         (mov_f+1, mov_f+2): New peephole2 patterns.
15209         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15210         (cstoresi4): Also allow re-use of zero result when doing a NE
15211         comparison to a non-zero operand.
15212         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15214         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15215         Use gen_addsi3_i / gen_subsi3_i.
15217 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15219         PR c++/34949
15220         PR c++/50243
15221         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15222         contain anything but clobbers, at most one __builtin_stack_restore,
15223         optionally debug stmts and final resx, and if it has at least one
15224         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15225         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15226         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15227         which isn't defaut definition, remove them.
15228         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15229         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15230         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15231         with MEM_REF LHS with SSA_NAME address.
15233 2013-04-08  Jeff Law  <law@redhat.com>
15235         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15237 2013-04-08  Richard Biener  <rguenther@suse.de>
15239         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15240         extra newline.
15241         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15242         determined vector type.
15243         (vect_analyze_data_refs): Likewise.
15244         (vect_get_new_vect_var): Adjust.
15245         (vect_create_destination_var): Preserve SSA name versions.
15246         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15247         not dump anything here.
15249 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15251         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15252         Add member lr_slot_known.
15253         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15254         if necessary.
15255         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15256         Remove code that sets lr_slot_offset according to what a previous
15257         version of epiphany_emit_save_restore used to do.
15258         (epiphany_emit_save_restore): When doing an lr save or restore,
15259         set/verify lr_slot_known and lr_slot_offset.
15261 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
15263         PR target/54338
15264         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15265         in ALL_REGS.
15267 2013-04-08  Richard Biener  <rguenther@suse.de>
15269         * alias.c (find_base_term): Fix thinko in previous change.
15271 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15273         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15274         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15275         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
15276         if possible to compute val.
15277         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15278         For QImode integers don't require anything about precision.  Use
15279         const_with_all_bytes_same to find out if the constant doesn't have
15280         repeated bytes in it.
15282 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15284         * config/s390/s390.c (s390_expand_insv): Only accept insertions
15285         within mode size.
15287 2013-04-08  Marek Polacek  <polacek@redhat.com>
15289         PR rtl-optimization/48182
15290         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15291         value to 1.
15293 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15295         PR target/55487
15296         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15297         nuses, make sure we have a label.
15299 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15301         PR target/56843
15302         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15303         (rs6000_emit_swdiv_low_precision): Remove.
15304         (rs6000_emit_swdiv): Rewrite to handle between one and four
15305         iterations of Newton-Raphson generally; modify required number of
15306         iterations for some cases.
15307         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15309 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
15311         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15312         set-but-unused variable.
15314         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15315         basic blocks of released function bodies garbage-collectable.
15317         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15318         (struct rtl_opt_pass): Add TODO_df_finish.
15320         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15322 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15324         * config/arm/constraints.md (q): New constraint.
15325         * config/arm/ldrdstrd.md: New file.
15326         * config/arm/arm.md (ldrdstrd.md) New include.
15327         (arm_movdi): Use "q" instead of "r" constraint
15328         for double-word memory access.
15329         (movdf_soft_insn): Likewise.
15330         * config/arm/vfp.md (movdi_vfp): Likewise.
15331         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15332         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15333         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15334         (mem_ok_for_ldrd_strd): Likewise.
15335         (output_move_double): Update assertion.
15337 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15339         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15341 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15343         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15344         define_insn_and_split.
15345         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15347 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15349         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15350         define_insn_and_split.
15351         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15352         (shiftsi3_compare): New pattern.
15353         (rrx): New pattern.
15354         * config/arm/unspecs.md (UNSPEC_RRX): New.
15356 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15358         * config/arm/arm.md (negdi_extendsidi): New pattern.
15359         (negdi_zero_extendsidi): Likewise.
15361 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15363         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15364         define_insn_and_split.
15365         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15366         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15368 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15370         * config/arm/arm.md (arm_subdi3): Convert define_insn into
15371         define_insn_and_split.
15372         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15373         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15375 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15377         * config/arm/arm.md (subsi3_carryin): New pattern.
15378         (subsi3_carryin_const): Likewise.
15379         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15380         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15382 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15384         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15386 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15388         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15389         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15391 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15393         * config/arm/arm.c (arm_expand_builtin): Change fcode
15394         type to unsigned int.
15396 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15398         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15400 2013-04-04  Ian Lance Taylor  <iant@google.com>
15402         * doc/standards.texi (Standards): The Go frontend supports the Go 1
15403         language standard.
15405 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
15407         PR middle-end/56729
15408         * df-scan.c (df_insn_delete): Disable failing assert.
15410 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15412         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15413         New function prototype.
15414         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15415         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15416         (arm_builtin_vectorized_function): New function.
15418 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15420         * config/arm/arm_neon_builtins.def: New file.
15421         * config/arm/arm.c (neon_builtin_data): Move contents to
15422         arm_neon_builtins.def.
15423         (enum arm_builtins): Include neon builtin definitions.
15424         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15425         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15427 2013-04-04  Marek Polacek  <polacek@redhat.com>
15429         PR tree-optimization/48186
15430         * predict.c (maybe_hot_frequency_p): Return false if
15431         HOT_BB_FREQUENCY_FRACTION is 0.
15432         (cgraph_maybe_hot_edge_p): Likewise.
15434 2013-04-04  Richard Biener  <rguenther@suse.de>
15436         PR tree-optimization/56826
15437         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15438         more accurately.
15440 2013-04-04  Richard Biener  <rguenther@suse.de>
15442         PR tree-optimization/56213
15443         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15444         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15446 2013-04-04  Richard Biener  <rguenther@suse.de>
15448         PR tree-optimization/56837
15449         * tree-loop-distribution.c (classify_partition): For non-zero
15450         values require that the value has the same precision as its
15451         mode to be useful as memset value.
15453 2013-04-03  Nick Clifton  <nickc@redhat.com>
15455         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15456         (fmssf4): Use fmsf.s on E3V5 architectures.
15457         (fnmasf4): Use fnmaf.s on E3V5 architectures.
15458         (fnmssf4): Use fnmsf.s on E3V5 architectures.
15460 2013-04-03  Jeff Law  <law@redhat.com>
15462         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15463         (lra-eliminations.o): Likewise.
15465 2013-04-03  Teresa Johnson  <tejohnson@google.com>
15467         * gcov-io.c (compute_working_sets): Moved most of body of old
15468         compute_working_sets here from profile.c.
15469         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15470         (gcov_working_set_t): Moved typedef here from basic-block.h
15471         (compute_working_set): Declare.
15472         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15473         (get_working_sets): Renamed from compute_working_set,
15474         replace most of body with call to new compute_working_sets.
15475         (get_exec_counts): Replace call to compute_working_sets
15476         to get_working_sets.
15477         * profile.h (get_working_sets): Renamed from compute_working_set.
15478         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15479         to get_working_sets.
15480         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15481         * gcov-dump.c (dump_working_sets): New function.
15483 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
15485         * hwint.c (sext_hwi, zext_hwi): New functions.
15486         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15487         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15488         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15489         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15490         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15491         (sext_hwi, zext_hwi): New functions.
15493 2013-04-03  Jeff Law  <law@redhat.com>
15495         PR tree-optimization/56799
15496         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15497         back test for widening conversion erroneously dropped in prior change.
15499 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15501         PR target/56809
15502         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15503         instead of next_real_insn.
15505 2013-04-03  Marek Polacek  <polacek@redhat.com>
15507         PR sanitizer/55702
15508         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15510 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15512         PR target/56809
15513         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15514         next_real_insn.
15515         (thumb1_output_casesi): Likewise.
15516         (thumb2_output_casesi): Likewise.
15518 2013-04-03  Richard Biener  <rguenther@suse.de>
15520         PR tree-optimization/56817
15521         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15522         Split out ...
15523         (tree_unroll_loops_completely_1): ... new function to manually
15524         walk the loop tree, properly defering outer loops of unrolled
15525         loops to later iterations.
15527 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15529         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15530         (vectorizable_load): Likewise.
15531         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15532         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15534 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15536         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15537         BIT_FIELD_REF.
15539 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15541         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15543 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
15545         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15547 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15549         PR tree-optimization/56790
15550         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15551         folding.
15553 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15555         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15556         Handle VEC_MERGE.
15557         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15558         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15559         equal arguments.
15561 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
15563         PR c/19449
15564         * tree.h (force_folding_builtin_constant_p): New decl.
15565         * builtins.c (force_folding_builtin_constant_p): New variable.
15566         (fold_builtin_constant_p): Fold immediately also if
15567         force_folding_builtin_constant_p.
15569 2013-04-03  Richard Biener  <rguenther@suse.de>
15571         PR tree-optimization/56812
15572         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15573         DRs of the same interleaving chain are independent.
15575 2013-04-02  Jason Merrill  <jason@redhat.com>
15577         * gdbinit.in (pbb): Use debug fn.
15579 2013-04-02  Lawrence Crowl  <crowl@google.com>
15581         * sese.h (struct ivtype_map_elt_s): Remove unused.
15582         (extern debug_ivtype_map): Remove unused.
15583         (extern eq_ivtype_map_elts): Remove unused.
15584         * sese.c (debug_ivtype_map): Removed unused.
15585         (debug_ivtype_map_1): Removed unused.
15586         (debug_ivtype_elt): Remove unused.
15587         (eq_ivtype_map_elts): Remove unused.
15590 2013-04-02  Kai Tietz  <ktietz@redhat.com>
15592         PR target/52790
15593         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
15594         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
15595         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
15596         function.
15597         (legitimize_pe_coff_symbol): Likewise.
15598         (is_imported_p): New helper-function.
15599         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
15600         for Windows x64 targets.
15601         (ix86_expand_prologue): Optimize for pe-coff targets.
15602         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
15603         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
15604         medium/large code-model.
15605         (legitimize_pic_address): Likewise.
15606         (legitimize_tls_address): Likewise.
15607         (ix86_expand_call): Likewise.
15608         (x86_output_mi_thunk): Likewise.
15609         (get_dllimport_decl): Add new beimport argument.
15610         (construct_plt_address): Don't assert for x64 pe-coff targets.
15611         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
15612         targets.
15613         (SYMBOL_FLAG_STUBVAR): New macro.
15614         (SYMBOL_REF_STUBVAR_P): Likewise.
15615         * config/i386/winnt.c (stub_list): New structure.
15616         (stub_head): New local variable.
15617         (i386_pe_record_stub): New function.
15618         (i386_pe_file_end): Emit refptr-stubs.
15620 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
15622         PR rtl-optimization/56745
15623         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
15624         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
15626         PR c++/34949
15627         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
15628         and both of them are MEM_REFs, just compare first argument for
15629         equality and attempt to deal even with differing offsets.
15631         PR c++/34949
15632         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
15633         of gimple_clobber_p to be MEM_REF.
15634         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
15635         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
15636         after gimplification.
15637         * asan.c (get_mem_ref_of_assignment): Don't instrument
15638         gimple_clobber_p stmts.
15639         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
15640         gimple_clobber_p stmt if they have MEM_REF lhs and
15641         are dead because of another gimple_clobber_p stmt.
15642         * tree-ssa-live.c (clear_unused_block_pointer): Treat
15643         gimple_clobber_p stmts like debug stmts.
15644         (remove_unused_locals): Remove clobbers with MEM_REF lhs
15645         that refer to unused VAR_DECLs or uninitialized values.
15646         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
15647         gimple_clobber_p stmts if they refer to removed parameters.
15648         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
15649         formatting.
15651 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
15653         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
15654         using SWI48 mode attribute.
15656 2013-04-02  Wei Mi  <wmi@google.com>
15658         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
15659         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
15660         *<rotate_insn><mode>3_mask in i386.md.
15662 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
15664         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
15666 2013-04-02  Richard Biener  <rguenther@suse.de>
15668         PR tree-optimization/56778
15669         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15670         Runtime alias tests are not supported for gather loads.
15671         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
15672         stmts referenced from SSA operands before updating SSA form.
15674 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
15675             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15677         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
15678         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
15679         * config/arm/cortex-a53.md: New file.
15680         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
15681         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
15682         * config/arm/arm.c (arm_issue_rate): Likewise.
15683         * config/arm/arm-tune.md: Regenerate
15684         * config/arm/arm-tables.opt: Regenerate.
15685         * config/arm/arm-cores.def: Add cortex-a53.
15687 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
15689         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
15690         non-static link.
15692 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15694         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
15695         scalar load/store operations using B/H registers.
15696         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
15698 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15700         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
15701         scalar move.
15702         * config/aarch64/aarch64.c
15703         (aarch64_simd_scalar_immediate_valid_for_move): New.
15704         * config/aarch64/aarch64-protos.h
15705         (aarch64_simd_scalar_immediate_valid_for_move): New.
15706         * config/aarch64/constraints.md (Dh, Dq): New.
15707         * config/aarch64/iterators.md (hq): New.
15709 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
15711         * reorg.c (get_branch_condition): Deal with conditional returns.
15712         (fill_simple_delay_slots): Remove dead code dealing with jumps.
15714 2013-04-01  Wei Mi  <wmi@google.com>
15716         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
15717         Truncate operand 2 using %b asm operand modifier.
15718         (*<shift_insn><mode>3_mask): Ditto.
15719         (*<rotate_insn><mode>3_mask): Ditto.
15721 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
15723         PR middle-end/56798
15724         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
15726 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
15728         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
15729         of next_real_insn.
15730         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
15732 2013-03-30  Lawrence Crowl  <crowl@google.com>
15734         * dse.c (clear_alias_sets): Remove never set.
15735         (disqualified_clear_alias_sets): Remove never set.
15736         (clear_alias_mode_pool): Remove never set.
15737         (dse_step0): Remove condition that is never true.
15738         (canon_address): Remove condition that is never true.
15739         (dse_step7): Remove condition that is never true.
15740         (rest_of_handle_dse): Remove condition that is never true.
15741         (rest_of_handle_dse::did_global): Remove never read from above.
15742         (dse_step2_spill): Remove never called from above.
15743         (dse_step5_spill): Remove never called from above.
15745 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
15747         * doc/md.texi (Standard Names) <casesi>: Update documentation for
15748         JUMP_TABLE_DATA changes.
15749         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
15750         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
15751         (Insns) <jump_table_data>: New entry.
15752         * doc/tm.texi: Regenerate.
15754         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
15756         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
15757         for table jump at the end of a basic block using tablejump_p.
15758         * targhooks.c (default_invalid_within_doloop): Likewise.
15759         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
15760         target hook implementation that is identical to the default hook.
15761         (rs6000_invalid_within_doloop): Remove.
15763         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
15764         unused variable from tablejump_p call.
15766         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
15767         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
15768         (INSN_DELETED_P): Likewise.
15769         (emit_jump_table_data): New prototype.
15770         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
15771         after 4th as unused.
15772         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
15773         * sched-vis.c (print_insn): Likewise.
15774         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
15775         insn for compatibility with back ends that use next_active_insn to
15776         identify jump table data.
15777         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
15778         (remove_insn): Likewise.
15779         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
15780         to be emitted.
15781         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
15782         (emit_jump_table_data): New function.
15784         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
15785         basic block, a JUMP_TABLE_DATA never is.
15786         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
15787         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
15788         off from code handling real insns.
15789         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
15790         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
15791         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
15792         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
15793         is not a NONDEBUG_INSN_P.
15794         * ira-costs.c (scan_one_insn): Likewise.
15795         * jump.c (mark_all_labels): Likewise.
15796         (mark_jump_label_1): Likewise.
15797         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15798         * lra.c (get_insn_freq): Expect all insns reaching here to be in
15799         a basic block.
15800         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
15801         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
15802         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
15803         JUMP_TABLE_DATA_P insns.
15804         (calculate_elim_costs_all_insns): Likewise.
15805         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
15806         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
15807         (delete_output_reload): Code style fixups.
15808         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
15809         insn flags on this non-insn.
15810         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
15811         as scheduling barriers, for pre-change compatibility.
15812         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
15813         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
15815         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
15816         redundant JUMP_TABLE_DATA_P test.
15817         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
15818         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15819         (frv_for_each_packet): Likewise.
15820         * config/i386/i386.c (min_insn_size): Likewise.
15821         (ix86_avoid_jump_mispredicts): Likewise.
15822         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15823         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
15824         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15825         (mips16_insn_length): Robustify.
15826         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
15827         (mips16_split_long_branches): Likewise.
15828         * config/pa/pa.c (pa_combine_instructions): Likewise.
15829         * config/rs6000/rs6000.c (get_next_active_insn): Treat
15830         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
15831         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
15832         as contributing to pool range lengths.
15833         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
15834         Remove redundant JUMP_TABLE_DATA_P test.
15835         (sh_loop_align): Likewise.
15836         (split_branches): Likewise.
15837         (sh_insn_length_adjustment): Likewise.
15838         * config/spu/spu.c (get_branch_target): Likewise.
15840 2013-03-29  Jan Hubicka  <jh@suse.cz>
15842         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
15843         gcov streaming; stream hot bb threshold to ltrans.
15844         * predict.c (get_hot_bb_threshold): Break out from ....
15845         (maybe_hot_count_p): ... here.
15846         (set_hot_bb_threshold): New function.
15847         * lto-section-in.c (lto_section_name): Add profile.
15848         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
15849         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
15850         and data-streamer.h
15851         (histogram_entry): New structure.
15852         (histogram, histogram_pool): New global vars.
15853         (histogram_hash): New structure.
15854         (histogram_hash::hash): New method.
15855         (histogram_hash::equal): Likewise.
15856         (account_time_size): New function.
15857         (cmp_counts): New function.
15858         (dump_histogram): New function.
15859         (ipa_profile_generate_summary): New function.
15860         (ipa_profile_write_summary): New function.
15861         (ipa_profile_read_summary): New function.
15862         (ipa_profile): Decide on threshold.
15863         (pass_ipa_profile): Add ipa_profile_write_summary and
15864         ipa_profile_read_summary.
15865         * Makefile.in (ipa.o): Update dependencies.
15866         * lto-streamer.h (LTO_section_ipa_profile): New section.
15868 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
15870         * tree.h (VAR_P): New.
15872 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
15874         PR lto/56777
15875         * doc/invoke.texi ([-fwhole-program]): Fix typo.
15877 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
15879         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
15880         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15881         (control_flow_insn_p): Likewise.
15882         * cfgrtl.c (duplicate_insn_chain): Likewise.
15883         * final.c (get_attr_length_1): Likewise.
15884         (shorten_branches): Likewise.
15885         (final_scan_insn): Likewise.
15886         * function.c (instantiate_virtual_regs): Likewise.
15887         * gcse.c (insert_insn_end_basic_block): Likewise.
15888         * ira-costs.c (scan_one_insn): Likewise.
15889         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15890         * lra.c (check_rtl): Likewise.
15891         * reload1.c (elimination_costs_in_insn): Likewise.
15892         * reorg.c (follow_jumps): Likewise.
15894         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
15895         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15896         (thumb_far_jump_used_p): Likewise.
15897         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
15898         (workaround_speculation): Likewise.
15899         (add_sched_insns_for_speculation): Likewise.
15900         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
15901         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15902         (frv_for_each_packet): Likewise.
15903         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
15904         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
15905         (final_emit_insn_group_barriers): Likewise.
15906         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15907         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15908         (mips16_insn_length): Likewise.
15909         * config/pa/pa.c (pa_reorg): Likewise.
15910         (pa_combine_instructions): Likewise.
15911         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
15912         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
15913         (sh_reorg): Likewise.
15914         (split_branches): Likewise.
15915         * config/spu/spu.c (get_branch_target): Likewise.
15917         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
15918         JUMP_TABLE_DATA_P.
15920 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
15922         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
15923         Fix declaration name.
15925 2013-03-28  Lawrence Crowl  <crowl@google.com>
15927         * graphds.h (struct graph.indicies): Remove unused.
15928         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
15929         (SCOP_ORIGINAL_PDDRS): Remove unused.
15930         * sese.h (extern insert_loop_close_phis): Removed unused.
15931         (extern insert_guard_phis): Removed unused.
15932         (extern ivtype_map_elt_info): Removed unused.
15933         (new_ivtype_map_elt): Removed unused.
15934         * sese.c (ivtype_map_elt_info): Removed unused.
15936 2013-03-28  Lawrence Crowl  <crowl@google.com>
15938         * Makefile.in: Add several missing include dependences.
15939         (DUMPFILE_H): New.
15940         (test-dump.o): New.  This object is not added to any executable,
15941         but is present for ad-hoc testing.
15942         * bitmap.c
15943         (debug (const bitmap_head_def &)): New.
15944         (debug (const bitmap_head_def *)): New.
15945         * bitmap.h
15946         (extern debug (const bitmap_head_def &)): New.
15947         (extern debug (const bitmap_head_def *)): New.
15948         * cfg.c
15949         (debug (edge_def &)): New.
15950         (debug (edge_def *)): New.
15951         * cfghooks.c
15952         (debug (basic_block_def &)): New.
15953         (debug (basic_block_def *)): New.
15954         * dumpfile.h
15955         (dump_node (const_tree, int, FILE *)): Correct source file.
15956         * dwarf2out.c
15957         (debug (die_struct &)): New.
15958         (debug (die_struct *)): New.
15959         * dwarf2out.h
15960         (extern debug (die_struct &)): New.
15961         (extern debug (die_struct *)): New.
15962         * gimple-pretty-print.c
15963         (debug (gimple_statement_d &)): New.
15964         (debug (gimple_statement_d *)): New.
15965         * gimple-pretty-print.h
15966         (extern debug (gimple_statement_d &)): New.
15967         (extern debug (gimple_statement_d *)): New.
15968         * ira-build.c
15969         (debug (ira_allocno_copy &)): New.
15970         (debug (ira_allocno_copy *)): New.
15971         (debug (ira_allocno &)): New.
15972         (debug (ira_allocno *)): New.
15973         * ira-int.h
15974         (extern debug (ira_allocno_copy &)): New.
15975         (extern debug (ira_allocno_copy *)): New.
15976         (extern debug (ira_allocno &)): New.
15977         (extern debug (ira_allocno *)): New.
15978         * ira-lives.c
15979         (debug (live_range &)): New.
15980         (debug (live_range *)): New.
15981         * lra-int.h
15982         (debug (lra_live_range &)): New.
15983         (debug (lra_live_range *)): New.
15984         * lra-lives.c
15985         (debug (lra_live_range &)): New.
15986         (debug (lra_live_range *)): New.
15987         * omega.c
15988         (debug (omega_pb_d &)): New.
15989         (debug (omega_pb_d *)): New.
15990         * omega.h
15991         (extern debug (omega_pb_d &)): New.
15992         (extern debug (omega_pb_d *)): New.
15993         * print-rtl.c
15994         (debug (const rtx_def &)): New.
15995         (debug (const rtx_def *)): New.
15996         * print-tree.c
15997         (debug_tree (tree): Move within file.
15998         (debug_raw (const tree_node &)): New.
15999         (debug_raw (const tree_node *)): New.
16000         (dump_tree_via_hooks (const tree_node *, int)): New.
16001         (debug (const tree_node &)): New.
16002         (debug (const tree_node *)): New.
16003         (debug_verbose (const tree_node &)): New.
16004         (debug_verbose (const tree_node *)): New.
16005         (debug_head (const tree_node &)): New.
16006         (debug_head (const tree_node *)): New.
16007         (debug_body (const tree_node &)): New.
16008         (debug_body (const tree_node *)): New.
16009         (debug_vec_tree (tree): Move and reimplement in terms of dump.
16010         (debug (vec<tree, va_gc> &)): New.
16011         (debug (vec<tree, va_gc> *)): New.
16012         * rtl.h
16013         (extern debug (const rtx_def &)): New.
16014         (extern debug (const rtx_def *)): New.
16015         * sbitmap.c
16016         (debug_raw (simple_bitmap_def &)): New.
16017         (debug_raw (simple_bitmap_def *)): New.
16018         (debug (simple_bitmap_def &)): New.
16019         (debug (simple_bitmap_def *)): New.
16020         * sbitmap.h
16021         (extern debug (simple_bitmap_def &)): New.
16022         (extern debug (simple_bitmap_def *)): New.
16023         (extern debug_raw (simple_bitmap_def &)): New.
16024         (extern debug_raw (simple_bitmap_def *)): New.
16025         * sel-sched-dump.c
16026         (debug (vinsn_def &)): New.
16027         (debug (vinsn_def *)): New.
16028         (debug_verbose (vinsn_def &)): New.
16029         (debug_verbose (vinsn_def *)): New.
16030         (debug (expr_def &)): New.
16031         (debug (expr_def *)): New.
16032         (debug_verbose (expr_def &)): New.
16033         (debug_verbose (expr_def *)): New.
16034         (debug (vec<rtx> &)): New.
16035         (debug (vec<rtx> *)): New.
16036         * sel-sched-dump.h
16037         (extern debug (vinsn_def &)): New.
16038         (extern debug (vinsn_def *)): New.
16039         (extern debug_verbose (vinsn_def &)): New.
16040         (extern debug_verbose (vinsn_def *)): New.
