2013-09-12 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob7a0d375fc33832ffb5daeea8b53b010a75f45ec4
1 2013-09-12  Richard Biener  <rguenther@suse.de>
3         PR tree-optimization/58404
4         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5         propagate non-invariant addresses into dereferences wrapped
6         in component references.
8 2013-09-12  Richard Biener  <rguenther@suse.de>
10         PR tree-optimization/58402
11         * passes.def: Move pass_late_warn_uninitialized later.
13 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
15         * tree-ssa.h: New.  Move content from tree-flow.h and
16         tree-flow-inline.h.
17         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
18         Move prototypes belonging to tree-ssa.c.
19         * tree-flow-inline.h (redirect_edge_var_map_def, 
20         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
21         tree-ssa.h.
22         * gimple.h: Adjust prototypes.
23         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move to...
24         * gimple.c (useless_type_conversion_p, types_compatible_p):  Here.
25         * tree.h: Move prototype to tree-ssa.h.
26         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
27         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
28         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
29         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
30         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
31         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
32         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
33         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
34         graphite-blocking.c, graphite-clast-to-gimple.c, graphite-dependences.c,
35         graphite-interchange.c, graphite-optimize-isl.c, graphite-poly.c,
36         graphite-scop-detection.c, graphite-sese-to-poly.c, graphite.c,
37         ipa-cp.c, ipa-inline-analysis.c, ipa-inline-transform.c, ipa-inline.c,
38         ipa-prop.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa-utils.c,
39         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
40         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
41         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
42         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
43         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
44         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
45         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
46         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
47         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
48         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
49         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
50         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
51         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
52         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
53         value-prof.c, var-tracking.c,
54         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
56 2013-09-12  Richard Biener  <rguenther@suse.de>
58         PR tree-optimization/58396
59         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
60         (build_rdg): Take a loop-nest parameter, fix memleaks.
61         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
63 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
65         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
66         for SLM.
68 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
70         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
71         The FMA instruction names should have a 'v' prefix.
73 2013-09-12  Richard Biener  <rguenther@suse.de>
75         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
76         (dot_rdg): Use popen instead of system in optional code.
77         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
78         (already_processed_vertex_p): Adjust.
79         (has_anti_or_output_dependence, predecessor_has_mem_write,
80         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
81         rdg_flag_uses): Remove.
82         (rdg_flag_vertex): Simplify.
83         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
84         remove recursion.
85         (build_rdg_partition_for_component): Process the first vertex
86         of a component only.
87         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
89 2013-09-12  Alan Modra  <amodra@gmail.com>
91         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
93 2013-09-11  DJ Delorie  <dj@redhat.com>
94             Nick Clifton  <nickc@redhat.com>
96         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
97         (rl78_stack_based_mem): New.
98         * config/rl78/constraints.md (Iv08): New.
99         (Iv16): New.
100         (Iv24): New.
101         (Is09): New.
102         (Is17): New.
103         (Is25): New.
104         (ISsi): New.
105         