16041         (extern debug (expr_def &)): New.
16042         (extern debug (expr_def *)): New.
16043         (extern debug_verbose (expr_def &)): New.
16044         (extern debug_verbose (expr_def *)): New.
16045         (extern debug (vec<rtx> &)): New.
16046         (extern debug (vec<rtx> *)): New.
16047         * sel-sched-ir.h
16048         (_list_iter_cond_expr): Make inline instead of static.
16049         * sreal.c
16050         (debug (sreal &)): New.
16051         (debug (sreal *)): New.
16052         * sreal.h
16053         (extern debug (sreal &)): New.
16054         (extern debug (sreal *)): New.
16055         * tree.h
16056         (extern debug_raw (const tree_node &)): New.
16057         (extern debug_raw (const tree_node *)): New.
16058         (extern debug (const tree_node &)): New.
16059         (extern debug (const tree_node *)): New.
16060         (extern debug_verbose (const tree_node &)): New.
16061         (extern debug_verbose (const tree_node *)): New.
16062         (extern debug_head (const tree_node &)): New.
16063         (extern debug_head (const tree_node *)): New.
16064         (extern debug_body (const tree_node &)): New.
16065         (extern debug_body (const tree_node *)): New.
16066         (extern debug (vec<tree, va_gc> &)): New.
16067         (extern debug (vec<tree, va_gc> *)): New.
16068         * tree-cfg.c
16069         (debug (struct loop &)): New.
16070         (debug (struct loop *)): New.
16071         (debug_verbose (struct loop &)): New.
16072         (debug_verbose (struct loop *)): New.
16073         * tree-dump.c: Add header dependence.
16074         * tree-flow.h
16075         (extern debug (struct loop &)): New.
16076         (extern debug (struct loop *)): New.
16077         (extern debug_verbose (struct loop &)): New.
16078         (extern debug_verbose (struct loop *)): New.
16079         * tree-data-ref.c
16080         (debug (data_reference &)): New.
16081         (debug (data_reference *)): New.
16082         (debug (vec<data_reference_p> &)): New.
16083         (debug (vec<data_reference_p> *)): New.
16084         (debug (vec<ddr_p> &)): New.
16085         (debug (vec<ddr_p> *)): New.
16086         * tree-data-ref.h
16087         (extern debug (data_reference &)): New.
16088         (extern debug (data_reference *)): New.
16089         (extern debug (vec<data_reference_p> &)): New.
16090         (extern debug (vec<data_reference_p> *)): New.
16091         (extern debug (vec<ddr_p> &)): New.
16092         (extern debug (vec<ddr_p> *)): New.
16093         * tree-ssa-alias.c
16094         (debug (pt_solution &)): New.
16095         (debug (pt_solution *)): New.
16096         * tree-ssa-alias.h
16097         (extern debug (pt_solution &)): New.
16098         (extern debug (pt_solution *)): New.
16099         * tree-ssa-alias.c
16100         (debug (_var_map &)): New.
16101         (debug (_var_map *)): New.
16102         (debug (tree_live_info_d &)): New.
16103         (debug (tree_live_info_d *)): New.
16104         * tree-ssa-alias.h
16105         (extern debug (_var_map &)): New.
16106         (extern debug (_var_map *)): New.
16107         (extern debug (tree_live_info_d &)): New.
16108         (extern debug (tree_live_info_d *)): New.
16110 2013-03-28  Jan Hubicka  <jh@suse.cz>
16112         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16114 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16116         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16117         record only when desired or required.
16119 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16121         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16122         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16124 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16126         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16127         (*andsi3_compare0_uxtw): New pattern.
16128         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16129         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16131 2013-03-28  Jan Hubicka  <jh@suse.cz>
16133         * data-streamer-in.c (streamer_read_gcov_count): New function.
16134         * gimple-streamer-out.c: Include value-prof.h.
16135         (output_gimple_stmt): Output histogram.
16136         (output_bb): Use streamer_write_gcov_count.
16137         * value-prof.c: Include data-streamer.h
16138         (dump_histogram_value): Add HIST_TYPE_MAX.
16139         (stream_out_histogram_value): New function.
16140         (stream_in_histogram_value): New function.
16141         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16142         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16143         * data-streamer-out.c (streamer_write_gcov_count): New function.
16144         (streamer_write_gcov_count_stream): New function.
16145         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16146         (lto_output_node): Likewise.
16147         (input_node, input_edge): Likewise.
16148         * lto-streamer-out.c (output_cfg): Update streaming.
16149         * lto-streamer-in.c (input_cfg): Likewise.
16150         * data-streamer.h (streamer_write_gcov_count,
16151         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16152         * gimple-streamer-in.c: Include value-prof.h
16153         (input_gimple_stmt): Input histograms.
16154         (input_bb): Update profile streaming.
16156 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16158         * genmodes.c (emit_max_int): New function.
16159         (emit_insn_modes_h): Added call to emit_max_function.
16160         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16161         Added doc.
16162         * machmode.def: Fixed comment.
16164 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16166         * combine.c (try_combine): Removed useless assert.
16167         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16169 2013-03-28  Marek Polacek  <polacek@redhat.com>
16170             Richard Biener  <rguenther@suse.de>
16172         PR tree-optimization/56695
16173         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16174         build signed result of a vector comparison.
16175         * tree-cfg.c (verify_gimple_comparison): Check that a result
16176         of a vector comparison has signed type.
16178 2013-03-28  Richard Biener  <rguenther@suse.de>
16180         PR tree-optimization/37021
16181         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16182         do not restrict gaps between groups.
16183         * tree-vect-stmts.c (vectorizable_load): Properly account for
16184         a gap between groups.
16186 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16188         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16189         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16190         is not enabled.
16192 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16194         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16195         * doc/extend.texi (Named Address Spaces): Ditto.
16196         (Variable Attributes): Ditto.
16198 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16200         * config.build: Add support for cygwin x64 target.
16201         * config.gcc: Likewise.
16202         * config.host: Likewise.
16203         * configure.ac: Likewise
16204         * configure: Regenerated.
16206 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16208         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16209         * config/i386/t-cygwin-w64: New file.
16210         * config/i386/cygwin-w64.h: New file.
16211         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16212         and add support for x64-cygwin target.
16213         (CPP_SPEC): Likewise.
16214         (CXX_WRAP_SPEC_LIST): Undefine before define.
16215         (LIBGCJ_SONAME): Use 15 as version.
16217 2013-03-27  Richard Biener  <rguenther@suse.de>
16219         PR tree-optimization/56716
16220         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16221         dumping for ref nodes.
16223 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16225         PR tree-optimization/55334
16226         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16227         restricted pointers to arrays.
16229 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16231         * Makefile.in (.SUFFIXES): Add .cc.
16232         (.c.o): Apply same recipe for implicit rule .cc.o.
16234 2013-03-27  Richard Biener  <rguenther@suse.de>
16236         PR tree-optimization/37021
16237         * tree-vect-data-refs.c (vect_check_strided_load): Allow
16238         REALPART/IMAGPART_EXPRs around the supported refs.
16239         * tree-ssa-structalias.c (find_func_aliases): Assume that
16240         floating-point values are not used to transfer pointers.
16242 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16244         * target.def (TARGET_HAS_IFUNC_P): New target hook.
16245         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16246         * doc/tm.texi: Regenerate.
16247         * targhooks.h (default_has_ifunc_p): New.
16248         * targhooks.c (default_has_ifunc_p): Ditto.
16249         * config/linux-protos.h: New file.
16250         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16251         hook for linux which disables support of indirect functions in android.
16252         * config/linux-android.c: New file.
16253         * config/t-linux-android.c: Ditto.
16254         * config.gcc: Added new object file linux-android.o.
16255         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16256         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16257         * varasm.c (do_assemble_alias): Likewise.
16258         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16259         doesn't support indirect functions.
16260         * configure: Regenerate.
16262 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
16264         PR target/56102
16265         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16266         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16267         mult-word mode.
16269 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16271         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16273 2013-03-27  Terry Guo  <terry.guo@arm.com>
16275         * config/arm/arm-cores.def: Added core cortex-r7.
16276         * config/arm/arm-tune.md: Regenerated.
16277         * config/arm/arm-tables.opt: Regenerated.
16278         * doc/invoke.texi: Added entry for core cortex-r7.
16280 2013-03-27  Walter Lee  <walt@tilera.com>
16282         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16283         double-decrement of next_scratch_regno.
16285 2013-03-27  Walter Lee  <walt@tilera.com>
16287         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16288         input operands.
16289         (insn_v1mulus): Ditto.
16290         (insn_v2muls): Ditto.
16292 2013-03-27  Walter Lee  <walt@tilera.com>
16294         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16295         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16297 2013-03-27  Walter Lee  <walt@tilera.com>
16299         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16300         (*sibcall_value): Ditto.
16302 2013-03-27  Walter Lee  <walt@tilera.com>
16304         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16305         (insn_mnz_v8qi): ... this ...
16306         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
16307         vector equivalent.
16308         (insn_v<n>mnz): Replaced by ...
16309         (insn_v1mnz): ... this ...
16310         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
16311         equivalent.
16312         (insn_mz_<mode>): Replaced by ...
16313         (insn_mz_v8qi): ... this ...
16314         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
16315         vector equivalent.
16316         (insn_v<n>mz): Replaced by ...
16317         (insn_v1mz): ... this ...
16318         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
16319         equivalent.
16321 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
16323         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16325 2013-03-26  Roland McGrath  <mcgrathr@google.com>
16327         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16328         than fprintf with a non-constant, non-format string.
16330 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16332         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16333         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
16334         operand 0 predicate.
16335         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16336         attribute.  Use general_x64nomem_operand as operand 1 predicate.
16337         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16338         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16339         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16340         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16341         (mov<mode>_insv_1): Remove expander.  Merge insn with
16342         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16343         Use general_x64nomem_operand as operand 1 predicate.
16344         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16345         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16346         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16347         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16348         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
16349         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16350         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16351         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16352         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
16353         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16354         (general_x64nomem_operand): Ditto.
16356 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
16358         * config/rtems.opt: Add -pthread option.
16360 2013-03-26  Richard Biener  <rguenther@suse.de>
16362         * alias.c (find_base_term): Avoid redundant and not used recursion.
16363         (base_alias_check): Get the initial base term from the caller.
16364         (true_dependence_1): Compute and pass base terms to base_alias_check.
16365         (write_dependence_p): Likewise.
16366         (may_alias_p): Likewise.
16368 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
16370         * config/aarch64/aarch64.c (aarch64_classify_address): Support
16371         PC-relative load in SI modes and above only.
16373 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
16375         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16376         * config/arm/iwmmxt.md (WCGR0): Update.
16377         (WCGR1, WCGR2, WCGR3): Likewise.
16379 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16381         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16382         Use x64 and nox64 isa attributes.
16384 2013-03-26  Richard Biener  <rguenther@suse.de>
16386         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16387         alignment computations and rely on get_object_alignment_1
16388         for the !TYPE_P case.
16389         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16391 2013-03-26  Walter Lee  <walt@tilera.com>
16393         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16394         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16396 2013-03-25  Jeff Law  <law@redhat.com>
16398         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16399         check for INTEGRAL_TYPE_P that was missing due to checking in
16400         wrong version of prior patch.
16402 2013-03-25  Walter Lee  <walt@tilera.com>
16404         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16405         TILEGX_INSN_SHUFFLEBYTES1.
16406         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16407         shufflebytes1.
16408         (tilegx_builtins): Ditto.
16409         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16411 2013-03-25  Walter Lee  <walt@tilera.com>
16413         * config/tilegx/tilegx.md (floatsisf2): New pattern.
16414         (floatunssisf2): New pattern.
16415         (floatsidf2): New pattern.
16416         (floatunssidf2): New pattern.
16418 2013-03-25  Walter Lee  <walt@tilera.com>
16420         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16421         tests for constraint J, K, N, P.
16423 2013-03-25  Walter Lee  <walt@tilera.com>
16425         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16426         Use indirect/pcrel encoding.
16427         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16428         Ditto.
16430 2013-03-25  Steve Ellcey  <sellcey@mips.com>
16432         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16433         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16434         * config/mips/mips.c (mips_option_override): Set IMADD default.
16435         * config/mips/mips.h (PTF_AVOID_IMADD): New.
16436         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16437         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16438         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16439         * doc/invoke.texi (-mimadd/-mno-imadd): New.
16441 2013-03-25  Jeff Law  <law@redhat.com>
16443         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16444         slightly to avoid creating and folding useless trees.  Simplify
16445         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16447 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
16449         * config/i386/i386.md (*zero_extendsidi2): Merge with
16450         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
16451         * config/i386/predicates.md (x86_64_zext_operand): Rename from
16452         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
16453         targets.  Clarify comment.
16455 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16457         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16458         pass-through jump functions differently.
16459         (ipa_read_jump_function): Likewise.  Also use setter functions to set
16460         up jump functions.
16462 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16464         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16465         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16466         process it.
16467         (ipa_get_indirect_edge_target): New function.
16468         (devirtualization_time_bonus): New parameter known_aggs, pass it to
16469         ipa_get_indirect_edge_target.  Update all callers.
16470         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
16471         ipa_get_indirect_edge_target_1 instead of calling
16472         ipa_get_indirect_edge_target.
16473         (create_specialized_node): Pass aggvlas to
16474         ipcp_discover_new_direct_edges.
16476 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16478         * config/arm/arm.md (f_sels, f_seld): New types.
16479         (*cmov<mode>): New pattern.
16480         * config/arm/predicates.md (arm_vsel_comparison_operator): New
16481         predicate.
16483 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16485         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16486         POSIX-printf for mingw-hosted  builds.
16488 2013-03-25  Richard Biener  <rguenther@suse.de>
16490         PR middle-end/56694
16491         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16492         must-not-throw stmt location.
16494 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16496         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16497         Emit load-acquire versions when acq is true.
16498         (arm_emit_store_exclusive): Add rel parameter.
16499         Emit store-release versions when rel is true.
16500         (arm_split_compare_and_swap): Use acquire-release instructions
16501         instead.
16502         of barriers when appropriate.
16503         (arm_split_atomic_op): Likewise.
16504         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16505         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16506         (VUNSPEC_SLX): Likewise.
16507         (VUNSPEC_LDA): Likewise.
16508         (VUNSPEC_STL): Likewise.
16509         * config/arm/sync.md (atomic_load<mode>): New pattern.
16510         (atomic_store<mode>): Likewise.
16511         (arm_load_acquire_exclusive<mode>): Likewise.
16512         (arm_load_acquire_exclusivesi): Likewise.
16513         (arm_load_acquire_exclusivedi): Likewise.
16514         (arm_store_release_exclusive<mode>): Likewise.
16516 2013-03-25  Catherine Moore  <clm@codesourcery.com>
16518         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16519         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16520         * config/mip/predicates.md (lwsp_swsp_operand,
16521         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16522         sb16_operand, db4_operand, db7_operand, ib3_operand,
16523         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16524         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16525         andi16_operand): New predicates.
16526         * config/mips/mips.md (compression): New attribute.
16527         (enabled): New attribute.
16528         (length): Consider compression in computing length.
16529         (shift_compression): New code attribute.
16530         (*add<mode>3): New operands. Record compression.
16531         (sub<mode>3): Likewise.
16532         (one_cmpl<mode>2): Likewise.
16533         (*and<mode>3): Likewise.
16534         (*ior<mode>3): Likewise.
16535         (unnamed pattern for xor): Likewise.
16536         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16537         (*<optab><mode>3): Likewise.
16538         (*mov<mode>_internal: Likewise.
16539         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16540         (mips_unsigned_immediate_p): New.
16541         (umips_lwsp_swsp_address_p): New.
16542         (m16_based_address_p): New.
16543         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16544         (mips_unsigned_immediate_p): New prototype.
16545         (lwsp_swsp_address_p): New prototype.
16546         (m16_based_address_p): New prototype.
16547         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16548         (mips_signed_immediate_p): New function.
16549         (m16_based_address_p): New function.
16550         (lwsp_swsp_address_p): New function.
16551         (mips_print_operand_punctuation): Recognize short delay slot insns
16552         for microMIPS.add<mode>3"
16554 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16556         PR target/56720
16557         * config/arm/iterators.md (v_cmp_result): New mode attribute.
16558         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16560 2013-03-25  Richard Biener  <rguenther@suse.de>
16562         PR tree-optimization/56689
16563         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16564         any edge.
16566 2013-03-25  Richard Biener  <rguenther@suse.de>
16568         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16569         of bitmap.
16570         (memory_references): Likewise.
16571         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16572         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16573         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16574         (gather_mem_refs_in_loops): Fold into ...
16575         (analyze_memory_references): ... this.  Move initialization
16576         to tree_ssa_lim_initialize.
16577         (fill_always_executed_in): Rename to ...
16578         (fill_always_executed_in_1): ... this.
16579         (fill_always_executed_in): Move contains_call computation to
16580         this new function from ...
16581         (tree_ssa_lim_initialize): ... here.
16582         (tree_ssa_lim): Call fill_always_executed_in.
16584 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16586         * postreload.c (reload_combine): Fix code detecting returns.
16588 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16590         * function.c (emit_use_return_register_into_block): On cc0 targets,
16591         do not emit the sequence between cc0 setter and user.
16593 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16595         * config/i386/predicates.md (local_symbolic_operand): Interpret
16596         dll-imported symbols as none-local.
16598 2013-03-25  Richard Biener  <rguenther@suse.de>
16600         * tree-ssa-loop-im.c (struct depend): Remove.
16601         (struct lim_aux_data): Make depends a vec of gimples.
16602         (free_lim_aux_data): Adjust.
16603         (add_dependency): Likewise.
16604         (set_level): Likewise.
16606 2013-03-25  Richard Biener  <rguenther@suse.de>
16608         PR middle-end/56434
16609         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
16610         the pointer returned by calls with ECF_MALLOC set.
16612 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16614         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
16616 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16618         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
16619         using MMXMODE mode iterator.
16620         (*move<mode>_internal): Merge with *movv2sf_internal and
16621         *movv2sf_internal_rex64 using MMXMODE mode iterator.
16623 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16625         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
16626         (record_last_mem_set_info): Likewise.
16628         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
16629         of XNEWVEC followed by memset.
16630         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
16632 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16634         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
16635         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
16636         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
16637         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
16638         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
16639         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
16640         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
16641         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
16642         BARRIER_P instead of GET_CODE.
16644 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16646         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
16647         inaccuracy in the probing code.
16649         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
16650         (ctrapdi4): Likewise.
16652 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16654         * calls.c (expand_call): Add missing guard to code handling return
16655         of non-BLKmode structures in MSB.
16656         * function.c (expand_function_end): Likewise.
16658 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16660         * combine.c (try_combine): Adjust comment.  Do not add the set of
16661         insn #0 if the destination indirectly is set or dies in insn #2.
16662         Tidy up code to distribute a new note.
16664 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16666         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
16667         also for alternatives 16 and 17.
16669 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16671         * config/i386/sse.md (*mov<mode>_internal): Merge with
16672         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
16673         Emit insn template depending on type attribute.  Use
16674         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16675         movd instead of movq mnemonic for interunit moves.  Rewrite mode
16676         attribute calculation.  Remove unit attribute calculation.
16677         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
16678         Set prefix_data16 attribute for DImode ssemov types.
16679         Use Ym instead of y for SSE-MMX conversion alternatives.
16680         Reorder operand constraints.
16682 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
16684         * df.h (df_insn_delete): Adjust prototype.
16685         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
16686         and let it decide whether mark the basic block dirty.
16687         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
16688         * df-scan.c (df_insn_info_delete): New helper function, split
16689         off from df_insn_delete.
16690         (df_scan_free_bb_info): Use it.
16691         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
16692         Likewise.
16693         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
16694         that the insn is actually an insn and it has a non-NULL basic block.
16695         Do not mark basic block dirty if only deleting a DEBUG_INSN.
16697 2013-03-22  Richard Biener  <rguenther@suse.de>
16699         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
16700         dep_ref members.
16701         (mem_ref_alloc): Do not allocate them.
16702         (refs_independent_p): Do not query or maintain a cache.
16704 2013-03-22  Richard Biener  <rguenther@suse.de>
16706         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
16707         (gather_mem_refs_in_loops): Do not compute it.
16708         (analyze_memory_references): Do not allocate it.
16709         (tree_ssa_lim_finalize): Do not free it.
16710         (for_all_locs_in_loop): Do not query all_refs_in_loop.
16712 2013-03-22  Richard Biener  <rguenther@suse.de>
16714         * is-a.h (as_a): Use gcc_checking_assert.
16716 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
16718         * config/aarch64/aarch64.c (aarch64_print_operand): New
16719         format specifier for printing a constant in hex.
16720         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
16721         format specifier for printing second operand.
16723 2013-03-22  Richard Biener  <rguenther@suse.de>
16725         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
16726         bitmaps.
16727         (gather_mem_refs_in_loops): Perform store accumulation here.
16728         (create_vop_ref_mapping_loop): Remove.
16729         (create_vop_ref_mapping): Likewise.
16730         (analyze_memory_references): Initialize refs_stored_in_loop.
16731         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
16732         (record_indep_loop): Remove.
16733         (record_dep_loop): New function.
16734         (ref_indep_loop_p_1): Adjust to only walk over references
16735         in the loop, not its subloops.
16736         (ref_indep_loop_p): Rename to ...
16737         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
16738         maintaining a more fine-grained cache.
16739         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
16740         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
16742 2013-03-22  Richard Biener  <rguenther@suse.de>
16744         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
16745         (struct mem_ref): Make accesses_in_loop a vec of a vec of
16746         aggregate mem_ref_loc.
16747         (free_mem_ref_locs): Inline into ...
16748         (memref_free): ... this and adjust.
16749         (mem_ref_alloc): Adjust.
16750         (mem_ref_locs_alloc): Remove.
16751         (record_mem_ref_loc): Adjust.
16752         (get_all_locs_in_loop): Rewrite into ...
16753         (for_all_locs_in_loop): ... this iterator.
16754         (rewrite_mem_ref_loc): New functor.
16755         (rewrite_mem_refs): Use for_all_locs_in_loop.
16756         (sm_set_flag_if_changed): New functor.
16757         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
16758         (ref_always_accessed): New functor.
16759         (ref_always_accessed_p): Use for_all_locs_in_loop.
16761 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16763         * tree-pass.h (PROP_gimple_lvec): New.
16764         * passes.c (dump_properties): Handle PROP_gimple_lvec.
16765         (init_optimization_passes): Move pass_lower_vector.
16766         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
16767         PROP_gimple_lvec.
16768         (pass_lower_vector): Provide PROP_gimple_lvec.
16769         (pass_lower_vector_ssa): Likewise.
16770         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
16772 2013-03-21  Mark Wielaard  <mjw@redhat.com>
16774         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
16776 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16778         * config/i386/i386.md (*movdi_internal): Disparage slightly
16779         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
16780         conversion alternatives.
16782 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
16784         PR middle-end/48087
16785         * diagnostic.def (DK_WERROR): New kind.
16786         * diagnostic.h (werrorcount): Define.
16787         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
16788         promoted to DK_ERROR, increment DK_WERROR counter instead of
16789         DK_ERROR counter.
16790         * toplev.c (toplev_main): Call print_ignored_options even if
16791         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
16792         even if just werrorcount is non-zero.
16794         PR debug/55608
16795         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
16796         on failure.
16797         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
16798         (string_cst_pool_decl): New function.
16799         (optimize_one_addr_into_implicit_ptr): New function.
16800         (resolve_addr_in_expr): Optimize DWARF location expression
16801         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
16802         which doesn't live in memory, but has DW_AT_location or
16803         DW_AT_const_value, or refers to a string literal, into
16804         DW_OP_GNU_implicit_pointer.
16805         (optimize_location_into_implicit_ptr): New function.
16806         (resolve_addr): If removing DW_AT_location of a variable because
16807         it was DW_OP_addr of address of the variable, but the variable doesn't
16808         live in memory, try to emit const value attribute for the initializer.
16810 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16812         * tree.h (VECTOR_TYPE_P): New macro.
16813         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
16814         TYPE_MODE): Use it.
16815         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
16816         VEC_COND_EXPR cannot be lvalues.
16817         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
16819 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16821         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
16822         Restrict the transformation to equal modes.
16824 2013-03-21  Richard Biener  <rguenther@suse.de>
16826         PR tree-optimization/39326
16827         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
16828         (MEM_ANALYZABLE): Adjust.
16829         (record_mem_ref_loc): Move bitmap ops ...
16830         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
16831         unanalyzable refs, do not record locations for it.
16832         (analyze_memory_references): Allocate ref zero as shared
16833         unanalyzable ref.
16834         (refs_independent_p): Do not test for unanalyzed mems here.
16835         (ref_indep_loop_p_1): Special-case disambiguation against
16836         the unanalyzed ref.
16837         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
16839 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
16841         * config/arm/arm-protos.h (tune_params): Add
16842         prefer_neon_for_64bits field.
16843         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
16844         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
16845         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
16846         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
16847         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
16848         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
16849         (arm_option_override): Handle -mneon-for-64bits new option.
16850         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
16851         (prefer_neon_for_64bits): Declare new variable.
16852         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
16853         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
16854         (arch_enabled): Handle new arch types. Remove support for onlya8
16855         and nota8.
16856         (one_cmpldi2): Use new arch names.
16857         (zero_extend<mode>di2, extend<mode>di2): Ditto.
16858         * config/arm/arm.opt (mneon-for-64bits): Add option.
16859         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
16860         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
16861         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
16862         of onlya8.
16863         * doc/invoke.texi (-mneon-for-64bits): Document.
16865 2013-03-21  Richard Biener  <rguenther@suse.de>
16867         PR tree-optimization/39326
16868         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
16869         (sort_bbs_in_loop_postorder_cmp): New function.
16870         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
16872 2013-03-21  Richard Biener  <rguenther@suse.de>
16874         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
16875         (vect_insert_into_interleaving_chain): Likewise.
16876         (vect_drs_dependent_in_basic_block): Inline ...
16877         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
16878         split out from ...
16879         (vect_analyze_data_ref_dependence): ... here.  Simplify.
16880         (vect_check_interleaving): Simplify.
16881         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
16882         (vect_slp_analyze_data_ref_dependences): ... this new function.
16883         (dr_group_sort_cmp): New function.
16884         (vect_analyze_data_ref_accesses): Compute data-reference groups
16885         here instead of in vect_analyze_data_ref_dependence.  Use
16886         a more efficient algorithm.
16887         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
16888         vect_slp_analyze_data_ref_dependences.  Call
16889         vect_analyze_data_ref_accesses earlier.
16890         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
16891         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
16892         (vect_slp_analyze_data_ref_dependences): New prototype.
16894 2013-03-21  Richard Biener  <rguenther@suse.de>
16896         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
16897         ref is stored in the loop.
16898         (find_refs_for_sm): Walk only over all stores.
16899         (store_motion_loop): Allocate from lim_bitmap_obstack.
16900         (store_motion): Likewise.
16902 2013-03-21  Richard Biener  <rguenther@suse.de>
16904         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
16905         Update virtual SSA form.
16907 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16909         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
16910         * configure: Regenerate.
16911         * config.in: Regenerate.
16912         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
16913         if !HAVE_LD_EH_FRAME_CIEV3.
16915 2013-03-21  Richard Biener  <rguenther@suse.de>
16917         * tree-cfg.c (verify_expr_no_block): New function.
16918         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
16919         nor DECL_VALUE_EXPR have locations with associated blocks.
16920         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
16921         (clear_unused_block_pointer): Remove code dealing with
16922         blocks in DECL_DEBUG_EXPR locations.
16924 2013-03-21  Richard Biener  <rguenther@suse.de>
16926         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
16927         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
16928         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
16929         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
16930         instead of DECL_DEBUG_EXPR_IS_FROM.
16931         * gimplify.c (gimplify_modify_expr): Likewise.
16932         * tree-cfg.c (verify_expr_location_1): Likewise.
16933         * tree-complex.c (create_one_component_var): Likewise.
16934         * tree-sra.c (create_access_replacement): Likewise.
16935         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
16936         (clear_unused_block_pointer): Likewise.
16937         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
16938         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
16939         * var-tracking.c (var_debug_decl): Likewise.
16940         (track_expr_p): Likewise.
16941         * tree-inline.c (add_local_variables): Likewise.  Set
16942         DECL_HAS_DEBUG_EXPR_P after copying it.
16943         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
16944         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
16946 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16948         PR bootstrap/56656
16949         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
16950         * configure: Regenerate.
16951         * config.in: Regenerate.
16952         * config/i386/i386.md (*movdf_internal): Use
16953         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16954         movd instead of movq mnemonic for interunit moves.
16955         (*movdi_internal): Ditto.
16957 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16959         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
16960         (abd<mode>_3): New pattern.
16961         (aba<mode>_3): New pattern.
16962         (fabd<mode>_3): New pattern.
16964 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16966         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
16967         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
16968         occurrence of REGISTER_PREFIX as its empty string.
16970 2013-03-20  Jeff Law  <law@redhat.com>
16972         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
16973         addititional equivalences for equality comparisons between an SSA_NAME
16974         and a constant where the SSA_NAME was set from a widening conversion.
16976 2013-03-20  Walter Lee  <walt@tilera.com>
16978         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
16980 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16982         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
16983         depending on type attribute.
16984         (*movti_internal): Ditto.
16985         (*movtf_internal): Ditto.
16986         (*movxf_internal): Ditto.
16987         (*movdf_internal): Ditto.
16988         (*movsf_internal): Ditto.
16990 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16992         * config/i386/i386.md (*movti_internal): Set prefix attribute to
16993         maybe_vex for sselog1 and ssemov types.
16994         (*movdi_internal): Reorder operand constraints.
16995         (*movsi_internal): Ditto.  Set prefix attribute to
16996         maybe_vex for sselog1 and ssemov types.
16997         (*movtf_internal): Set prefix attribute to maybe_vex
16998         for sselog1 and ssemov types.
16999         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
17000         DImode ssemov types.  Reorder operand constraints.
17001         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
17002         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
17003         attribute for SImode ssemov types.  Reorder operand constraints.
17005 2013-03-20  Martin Jambor  <mjambor@suse.cz>
17007         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17008         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17010 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
17012         * config/rs6000/predicates.md (indexed_address, update_address_mem
17013         update_indexed_address_mem): New predicates.
17014         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17015         attribute for load/store instructions.
17016         * config/rs6000/dfp.md (movsd_store): Likewise.
17017         (movsd_load): Likewise.
17018         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17019         (unnamed HI->DI extend define_insn): Likewise.
17020         (unnamed SI->DI extend define_insn): Likewise.
17021         (unnamed QI->SI extend define_insn): Likewise.
17022         (unnamed QI->HI extend define_insn): Likewise.
17023         (unnamed HI->SI extend define_insn): Likewise.
17024         (unnamed HI->SI extend define_insn): Likewise.
17025         (extendsfdf2_fpr): Likewise.
17026         (movsi_internal1): Likewise.
17027         (movsi_internal1_single): Likewise.
17028         (movhi_internal): Likewise.
17029         (movqi_internal): Likewise.
17030         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17031         attribute for load/store instructions.
17032         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17033         instructions.
17034         (mov<mode>_softfloat): Likewise.
17035         (mov<mode>_hardfloat32): Likewise.
17036         (mov<mode>_hardfloat64): Likewise.
17037         (mov<mode>_softfloat64): Likewise.
17038         (movdi_internal32): Likewise.
17039         (movdi_internal64): Likewise.
17040         (probe_stack_<mode>): Likewise.
17042 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17044         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17045         floating point, and decimal floating point to reload iterator.
17047         * config/rs6000/constraints.md (wl constraint): New constraints to
17048         return FLOAT_REGS if certain options are used to reduce the number
17049         of separate patterns that exist in the file.
17050         (wx constraint): Likewise.
17051         (wz constraint): Likewise.
17053         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17054         -mdebug=reg, print wg, wl, wx, and wz constraints.
17055         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17056         Initialize the reload functions for 64-bit binary/decimal floating
17057         point types.
17058         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17059         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17060         create the buffer on the stack to overcome not having a 32-bit
17061         load and store.
17062         (rs6000_emit_move): Likewise.
17063         (rs6000_secondary_memory_needed_rtx): Likewise.
17064         (rs6000_alloc_sdmode_stack_slot): Likewise.
17065         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17066         via xxlxor, just like DFmode 0.0.
17068         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17069         define as 1 if we are running on a power7 or newer.
17070         (enum r6000_reg_class_enum): Add new constraints.
17072         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17073         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17074         with other moves by using conditional constraits (wg).  Use LFIWZX
17075         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17076         (movsd splitter): Likewise.
17077         (movsd_hardfloat): Likewise.
17078         (movsd_softfloat): Likewise.
17080         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17081         binary and decimal floating point moves.
17082         (fmove_ok): New attributes to combine binary and decimal floating
17083         point moves, and to combine power6x (mfpgpr) moves along normal
17084         floating moves.
17085         (real_value_to_target): Likewise.
17086         (f32_lr): Likewise.
17087         (f32_lm): Likewise.
17088         (f32_li): Likewise.
17089         (f32_sr): Likewise.
17090         (f32_sm): Likewise.
17091         (f32_si): Likewise.
17092         (movsf): Combine binary and decimal floating point moves.  Combine
17093         power6x (mfpgpr) moves with other moves by using conditional
17094         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17095         (mov<mode> for SFmode/SDmode); Likewise.
17096         (SFmode/SDmode splitters): Likewise.
17097         (movsf_hardfloat): Likewise.
17098         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17099         (movsf_softfloat): Likewise.
17100         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17102         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17103         wx and wz constraints.
17105         * config/rs6000/constraints.md (wg constraint): New constraint to
17106         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17108         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17109         constraint.
17111         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17112         -mdebug=reg, print wg, wl, wx, and wz constraints.
17113         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17114         Initialize the reload functions for 64-bit binary/decimal floating
17115         point types.
17116         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17117         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17118         create the buffer on the stack to overcome not having a 32-bit
17119         load and store.
17120         (rs6000_emit_move): Likewise.
17121         (rs6000_secondary_memory_needed_rtx): Likewise.
17122         (rs6000_alloc_sdmode_stack_slot): Likewise.
17123         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17124         via xxlxor, just like DFmode 0.0.
17126         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17127         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17128         with other moves by using conditional constraits (wg).  Use LFIWZX
17129         and STFIWX for loading SDmode on power7.
17130         (movdd splitters): Likewise.
17131         (movdd_hardfloat32): Likewise.
17132         (movdd_softfloat32): Likewise.
17133         (movdd_hardfloat64_mfpgpr): Likewise.
17134         (movdd_hardfloat64): Likewise.
17135         (movdd_softfloat64): Likewise.
17137         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17138         64-bit binary and decimal floating point moves.
17139         (FMOVE64X): Likewise.
17140         (movdf): Combine 64-bit binary and decimal floating point moves.
17141         Combine power6x (mfpgpr) moves with other moves by using
17142         conditional constraits (wg).
17143         (mov<mode> for DFmode/DDmode): Likewise.
17144         (DFmode/DDmode splitters): Likewise.
17145         (movdf_hardfloat32): Likewise.
17146         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17147         (movdf_softfloat32): Likewise.
17148         (movdf_hardfloat64_mfpgpr): Likewise.
17149         (movdf_hardfloat64): Likewise.
17150         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17151         (movdf_softfloat64): Likewise.
17152         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17153         (reload_<mode>_load): Move to later in the file so they aren't in
17154         the middle of the floating point move insns.
17155         (reload_<mode>_store): Likewise.
17157         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17158         constraint.
17160         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17161         constraint if -mdebug=reg.
17162         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17163         Enable using dd reload support if needed.
17165         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17166         binary and decimal floating point moves in rs6000.md.
17167         (movtd_internal): Likewise.
17169         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17170         decimal floating point moves.
17171         (movtf): Likewise.
17172         (movtf_internal): Likewise.
17173         (mov<mode>_internal, TDmode/TFmode): Likewise.
17174         (movtf_softfloat): Likewise.
17175         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17177         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17178         movdi_internal64, using wg constraint for move direct operations.
17179         (movdi_internal64): Likewise.
17181         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17182         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17183         the various virtual registers. Use GPR/FPR first/last values,
17184         instead of hard coding the register numbers.  Print which modes
17185         have reload functions registered.
17186         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17187         settings before/after setting cpu, target and subtarget settings.
17188         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17189         and for secondary reload failures in rs6000_secondary_reload_inner.
17190         (rs6000_secondary_reload_fail): Likewise.
17191         (rs6000_secondary_reload_inner): Likewise.
17193         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17194         macros for first/last GPR and FPR registers.
17195         (LAST_GPR_REGNO): Likewise.
17196         (FIRST_FPR_REGNO): Likewise.
17197         (LAST_FPR_REGNO): Likewise.
17199         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17200         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17201         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17202         (vcond<mode><mode>): Likewise.
17203         (vcondu<mode><mode>): Likewise.
17204         (vector_gtu<mode>): Likewise.
17205         (vector_gte<mode>): Likewise.
17206         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17207         to prevent the compiler from converting DImode operations to TImode.
17208         (ior<mode>3): Likewise.
17209         (and<mode>3): Likewise.
17210         (one_cmpl<mode>2): Likewise.
17211         (nor<mode>3): Likewise.
17212         (andc<mode>3): Likewise.
17214         * config/rs6000/constraints.md (wt constraint): New constraint
17215         that returns VSX_REGS if TImode is allowed in VSX registers.
17217         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17218         constant under VSX.
17220         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17221         similar to TImode, but it is restricted to being in the GPRs.
17223         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17224         TImode to occupy a single VSX register.
17226         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17227         -mvsx-timode for power7/power8.
17228         (power7 cpu): Likewise.
17229         (power8 cpu): Likewise.
17231         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17232         sure that TFmode/TDmode take up two registers if they are ever
17233         allowed in the upper VSX registers.
17234         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17235         registers.
17236         (rs6000_init_hard_regno_mode_ok): Likewise.
17237         (rs6000_debug_reg_global): Add debugging for PTImode and wt
17238         constraint.  Print if LRA is turned on.
17239         (rs6000_option_override_internal): Give an error if -mvsx-timode
17240         and VSX is not enabled.
17241         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
17242         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17243         to reg+offset addressing.  Use PTImode when checking offset
17244         addresses for validity.
17245         (reg_offset_addressing_ok_p): Likewise.
17246         (rs6000_legitimate_offset_address_p): Likewise.
17247         (rs6000_legitimize_address): Likewise.
17248         (rs6000_legitimize_reload_address): Likewise.
17249         (rs6000_legitimate_address_p): Likewise.
17250         (rs6000_eliminate_indexed_memrefs): Likewise.
17251         (rs6000_emit_move): Likewise.
17252         (rs6000_secondary_reload): Likewise.
17253         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
17254         reloads to fpr registers to continue to use reg+offset addressing,
17255         but 64-bit reloads to altivec registers need reg+reg addressing.
17256         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17257         it.  Treat LO_SUM like a PLUS operation.
17258         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17259         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17260         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17261         registers to share a register with a smaller sized type, since VSX
17262         puts scalars in the upper 64-bits.
17263         (print_operand): Add support for PTImode.
17264         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17265         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17266         registers, but don't have arithmetic support.
17267         (rs6000_memory_move_cost): Add test for VSX.
17268         (rs6000_opt_masks): Add -mvsx-timode.
17270         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17271         for TImode.
17272         (VSs): Likewise.
17273         (VSr): Use wt constraint for TImode.
17274         (VSv): Drop TImode support.
17275         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17276         (vsx_movti_64bit): Likewise.
17277         (vsx_movti_32bit): Likewise.
17278         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17279         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17280         one '?' on the appropriate output constraint.  Do not allow TImode
17281         logical operations on 32-bit systems.
17282         (vsx_ior<mode>3): Likewise.
17283         (vsx_xor<mode>3): Likewise.
17284         (vsx_one_cmpl<mode>2): Likewise.
17285         (vsx_nor<mode>3): Likewise.
17286         (vsx_andc<mode>3): Likewise.
17287         (vsx_concat_<mode>): Likewise.
17288         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17290         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17291         OPTION_MASK_VSX_TIMODE.
17292         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17293         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17295         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17296         (TI2 iterator): New iterator for TImode, PTImode.
17297         (wd mode attribute): Add values for vector types.
17298         (movti_string): Replace TI move operations with operations for TImode
17299         and PTImode.  Add support for TImode being allowed in VSX registers.
17300         (mov<mode>_string, TImode/PTImode): Likewise.
17301         (movti_ppc64): Likewise.
17302         (mov<mode>_ppc64, TImode/PTImode): Likewise.
17303         (TI mode splitters): Likewise.
17305         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17306         constraint.
17308 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
17310         PR tree-optimization/56355
17311         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17312         Also handle integers with undefined overflow.
17314 2013-03-20  Catherine Moore  <clm@codesourcery.com>
17315             Maciej W. Rozycki  <macro@codesourcery.com>
17316             Tom de Vries  <tom@codesourcery.com>
17317             Nathan Sidwell  <nathan@codesourcery.com>
17318             Iain Sandoe  <iain@codesourcery.com>
17319             Nathan Froyd  <froydnj@codesourcery.com>
17320             Chao-ying Fu  <fu@mips.com>
17322         * doc/extend.texi: (micromips, nomicromips, nocompression):
17323         Document new function attributes.
17324         * doc/invoke.texi (minterlink-compressed, mmicromips,
17325         m14k, m14ke, m14kec): Document new options.
17326         (minterlink-mips16): Update documentation.
17327         * doc/md.texi (ZC, ZD): Document new constraints.
17328         * configure.ac (gcc_cv_as_micromips): Check if linker
17329         supports the .set micromips directive.
17330         * configure: Regenerate.
17331         * config.in: Regenerate.
17332         * config/mips/mips-tables.opt: Regenerate.
17333         * config/mips/micromips.md: New file.
17334         * constraints.md (ZC, ZD): New constraints.
17335         * config/mips/predicates.md (movep_src_register): New predicate.
17336         (movep_src_operand): New predicate.
17337         (non_volatile_mem_operand): New predicate.
17338         * config/mips/mips.md (multimem): New type.
17339         (length): Differentiate between 17-bit and 18-bit branch offsets.
17340         (MOVEP1, MOVEP2): New mode iterator.
17341         (mov_<load>l): Use ZC constraint.
17342         (mov_<load>r): Likewise.
17343         (mov_<store>l): Likewise.
17344         (mov_<store>r): Likewise.
17345         (*branch_equality<mode>_inverted): Add microMIPS support.
17346         (*branch_equality<mode>): Likewise.
17347         (*jump_absolute): Likewise.
17348         (indirect_jump_<mode>): Likewise.
17349         (tablejump_<mode>): Likewise.
17350         (<optab>_internal): Likewise.
17351         (sibcall_internal): Likewise.
17352         (sibcall_value_internal): Likewise.
17353         (prefetch): Use constraint ZD.
17354         * config/mips/mips.opt (minterlink-compressed): New option.
17355         (minterlink-mips16): Now an alias for minterlink-compressed.
17356         (mmicromips): New option.
17357         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17358         (compare_and_swap_12): Likewise.
17359         (sync_add<mode>): Likewise.
17360         (sync_<optab>_12): Likewise.
17361         (sync_old_<optab>_12): Likewise.
17362         (sync_new_<optab>_12): Likewise.
17363         (sync_nand_12): Likewise.
17364         (sync_old_nand_12): Likewise.
17365         (sync_new_nand_12): Likewise.
17366         (sync_sub<mode>): Likewise.
17367         (sync_old_add<mode>): Likewise.
17368         (sync_old_sub<mode>): Likewise.
17369         (sync_new_add<mode>): Likewise.
17370         (sync_new_sub<mode>): Likewise.
17371         (sync_<optab><mode>): Likewise.
17372         (sync_old_<optab><mode>): Likewise.
17373         (sync_new_<optab><mode>): Likewise.
17374         (sync_nand<mode>): Likewise.
17375         (sync_old_nand<mode>): Likewise.
17376         (sync_new_nand<mode>): Likewise.
17377         (sync_lock_test_and_set<mode>): Likewise.
17378         (test_and_set_12): Likewise.
17379         (atomic_compare_and_swap<mode>): Likewise.
17380         (atomic_exchange<mode>_llsc): Likewise.
17381         (atomic_fetch_add<mode>_llsc): Likewise.
17382         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17383         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17384         (umips_save_restore_pattern_p): Likewise.
17385         (umips_load_store_pair_p): Likewise.
17386         (umips_output_load_store_pair): Likewise.
17387         (umips_movep_target_p): Likewise.
17388         (umips_12bit_offset_address_p): Likewise.
17389         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17390         (mips_base_mips16): Rename this...
17391         (mips_base_compression_flags): ...to this. Update all uses.
17392         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17393         (mips_mips16_decl_p): Delete.
17394         (mips_nomips16_decl_p): Delete.
17395         (mips_get_compress_on_flags): New function.
17396         (mips_get_compress_off_flags): New function.
17397         (mips_get_compress_mode): New function.
17398         (mips_get_compress_on_name): New function.
17399         (mips_get_compress_off_name): New function.
17400         (mips_insert_attributes): Support multiple compression types.
17401         (mips_merge_decl_attributes): Likewise.
17402         (umips_12bit_offset_address_p): New function.
17403         (mips_start_function_definition): Emit .set micromips directive.
17404         (mips_call_may_need_jalx_p): New function.
17405         (mips_function_ok_for_sibcall): Add microMIPS support.
17406         (mips_print_operand_punctuation): Support short delay slots and
17407         compact jumps.
17408         (umips_swm_mask, umips_swm_encoding): New.
17409         (umips_build_save_restore): New function.
17410         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17411         (was_mips16_p): Remove.
17412         (old_compression_mode): New.
17413         (mips_set_compression_mode): New function.
17414         (mips_set_current_function): Add microMIPS support.
17415         (mips_option_override): Likewise.
17416         (umips_save_restore_pattern_p): New function.
17417         (umips_output_save_restore): New function.
17418         (umips_load_store_pair_p_1): New function.
17419         (umips_load_store_pair_p): New function.
17420         (umips_output_load_store_pair_1): New function.
17421         (umips_output_load_store_pair): New function.
17422         (umips_movep_target_p) New function.
17423         (mips_prepare_pch_save): Add microMIPS support.
17424         * config/mips/mips.h (TARGET_COMPRESSION): New.
17425         (TARGET_CPU_CPP_BUILTINS): Update macro
17426         to use new compression flags and to support microMIPS.
17427         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17428         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17429         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17430         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17431         (ASM_SPEC): Support mmicromips and mno-micromips.
17432         (M16STORE_REG_P): New macro.
17433         (MIPS_CALL): Support TARGET_MICROMIPS.
17434         (MICROMIPS_J): New macro.
17435         (mips_base_mips16): Rename this...
17436         (mips_base_compression_flags): ...to this.
17437         (UMIPS_12BIT_OFFSET_P): New macro.
17438         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17439         (MULTILIB_DIRNAMES): Likewise.
17440 2013-03-20  Richard Biener  <rguenther@suse.de>
17442         PR tree-optimization/56661
17443         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17444         the result does not have to be distinct.
17446 2013-03-20  Richard Biener  <rguenther@suse.de>
17448         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17449         remap_gimple_op_r.
17451 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17452             Steven Bosscher  <steven@gcc.gnu.org>
17454         PR rtl-optimization/56605
17455         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17457 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17459         PR bootstrap/56656
17460         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17461         that require movd instead of movq.
17463 2013-03-20  Richard Biener  <rguenther@suse.de>
17465         * tree-ssa-structalias.c (struct variable_info): Add pointer
17466         to the first field of an aggregate with sub-vars.  Make
17467         this and the pointer to the next subfield its ID.
17468         (vi_next): New function.
17469         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17470         storedanything_id, integer_id): Increment by one.
17471         (new_var_info, get_call_vi, lookup_call_clobber_vi,
17472         get_call_clobber_vi): Adjust.
17473         (solution_set_expand): Simplify and speedup.
17474         (solution_set_add): Inline into ...
17475         (set_union_with_increment): ... this.  Adjust accordingly.
17476         (do_sd_constraint): Likewise.
17477         (do_ds_constraint): Likewise.
17478         (do_complex_constraint): Simplify.
17479         (build_pred_graph): Adjust.
17480         (solve_graph): Likewise.  Simplify and speedup.
17481         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17482         get_constraint_for_component_ref, get_constraint_for_1,
17483         first_vi_for_offset, first_or_preceding_vi_for_offset,
17484         create_function_info_for, create_variable_info_for_1,
17485         create_variable_info_for, intra_create_variable_infos): Adjust.
17486         (init_base_vars): Push NULL for ID zero.
17487         (compute_points_to_sets): Adjust.
17489 2013-03-20  Richard Biener  <rguenther@suse.de>
17491         * cfgloop.c (verify_loop_structure): Streamline and avoid
17492         ICEing on corrupt loop tree.
17493         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17494         loop tree.
17496 2013-03-20  Richard Biener  <rguenther@suse.de>
17498         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17499         check whether an SSA update is needed.
17501 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
17503         * config/mips/constraints.md (T): Rename to...
17504         (Yf): ...this.
17505         (U): Rename to...
17506         (Yd): ...this.
17507         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17508         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17510 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17512         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17513         (*subsi3_carryin_uxtw): Likewise.
17515 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17517         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17518         (*rorsi3_insn_uxtw): Likewise.
17520 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17522         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17523         (*extrsi5_insn_uxtw): Likewise.
17525 2013-03-19  Richard Biener  <rguenther@suse.de>
17527         PR tree-optimization/56273
17528         * passes.c (init_optimization_passes): Move second VRP after DOM.
17530 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
17532         * config/i386/i386.md (*movti_internal): Merge from
17533         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
17534         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
17535         nox64 isa attributes.
17537 2013-03-18  Richard Biener  <rguenther@suse.de>
17539         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17540         (unite): Likewise.
17541         (merge_node_constraints): Likewise.
17542         (build_succ_graph): Likewise.
17543         (valid_graph_edge): Inline into single caller.
17544         (unify_nodes): Likewise.  Use bitmap_set_bit return value
17545         and cache varinfo.
17546         (scc_visit): Fix formatting and variable use.
17547         (do_sd_constraint): Use gcc_checking_assert.
17548         (do_ds_constraint): Likewise.
17549         (do_complex_constraint): Likewise.
17550         (condense_visit): Likewise.  Cleanup.
17551         (dump_pred_graph): New function.
17552         (perform_var_substitution): Dump the pred-graph before
17553         variable substitution.
17554         (find_equivalent_node): Use gcc_checking_assert.
17555         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17557 2013-03-18  Richard Biener  <rguenther@suse.de>
17559         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17560         Remove cond_expr_stmt_list argument and do not gimplify the
17561         built expression.
17562         (vect_loop_versioning): Adjust.
17563         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17564         Cleanup to use less temporaries.
17565         (vect_create_data_ref_ptr): Cleanup.
17567 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17569         PR tree-optimization/56635
17570         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17571         require types_compatible_p types.
17573 2013-03-18  Nick Clifton  <nickc@redhat.com>
17575         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
17576         spurious backslash.
17578         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
17579         Add missing line to comment describing function.
17581 2013-03-18  Richard Biener  <rguenther@suse.de>
17583         PR tree-optimization/56210
17584         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17585         Handle string / character search functions.
17586         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
17588 2013-03-18  Richard Biener  <rguenther@suse.de>
17590         PR middle-end/56483
17591         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
17592         and implement properly.
17593         * gimple.h (gimple_cond_single_var_p): Remove.
17595 2013-03-18  Richard Biener  <rguenther@suse.de>
17597         * tree-data-ref.h (find_data_references_in_loop): Declare.
17598         * tree-data-ref.c (get_references_in_stmt): Use a stack
17599         vector pre-allocated in the callers.
17600         (find_data_references_in_stmt): Adjust.
17601         (graphite_find_data_references_in_stmt): Likewise.
17602         (create_rdg_vertices): Likewise.
17603         (find_data_references_in_loop): Export.
17604         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17605         Compute dependences here...
17606         (vect_analyze_data_refs): ...not here.  When we encounter
17607         a non-vectorizable data reference in basic-block vectorization
17608         truncate the data reference vector.  Do not bother to
17609         fixup data-dependence information for gather loads.
17610         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
17611         of data references, as reported.
17613 2013-03-18  Richard Biener  <rguenther@suse.de>
17615         PR tree-optimization/3713
17616         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
17617         has_constants and expr.
17618         (stmt_has_constants): Properly valueize SSA names when deciding
17619         whether the stmt has constants.
17621 2013-03-18  Richard Biener  <rguenther@suse.de>
17623         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
17624         whole function when there is nothing to do.
17625         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
17626         * tree-vectorizer.c (vectorize_loops): Update virtual and
17627         loop-closed SSA once.
17628         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
17630 2013-03-18  Richard Biener  <rguenther@suse.de>
17632         PR middle-end/56113
17633         * domwalk.c (bb_postorder): New global static.
17634         (cmp_bb_postorder): New function.
17635         (walk_dominator_tree): Replace scheme imposing an order for
17636         visiting dominator sons by one sorting them at the time they
17637         are pushed on the stack.
17639 2013-03-18  Richard Biener  <rguenther@suse.de>
17641         PR tree-optimization/39326
17642         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
17643         (struct mem_ref): Replace mem member with ao_ref typed member.
17644         (MEM_ANALYZABLE): Adjust.
17645         (memref_eq): Likewise.
17646         (mem_ref_alloc): Likewise.
17647         (gather_mem_refs_stmt): Likewise.
17648         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
17649         (execute_sm_if_changed_flag_set): Adjust.
17650         (execute_sm): Likewise.
17651         (ref_always_accessed_p): Likewise.
17652         (refs_independent_p): Likewise.
17653         (can_sm_ref_p): Likewise.
17655 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17657         PR c/56566
17658         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
17659         return 1 even for !unsignedp.
17661 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17663         * config/i386/i386.md (isa): Add x64 and nox64.
17664         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
17665         (*pushtf): Enable *roF alternative for x64 isa only.
17666         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
17667         mode attribute of integer alternatives to DImode for TARGET_64BIT.
17668         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
17669         (*movtf_internal): Merge from *movtf_internal_rex64 and
17670         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
17671         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
17672         nox64 isa attributes.
17673         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
17674         nox64 isa attributes.
17675         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
17677 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17679         * config/alpha/alpha.c (TARGET_LRA_P): New define.
17681 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
17683         PR target/56640
17684         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
17685         class names.  Remove trailing comma after "ALL_REGS".
17687 2013-03-16  Jan Hubicka  <jh@suse.cz>
17689         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
17690         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
17691         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
17692         of cgraph_get_create_node.
17693         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17695 2013-03-16  Jason Merrill  <jason@redhat.com>
17697         PR debug/49090
17698         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
17699         with DW_AT_default_value.
17701 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
17703         * BASE-VER: Set to 4.9.0.
17705 2013-03-14  Andi Kleen  <ak@linux.intel.com>
17707         PR target/56619
17708         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
17709         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
17710         Document _x* TSX intrinsics.
17712 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
17713             David Holsgrove  <david.holsgrove@xilinx.com>
17715         * configure.ac: Add MicroBlaze TLS support detection.
17716         * configure: Regenerate.
17717         * config/microblaze/microblaze-protos.h
17718         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
17719         symbol_mentioned_p, label_mentioned_p): Add prototypes.
17720         * config/microblaze/microblaze.c (microblaze_address_type): Add
17721         ADDRESS_TLS and tls_reloc address types.
17722         (microblaze_address_info): Add tls_reloc.
17723         (TARGET_HAVE_TLS): Define.
17724         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
17725          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
17726          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
17727         load_tls_operand,  microblaze_call_tls_get_addr,
17728         microblaze_legitimize_tls_address): New functions.
17729         (microblaze_classify_unspec): Handle UNSPEC_TLS.
17730         (get_base_reg): Use microblaze_tls_symbol_p.
17731         (microblaze_classify_address): Handle TLS.
17732         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
17733         label_mentioned_p and microblaze_tls_referenced_p.
17734         (microblaze_legitimize_address): Handle TLS.
17735         (microblaze_address_insns): Handle ADDRESS_TLS.
17736         (pic_address_needs_scratch): Handle TLS.
17737         (print_operand_address): Handle TLS.
17738         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
17739         (microblaze_expand_move): Handle TLS.
17740         (microblaze_legitimate_constant_p): Check
17741         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
17742         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
17743         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
17744         (PIC_OFFSET_TABLE_REGNUM): Set.
17745         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
17746         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
17747         (addsi3, movsi_internal2, movdf_internal): Update constraints
17748         * config/microblaze/predicates.md (arith_plus_operand): Define
17749         (move_operand): Redefine as move_src_operand,
17750         check microblaze_tls_referenced_p.
17752 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17754         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
17755         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
17757 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17759         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
17760         CC mode for AND.
17762 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
17764         PR tree-optimization/53265
17765         * common.opt (Waggressive-loop-optimizations): New option.
17766         * tree-ssa-loop-niter.c: Include tree-pass.h.
17767         (do_warn_aggressive_loop_optimizations): New function.
17768         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
17769         if number_of_latch_executions returned constant.
17770         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
17771         early.  If number_of_latch_executions returned constant, set
17772         nb_iterations_upper_bound back to it.
17773         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
17774         field.
17775         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
17776         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
17778         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
17779         (MULTILIB_OSDIRNAMES): Set.
17780         * genmultilib: If defaultosdirname doesn't start with :: , set
17781         defaultosdirname2 instead, clear it and emit two . multilib_raw
17782         entries instead of just one.
17784 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
17786         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17787         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
17788         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17789         (SUBTARGET_OVERRIDE_OPTIONS): New.
17791 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
17793         PR target/49880
17794         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
17795         (musermode): Convert to Var(TARGET_USERMODE).
17796         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
17797         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
17798         * config/sh/sh.c (sh_option_override): Use
17799         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
17800         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
17801         condition.
17802         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
17803         TARGET_SH4.
17804         (udivsi3_i4_single, divsi3_i4_single): Use
17805         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
17807 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
17809         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
17810         default setting.
17812 2013-03-13  Richard Biener  <rguenther@suse.de>
17814         PR tree-optimization/56608
17815         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
17816         calls when vectorizing basic-blocks.
17818 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
17820         PR plugins/45078
17821         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
17822         tm_file.
17824 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
17826         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
17828 2013-03-11  Jan Hubicka  <jh@suse.cz>
17830         PR lto/56557
17831         * lto-streamer-out.c (output_symbol_p): Skip references from
17832         constructors of external variables.
17834 2013-03-11  Jan Hubicka  <jh@suse.cz>
17836         PR middle-end/56571
17837         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
17838         from pseudos.
17839         * emit-rtl.c (verify_rtx_sharing): Likewise.
17840         (copy_insn_1): Likewise.
17841         * rtl.c (copy_rtx): Likewise.
17843 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
17845         PR target/56591
17846         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
17847         output_operand_lossage message.
17849 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
17851         PR target/56470
17852         * arm.c (shift_op): Validate RTL pattern on the fly.
17853         (arm_print_operand, case 'S'): Don't use shift_operator to validate
17854         the RTL.
17856 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17858         PR target/56347
17859         * config/pa/pa.md (call_value): Check for calls to powf and direct to
17860         new call patterns that clobber %fr12.
17861         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
17862         split and postreload patterns.
17863         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
17864         registers %fr12 and %fr12R as call used.
17866 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
17868         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
17869         (canon_address, record_store, replace_read, check_mem_read_rtx,
17870         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
17871         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
17872         rest_of_handle_dse): Likewise.
17874 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
17876         PR middle-end/56524
17877         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
17878         Add base_optabs.
17879         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
17880         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
17881         (save_optabs_if_changed): Replace with...
17882         (init_tree_optimization_optabs): ...this.
17883         * optabs.c (save_optabs_if_changed): Rename to...
17884         (init_tree_optimization_optabs): ...this.  Take the optimization node
17885         as argument.  Do nothing if the base optabs are already correct.
17886         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
17887         to recompute optabs.
17888         * function.h (function): Remove optabs field.
17889         * function.c (invoke_set_current_function_hook): Call
17890         init_tree_optimization_optabs.  Use the result to initialize
17891         this_fn_optabs.
17893 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
17895         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
17896         if GTMA_HAS_NO_INSTRUMENTATION.
17897         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
17898         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
17899         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
17900         * gimple-pretty-print.c (dump_gimple_transaction): Handle
17901         GTMA_HAS_NO_INSTRUMENTATION.
17903 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
17905         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
17906         libasan_preinit.o.
17908 2013-03-08  Marek Polacek  <polacek@redhat.com>
17909             Jakub Jelinek  <jakub@redhat.com>
17911         PR tree-optimization/56478
17912         * predict.c (is_comparison_with_loop_invariant_p): Change the
17913         type of loop_step to tree.
17914         (predict_loops): Adjust.
17915         (predict_iv_comparison): Perform the computations on double_ints.
17917 2013-03-08  Richard Biener  <rguenther@suse.de>
17919         PR tree-optimization/56570
17920         * tree-cfg.c (verify_expr_location_1): Verify locations for
17921         DECL_DEBUG_EXPR.
17922         * tree-sra.c (create_access_replacement): Strip locations
17923         from DECL_DEBUG_EXPRs.
17925 2013-03-08  Richard Biener  <rguenther@suse.de>
17927         * tree-inline.c (expand_call_inline): Do not associate
17928         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
17929         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
17931 2013-03-08  Richard Biener  <rguenther@suse.de>
17933         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
17934         or block changes with -Og.  Fix for location / block encoding
17935         changes and PHI arguments with locations.
17937 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
17939         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
17940         for all counters.
17941         (struct output_info): Likewise.
17942         (register_overhead): Remove bad gcc_assert.
17943         (bitmap_find_bit): If there is only a single bitmap element, do not
17944         count a miss as a search.
17945         (print_statistics): Update for counter type changes.
17946         (dump_bitmap_statistics): Likewise.  Print headers such that they
17947         are properly lined up with the printed counters.
17949 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
17951         PR tree-optimization/56559
17952         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
17953         check that it has only a single use.
17955 2013-03-07  Richard Biener  <rguenther@suse.de>
17957         * doc/invoke.texi (fwhole-program): Discourage use in combination
17958         with -flto.
17960 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
17962         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
17964         PR tree-optimization/56539
17965         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
17966         instead of GSI_CONTINUE_LINKING as last argument to
17967         force_gimple_operand_gsi.  Adjust function comment.
17969         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
17970         aarch64-cores.def.
17972         PR middle-end/56548
17973         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
17974         promoted mode, convert the result back to the original mode.
17976 2013-03-06  Richard Biener  <rguenther@suse.de>
17978         PR middle-end/56294
17979         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
17980         (insert_updated_phi_nodes_compare_uids): New function.
17981         (update_ssa): Sort symbols_to_rename after UID before
17982         traversing it to insert PHI nodes.
17984 2013-03-06  Richard Biener  <rguenther@suse.de>
17986         PR middle-end/50494
17987         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
17988         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
17990         Revert
17991         2013-02-13  Richard Biener  <rguenther@suse.de>
17993         PR lto/50494
17994         * varasm.c (output_constant_def_1): Get the decl representing
17995         the constant as argument.
17996         (output_constant_def): Wrap output_constant_def_1.
17997         (make_decl_rtl): Use output_constant_def_1 with the decl
17998         representing the constant.
17999         (build_constant_desc): Optionally re-use a decl already
18000         representing the constant.
18001         (tree_output_constant_def): Adjust.
18003 2013-03-06  Joey Ye  <joey.ye@arm.com>
18005         PR lto/50293
18006         * gcc.c (convert_white_space): New function.
18007         (main): Handles white space in function name.
18009 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
18011         PR target/56529
18012         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18013         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
18014         to SH_DIV_CALL_TABLE for TARGET_SH2.
18015         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18016         list.
18017         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18018         call-table options.
18020 2013-03-05  Sterling Augustine  <saugustine@google.com>
18021             Cary Coutant  <ccoutant@google.com>
18023         PR debug/55364
18024         * dwarf2out.c (resolve_addr): Don't call
18025         remove_loc_list_addr_table_entries a second time for the same
18026         expression.
18028 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18030         PR debug/56510
18031         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18032         (avoid_complex_debug_insns): New function.
18033         (expand_debug_locations): Call it.
18035         PR rtl-optimization/56484
18036         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18037         lifetimes of hard registers on small register class machines.
18039 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18041         * config/microblaze/microblaze-protos.h: Rename
18042         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18043         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18044         fast_interrupt.
18045         (microblaze_fast_interrupt_function_p): New function.
18046         (microblaze_is_interrupt_handler): Rename to
18047         microblaze_is_interrupt_variant and add fast_interrupt check.
18048         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18049         (save_restore_insns): Likewise.
18050         (compute_frame_size): Likewise.
18051         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18052         (microblaze_globalize_label): Likewise.
18053         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18054         * config/microblaze/microblaze.md: Use wrapper
18055         microblaze_is_interrupt_variant.
18057 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18059         * sdbout.c (sdbout_one_type): Switch to current function's section
18060         supporting cold/hot.
18062 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18064         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18065         -mxl-reorder.
18067 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18069         PR middle-end/56461
18070         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18071         if VALGRIND_GET_VBITS is defined, temporarily make object
18072         memory all defined, and restore previous valgrind addressability
18073         and definability afterwards.  Free this_object at the end.
18075         PR middle-end/56461
18076         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18077         right before calling lra_create_live_ranges, also call it
18078         when clearing live_p.  Only call lra_clear_live_ranges
18079         at the end if live_p.
18081         PR middle-end/56461
18082         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18084 2013-03-05  Richard Biener  <rguenther@suse.de>
18086         PR tree-optimization/56521
18087         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18088         value-id.
18090 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18092         PR c++/55135
18093         * except.h (remove_unreachable_eh_regions): New prototype.
18094         * except.c (remove_eh_handler_splicer): New function, split out
18095         of remove_eh_handler.
18096         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18097         warning about running it on many EH regions one at a time.
18098         (remove_unreachable_eh_regions_worker): New function, walk the
18099         EH tree in depth-first order and remove non-marked regions.
18100         (remove_unreachable_eh_regions): New function.
18101         * tree-eh.c (mark_reachable_handlers): New function, split out
18102         from remove_unreachable_handlers.
18103         (remove_unreachable_handlers): Use mark_reachable_handlers and
18104         remove_unreachable_eh_regions.
18105         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18106         and remove_unreachable_eh_regions.
18108 2013-03-05  Richard Biener  <rguenther@suse.de>
18110         PR middle-end/56525
18111         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18112         not freeing them until the end.
18114 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18116         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18118 2013-03-05  Richard Biener  <rguenther@suse.de>
18120         PR tree-optimization/56270
18121         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18122         of loads after scheduling an SLP instance.
18124 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18126         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18127         tic6x.exp.
18128         (check_gcc_parallelize): Run guality.exp as a separate job from
18129         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18130         struct-layout-1.exp with stackalign.exp.
18132         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18134         PR middle-end/56461
18135         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18136         load_index sbitmap even if some bit in it isn't set.
18138         PR middle-end/56461
18139         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18140         (discover_iteration_bound_by_body_walk): Change queues to
18141         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18142         spelling in comment.  Call safe_push on queues[bound_index] directly.
18143         Release queues[queue_index] in every iteration unconditionally.
18144         Release bounds vector.
18146         PR middle-end/56461
18147         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18148         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18149         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18150         inner_phis vector.
18152 2013-03-05  Richard Biener  <rguenther@suse.de>
18154         PR lto/56515
18155         * tree-inline.c (remap_blocks_to_null): New function.
18156         (expand_call_inline): When expanding a call stmt without
18157         an associated block inline remap all callee blocks to NULL.
18159 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18161         PR rtl-optimization/56494
18162         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18163         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18164         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18166         PR middle-end/56461
18167         * sel-sched-ir.c (free_sched_pools): Release
18168         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18169         if succs_info_pool.max_top isn't -1.
18171         PR bootstrap/56509
18172         * opts.c (opts_obstack, opts_concat): Moved to...
18173         * opts-common.c (opts_obstack, opts_concat): ... here.
18175 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18177         PR middle-end/56461
18178         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18180 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18182         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18183         all appropriate places.
18185 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18187         PR tree-optimization/56424
18188         * ipa-split.c (split_function): Do not set the RSO flag if result is
18189         not by reference and its type is a register type.
18191 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18193         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18194         (microblaze_legitimate_pic_operand): Likewise
18195         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18196         new function microblaze_legitimate_pic_operand
18197         * config/microblaze/microblaze-protos.h
18198         (microblaze_legitimate_pic_operand): Declare.
18200 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18202         * config/microblaze/predicates.md (call_insn_simple_operand):
18203         New predicate for supported rtx code types.
18204         * config/microblaze/microblaze.md (call_internal1): Use
18205         call_insn_simple_operand predicate.
18207 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18209         PR middle-end/56461
18210         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18211         partitions.ordered_remove.
18213         PR middle-end/56461
18214         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18215         vec_oprnds0.create (1) for modifier == NONE.
18217         PR middle-end/56461
18218         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18219         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18220         vec_oprnds1 right before pushing anything to it for
18221         scalar_shift_arg.
18223         PR middle-end/56461
18224         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18225         set nbbs to 0 instead of having separate code path.
18226         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18227         instead of false as last argument if returning NULL.
18229 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18231         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18232         the attribute is now called "target" instead of "option".
18233         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18234         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18235         attribute/pragma name for TARGET_OPTION_VALID_P and
18236         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18237         * doc/tm.texi: Regenerated.
18239 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
18241         * config/microblaze/microblaze.c:
18242         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18243         * config/microblaze/microblaze.h: Add -mxl-reorder to
18244         DRIVER_SELF_SPECS.
18245         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18246         instructions emitted if TARGET_REORDER.
18247         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18248         or 0 for -m/-mno case, but initialises as 2 to detect default use case
18249         separately.
18251 2013-03-01  Xinliang David Li  <davidxl@google.com>
18253         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18254         walk length.
18256 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
18258         PR middle-end/56461
18259         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18260         vector even when returning true.  Fix up function comment formatting.
18262         PR middle-end/56461
18263         * ira-build.c (ira_loop_nodes_count): New variable.
18264         (create_loop_tree_nodes): Initialize it.
18265         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18267         PR middle-end/56461
18268         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18269         method on dr_chain and result_chain.
18270         * tree-vect-stmts.c (vectorizable_store): Only call
18271         result_chain.create if j == 0.
18273         PR middle-end/56461
18274         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18275         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18276         before overwriting it.
18278 2013-03-01  Tobias Burnus  <burnus@net-b.de>
18280         * doc/extended.texi (C Extensions): Change order in @menu
18281         to match @node.
18282         (Other MIPS Built-in Functions): Move last MIPS entry before
18283         "picoChip Built-in Functions".
18284         (SH Built-in Functions): Move after RX Built-in Functions.
18285         * doc/gcc.texi (Introduction): Change order in @menu
18286         to match @node.
18287         * doc/md.texi (Constraints): Ditto.
18288         * gty.texi (Type Information): Ditto.
18289         (User-provided marking routines for template types): Make
18290         subsection.
18291         * doc/invoke.texi (AArch64 Options): Move before
18292         "Adapteva Epiphany Options".
18294 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
18295             Jakub Jelinek  <jakub@redhat.com>
18297         PR sanitizer/56454
18298         * asan.c (gate_asan): Lookup no_sanitize_address instead of
18299         no_address_safety_analysis attribute.
18300         * doc/extend.texi (no_address_safety_attribute): Rename to
18301         no_sanitize_address attribute, mention no_address_safety_analysis
18302         attribute as deprecated alias.
18304 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
18306         PR middle-end/56461
18307         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18308         type to vec<vec<tree> > *.
18309         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
18310         to be vec<tree> instead of vec<tree> *, set vec_defs
18311         to vNULL and call vec_defs.create (number_of_vects), adjust other
18312         uses of vec_defs.
18313         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18314         vectorizable_condition): Adjust vect_get_slp_defs callers.
18316 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18318         * config/aarch64/aarch64.c
18319         (aarch64_float_const_representable): Remove unused variable.
18321 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18323         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18325 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18327         * config/aarch64/aarch64-builtins.c
18328         (aarch64_init_simd_builtins): Make static.
18330 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18332         * config/aarch64/aarch64.c
18333         (aarch64_simd_make_constant): Make static.
18335 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18337         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18338         with no initialization to the RHS of debug statements.
18340 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18342         PR tree-optimization/56294
18343         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18344         Adjust dumping.
18345         (get_access_replacement): Do not call create_access_replacement.
18346         Assert a replacement exists.
18347         (get_repl_default_def_ssa_name): Create the replacement declaration
18348         itself.
18350 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18352         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18353         final_end_function.
18355 2013-02-28  Marek Polacek  <polacek@redhat.com>
18357         PR rtl-optimization/56466
18358         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18359         if we're changing a loop.
18360         (peel_loops_completely): Likewise.
18362 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
18364         PR c++/55813
18365         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18367 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18369         PR target/56445
18370         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18371         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18372         INTX_FTYPE_FX, FX_FTYPE_INTX.
18373         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18375 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18377         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18378         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18379         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18380         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18381         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18382         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18383         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18384         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18385         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18386         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18387         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18388         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18389         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18390         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18391         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18392         (avrxmega6): Increase max flash segments from 5 to 6.
18393         * config/avr/t-multilib: Regenerate.
18394         * config/avr/avr-tables.opt: Regenerate.
18395         * doc/avr-mmcu.texi: Regenerate.
18397 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18399         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18400         (avr_device_to_arch): Rename to avr_device_to_ld.
18401         (avr_device_to_as): New prototype.
18402         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18403         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18404         * config/avr/driver-avr.c (avr_device_to_as): New.
18405         (avr_device_to_arch): Rename to avr_device_to_ld.
18407 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18409         PR middle-end/56461
18410         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18411         method on dr_chain and result_chain.
18413         PR middle-end/56461
18414         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18415         pointer_set_destroy on not_executed_last_iteration.
18417         PR middle-end/56461
18418         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18420         PR middle-end/56461
18421         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18422         FOR_EACH_DEFINED_FUNCTION when freeing state.
18424         PR middle-end/56461
18425         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18426         pool_free.
18427         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18428         overwriting it.
18430         PR middle-end/56461
18431         * ipa-cp.c (decide_whether_version_node): Call vec_free on
18432         known_aggs[i].items and release known_aggs vector.
18434         PR middle-end/56461
18435         * ipa-reference.c (propagate): Free node_info even for alias nodes.
18437 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18439         * config/microblaze/microblaze.c (microblaze_emit_compare):
18440         Use xor for EQ/NE comparisions.
18441         * config/microblaze/microblaze.md (cstoresf4): Add constraints
18442         (cbranchsf4): Adjust operator to comparison_operator.
18444 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18446         PR middle-end/56461
18447         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18448         vector.
18449         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18450         vec_safe_push, always update *slot.
18451         (redirect_edge_var_map_clear): Use vec_free.
18452         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18453         (free_var_map_entry): Use vec_free.
18454         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18455         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18457 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
18459         PR middle-end/45472
18460         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18461         when the may_trap_p bit of the exprs being merged differs.
18462         Reorder tests for speculativeness in the logical and operator.
18464 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18466         * incpath.c (add_standard_paths): Use reconcat instead of concat
18467         where appropriate and avoid leaking memory.
18469         * opts.h: Include obstack.h.
18470         (opts_concat): New prototype.
18471         (opts_obstack): New declaration.
18472         * opts.c (opts_concat): New function.
18473         (opts_obstack): New variable.
18474         (init_options_struct): Call gcc_init_obstack on opts_obstack.
18475         (finish_options): Use opts_concat instead of concat
18476         and XOBNEWVEC instead of XNEWVEC.
18477         * opts-common.c (generate_canonical_option, decode_cmdline_option,
18478         generate_option): Likewise.
18479         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18480         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18482         PR target/56455
18483         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18484         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18486 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18488         PR middle-end/56461
18489         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18491 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18493         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18494         (arm_block_move_unaligned_straight): Likewise.
18495         (arm_adjust_block_mem): Likewise.
18497 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18499         PR target/48901
18500         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18501         temp, cond and label.
18502         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18504         PR target/52500
18505         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18506         * config/c6x/c6x.h (dbx_register_map): Update declaration.
18508         PR target/52501
18509         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18510         of prologue/epilogue functions.
18512         PR target/52550
18513         * config/tilegx/tilegx.c (tilegx_expand_prologue):
18514         Remove unused variable cfa_offset.
18515         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18517         PR target/54639
18518         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18519         type promotion to unsigned.
18521         PR target/54640
18522         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18523         for HOST_WIDE_INT of 32 bit / same size as int.
18524         (arm_block_move_unaligned_straight): Likewise.
18525         (arm_adjust_block_mem): Likewise.
18527         PR target/54662
18528         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18529         ALL_CFLAGS.
18531 2013-02-26  Marek Polacek  <polacek@redhat.com>
18533         PR tree-optimization/56426
18534         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18536 2013-02-26  Richard Biener  <rguenther@suse.de>
18538         PR target/56444
18539         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18540         unused variable loops.
18542 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18544         PR tree-optimization/56448
18545         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18546         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18547         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18548         later operands of the references, or even first operand for
18549         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18551         PR tree-optimization/56443
18552         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18553         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18554         to type_for_mode langhook.
18556 2013-02-25  Matt Turner  <mattst88@gmail.com>
18558         * doc/invoke.texi: Document r4700.
18560 2013-02-25  Richard Biener  <rguenther@suse.de>
18562         PR tree-optimization/56175
18563         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18564         split out from ...
18565         (simplify_bitwise_binary): ... here.  Also guard the conversion
18566         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18568 2013-02-25  Catherine Moore  <clm@codesourcery.com>
18570         Revert:
18571         2013-02-24  Catherine Moore  <clm@codesourcery.com>
18572             Maciej W. Rozycki  <macro@codesourcery.com>
18573             Tom de Vries  <tom@codesourcery.com>
18574             Nathan Sidwell  <nathan@codesourcery.com>
18575             Iain Sandoe  <iain@codesourcery.com>
18576             Nathan Froyd  <froydnj@codesourcery.com>
18577             Chao-ying Fu  <fu@mips.com>
18579         * doc/extend.texi: (micromips, nomicromips, nocompression):
18580         Document new function attributes.
18581         * doc/invoke.texi (minterlink-compressed, mmicromips,
18582         m14k, m14ke, m14kec): Document new options.
18583         (minterlink-mips16): Update documentation.
18584         * doc/md.texi (ZC, ZD): Document new constraints.
18585         * configure.ac (gcc_cv_as_micromips): Check if linker
18586         supports the .set micromips directive.
18587         * configure: Regenerate.
18588         * config.in: Regenerate.
18589         * config/mips/mips-tables.opt: Regenerate.
18590         * config/mips/micromips.md: New file.
18591         * constraints.md (ZC, AD): New constraints.
18592         * config/mips/predicates.md (movep_src_register): New predicate.
18593         (movep_src_operand): New predicate.
18594         (non_volatile_mem_operand): New predicate.
18595         * config/mips/mips.md (multimem): New type.
18596         (length): Differentiate between 17-bit and 18-bit branch offsets.
18597         (MOVEP1, MOVEP2): New mode iterator.
18598         (mov_<load>l): Use ZC constraint.
18599         (mov_<load>r): Likewise.
18600         (mov_<store>l): Likewise.
18601         (mov_<store>r): Likewise.
18602         (*branch_equality<mode>_inverted): Add microMIPS support.
18603         (*branch_equality<mode>): Likewise.
18604         (*jump_absolute): Likewise.
18605         (indirect_jump_<mode>): Likewise.
18606         (tablejump_<mode>): Likewise.
18607         (<optab>_internal): Likewise.
18608         (sibcall_internal): Likewise.
18609         (sibcall_value_internal): Likewise.
18610         (prefetch): Use constraint ZD.
18611         * config/mips/mips.opt (minterlink-compressed): New option.
18612         (minterlink-mips16): Now an alias for minterlink-compressed.
18613         (mmicromips): New option.
18614         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18615         (compare_and_swap_12): Likewise.
18616         (sync_add<mode>): Likewise.
18617         (sync_<optab>_12): Likewise.
18618         (sync_old_<optab>_12): Likewise.
18619         (sync_new_<optab>_12): Likewise.
18620         (sync_nand_12): Likewise.
18621         (sync_old_nand_12): Likewise.
18622         (sync_new_nand_12): Likewise.
18623         (sync_sub<mode>): Likewise.
18624         (sync_old_add<mode>): Likewise.
18625         (sync_old_sub<mode>): Likewise.
18626         (sync_new_add<mode>): Likewise.
18627         (sync_new_sub<mode>): Likewise.
18628         (sync_<optab><mode>): Likewise.
18629         (sync_old_<optab><mode>): Likewise.
18630         (sync_new_<optab><mode>): Likewise.
18631         (sync_nand<mode>): Likewise.
18632         (sync_old_nand<mode>): Likewise.
18633         (sync_new_nand<mode>): Likewise.
18634         (sync_lock_test_and_set<mode>): Likewise.
18635         (test_and_set_12): Likewise.
18636         (atomic_compare_and_swap<mode>): Likewise.
18637         (atomic_exchange<mode>_llsc): Likewise.
18638         (atomic_fetch_add<mode>_llsc): Likewise.
18639         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18640         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18641         (umips_save_restore_pattern_p): Likewise.
18642         (umips_load_store_pair_p): Likewise.
18643         (umips_output_load_store_pair): Likewise.
18644         (umips_movep_target_p): Likewise.
18645         (umips_12bit_offset_address_p): Likewise.
18646         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18647         (mips_base_mips16): Rename this...
18648         (mips_base_compression_flags): ...to this. Update all uses.
18649         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18650         (mips_mips16_decl_p): Delete.
18651         (mips_nomips16_decl_p): Delete.
18652         (mips_get_compress_on_flags): New function.
18653         (mips_get_compress_off_flags): New function.
18654         (mips_get_compress_mode): New function.
18655         (mips_get_compress_on_name): New function.
18656         (mips_get_compress_off_name): New function.
18657         (mips_insert_attributes): Support multiple compression types.
18658         (mips_merge_decl_attributes): Likewise.
18659         (umips_12bit_offset_address_p): New function.
18660         (mips_start_function_definition): Emit .set micromips directive.
18661         (mips_call_may_need_jalx_p): New function.
18662         (mips_function_ok_for_sibcall): Add microMIPS support.
18663         (mips_print_operand_punctuation): Support short delay slots and
18664         compact jumps.
18665         (umips_swm_mask, umips_swm_encoding): New.
18666         (umips_build_save_restore): New function.
18667         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18668         (was_mips16_p): Remove.
18669         (old_compression_mode): New.
18670         (mips_set_compression_mode): New function.
18671         (mips_set_current_function): Add microMIPS support.
18672         (mips_option_override): Likewise.
18673         (umips_save_restore_pattern_p): New function.
18674         (umips_output_save_restore): New function.
18675         (umips_load_store_pair_p_1): New function.
18676         (umips_load_store_pair_p): New function.
18677         (umips_output_load_store_pair_1): New function.
18678         (umips_output_load_store_pair): New function.
18679         (umips_movep_target_p) New function.
18680         (mips_prepare_pch_save): Add microMIPS support.
18681         * config/mips/mips.h (TARGET_COMPRESSION): New.
18682         (TARGET_CPU_CPP_BUILTINS): Update macro
18683         to use new compression flags and to support microMIPS.
18684         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18685         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18686         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18687         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18688         (ASM_SPEC): Support mmicromips and mno-micromips.
18689         (M16STORE_REG_P): New macro.
18690         (MIPS_CALL): Support TARGET_MICROMIPS.
18691         (MICROMIPS_J): New macro.
18692         (mips_base_mips16): Rename this...
18693         (mips_base_compression_flags): ...to this.
18694         (UMIPS_12BIT_OFFSET_P): New macro.
18695         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18696         (MULTILIB_DIRNAMES): Likewise.
18698 2013-02-25  Tom de Vries  <tom@codesourcery.com>
18700         PR rtl-optimization/56131
18701         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
18702         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
18703         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
18705 2013-02-25  Tobias Burnus  <burnus@net-b.de>
18707         * doc/invoke.texi (-fsanitize=): Move from optimization
18708         to debugging options.
18710 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18712         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
18714 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18715             Alexander Monakov  <amonakov@ispras.ru>
18717         PR middle-end/56077
18718         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
18719         flush pending lists also on non-jumps.  Adjust comment.
18721 2013-02-24  Catherine Moore  <clm@codesourcery.com>
18722             Maciej W. Rozycki  <macro@codesourcery.com>
18723             Tom de Vries  <tom@codesourcery.com>
18724             Nathan Sidwell  <nathan@codesourcery.com>
18725             Iain Sandoe  <iain@codesourcery.com>
18726             Nathan Froyd  <froydnj@codesourcery.com>
18727             Chao-ying Fu  <fu@mips.com>
18729         * doc/extend.texi: (micromips, nomicromips, nocompression):
18730         Document new function attributes.
18731         * doc/invoke.texi (minterlink-compressed, mmicromips,
18732         m14k, m14ke, m14kec): Document new options.
18733         (minterlink-mips16): Update documentation.
18734         * doc/md.texi (ZC, ZD): Document new constraints.
18735         * configure.ac (gcc_cv_as_micromips): Check if linker
18736         supports the .set micromips directive.
18737         * configure: Regenerate.
18738         * config.in: Regenerate.
18739         * config/mips/mips-tables.opt: Regenerate.
18740         * config/mips/micromips.md: New file.
18741         * constraints.md (ZC, AD): New constraints.
18742         * config/mips/predicates.md (movep_src_register): New predicate.
18743         (movep_src_operand): New predicate.
18744         (non_volatile_mem_operand): New predicate.
18745         * config/mips/mips.md (multimem): New type.
18746         (length): Differentiate between 17-bit and 18-bit branch offsets.
18747         (MOVEP1, MOVEP2): New mode iterator.
18748         (mov_<load>l): Use ZC constraint.
18749         (mov_<load>r): Likewise.
18750         (mov_<store>l): Likewise.
18751         (mov_<store>r): Likewise.
18752         (*branch_equality<mode>_inverted): Add microMIPS support.
18753         (*branch_equality<mode>): Likewise.
18754         (*jump_absolute): Likewise.
18755         (indirect_jump_<mode>): Likewise.
18756         (tablejump_<mode>): Likewise.
18757         (<optab>_internal): Likewise.
18758         (sibcall_internal): Likewise.
18759         (sibcall_value_internal): Likewise.
18760         (prefetch): Use constraint ZD.
18761         * config/mips/mips.opt (minterlink-compressed): New option.
18762         (minterlink-mips16): Now an alias for minterlink-compressed.
18763         (mmicromips): New option.
18764         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18765         (compare_and_swap_12): Likewise.
18766         (sync_add<mode>): Likewise.
18767         (sync_<optab>_12): Likewise.
18768         (sync_old_<optab>_12): Likewise.
18769         (sync_new_<optab>_12): Likewise.
18770         (sync_nand_12): Likewise.
18771         (sync_old_nand_12): Likewise.
18772         (sync_new_nand_12): Likewise.
18773         (sync_sub<mode>): Likewise.
18774         (sync_old_add<mode>): Likewise.
18775         (sync_old_sub<mode>): Likewise.
18776         (sync_new_add<mode>): Likewise.
18777         (sync_new_sub<mode>): Likewise.
18778         (sync_<optab><mode>): Likewise.
18779         (sync_old_<optab><mode>): Likewise.
18780         (sync_new_<optab><mode>): Likewise.
18781         (sync_nand<mode>): Likewise.
18782         (sync_old_nand<mode>): Likewise.
18783         (sync_new_nand<mode>): Likewise.
18784         (sync_lock_test_and_set<mode>): Likewise.
18785         (test_and_set_12): Likewise.
18786         (atomic_compare_and_swap<mode>): Likewise.
18787         (atomic_exchange<mode>_llsc): Likewise.
18788         (atomic_fetch_add<mode>_llsc): Likewise.
18789         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18790         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18791         (umips_save_restore_pattern_p): Likewise.
18792         (umips_load_store_pair_p): Likewise.
18793         (umips_output_load_store_pair): Likewise.
18794         (umips_movep_target_p): Likewise.
18795         (umips_12bit_offset_address_p): Likewise.
18796         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18797         (mips_base_mips16): Rename this...
18798         (mips_base_compression_flags): ...to this. Update all uses.
18799         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18800         (mips_mips16_decl_p): Delete.
18801         (mips_nomips16_decl_p): Delete.
18802         (mips_get_compress_on_flags): New function.
18803         (mips_get_compress_off_flags): New function.
18804         (mips_get_compress_mode): New function.
18805         (mips_get_compress_on_name): New function.
18806         (mips_get_compress_off_name): New function.
18807         (mips_insert_attributes): Support multiple compression types.
18808         (mips_merge_decl_attributes): Likewise.
18809         (umips_12bit_offset_address_p): New function.
18810         (mips_start_function_definition): Emit .set micromips directive.
18811         (mips_call_may_need_jalx_p): New function.
18812         (mips_function_ok_for_sibcall): Add microMIPS support.
18813         (mips_print_operand_punctuation): Support short delay slots and
18814         compact jumps.
18815         (umips_swm_mask, umips_swm_encoding): New.
18816         (umips_build_save_restore): New function.
18817         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18818         (was_mips16_p): Remove.
18819         (old_compression_mode): New.
18820         (mips_set_compression_mode): New function.
18821         (mips_set_current_function): Add microMIPS support.
18822         (mips_option_override): Likewise.
18823         (umips_save_restore_pattern_p): New function.
18824         (umips_output_save_restore): New function.
18825         (umips_load_store_pair_p_1): New function.
18826         (umips_load_store_pair_p): New function.
18827         (umips_output_load_store_pair_1): New function.
18828         (umips_output_load_store_pair): New function.
18829         (umips_movep_target_p) New function.
18830         (mips_prepare_pch_save): Add microMIPS support.
18831         * config/mips/mips.h (TARGET_COMPRESSION): New.
18832         (TARGET_CPU_CPP_BUILTINS): Update macro
18833         to use new compression flags and to support microMIPS.
18834         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18835         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18836         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18837         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18838         (ASM_SPEC): Support mmicromips and mno-micromips.
18839         (M16STORE_REG_P): New macro.
18840         (MIPS_CALL): Support TARGET_MICROMIPS.
18841         (MICROMIPS_J): New macro.
18842         (mips_base_mips16): Rename this...
18843         (mips_base_compression_flags): ...to this.
18844         (UMIPS_12BIT_OFFSET_P): New macro.
18845         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18846         (MULTILIB_DIRNAMES): Likewise.
18848 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
18850         PR target/52555
18851         * target-globals.c (save_target_globals): For init_reg_sets and
18852         target_reinit remporarily set this_fn_optabs to this_target_optabs.
18854 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
18856         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
18857         * config/aarch64/t-aarch64
18858         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
18860 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
18862         PR inline-asm/56148
18863         * lra-constraints.c (process_alt_operands): Reload operand
18864         conflicting with earlier clobber only if no more other conflicting
18865         operands.
18867 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
18869         PR sanitizer/56393
18870         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
18871         if not linking a shared library.
18873 2013-02-22  Seth LaForge  <sethml@google.com>
18875         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
18877 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
18879         * config/arm/arm.md (split for extendsidi): Update condition.
18880         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
18881         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
18882         (qhs_zextenddi_cstr): Likewise.
18884 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18886         PR middle-end/56420
18887         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
18888         avoid signed wrapping.
18889         (expand_mult): Handle properly multiplication by
18890         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
18891         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
18892         in the compiler if coeff is HOST_WIDE_INT_MIN.
18893         (expand_divmod): Don't make ext_op1 static, change it's type to
18894         uhwi.  Avoid undefined behavior in -INTVAL (op1).
18896         PR rtl-optimization/50339
18897         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
18898         field.
18899         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
18900         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
18901         into splitting_ashiftrt field.
18902         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
18903         ASHIFTRT.
18904         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
18905         choices.
18907 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
18909         PR middle-end/56108
18910         * trans-mem.c (execute_tm_mark): Do not expand transactions that
18911         are sure to go irrevocable.
18913 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
18915         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
18916         scalars are valid operands.
18918 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18920         PR tree-optimization/56310
18921         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
18922         only matching indices and non-negative final offsets.
18923         (intersect_aggregates_with_edge): Pass src_idx to
18924         agg_replacements_to_vector.  Pass src_idx insstead of index to
18925         intersect_with_agg_replacements.
18927 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18929         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
18930         instead of hard-wired defaults.
18932 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
18934         * doc/invoke.texi (MIPS Options): Update documentation of the
18935         floating-point multiply-accumulate instruction restrictions.
18937 2013-02-21  Kostya Serebryany  <kcc@google.com>
18939         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
18940         asan_shadow_offset on x86_64 linux.
18942 2013-02-21  Richard Biener  <rguenther@suse.de>
18944         PR tree-optimization/56415
18945         Revert
18946         2013-02-11  Richard Biener  <rguenther@suse.de>
18948         PR tree-optimization/56273
18949         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
18950         first VRP run.
18952 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18954         PR bootstrap/56258
18955         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
18956         instead of @itemx.
18958         PR inline-asm/56405
18959         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
18960         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
18962 2013-02-20  Jan Hubicka  <jh@suse.cz>
18964         PR tree-optimization/56265
18965         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
18966         when target is referenced for first time.
18968 2013-02-20  Richard Biener  <rguenther@suse.de>
18970         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
18971         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
18972         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
18973         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
18974         not return anything.
18975         (rename_ssa_copies): Do not remove unused locals.
18976         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
18977         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
18978         * passes.c (execute_function_todo): Do not schedule unused locals
18979         removal if cleanup_tree_cfg did something.
18980         * tree-ssa-live.c (remove_unused_locals): Dump statistics
18981         about the number of removed locals.
18983 2013-02-20  Richard Biener  <rguenther@suse.de>
18985         PR tree-optimization/56398
18986         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
18988 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18990         PR tree-optimization/55334
18991         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
18992         restricted pointers to arrays.
18994 2013-02-20  Richard Biener  <rguenther@suse.de>
18995             Jakub Jelinek  <jakub@redhat.com>
18997         PR tree-optimization/56396
18998         * tree-ssa-ccp.c (n_const_val): New static variable.
18999         (get_value): Return NULL for SSA names we don't have a lattice
19000         entry for.
19001         (ccp_initialize): Initialize n_const_val.
19002         * tree-ssa-copy.c (n_copy_of): New static variable.
19003         (init_copy_prop): Initialize n_copy_of.
19004         (get_value): Return NULL_TREE for SSA names we don't have a
19005         lattice entry for.
19007 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19009         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19011 2013-02-20  Richard Biener  <rguenther@suse.de>
19013         * genpreds.c (write_lookup_constraint): Do not compare first
19014         letter of the constraint again.
19016 2013-02-20  Richard Biener  <rguenther@suse.de>
19018         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19019         and ceil_log2.
19020         (get_use_iv_cost): Terminate hashtable walk when coming across
19021         an empty entry.
19023 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19025         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19026         reassociation for avx2 targets.
19028 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19030         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19031         Add version check for v8.10.a to enable microblaze_has_clz
19032         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19033         version and TARGET_PATTERN_COMPARE check
19034         * config/microblaze/microblaze.md: New clzsi2 instruction
19036 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19038         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19039         function before branching.
19041 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19043         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19044         DUMP_INSN_RTX_UID.
19045         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19047 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19049         PR middle-end/55889
19050         * sel-sched.c: Include ira.h.
19051         (implicit_clobber_conflict_p): New function.
19052         (moveup_expr): Use it.
19053         * Makefile.in (sel-sched.o): Depend on ira.h.
19055 2013-02-19  Richard Biener  <rguenther@suse.de>
19057         PR tree-optimization/56384
19058         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19059         (vn_hash_type): Split out from ...
19060         (vn_hash_constant_with_type): ... here.
19061         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19062         (vn_phi_eq): Compare types from vn_phi_s structure.
19063         (vn_phi_lookup): Populate vn_phi_s type.
19064         (vn_phi_insert): Likewise.
19066 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19068         PR tree-optimization/56350
19069         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19070         if haven't found reduction or nested cycle operand, rather than
19071         asserting we must find it.
19073         PR tree-optimization/56381
19074         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19075         to fold_build3.
19077 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19078             Jakub Jelinek  <jakub@redhat.com>
19080         PR target/52555
19081         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19082         (swap_optab_enable): Same.
19083         (init_all_optabs): Use argument instead of global.
19084         * tree.h (struct tree_optimization_option): New field target_optabs.
19085         * expr.h (init_all_optabs): Add argument to prototype.
19086         (TREE_OPTIMIZATION_OPTABS): New.
19087         (save_optabs_if_changed): Protoize.
19088         * optabs.h: Declare this_fn_optabs.
19089         * optabs.c (save_optabs_if_changed): New.
19090         Declare this_fn_optabs.
19091         (init_optabs): Add argument to init_all_optabs() call.
19092         * function.c (invoke_set_current_function_hook): Handle per
19093         function optabs.
19094         * function.h (struct function): New field optabs.
19095         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19096         optimization_current_node has changed.
19097         * target-globals.h (save_target_globals_default_opts): Protoize.
19098         * target-globals.c (save_target_globals_default_opts): New.
19100 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19102         PR target/56347
19103         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19104         registers %fr12 and %fr12R as call used.
19106         PR target/56214
19107         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19108         and HImode, require all displacements to be an integer multiple of
19109         their mode size.
19110         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19111         only allow QImode and HImode when reload is in progress and strict is
19112         true.  Likewise for symbolic addresses.  Use base14_operand to check
19113         displacements in REG+BASE addresses.
19115 2013-02-18  Richard Biener  <rguenther@suse.de>
19117         PR tree-optimization/56366
19118         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19119         handle sign-conversion of outer-loop initial induction value.
19121 2013-02-18  Richard Biener  <rguenther@suse.de>
19123         PR middle-end/56349
19124         * cfghooks.c (merge_blocks): If we merge a latch into another
19125         block adjust references to it.
19126         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19127         (verify_loop_structure): Verify that a recorded latch is in fact
19128         a latch.
19130 2013-02-18  Richard Biener  <rguenther@suse.de>
19132         PR tree-optimization/56321
19133         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19134         order SSA name release and virtual operand unlinking.
19136 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19138         * config/microblaze/microblaze.md (save_stack_block): Define.
19139         (restore_stack_block): Likewise.
19141 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19143         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19144         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19145         * config/microblaze/microblaze.c (microblaze_option_override):
19146         Bail out early for PIC modes when target does not support PIC.
19148 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19150         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19151         Replace with a microblaze version.
19152         (microblaze_trampoline_init): Adapt for microblaze.
19153         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19154         microblaze.
19156 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19157             Dodji Seketeli  <dodji@redhat.com>
19159         PR asan/56330
19160         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19161         (instrument_mem_region_access): Do not forget to always put
19162         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19163         0) statement, even for cases where either 'base' or 'base + len'
19164         are not instrumented -- because they have been previously
19165         instrumented.  Simplify the logic by putting all the statements
19166         instrument 'base + len' inside a sequence, and then insert that
19167         sequence right before the current insertion point.  Then, to
19168         instrument 'base + len', just get an iterator on that statement.
19169         And do not forget to update the pointer to iterator the function
19170         received as argument.
19172 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19174         PR rtl-optimization/56348
19175         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19177 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19179         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19180         (clean_graph_dump_file): Pass base to start_graph_dump.
19182 2013-02-14  Richard Henderson  <rth@redhat.com>
19184         PR target/55941
19185         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19187 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19189         * collect2-aix.h: Define F_LOADONLY.
19191 2013-02-14  Richard Biener  <rguenther@suse.de>
19193         PR lto/50494
19194         * varasm.c (output_constant_def_1): Get the decl representing
19195         the constant as argument.
19196         (output_constant_def): Wrap output_constant_def_1.
19197         (make_decl_rtl): Use output_constant_def_1 with the decl
19198         representing the constant.
19199         (build_constant_desc): Optionally re-use a decl already
19200         representing the constant.
19201         (tree_output_constant_def): Adjust.
19203 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19205         Fix an asan crash
19206         * asan.c (instrument_builtin_call):  Really put the length of the
19207         second source argument into src1_len.
19209 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19211         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19212         argument.  If it is false, don't create edge from then_bb to
19213         fallthru_bb.
19214         (insert_if_then_before_iter): Pass true to it.
19215         (build_check_stmt): Pass false to it.
19216         (transform_statements): Flush hash table only on extended basic
19217         block boundaries, rather than at the beginning of every bb.
19218         Don't flush hash table on nonfreeing_call_p calls.
19219         * tree-flow.h (nonfreeing_call_p): New prototype.
19220         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19222 2013-02-13  David S. Miller  <davem@davemloft.net>
19224         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19226 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19228         PR target/56184
19229         * ira.c (max_regno_before_ira): Move from ...
19230         (ira): ... here.
19231         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19232         vec_safe_length.
19234 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19236         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19238 2013-02-13  Richard Biener  <rguenther@suse.de>
19240         PR lto/56295
19241         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19242         globals in MEM_REFs.
19244 2013-02-13  Richard Biener  <rguenther@suse.de>
19246         * loop-init.c (loop_optimizer_init): Clear loop state when
19247         re-initializing preserved loops.
19248         * loop-unswitch.c (unswitch_single_loop): Return whether
19249         we unswitched the loop.  Do not verify loop state here.
19250         (unswitch_loops): When we unswitched a loop discover new loops.
19252 2013-02-13  Kostya Serebryany  <kcc@google.com>
19254         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19255         on x86_64 linux.
19256         * sanitizer.def: Rename __asan_init to __asan_init_v1.
19258 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
19260         Avoid instrumenting duplicated memory access in the same basic block
19261         * Makefile.in (asan.o): Add new dependency on hash-table.h
19262         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19263         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19264         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19265         (free_mem_ref_resources, has_mem_ref_been_instrumented)
19266         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19267         (get_mem_ref_of_assignment): New functions.
19268         (get_mem_refs_of_builtin_call): Extract from
19269         instrument_builtin_call and tweak a little bit to make it fit with
19270         the new signature.
19271         (instrument_builtin_call): Use the new
19272         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
19273         of is_gimple_builtin_call.
19274         (instrument_derefs, instrument_mem_region_access): Insert the
19275         instrumented memory reference into the hash table.
19276         (maybe_instrument_assignment): Renamed instrument_assignment into
19277         this, and change it to advance the iterator when instrumentation
19278         actually happened and return true in that case.  This makes it
19279         homogeneous with maybe_instrument_assignment, and thus give a
19280         chance to callers to be more 'regular'.
19281         (transform_statements): Clear the memory reference hash table
19282         whenever we enter a new BB, when we cross a function call, or when
19283         we are done transforming statements.  Use
19284         maybe_instrument_assignment instead of instrumentation.  No more
19285         need to special case maybe_instrument_assignment and advance the
19286         iterator after calling it; it's now handled just like
19287         maybe_instrument_call.  Update comment.
19289 2013-02-13  Richard Biener  <rguenther@suse.de>
19291         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19292         Fix loop discovery code.
19294 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
19296         PR inline-asm/56148
19297         * lra-constraints.c (process_alt_operands): Match early clobber
19298         operand with itself.  Check conflicts with earlyclobber only if
19299         the operand is not reloaded.  Prefer to reload conflicting operand
19300         if earlyclobber and matching operands are the same.
19302 2013-02-12  Richard Biener  <rguenther@suse.de>
19304         PR lto/56297
19305         * lto-streamer-out.c (write_symbol): Do not output symbols
19306         for hard register variables.
19308 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
19310         PR target/54222
19311         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19312         (umulsidi3_insn, mulsidi3_insn): New insns.
19314 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
19316         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19317         (struct tune_params): Add vec_costs field.
19318         * config/arm/arm.c (arm_builtin_vectorization_cost)
19319         (arm_add_stmt_cost): New functions.
19320         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19321         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19322         (arm_default_vec_cost): New struct of type cpu_vec_costs.
19323         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19324         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19325         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19326         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19328 2013-02-12  Richard Biener  <rguenther@suse.de>
19330         PR lto/56295
19331         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19332         decls again if possible.
19334 2013-02-12  Richard Biener  <rguenther@suse.de>
19336         PR middle-end/56288
19337         * tree-ssa.c (verify_ssa_name): Fix check, move
19338         SSA_NAME_IN_FREE_LIST check up.
19340 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
19341             Steven Bosscher   <steven@gcc.gnu.org>
19343         PR rtl-optimization/56151
19344         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19345         equal to op0 or op1, and last_insn pattern is CODE operation
19346         with MEM dest and one of the operands matches that MEM.
19348 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
19350         * doc/extend.texi: Document Function Multiversioning and "default"
19351         parameter string to target attribute.
19352         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19353         target attribute parameter is "default".
19354         (ix86_compare_version_priority): Remove checks for target attribute.
19355         (ix86_mangle_function_version_assembler_name): Change error to sorry.
19356         Remove check for target attribute equal to NULL. Add assert.
19357         (ix86_generate_version_dispatcher_body): Change error to sorry.
19359 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
19360             Jack Howarth  <howarth@bromo.med.uc.edu>
19361             Patrick Marlier  <patrick.marlier@gmail.com>
19363         PR libitm/55693
19364         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19365         define ENDFILE_SPEC as TM_DESTRUCTOR.
19366         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19368 2013-02-11  Alexander Potapenko  <glider@google.com>
19369             Jack Howarth  <howarth@bromo.med.uc.edu>
19370             Jakub Jelinek  <jakub@redhat.com>
19372         PR sanitizer/55617
19373         * config/darwin.c (cdtor_record): Rename ctor_record.
19374         (sort_cdtor_records): Rename sort_ctor_records.
19375         (finalize_dtors): New routine to sort destructors by
19376         priority before use in assemble_integer.
19377         (machopic_asm_out_destructor): Use finalize_dtors if needed.
19379 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
19381         PR rtl-optimization/56275
19382         * simplify-rtx.c (avoid_constant_pool_reference): Check that
19383         offset is non-negative and less than cmode size before
19384         calling simplify_subreg.
19386 2013-02-11  Richard Biener  <rguenther@suse.de>
19388         PR tree-optimization/56264
19389         * cfgloop.h (fix_loop_structure): Adjust prototype.
19390         * loop-init.c (fix_loop_structure): Return the number of
19391         newly discovered loops.
19392         * tree-cfgcleanup.c (repair_loop_structures): When new loops
19393         are discovered, do a full loop-closed SSA rewrite.
19395 2013-02-11  Richard Biener  <rguenther@suse.de>
19397         PR tree-optimization/56273
19398         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19399         first VRP run.
19400         (check_array_ref): Fix missing newline in dumps.
19401         (search_for_addr_array): Likewise.
19403 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
19405         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19407 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
19409         PR target/56256
19410         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19412 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
19414         PR rtl-optimization/56246
19415         * lra-constraints.c (simplify_operand_subreg): Try to reuse
19416         reload pseudo.
19417         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19418         constraints are satisfied.
19420 2013-02-08  Jeff Law  <law@redhat.com>
19422         PR debug/53948
19423         * emit-rtl.c (reg_is_parm_p): New function.
19424         * regs.h (reg_is_parm_p): New prototype.
19425         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19426         callee-clobbered registers.
19428 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
19430         PR target/56043
19431         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19432         If there is no implicit builtin declaration, just return NULL.
19434 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19436         * config/i386/sse.md (FMAMODEM): New mode iterator.
19437         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19438         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19440 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19442         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19443         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19444         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19446 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19448         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19449         (microblaze*-*-elf): Likewise.
19450         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19451         LINK_SPEC.
19452         * config/microblaze/microblaze-c.c: Add builtin defines for
19453         _LITTLE_ENDIAN and _BIG_ENDIAN.
19454         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19455         add to TARGET_DEFAULT flags.
19456         Expand ASM_SPEC and LINK_SPEC.
19457         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19458         * config/microblaze/microblaze.md: Update extendsidi2 and
19459         movdi_internal instructions to use low-order / high-order reg
19460         print_operands.
19461         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19462         options and inversemask / mask of LITTLE_ENDIAN.
19463         * config/microblaze/t-microblaze: Expand multilib options to
19464         include mlittle-endian (le) and update exceptions patterns.
19466 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
19468         PR rtl-optimization/56195
19469         * lra-constraints.c (get_reload_reg): Don't reuse regs
19470         if they have smaller mode than requested, if they have
19471         wider mode than requested, try to return a SUBREG.
19473         PR tree-optimization/56250
19474         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19475         if type is unsigned and code isn't MULT_EXPR.
19477 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19479         PR tree-optimization/56064
19480         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19481         bits according to mode.
19482         * fixed-value.h (fixed_from_double_int)
19483         (const_fixed_from_double_int): Adjust comments.
19485 2013-02-08  Richard Biener  <rguenther@suse.de>
19487         PR lto/56231
19488         * lto-streamer.h (struct data_in): Remove current_file, current_line
19489         and current_col members.
19490         * lto-streamer-out.c (lto_output_location): Stream changed bits
19491         en-block for efficiency.
19492         * lto-streamer-in.c (clear_line_info): Remove.
19493         (lto_input_location): Cache current file, line and column
19494         globally via local statics.  Read changed bits en-block.
19495         (input_function): Do not call clear_line_info.
19496         (lto_read_body): Likewise.
19497         (lto_input_toplevel_asms): Likewise.
19499 2013-02-08  Michael Matz  <matz@suse.de>
19501         PR tree-optimization/52448
19502         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19503         (nt_call_phase): New static.
19504         (add_or_mark_expr): Only mark accesses with newer phase than any
19505         call seen.
19506         (nonfreeing_call_p): New.
19507         (nt_init_block): Update nt_call_phase, mark blocks as visited.
19508         (nt_fini_block): Keep blocks marked as visited.
19509         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19511 2013-02-08  Richard Biener  <rguenther@suse.de>
19513         * ira.c (ira): Free broken dominator information.
19515 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19517         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19519 2013-02-08  Marek Polacek  <polacek@redhat.com>
19521         * cfgloop.c (verify_loop_structure): Add more checking of headers.
19523 2013-02-08  Richard Biener  <rguenther@suse.de>
19525         PR middle-end/56181
19526         * cfgloop.h (flow_loops_find): Adjust.
19527         (bb_loop_header_p): Declare.
19528         * cfgloop.c (bb_loop_header_p): New function split out from ...
19529         (flow_loops_find): ... here.  Adjust function signature,
19530         support incremental loop structure update.
19531         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
19532         * cfgloopmanip.c (fix_loop_structure): Move ...
19533         * loop-init.c (fix_loop_structure): ... here.
19534         (apply_loop_flags): Split out from ...
19535         (loop_optimizer_init): ... here.
19536         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
19537         in incremental mode, only remove dead loops here.
19539 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19541         PR target/54222
19542         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19543         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19544         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19545         (*round<mode>3.libgcc): New insns for fixed-modes.
19546         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19547         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19548         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19549         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19550         implementations.  Define to __builtin_avr_absFX,
19551         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19552         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19553         __builtin_avr_countlsFX, respectively.
19554         * config/avr/avr-c.c (target.h): Include it.
19555         (enum avr_builtin_id): New enum.
19556         (avr_resolve_overloaded_builtin): New static function.
19557         (avr_register_target_pragmas): Use it to set
19558         targetm.resolve_overloaded_builtin.
19559         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19560         tree nodes used by DEF_BUILTIN.
19561         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19562         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19563         <AVR_BUILTIN_xxBITS>: Same.
19565 2013-02-08  Richard Biener  <rguenther@suse.de>
19567         * cfgloop.c (verify_loop_structure): Properly handle
19568         a loop exiting to another loop header.
19569         * ira-int.h (ira_loops): Remove.
19570         * ira.c (ira_loops): Remove.
19571         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19572         (do_reload): Use loop_optimizer_finalize.
19573         * ira-build.c (create_loop_tree_nodes): Use get_loops and
19574         number_of_loops to access the loop tree.
19575         (more_one_region_p): Likewise.
19576         (finish_loop_tree_nodes): Likewise.
19577         (rebuild_regno_allocno_maps): Likewise.
19578         (mark_loops_for_removal): Likewise.
19579         (mark_all_loops_for_removal): Likewise.
19580         (remove_unnecessary_regions): Likewise.
19581         (ira_build): Likewise.
19582         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19584 2013-02-08  Richard Biener  <rguenther@suse.de>
19586         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
19587         * ipa-pure-const.c (analyze_function): Avoid calling
19588         mark_irreducible_loops twice.
19589         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
19591 2013-02-07  David S. Miller  <davem@davemloft.net>
19593         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
19594         on 'reg'.
19595         * var-tracking.c (vt_add_function_parameter): Test the presence of
19596         HAVE_window_save properly and do not remap argument registers when
19597         we have a leaf function.
19599 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
19601         PR bootstrap/56227
19602         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
19603         instead of "ll".
19604         * config/i386/i386.c (ix86_print_operand): Ditto.
19606 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19608         * lra-constraints.c (process_alt_operands): Fix recently added comment.
19610 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19612         PR rtl-optimization/56225
19613         * lra-constraints.c (process_alt_operands): Check that reload hard
19614         reg can hold value for strict_low_part.
19616 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19618         PR debug/56154
19619         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
19620         dwarf2out_end_function.
19621         (in_first_function_p, maybe_at_text_label_p,
19622         first_loclabel_num_not_at_text_label): New variables.
19623         (dwarf2out_var_location): In the first function find out
19624         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
19625         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
19626         functions.
19628 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
19630         PR rtl-optimization/56178
19631         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
19632         SUBREG of a register.  Tidy up related block of code.
19633         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
19634         note if the source is a register or a SUBREG of a register.
19636 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19638         PR target/56228
19639         * config/rs6000/rs6000.md (ptrm): New mode attr.
19640         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
19641         call_value_indirect_aix<pttrsize>,
19642         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
19643         m in constraints.
19645 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
19647         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
19648         if -bnortl. Convert to strcmp and strncmp.
19650 2013-02-07  Alan Modra  <amodra@gmail.com>
19652         PR target/54009
19653         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
19654         addresses won't wrap when offsetting.
19655         (rs6000_secondary_reload): Provide secondary reloads needed for
19656         wrapping LO_SUM addresses.
19658 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
19660         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
19661         MACH, just __MACH__.
19663 2013-02-06  Richard Biener  <rguenther@suse.de>
19665         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
19666         instead of calling fix_loop_structure.
19668 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
19670         PR middle-end/56217
19671         * omp-low.c (use_pointer_for_field): Return false if
19672         lower_send_shared_vars doesn't generate any copy-out code.
19674 2013-02-06  Tom de Vries  <tom@codesourcery.com>
19676         PR rtl-optimization/56131
19677         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
19678         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
19679         of the label is NULL.  Add comment.
19681 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
19683         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
19685         PR sanitizer/55374
19686         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
19687         (STATIC_LIBTSAN_LIBS): Likewise.
19688         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
19689         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
19690         is defined, don't add anything else beyond that.
19691         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
19692         (LINK_COMMAND_SPEC): Use them.
19694         PR tree-optimization/56205
19695         * tree-stdarg.c (check_all_va_list_escapes): Return true if
19696         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
19697         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
19699 2013-02-05  Richard Biener  <rguenther@suse.de>
19701         PR tree-optimization/53342
19702         PR tree-optimization/53185
19703         * tree-vectorizer.h (vect_check_strided_load): Remove.
19704         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
19705         not disallow peeling for vectorized strided loads.
19706         (vect_check_strided_load): Make static and simplify.
19707         (vect_analyze_data_refs): Adjust.
19708         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
19709         correctly when vectorizing strided loads.
19711 2013-02-05  Richard Biener  <rguenther@suse.de>
19713         * doc/install.texi: Refer to ISL, not PPL.
19715 2013-02-05  Jan Hubicka  <jh@suse.cz>
19717         PR tree-optimization/55789
19718         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
19720 2013-02-05  Jan Hubicka  <jh@suse.cz>
19722         PR tree-optimization/55789
19723         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
19724         the dead call anyway.
19726 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
19728         PR sanitizer/55374
19729         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
19731 2013-02-04  Alexander Potapenko  <glider@google.com>
19732             Jack Howarth  <howarth@bromo.med.uc.edu>
19733             Jakub Jelinek  <jakub@redhat.com>
19735         PR sanitizer/55617
19736         * config/darwin.c (sort_ctor_records): Stabilized qsort
19737         on constructor priority by using original position.
19738         (finalize_ctors): New routine to sort constructors by
19739         priority before use in assemble_integer.
19740         (machopic_asm_out_constructor): Use finalize_ctors if needed.
19742 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
19744         PR libstdc++/54314
19745         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
19746         about visibility on artificial decls.
19747         * config/sol2.c (solaris_assemble_visibility): Likewise.
19749 2013-02-04  Kai Tietz  <ktietz@redhat.com>
19751         PR target/56186
19752         * config/i386/i386.c (function_value_ms_64): Add additional valtype
19753         argument and improve checking of return-argument types for 16-byte
19754         modes.
19755         (ix86_function_value_1): Add additional valtype argument on call
19756         of function_value_64.
19757         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
19758         handling infunction_value_64 function.
19760 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
19762         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
19764 2013-02-04  Richard Biener  <rguenther@suse.de>
19766         PR tree-optimization/56188
19767         * tree-ssa-structalias.c (label_visit): Consider case with
19768         initially non-empty points-to set.
19769         (perform_var_substitution): Dump node mapping and clean up.
19771 2013-02-04  Richard Guenther  <rguenther@suse.de>
19773         PR lto/56168
19774         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
19775         node prevail as last resort.
19776         (lto_symtab_merge_decls): Remove guard on LTRANS here.
19777         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
19779 2013-02-04  Richard Biener  <rguenther@suse.de>
19781         PR tree-optimization/56113
19782         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
19783         Merge into ...
19784         (equiv_class_lookup_or_add): ... this.
19785         (label_visit): Adjust and fix error in previous patch.
19786         (perform_var_substitution): Adjust.
19788 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
19790         * config/sh/divtab.c: Fix formatting and comments throughout the file.
19791         * config/sh/sh4-300.md: Likewise.
19792         * config/sh/sh4a.md: Likewise.
19793         * config/sh/constraints.md: Likewise.
19794         * config/sh/sh.md: Likewise.
19795         * config/sh/netbsd-elf.h: Likewise.
19796         * config/sh/predicates.md: Likewise.
19797         * config/sh/sh-protos.h: Likewise.
19798         * config/sh/ushmedia.h: Likewise.
19799         * config/sh/linux.h: Likewise.
19800         * config/sh/sh.c: Likewise.
19801         * config/sh/superh.h: Likewise.
19802         * config/sh/elf.h: Likewise.
19803         * config/sh/sh4.md: Likewise.
19804         * config/sh/sh.h: Likewise.
19806 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19808         * config/pa/constraints.md: Adjust unused letters.  Change "T"
19809         constraint to match_test floating_point_store_memory_operand().
19810         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
19811         (base14_operand): New.
19812         (floating_point_store_memory_operand): New.
19813         (integer_store_memory_operand): Revise to use base14_operand and
19814         reg_plus_base_memory_operand.
19815         (move_dest_operand): Allow symbolic_memory_operands.
19816         (symbolic_memory_operand): Check for LO_SOM.
19817         (symbolic_operand): Change default case to break.
19818         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
19819         CONST_DOUBLE values to be reloaded by putting them into memory when
19820         the destination is a floating point register.
19821         (movdf): Remove code to handle CONST_DOUBLE.
19822         (movsf): Likewise.
19823         (reload_indf_r1): New.
19824         (reload_insf_r1): New.
19825         Consistently use "Q" and "T" constraints with integer and floating
19826         point move instructions, respectively.
19827         (movdi): Remove FAIL.
19828         Change predicate for source operand unamed DImode move from
19829         general_operand to move_src_operand.
19830         (umulsidi3): Change predicate for destination operand to
19831         register_operand.
19832         Likewise for similar unamed patterns.
19833         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
19834         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
19835         (hppa_legitimize_address): Simplify mask calculation.
19836         (pa_emit_move_sequence): Revised handling of secondary reloads from
19837         REG+D addresses for floating point loads and stores.  Directly handle
19838         loading CONST0_RTX (mode) to a floating point register.
19839         (pa_secondary_reload): Handle reloading DF and SFmode constant values
19840         to floating point registers.  Don't restrict secondary reloads to
19841         floating point registers to integer modes.  Revise some comments and
19842         cleanup some code.
19843         (TARGET_LEGITIMATE_ADDRESS_P): Define.
19844         (pa_legitimate_address_p): New.
19845         (pa_legitimize_reload_address): New.
19846         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
19847         (STRICT_REG_OK_FOR_BASE_P): New.
19848         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
19849         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
19851 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
19852             Andrew Dixie  <andrewd@gentrack.com>
19854         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
19855         flag set.
19857 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
19859         * expmed.c (extract_bit_field_1): Pass the full width of the
19860         structure to get_best_reg_extraction_insn.
19862 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19864         PR target/54601
19865         * configure.ac (use_cxa_atexit): Add AIX.
19866         * configure: Regenerate.
19868         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
19870 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
19872         PR debug/54793
19873         * final.c (need_profile_function): New variable.
19874         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
19875         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
19876         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
19877         notes, targetm.asm_out.function_prologue doesn't emit anything,
19878         HAVE_prologue and profiler should be emitted before prologue,
19879         set need_profile_function instead of emitting it.
19880         (final_scan_insn): If need_profile_function, emit
19881         profile_function on the first NOTE_INSN_BASIC_BLOCK or
19882         NOTE_INSN_FUNCTION_BEG note.
19884 2013-02-01  Richard Henderson  <rth@redhat.com>
19886         * config/rs6000/rs6000.md (smulditi3): New.
19887         (umulditi3): New.
19889         * config/alpha/alpha.md (umulditi3): New.
19891 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19893         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
19894         (ASM_OUTPUT_ALIGNED_LOCAL): New.
19896 2013-02-01  Richard Biener  <rguenther@suse.de>
19898         PR tree-optimization/56113
19899         * tree-ssa-structalias.c (label_visit): Reduce work for
19900         single-predecessor nodes.
19902 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
19904         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
19905         range isn't testing for zero.
19907 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
19909         PR middle-end/56113
19910         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
19912 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
19913             Nick Clifton  <nickc@redhat.com>
19915         * config/v850/constraints.md (Q): Define as a memory constraint.
19916         * config/v850/predicates.md (label_ref_operand): New predicate.
19917         (e3v5_shift_operand): New predicate.
19918         (ior_operator): New predicate.
19919         * config/v850/t-v850: Add e3v5 multilib.
19920         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
19921         (v850_gen_movdi): Prototype.
19922         * config/v850/v850.c: Add support for e3v5 architecture.
19923         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
19924         TARGET_V850E_UP.
19925         (construct_save_jarl): Add e3v5 long JARL support.
19926         (v850_adjust_insn_length): New function.  Adjust length of call
19927         insns when using e3v5 instructions.
19928         (v850_gen_movdi): New function: Generate instructions to move a
19929         DImode value.
19930         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
19931         (CPP_SPEC): Define __v850e3v5__ as appropriate.
19932         (TARGET_USE_FPU): Enable for e3v5.
19933         (CONST_OK_FOR_W): New macro.
19934         (ADJUST_INSN_LENGTH): Define.
19935         * config/v850/v850.md (UNSPEC_LOOP): Define.
19936         (attr cpu): Add v850e3v5.
19937         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
19938         (movdi): New pattern.
19939         (movdi_internal): New pattern.
19940         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
19941         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
19942         (cstoresf4): Likewise.
19943         (cstoredf4): Likewise.
19944         (insv): New pattern.
19945         (rotlso3_a): New pattern.
19946         (rotlsi3_b): New pattern
19947         (rotlsi3_v850e3v5): New pattern.
19948         (doloop_begin): New pattern.
19949         (fix_loop_counter): New pattern.
19950         (doloop_end): New pattern.
19951         (branch_normal): Add e3v5 long branch support.
19952         (branch_invert): Likewise.
19953         (branch_z_normal): Likewise.
19954         (branch_z_invert): Likewise.
19955         (branch_nz_normal): Likewise.
19956         (branch_nz_invert): Likewise.
19957         (call_internal_short): Add e3v5 register-indirect JARL support.
19958         (call_internal_long): Likewise.
19959         (call_value_internal_short): Likewise.
19960         (call_value_internal_long): Likewise.
19961         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
19962         (mloop): New option.
19963         * config.gcc: Add support for configuring v840e3v5 target.
19964         * doc/invoke.texi: Document new v850 specific command line options.
19966 2013-01-31  Paul Koning  <ni1d@arrl.net>
19968         PR debug/55059
19969         PR debug/54508
19970         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
19971         children if parent is a class.
19972         (prune_unused_types_prune): Don't add DW_AT_declaration.
19974 2013-01-31  Richard Biener  <rguenther@suse.de>
19976         PR tree-optimization/56157
19977         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
19978         match up operand with SLP child.
19980 2013-01-31  Jason Merrill  <jason@redhat.com>
19982         PR debug/54410
19983         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
19984         parameters the first time.
19985         (gen_scheduled_generic_parms_dies): Check completeness here.
19987 2013-01-31  Richard Biener  <rguenther@suse.de>
19989         PR middle-end/53073
19990         * common.opt (faggressive-loop-optimizations): New flag,
19991         enabled by default.
19992         * doc/invoke.texi (faggressive-loop-optimizations): Document.
19993         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
19994         infer_loop_bounds_from_undefined by it.
19996 2013-01-31  Richard Biener  <rguenther@suse.de>
19998         PR tree-optimization/56150
19999         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
20000         visit virtual operands.
20001         (find_uses_to_rename_bb): Likewise.
20003 2013-01-31  Richard Biener  <rguenther@suse.de>
20005         PR tree-optimization/56150
20006         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20007         mixed store non-store stmts.
20009 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
20011         PR sanitizer/55374
20012         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20013         LIBASAN_EARLY_SPEC is defined.
20014         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20015         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20016         before %o.
20017         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20019         PR c++/55742
20020         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20021         invalid args instead of ICEing on it.
20022         (ix86_valid_target_attribute_tree): Return error_mark_node if
20023         ix86_valid_target_attribute_inner_p failed.
20024         (ix86_valid_target_attribute_p): Return false only if
20025         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20026         target("default") attribute.
20027         (sorted_attr_string): Change argument from const char * to tree,
20028         merge in all target attribute arguments rather than just one.
20029         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20030         instead of free.  Avoid using strcat.
20031         (ix86_mangle_function_version_assembler_name): Mangle
20032         target("default") as if no target attribute is present.  Adjust
20033         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20034         instead of xmalloc and XDELETEVEC instead of free.
20035         (ix86_function_versions): Don't return true if one of the decls
20036         doesn't have target attribute.  If they don't and one of the decls
20037         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20038         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20039         (ix86_supports_function_versions): Remove.
20040         (make_name): Fix up formatting.
20041         (make_dispatcher_decl): Remove resolver_name and its initialization.
20042         Avoid leaking memory.
20043         (is_function_default_version): Return true if there is
20044         target("default") attribute rather than no target attribute at all.
20045         (make_resolver_func): Avoid leaking memory.
20046         (ix86_generate_version_dispatcher_body): Likewise.
20047         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20048         * target.def (supports_function_versions): Remove.
20049         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20050         * doc/tm.texi: Regenerated.
20052 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20054         PR rtl-optimization/56144
20055         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20056         for values with side effects.
20058 2013-01-30  Richard Biener  <rguenther@suse.de>
20060         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20061         (sparseset_pop): Likewise.
20062         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20063         to be able to use quick_push in the worker loop.
20065 2013-01-30  Marek Polacek  <polacek@redhat.com>
20067         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20069 2013-01-30  Richard Biener  <rguenther@suse.de>
20071         PR lto/56147
20072         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20074 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20076         PR tree-optimization/56064
20077         * fixed-value.c (fixed_from_double_int): New function.
20078         * fixed-value.h (fixed_from_double_int): New prototype.
20079         (const_fixed_from_double_int): New static inline function.
20080         * fold-const.c (native_interpret_fixed): New static function.
20081         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20082         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20083         (native_encode_fixed): New static function.
20084         (native_encode_expr) <FIXED_CST>: Use it.
20085         (native_interpret_int): Move double_int worker code to...
20086         * double-int.c (double_int::from_buffer): ...this new static method.
20087         * double-int.h (double_int::from_buffer): Prototype it.
20089 2013-01-30  Richard Biener  <rguenther@suse.de>
20091         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20092         New pointer-map and obstack.
20093         (init_alias_vars): Allocate pointer-map and obstack.
20094         (delete_points_to_sets): Free them.
20095         (find_what_var_points_to): Cache result.
20096         (find_what_p_points_to): Adjust for changed interface of
20097         find_what_var_points_to.
20098         (compute_points_to_sets): Likewise.
20099         (ipa_pta_execute): Likewise.
20101 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20103         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20104         * configure: Regenerate.
20105         * config.in: Regenerate.
20106         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20107         #nobits/#progbits if supported.
20109 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20111         PR target/56121
20112         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20113         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20114         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20116 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20118         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20119         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20121 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20123         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20124         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20126 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20128         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20129         declaration.
20130         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20131         * config/arm/cortex-a7.md: New bypasses using
20132         arm_mac_accumulator_is_result.
20134 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20136         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20137         (cortex_a7_neon_mla): Likewise.
20138         (cortex_a7_fpfmad): New reservation.
20139         (cortex_a7_fpmacs): Use ffmas and update required units.
20140         (cortex_a7_fpmuld): Update required units and latency.
20141         (cortex_a7_fpmacd): Likewise.
20142         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20143         (cortex_a7_neon). Likewise.
20144         (bypass) Update participating units.
20146 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20148         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20149         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20150         from fmac to ffma.
20151         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20152         (vfp_fmul): Use ffmad.
20153         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20154         (cortex_r4_fmacd): Use ffmad.
20155         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20156         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20157         (cortex_a9_fmacd): Use ffmad.
20158         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20159         (cortex_a8_vfp_macd): Use ffmad.
20160         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20161         (cortex_a5_fpmacd): Use ffmad.
20162         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20163         (cortex_a15_vfp_macd): Use ffmad.
20164         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20166 2013-01-29  Jason Merrill  <jason@redhat.com>
20168         PR libstdc++/54314
20169         * varasm.c (default_assemble_visibility): Don't warn about
20170         visibility on artificial decls.
20172 2013-01-29  Richard Biener  <rguenther@suse.de>
20174         PR tree-optimization/56113
20175         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20176         the bitmap leader.
20177         (label_visit): Free duplicate bitmaps and record the leader instead.
20178         (perform_var_substitution): Adjust.
20180 2013-01-29  Richard Biener  <rguenther@suse.de>
20182         PR tree-optimization/55270
20183         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20184         the CFG, schedule loops for fixup.
20186 2013-01-29  Nick Clifton  <nickc@redhat.com>
20188         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20189         SP_REG.
20191 2013-01-28  Leif Ekblad  <leif@rdos.net>
20193         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20194         * config/i386/i386.h (TARGET_RDOS): New macro.
20195         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20196         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20197         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20198         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20199         DEFAULT_LARGE_SECTION_THRESHOLD.
20200         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20201         * config/i386/rdos.h: New file.
20202         * config/i386/rdos64.h: New file.
20204 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20206         PR other/54814
20207         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20208         TEST_HARD_REG_BIT.
20210 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20212         PR rtl-optimization/56117
20213         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20214         call cselib_lookup_from_insn on the MEM before calling
20215         add_insn_mem_dependence.
20217 2013-01-28  Richard Biener  <rguenther@suse.de>
20219         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20220         to a stmt that didn't have one.
20221         (copy_phis_for_bb): Likewise for PHI arguments.
20222         (copy_debug_stmt): Likewise for debug stmts.
20224 2013-01-28  Richard Biener  <rguenther@suse.de>
20226         PR tree-optimization/56034
20227         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20228         (partition_builtin_p): Adjust.
20229         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20230         it is the last partition.
20231         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20232         up the vertex for the definition.
20233         (classify_partition): Classify whether a partition is a
20234         PKIND_REDUCTION, thus has uses outside of the loop.
20235         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20236         Merge all PKIND_REDUCTION partitions into the last partition.
20237         (tree_loop_distribution): Seed partitions from reductions as well.
20239 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20241         PR tree-optimization/56125
20242         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20243         pow(x,c) into sqrt(x) * powi(x, n/2) or
20244         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20245         optimizing for size.
20246         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20247         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20248         integer.
20250         PR tree-optimization/56094
20251         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20252         to UNKNOWN_LOCATION while gimplifying expr.
20254 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
20256         PR target/56114
20257         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20258         operand 0 in movabs insn template for -masm=intel asm alternative.
20259         (*movabs<mode>_2): Ditto for operand 1.
20261 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
20263         PR target/54663
20264         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20265         of microblaze-c.o
20267 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20269         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20270         tm_file.
20272 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20274         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20275         Undef to avoid warning.
20277 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20279         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20280         * configure: Regenerate.
20282 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
20284         PR tree-optimization/56098
20285         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20286         for stmts with volatile ops.
20287         (cond_store_replacement): Don't optimize if assign has volatile ops.
20288         (cond_if_else_store_replacement_1): Don't optimize if either
20289         then_assign or else_assign have volatile ops.
20290         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20291         volatile ops.
20293 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20295         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20297 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20299         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20300         missing ':' in asm example.
20302 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
20304         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20305         entries into lane and laneq entries.
20306         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20307         Remove AdvSIMD scalar modes.
20308         (aarch64_sq<r>dmulh_laneq<mode>): New.
20309         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20310         modes.
20311         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20312         builtin implementations to relfect changes in RTL in aarch64-simd.md.
20313         * config/aarch64/iterators.md (VCOND): New.
20314         (VCONQ): New.
20316 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20318         PR target/54222
20319         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20320         Add NULL LIBNAME argument to existing definitions.
20321         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20322         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20323         * config/avr/avr.c (DEF_BUILTIN): Same.
20324         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20325         (avr_expand_builtin): Expand to a vanilla call if a libgcc
20326         implementation is available (DECL_ASSEMBLER_NAME is set).
20327         (avr_fold_absfx): New static function.
20328         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20329         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20330         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20331         AVR_BUILTIN_ABSLLK.
20332         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20333         (abshk, absk, abslk, absllk): Provide as static inline functions.
20335 2013-01-25  Marek Polacek  <polacek@redhat.com>
20337         PR tree-optimization/56035
20338         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20340 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
20342         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20343         (*movtf_internal_rex64): Add (!o,C) alternative
20344         (*movxf_internal_rex64): Ditto.
20345         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20347 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
20349         * doc/invoke.texi: fix typo.
20350         * doc/objc.texi: fix typo.
20352 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
20354         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20355         for the first two alternatives.
20357 2013-01-24  Diego Novillo  <dnovillo@google.com>
20359         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
20360         (ggc-zone.o): Remove.
20361         * configure.ac: Remove option --with-gc.
20362         * configure: Re-generate.
20363         * doc/install.texi: Remove documentation for --with-gc.
20364         * gengtype.c (write_enum_defn): Remove.  Update all users.
20365         (write_Types_process_field): Remove generation of gt_e_* argument.
20366         (output_type_enum): Remove.  Update all users.
20367         (write_enum_defn): Remove.  Update all users.
20368         (enum alloc_zone): Remove.  Update all users.
20369         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20370         * ggc-common.c (ggc_splay_alloc): Remove first argument.
20371         Update all callers.
20372         (struct ptr_data): Remove field TYPE.  Update all users.
20373         (gt_pch_note_object): Remove argument TYPE.  Update all users.
20374         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20375         Update all users.
20376         * ggc-none.c (ggc_alloc_typed_stat): Remove.
20377         (struct alloc_zone): Remove.
20378         (ggc_internal_alloc_zone_stat): Remove.
20379         (ggc_internal_cleared_alloc_zone_stat): Remove.
20380         * ggc-page.c (ggc_alloc_typed_stat): Remove.
20381         (ggc_pch_count_object): Remove last argument.  Update all users.
20382         (ggc_pch_alloc_object): Remove last argument.  Update all users.
20383         (struct alloc_zone): Remove.
20384         * ggc-zone.c: Remove.
20385         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
20386         (struct alloc_zone): Remove.
20387         (ggc_alloc_typed_stat): Remove.
20388         (ggc_alloc_typed): Remove.
20389         (ggc_splay_alloc): Remove first argument.
20390         (rtl_zone): Remove.  Update all users.
20391         (tree_zone): Remove.  Update all users.
20392         (tree_id_zone): Remove.  Update all users.
20393         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
20394         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
20395         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
20396         * tree-ssanames.c: Remove references to zone allocator in comments.
20398 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
20400         * config/avr/avr.c (avr_out_fract): Make register numbers that
20401         might be outside of source operand signed.
20403 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
20405         * config/i386/constraints.md (Yf): New constraint.
20406         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20407         of f constraint to conditionaly disable x87 register preferences.
20408         (*movdf_internal): Ditto.
20409         (*movsf_internal): Ditto.
20411 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
20413         PR inline-asm/55934
20414         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20415         that have operands with impossible constraints.
20416         Add a FIXME for a speed-up opportunity.
20417         * lra-constraints.c (process_alt_operands): Verify that a class
20418         selected from constraints on asms is valid for the operand mode.
20419         (curr_insn_transform): Remove incorrect comment.
20421 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
20423         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20424         TOC operand is a valid symbol ref in the constant pool.
20426 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20428         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20430 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
20432         PR target/54222
20433         * config/avr/stdfix.h: New file.
20434         * t-avr (stdfix-gcc.h): New rule to build it.
20435         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20437 2013-01-23  Kostya Serebryany  <kcc@google.com>
20439         * config/darwin.h: remove dependency on
20440         CoreFoundation (asan on Mac OS).
20442 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
20444         PR target/49069
20445         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20446         instead of cmpdi_operand for first comparison operand.
20447         Don't assert that comparison operands aren't both constants.
20449 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20451         * doc/install.texi (Downloading the Source): Update references to
20452         downloading separate components.
20454 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20456         * doc/extend.texi (__int128): Improve grammar.
20458 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
20460         PR target/56028
20461         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20462         alternative to (o,r).
20463         (*movdi_internal_rex64): Remove (!o,n) alternative.
20464         (DImode immediate->memory splitter): Remove.
20465         (DImode immediate->memory peephole2): Remove.
20466         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20467         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20468         alternative to (!o,*r).
20469         (*movtf_internal_sse): New pattern.
20470         (*movxf_internal_rex64): New pattern.
20471         (*movxf_internal): Disable for TARGET_64BIT.
20472         (*movdf_internal_rex64): Remove (!o,F) alternative.
20474 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
20476         PR middle-end/56074
20477         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20478         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20479         * tree-vect-loop-manip.c (find_loop_location): Also ignore
20480         stmt locations where LOCATION_LOCUS of the stmt location is
20481         UNKNOWN_LOCATION or BUILTINS_LOCATION.
20483         PR target/55686
20484         * config/i386/i386.md (UNSPEC_STOS): New.
20485         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20486         *strsetqi_1): Add UNSPEC_STOS.
20488 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
20490         PR c++/56067
20491         * doc/invoke.texi: Remove left over -Wsynth example.
20493 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20495         PR tree-optimization/56051
20496         * fold-const.c (fold_binary_loc): Don't fold
20497         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20498         a narrowing conversion, or widening conversion from signed
20499         to unsigned.
20501 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20503         PR rtl-optimization/56023
20504         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20505         dependent on debug instruction.
20507 2013-01-21  Martin Jambor  <mjambor@suse.cz>
20509         PR middle-end/56022
20510         * function.c (allocate_struct_function): Call
20511         invoke_set_current_function_hook earlier.
20513 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20515         * reload1.c (init_reload): Only initialize reload_obstack
20516         during the first call.
20518 2013-01-21  Marek Polacek  <polacek@redhat.com>
20520         * cfgloop.c (verify_loop_structure): Fix up grammar.
20522 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
20524         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20525         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20527 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20529         PR target/56058
20530         * config/arm/marvell-pj4.md: Update copyright year.
20531         Fix up use of alu to alu_reg and simple_alu_imm.
20533 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20535         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20537 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
20539         PR target/55433
20540         * lra-constraints.c (curr_insn_transform): Don't reuse original
20541         insn for secondary memory move when memory mode should be different.
20543 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20545         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20546         atomic_storedi_1): New patterns.
20548 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
20550         btver2 pipeline descriptions.
20551         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20552         descriptions.
20553         * config/i386/i386.md (btver2_decode): New type attributes.
20554         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20555         type attributes.
20556         * config/i386/btver2.md: New file describing btver2 pipelines.
20558 2013-01-19  Andrew Pinski  <apinski@cavium.com>
20560         PR tree-optimization/52631
20561         * tree-ssa-sccvn (visit_use): Before looking up the original
20562         statement, try looking up the simplified expression.
20564 2013-01-19  Anthony Green  <green@moxielogic.com>
20566         * config/moxie/moxie.c (moxie_expand_prologue): Set
20567         current_function_static_stack_size.
20569 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20571         PR tree-optimization/56029
20572         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20573         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20575 2013-01-18  Sharad Singhai  <singhai@google.com>
20577         PR tree-optimization/55995
20578         * dumpfile.c (dump_loc): Print location only if available.
20579         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
20581 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
20583         PR target/55433
20584         * lra-constraints.c (curr_insn_transform): Reuse original insn for
20585         secondary memory move.
20586         (inherit_reload_reg): Use rclass instead of cl for
20587         check_secondary_memory_needed_p.
20589 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20591         PR middle-end/56015
20592         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
20593         the case where writing real complex part of target modifies op1.
20595 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
20597         * config/aarch64/aarch64-simd.md
20598         (aarch64_vcond_internal<mode>): Handle unordered cases.
20599         * config/aarch64/iterators.md (v_cmp_result): New.
20601 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
20602             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20604         * config/arm/marvell-pj4.md: New file.
20605         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
20606         * config/arm/arm.md (generic_sched): Add marvell_pj4.
20607         (generic_vfp): Likewise.
20608         * config/arm/arm-cores.def: Add marvell-pj4.
20609         * config/arm/arm-tune.md: Regenerate.
20610         * config/arm/arm-tables.opt: Regenerate.
20611         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
20612         * doc/invoke.texi: Document marvell-pj4.
20614 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
20616         * config/aarch64/arm_neon.h: Map scalar types to standard types.
20618 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20620         PR debug/54114
20621         PR debug/54402
20622         PR debug/49888
20623         * var-tracking.c (negative_power_of_two_p): New.
20624         (global_get_addr_cache, local_get_addr_cache): New.
20625         (get_addr_from_global_cache, get_addr_from_local_cache): New.
20626         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
20627         heading comment.
20628         (vt_stack_offset_p): Remove.
20629         (vt_canon_true_dep): Always canonicalize loc's address.
20630         (clobber_overlapping_mems): Make sure we have a MEM.
20631         (local_get_addr_clear_given_value): New.
20632         (val_reset): Clear local cached entries.
20633         (compute_bb_dataflow): Create and release the local cache.
20634         Disable duplicate MEMs clobbering.
20635         (emit_notes_in_bb): Clobber MEMs likewise.
20636         (vt_emit_notes): Create and release the local cache.
20637         (vt_initialize, vt_finalize): Create and release the global
20638         cache, respectively.
20639         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
20641 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20643         PR libmudflap/53359
20644         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
20645         not found in the symtab.
20647 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20649         PR debug/56006
20650         PR rtl-optimization/55547
20651         PR rtl-optimization/53827
20652         PR debug/53671
20653         PR debug/49888
20654         * alias.c (offset_overlap_p): New, factored out of...
20655         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
20656         the conservative special case for symbolic constants.  Don't
20657         adjust zero sizes on alignment.
20659 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
20661         PR rtl-optimization/52573
20662         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
20663         REG_UNUSED for the same register.
20665 2013-01-17  Richard Biener  <rguenther@suse.de>
20666             Marek Polacek  <polacek@redhat.com>
20668         PR rtl-optimization/55833
20669         * loop-unswitch.c (unswitch_loops): Move loop verification...
20670         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
20671         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
20672         Set it to true when we're removing a loop from hierarchy tree in
20673         an irreducible region.
20674         (fix_bb_placements): Adjust caller.
20675         (fix_loop_placements): Likewise.
20677 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
20679         * config/avr/builtins.def (DEF_BUILTIN): Factor out
20680         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
20681         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
20682         Remove ID.  Adjust comments.
20683         * config/avr/avr-c.c (avr_builtin_name): Remove.
20684         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
20685         * config/avr/avr.c (avr_tolower): New static function.
20686         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
20687         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
20688         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
20689         default expansion.
20691 2013-01-17  Jan Hubicka  <jh@suse.cz>
20693         PR tree-optimization/55273
20694         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
20696 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
20698         PR target/55981
20699         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
20700         store through atomic_store<mode>_1.
20701         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
20703 2013-01-17  Martin Jambor  <mjambor@suse.cz>
20705         PR tree-optimizations/55264
20706         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
20707         for virtual methods.
20708         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
20709         virtual methods before inlining is over.
20710         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
20711         virtual functions.
20712         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
20713         non-virtual.
20715 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
20717         PR rtl-optimization/56005
20718         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
20719         pending reads for prefetch.
20721 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
20723         * config/aarch64/aarch64.md
20724         (*cstoresi_neg_uxtw): New pattern.
20725         (*cmovsi_insn_uxtw): New pattern.
20726         (*<optab>si3_uxtw): New pattern.
20727         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
20728         (*<optab>si3_insn_uxtw): New pattern.
20729         (*bswapsi2_uxtw): New pattern.
20731 2013-01-16  Richard Biener  <rguenther@suse.de>
20733         * tree-inline.c (tree_function_versioning): Remove set but
20734         never used variable.
20736 2013-01-16  Richard Biener  <rguenther@suse.de>
20738         PR tree-optimization/55964
20739         * tree-flow.h (rename_variables_in_loop): Remove.
20740         (rename_variables_in_bb): Likewise.
20741         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
20742         (copy_loop_before): Adjust and delete update-ssa status.
20743         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
20744         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
20745         (rename_variables_in_loop): Remove.
20746         (slpeel_update_phis_for_duplicate_loop): Likewise.
20747         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
20748         use available cfg machinery instead of duplicating it.
20749         Update PHI nodes and perform poor-mans SSA update here.
20750         (slpeel_tree_peel_loop_to_edge): Adjust.
20752 2013-01-16  Richard Biener  <rguenther@suse.de>
20754         PR tree-optimization/54767
20755         PR tree-optimization/53465
20756         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
20757         (vrp_visit_phi_node): For PHI arguments coming via backedges
20758         drop all symbolical range information.
20759         (execute_vrp): Compute backedges.
20761 2013-01-16  Richard Biener  <rguenther@suse.de>
20763         * doc/install.texi: Update CLooG and ISL requirements to
20764         0.18.0 and 0.11.1.
20766 2013-01-16  Christian Bruel  <christian.bruel@st.com>
20768         PR target/55301
20769         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
20770         (broken_move): Handle UNSPECV_SP_SWITCH_B.
20771         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
20773 2013-01-16  DJ Delorie  <dj@redhat.com>
20775         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
20776         (UNSPECV_SP_SWITCH_E): New.
20777         (sp_switch_1): Change to an unspec.
20778         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
20779         replace $r15.
20781 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
20783         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
20784         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
20785         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
20786         (expand_mem_thread_fence): Ditto.
20787         (expand_mem_signal_fence): Ditto.
20788         (expand_atomic_load): Ditto.
20789         (expand_atomic_store): Ditto.
20791 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
20793         PR rtl-optimization/55547
20794         PR rtl-optimization/53827
20795         PR debug/53671
20796         PR debug/49888
20797         * alias.c (memrefs_conflict_p): Set sizes to negative after
20798         AND adjustments.
20800 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20802         PR target/55940
20803         * function.c (thread_prologue_and_epilogue_insns): Always
20804         add crtl->drap_reg to set_up_by_prologue.set, even if
20805         stack_realign_drap is false.
20807 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
20809         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
20810         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
20811         *call): Fix indention.
20813 2013-01-15  Tom de Vries  <tom@codesourcery.com>
20815         PR target/55876
20816         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
20817         Update comment.
20819 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
20821         PR rtl-optimization/55153
20822         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
20824 2013-01-15  Martin Jambor  <mjambor@suse.cz>
20826         PR tree-optimization/55920
20827         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
20828         accesses as grp_to_be_debug_replaced.
20830 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20832         PR tree-optimization/55920
20833         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
20834         there is non-useless type conversion needed from debug rhs to lhs,
20835         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
20837 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
20838             Mikael Pettersson  <mikpe@it.uu.se>
20840         PR target/43961
20841         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
20842         Thumb.
20843         (ASM_OUTPUT_CASE_LABEL): Remove.
20844         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
20845         * final.c (shorten_branches): Update alignment of labels before
20846         jump tables if CASE_VECTOR_SHORTEN_MODE.
20848 2013-01-15  Richard Biener  <rguenther@suse.de>
20850         PR bootstrap/55961
20851         * system.h: Do not include gmp.h for building host tools.
20853 2013-01-15  Richard Biener  <rguenther@suse.de>
20855         PR middle-end/55882
20856         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
20857         account for bitpos when computing alignment.
20859 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
20861         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
20862         (ix86_target_macros_internal): Likewise.
20864         * config/i386/i386.c (m_CORE2I7): Removed.
20865         (m_CORE_HASWELL): New macro.
20866         (m_CORE_ALL): Likewise.
20867         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
20868         (initial_ix86_arch_features): Likewise.
20869         (processor_target_table): Initializations for Core avx2.
20870         (cpu_names): New names "core-avx2".
20871         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
20872         PROCESSOR_CORE_HASWELL.
20873         (ix86_issue_rate): New case.
20874         (ia32_multipass_dfa_lookahead): Likewise.
20875         (ix86_sched_init_global): Likewise.
20877         * config/i386/i386.h (TARGET_HASWELL): New macro.
20878         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
20879         (processor_type): New PROCESSOR_HASWELL.
20881 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20883         PR tree-optimization/55955
20884         * tree-vect-loop.c (vectorizable_reduction): Give up early on
20885         *SHIFT_EXPR and *ROTATE_EXPR codes.
20887         PR tree-optimization/48766
20888         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
20889         -ftrapv disable -fwrapv.
20891 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20893         PR target/55974
20894         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
20895         etc. to 1 and not to __flash.
20896         Use LL suffix for __INT24_MAX__ with -mint8.
20897         Use ULL suffix for __UINT24_MAX__ with -mint8.
20899 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20901         * config/avr/avr-arch.h
20902         (struct base_arch_s): Use typedef avr_arch_t instead.
20903         (struct arch_info_s): Use typedef avr_arch_info_t instead.
20904         (struct mcu_type_s): Use typedef avr_mcu_t instead.
20905         * config/avr/avr.c: Same.
20906         * config/avr/avr-devices.c: Same.
20907         * config/avr/driver-avr.c: Same.
20908         * config/avr/gen-avr-mmcu-texi.c: Same.
20909         * config/avr/avr-mcus.def: Adjust comment.
20911 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
20913         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
20914         * config/aarch64/iterators.md (VALLDI): New.
20916 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
20917             Andi Kleen  <ak@linux.intel.com>
20919         PR target/55948
20920         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
20921         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
20922         memmodel flag.
20924 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20926         * config/avr/avr-stdint.h: Remove trailing blanks.
20927         * config/avr/avr-log.h: Same.
20928         * config/avr/avr-arch.h: Same.
20929         * config/avr/avr-devices.c: Same.
20930         * config/avr/avr-dimode.md: Same.
20931         * config/avr/predicates.md: Same.
20932         * config/avr/avr-c.c: Same.  And fix typo.
20934         * config/avr/avr-protos.h: Same.  And:
20935         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
20936         (init_cumulative_args): Rename to avr_init_cumulative_args.
20937         (expand_prologue): Rename to avr_expand_prologue.
20938         (expand_epilogue): Rename to avr_expand_epilogue.
20939         (adjust_insn_length): Rename to avr_adjust_insn_length.
20940         (notice_update_cc): Rename to avr_notice_update_cc.
20941         (final_prescan_insn): Rename to avr_final_prescan_insn.
20942         * config/avr/avr.c: Same.
20943         * config/avr/avr.h: Same.
20944         * config/avr/avr.md: Remove trailing blanks.
20945         (prologue): Use avr_expand_prologue.
20946         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
20948 2013-01-14  Richard Biener  <rguenther@suse.de>
20950         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
20951         verify_location, collect_subblocks): New functions.
20952         (verify_gimple_in_cfg): Verify that locations only reference
20953         BLOCKs in the functions BLOCK tree.
20955 2013-01-14  Richard Biener  <rguenther@suse.de>
20957         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
20958         PHI argument.
20959         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
20960         unshare reference.
20961         (insert_out_of_ssa_copy_on_edge): Likewise.
20962         (rewrite_close_phi_out_of_ssa): Likewise.
20963         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
20964         debug expressions.
20965         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
20966         propagated constants.
20967         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
20968         can not be shared.
20970 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20972         * config/avr/avr-modes.def: Add GPL copyright notice.
20974 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
20976         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
20977         MEMMODEL_MASK to determine memory model.
20978         (atomic_store<mode>): Ditto from operands[2].
20979         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
20981 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
20983         PR fortran/55935
20984         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
20985         (fold_gimple_assign): Don't call unshare_expr here.
20986         (fold_ctor_reference): Call unshare_expr.
20988 2013-01-13  Terry Guo  <terry.guo@arm.com>
20990         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
20991         * doc/fragments.texi: Document MULTILIB_REUSE.
20992         * gcc.c (multilib_reuse): New internal spec.
20993         (set_multilib_dir): Also search multilib from multilib_reuse.
20994         * genmultilib (tmpmultilib3): Refactor code.
20995         (tmpmultilib4): Ditto.
20996         (multilib_reuse): New multilib argument.
20998 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
21000         * Makefile.in: Update copyright.
21002 2013-01-12  Tom de Vries  <tom@codesourcery.com>
21004         PR middle-end/55890
21005         * calls.c (expand_call): Check if arg_nr is valid.
21007 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
21009         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21010         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21011         documentation.  Add missing '__' in front of
21012         __builtin_ia32_packssdw256.
21014 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21016         PR target/55719
21017         * config/s390/s390.c (s390_preferred_reload_class): Do not return
21018         NO_REGS for larl operands.
21019         (s390_reload_larl_operand): Use s390_load_address instead of
21020         emit_move_insn.
21022 2013-01-11  Richard Biener  <rguenther@suse.de>
21024         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21025         (verify_node_sharing): ... here.
21026         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21028 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21030         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21031         Substitute TREECHECKING.
21032         * configure: Regenerate.
21033         * Makefile.in (TREECHECKING): New.
21035 2013-01-11  Richard Guenther  <rguenther@suse.de>
21037         PR tree-optimization/44061
21038         * tree-vrp.c (extract_range_basic): Compute zero as
21039         value-range for __builtin_constant_p of function parameters.
21041 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21043         Update copyright years.
21045 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21047         PR rtl-optimization/55672
21048         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21049         const to be eliminable.
21051 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21053         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21054         * configure: Regenerate.
21056 2013-01-10  Richard Biener  <rguenther@suse.de>
21058         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21060 2013-01-10  Richard Biener  <rguenther@suse.de>
21062         PR bootstrap/55792
21063         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21064         locations for virtual PHI arguments.
21065         (rewrite_update_phi_arguments): Likewise.
21067 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21069         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21070         on to assembler.
21072 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21074         PR tree-optimization/55921
21075         * tree-complex.c (expand_complex_asm): New function.
21076         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21078 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21080         PR target/55718
21081         * config/s390/s390.c (s390_symref_operand_p)
21082         (s390_loadrelative_operand_p): Merge the two functions.
21083         (s390_check_qrst_address, print_operand_address): Add parameters
21084         to s390_loadrelative_operand_p invokation.
21085         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21086         (s390_reload_larl_operand, s390_secondary_reload): Use
21087         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21088         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21090 2013-01-09  Mike Stump  <mikestump@comcast.net>
21092         * dse.c (record_store): Remove unnecessary assert.
21094 2013-01-09  Jan Hubicka  <jh@suse.cz>
21096         PR tree-optimization/55569
21097         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21098         * cfgloop.h (scale_loop_profile): Likewise.
21100 2013-01-09  Jan Hubicka  <jh@suse.cz>
21102         PR lto/45375
21103         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21104         functions.
21105         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21107 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21109         PR middle-end/55114
21110         * expr.h (maybe_emit_group_store): Declare.
21111         * expr.c (maybe_emit_group_store): New function.
21112         * builtins.c (expand_builtin_int_roundingfn): Call it.
21113         (expand_builtin_int_roundingfn_2): Likewise.
21115 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21117         PR rtl-optimization/55829
21118         * lra-constraints.c (match_reload): Add code for absent output.
21119         (curr_insn_transform): Add code for reloads of matched inputs
21120         without output.
21122 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21124         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21125         attribute of movddup insn to DF.
21126         (*vec_interleave_lowv2df): Ditto.
21127         (vec_dupv2df): Ditto.
21129 2013-01-09  Jan Hubicka  <jh@suse.cz>
21131         PR tree-optimiation/55875
21132         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21133         EVERY_ITERATION parameter.
21134         (number_of_iterations_exit): Check if exit is executed every iteration.
21135         (idx_infer_loop_bounds): Similarly here.
21136         (n_of_executions_at_most): Simplify
21137         to only test for cases where statement is dominated by the
21138         particular bound; handle correctly the "postdominance" test.
21139         (scev_probably_wraps_p): Use max loop iterations info
21140         as a global bound first.
21142 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21143             Nick Clifton  <nickc@redhat.com>
21145         * config/v850/v850.md (cbranchsf4): New pattern.
21146         (cstoresf4): New pattern.
21147         (cbranchdf4): New pattern.
21148         (cstoredf4): New pattern.
21149         (movsicc): Disallow floating point comparisons.
21150         (cmpsf_le_insn): Fix order of operators.
21151         (cmpsf_lt_insn): Likewise.
21152         (cmpsf_eq_insn): Likewise.
21153         (cmpdf_le_insn): Likewise.
21154         (cmpdf_lt_insn): Likewise.
21155         (cmpdf_eq_insn): Likewise.
21156         (cmpsf_ge_insn): Use LE comparison.
21157         (cmpdf_ge_insn): Likewise.
21158         (cmpsf_gt_insn): Use LT comparison.
21159         (cmpdf_gt_insn): Likewise.
21160         (cmpsf_ne_insn): Delete pattern.
21161         (cmpdf_ne_insn): Delete pattern.
21162         * config/v850/v850.c (v850_gen_float_compare): Use
21163         gen_cmpdf_eq_insn for NE comparison.
21164         (v850_float_z_comparison_operator)
21165         (v850_float_nz_comparison_operator): Move from here ...
21166         * config/v850/predicates.md: ... to here.  Move GT and GE
21167         comparisons into v850_float_z_comparison_operator.
21168         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21169         Delete prototype.
21170         (v850_float_nz_comparison_operator): Likewise.
21172 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21174         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21175         with calls to gen_insvsi/gen_insvdi.
21177 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21179         * config/i386/i386.c (initial_ix86_tune_features): Set up
21180         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21182 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21183             Jakub Jelinek  <jakub@redhat.com>
21185         PR tree-optimization/48189
21186         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21187         If nitercst is 0, don't predict the exit edge.
21189 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21191         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21192         in asm_fprintf with reg_names.
21193         (aarch64_print_operand_address): Likewise.
21194         (aarch64_return_addr): Likewise.
21195         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21197 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21199         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21200         (INT_U6_BITS): Likewise.
21201         * config/pa/predicates.md (uint6_operand): New predicate.
21202         (shift5_operand, shift6_operand): Likewise.
21203         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21204         arith32_operand.
21205         (lshrdi3): Use shift6_operand.
21206         (shrpsi4, shrpdi4): New insn patterns.
21207         (extzv): Delete expander.
21208         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21209         predicates in unamed zero extract patterns.  Tighten common constraint.
21210         (extv): Delete expander.
21211         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21212         predicates in unamed sign extract patterns.  Tighten common constraint.
21213         (insv): Delete expander.
21214         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21215         predicates in unamed insert patterns.  Tighten common constraint.
21216         Change uint32_operand predicate to uint6_operand predicate in unamed
21217         DImode pattern to insert constant values of type 1...1xxxx.
21219 2013-01-04  Jan Hubicka  <jh@suse.cz>
21221         PR tree-optimization/55823
21222         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21223         issue.
21225 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21226             Uros Bizjak  <ubizjak@gmail.com>
21228         PR rtl-optimization/55845
21229         * df-problems.c (can_move_insns_across): Stop scanning at
21230         volatile_insn_p source instruction or give up if
21231         across_from .. across_to range contains any volatile_insn_p
21232         instructions.
21234 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21236         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21237         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21238         Declare.
21239         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21240         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21242 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21244         PR fortran/55341
21245         * asan.c (asan_clear_shadow): New function.
21246         (asan_emit_stack_protection): Use it.
21248 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21250         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21251         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21252         with tab instead of space.
21254 2013-01-08  Nick Clifton  <nickc@redhat.com>
21256         * config/rl78/rl78.c (rl78_expand_prologue): Always select
21257         register bank 0 at the start of an interrupt handler.
21258         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21259         MDBH registers.
21261 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21263         * config/aarch64/aarch64-simd.md
21264         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21265         (aarch64_simd_bsl): Likewise.
21266         (aarch64_vcond_internal<mode>): Likewise.
21267         (vcond<mode><mode>): Likewise.
21268         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21269         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21271 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21273         * config/aarch64/aarch64-builtins.c
21274         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21276 2013-01-08  Martin Jambor  <mjambor@suse.cz>
21278         PR debug/55579
21279         * tree-sra.c (analyze_access_subtree): Return true also after
21280         potentially creating a debug-only replacement.
21282 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21284         PR middle-end/55890
21285         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21287         PR tree-optimization/54120
21288         * tree-vrp.c (range_fits_type_p): Don't allow
21289         src_precision < precision from signed vr to unsigned_p
21290         if vr->min or vr->max is negative.
21291         (simplify_float_conversion_using_ranges): Test can_float_p
21292         against CODE_FOR_nothing.
21294 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21295             Richard Biener  <rguenther@suse.de>
21297         PR middle-end/55851
21298         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21299         types instead of just INTEGER_TYPE types.
21301 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
21303         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21304         TF_SIZE): Define.
21306 2013-01-07  Steve Ellcey  <sellcey@mips.com>
21308         PR target/42661
21309         * config/mips/mips.opt: Change mad to mmad to match documentation.
21311 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21313         PR target/55897
21314         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21315         .progmemx.data now.
21317 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21319         PR target/55897
21320         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21321         (avr_addrspace_t): Add .section_name field.
21322         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21323         array size.
21324         (avr_addrspace): Same.  Initialize .section_name.  Remove last
21325         NULL entry.  Put __memx into .progmemx.data.
21326         (progmem_section_prefix): Remove.
21327         (avr_asm_init_sections): No need to initialize progmem_section.
21328         (avr_asm_named_section): Use avr_addrspace[].section_name to get
21329         section name prefix.
21330         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
21331         retrieve the progmem section.
21332         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21333         boundary to run over avr_addrspace[].
21334         (avr_register_target_pragmas): Ditto.
21336 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
21338         * varasm.c (output_constant_def_contents): For asan_protect_global
21339         protected strings, adjust DECL_ALIGN if needed, before testing for
21340         anchored symbols.
21341         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21342         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
21343         normal decls.
21344         (output_object_block): For asan protected decls, emit asan padding
21345         after their contents.
21346         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21347         (asan_finish_file): Test it here instead.
21349 2013-01-07  Nick Clifton  <nickc@redhat.com>
21350             Matthias Klose  <doko@debian.org>
21351             Doug Kwan  <dougkwan@google.com>
21352             H.J. Lu  <hongjiu.lu@intel.com>
21354         PR driver/55470
21355         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21357         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21359         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21361         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21363         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21365 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21367         PR target/54461
21368         * doc/install.texi (Cross-Compiler-Specific Options): Document
21369         --with-avrlibc.
21371 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
21373         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21374         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21375         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21376         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21377         vqmovun_high_s64): Fix source operand number and update copyright.
21379 2013-01-07  Richard Biener  <rguenther@suse.de>
21381         PR middle-end/55890
21382         * gimple.h (gimple_call_builtin_p): New overload.
21383         * gimple.c (validate_call): New function.
21384         (gimple_call_builtin_p): Likewise.
21385         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21386         Use gimple_call_builtin_p.
21387         (find_func_clobbers): Likewise.
21388         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21389         (strlen_optimize_stmt): Likewise.
21391 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21393         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21394         (vld1q_dup_*): Likewise.
21395         (vld1_*): Likewise.
21396         (vld1q_*): Likewise.
21397         (vld1_lane_*): Likewise.
21398         (vld1q_lane_*): Likewise.
21400 2013-01-07  Richard Biener  <rguenther@suse.de>
21402         * lto-streamer.h (LTO_minor_version): Bump to 2.
21404 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21406         * config/aarch64/aarch64-protos.h
21407         (aarch64_const_double_zero_rtx_p): Rename to...
21408         (aarch64_float_const_zero_rtx_p): ...this.
21409         (aarch64_float_const_representable_p): New.
21410         (aarch64_output_simd_mov_immediate): Likewise.
21411         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21412         move immediate case.
21413         * config/aarch64/aarch64.c
21414         (aarch64_const_double_zero_rtx_p): Rename to...
21415         (aarch64_float_const_zero_rtx_p): ...this.
21416         (aarch64_print_operand): Allow printing of new constants.
21417         (aarch64_valid_floating_const): New.
21418         (aarch64_legitimate_constant_p): Check for valid floating-point
21419         constants.
21420         (aarch64_simd_valid_immediate): Likewise.
21421         (aarch64_vect_float_const_representable_p): New.
21422         (aarch64_float_const_representable_p): Likewise.
21423         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21424         (aarch64_output_simd_mov_immediate): New.
21425         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21426         (*movdf_aarch64): Likewise.
21427         * config/aarch64/constraints.md (Ufc): New.
21428         (Y): call aarch64_float_const_zero_rtx.
21429         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21431 2013-01-07  Richard Biener  <rguenther@suse.de>
21433         PR tree-optimization/55888
21434         PR tree-optimization/55862
21435         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21436         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21437         not if it is contained therein.
21439 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21441         * config/avr/t-avr: Typo.
21443 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21445         PR55243
21446         * config/avr/t-avr: Don't automatically rebuild
21447         $(srcdir)/config/avr/t-multilib
21448         $(srcdir)/config/avr/avr-tables.opt
21449         $(srcdir)/doc/avr-mmcu.texi
21450         (avr-mcus): New phony target to build them on request.
21451         (s-avr-mlib, s-avr-mmcu-texi): Remove.
21452         * avr/avr-mcus.def: Adjust comments.
21454 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
21456         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21458 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21460         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21462 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21464         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21466 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
21468         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21469         to generate profiling.
21470         * config/rs6000/aix64.h (LIB_SPEC): Same.
21472 2013-01-04  Andrew Pinski  <apinski@cavium.com>
21474         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21475         New function.
21476         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21478 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
21480         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21481         unconditionally.
21482         (ix86_expand_move): Ditto.
21483         (ix86_zero_extend_to_Pmode): Ditto.
21484         (ix86_expand_call): Ditto.
21485         (ix86_expand_special_args_builtin): Ditto.
21486         (ix86_expand_builtin): Ditto.
21488 2013-01-04  Richard Biener  <rguenther@suse.de>
21490         PR tree-optimization/55862
21491         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21492         translating them through PHI nodes.
21494 2013-01-04  Martin Jambor  <mjambor@suse.cz>
21496         PR tree-optimization/55755
21497         * tree-sra.c (sra_modify_assign): Do not check that an access has no
21498         children when trying to avoid producing a VIEW_CONVERT_EXPR.
21500 2013-01-04  Marek Polacek  <polacek@redhat.com>
21502         PR middle-end/55859
21503         * opts.c (default_options_optimization): Clarify error message.
21505 2013-01-04  Richard Biener  <rguenther@suse.de>
21507         PR middle-end/55863
21508         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21509         reassociation.
21511 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21513         PR target/53789
21514         * config/pa/pa.md (movsi): Revert previous change.
21515         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21516         references.
21518 2013-01-03  Richard Henderson  <rth@redhat.com>
21520         * config/i386/i386.c (ix86_expand_move): Always assign to op1
21521         after eliminating TLS symbols.
21523 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
21525         PR bootstrap/50167
21526         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21527         * graphite-poly.c (debug_gmp_value): Likewise.
21529 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
21531         PR target/55712
21532         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21533         selected code model, define __code_mode_small__, __code_model_medium__,
21534         __code_model_large__, __code_model_32__ or __code_model_kernel__.
21535         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21536         xchg temporary register with %k.  Declare temporary register as
21537         early clobbered.
21538         [__x86_64__]: For medium and large code models, preserve %rbx register.
21540 2013-01-03  Richard Biener  <rguenther@suse.de>
21542         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21543         (dump_subscript): Adjust.
21544         (finalize_ddr_dependent): Do not dump redundant info.
21545         (analyze_siv_subscript): Adjust.
21546         (subscript_dependence_tester): Likewise.
21547         (compute_affine_dependence): Likewise.
21549 2013-01-03  Richard Biener  <rguenther@suse.de>
21551         Revert
21552         2013-01-03  Richard Biener  <rguenther@suse.de>
21554         PR tree-optimization/55857
21555         * tree-vect-stmts.c (vectorizable_load): Do not setup
21556         re-alignment for invariant loads.
21558         2013-01-02  Richard Biener  <rguenther@suse.de>
21560         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21561         invariant load do not generate a vector load from the scalar location.
21563 2013-01-03  Richard Biener  <rguenther@suse.de>
21565         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21566         for not vectorizing.
21567         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21568         not build INDIRECT_REFs, call get_name once only.
21569         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
21570         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21572 2013-01-03  Richard Biener  <rguenther@suse.de>
21574         PR tree-optimization/55857
21575         * tree-vect-stmts.c (vectorizable_load): Do not setup
21576         re-alignment for invariant loads.
21578 2013-01-03  Richard Biener  <rguenther@suse.de>
21580         PR lto/55848
21581         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
21582         prefer a built-in decl.
21584 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21586         * gcc.c (process_command): Update copyright notice dates.
21587         * gcov.c (print_version): Likewise.
21588         * gcov-dump.c (print_version): Likewise.
21590         PR rtl-optimization/55838
21591         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
21592         iv0.step, iv1.step and step.
21594 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21595             Marc Glisse  <marc.glisse@inria.fr>
21597         PR tree-optimization/55832
21598         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
21599         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
21600         integer_{one,zero}_node.
21602 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21604         PR debug/54402
21605         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
21606         * var-tracking.c (reverse_op): Don't add reverse ops to
21607         VALUEs that have already
21608         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
21610 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
21612         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
21614 2013-01-02  Teresa Johnson  <tejohnson@google.com>
21616         * dumpfile.c (dump_loc): Print filename with location.
21617         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
21618         new location_t parameter to emit complete unroll message with
21619         new dump framework.
21620         (canonicalize_loop_induction_variables): Compute loops location
21621         and pass to try_unroll_loop_completely.
21622         * loop-unroll.c (report_unroll_peel): New function.
21623         (peel_loops_completely): Use new dump format with location
21624         for main dumpfile message, and invoke report_unroll_peel on success.
21625         (decide_unrolling_and_peeling): Ditto.
21626         (decide_peel_once_rolling): Remove old dumpfile message subsumed
21627         by report_unroll_peel.
21628         (decide_peel_completely): Ditto.
21629         (decide_unroll_constant_iterations): Ditto.
21630         (decide_unroll_runtime_iterations): Ditto.
21631         (decide_peel_simple): Ditto.
21632         (decide_unroll_stupid): Ditto.
21633         * cfgloop.c (get_loop_location): New function.
21634         * cfgloop.h (get_loop_location): Declare.
21636 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21638         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
21639         NULL.
21641 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21643         PR middle-end/55198
21644         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
21645         BLKmode objects when EXPAND_MEMORY is specified.
21647 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21649         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
21650         in loop predicate.
21651         (fold_builtin_cpu): Do not share cpu model decls across statements.
21653 2013-01-02  Jason Merrill  <jason@redhat.com>
21655         PR c++/55804
21656         * tree.c (build_array_type_1): Revert earlier change.
21658 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
21660         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
21661         "cortex-a57".
21662         * config/aarch64/aarch64-tune.md: Re-generate.
21664 2013-01-02  Richard Biener  <rguenther@suse.de>
21666         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21667         invariant load do not generate a vector load from the scalar location.
21669 2013-01-02  Richard Biener  <rguenther@suse.de>
21671         PR bootstrap/55784
21672         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
21673         * configure: Regenerate.
21675 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21677         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
21678         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
21679         (expand_builtin_int_roundingfn_2): Keep the original target around
21680         for the fallback case.
21682 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21684         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
21685         to be clear for sign changes.
21687 2013-01-01  Jan Hubicka  <jh@suse.cz>
21689         * ipa-inline-analysis.c: Fix formatting.
21691 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
21693         PR tree-optimization/55831
21694         * tree-vect-loop.c (get_initial_def_for_induction): Use
21695         gsi_after_labels instead of gsi_start_bb.
21697 Copyright (C) 2013 Free Software Foundation, Inc.
21699 Copying and distribution of this file, with or without modification,
21700 are permitted in any medium without royalty provided the copyright
21701 notice and this notice are preserved.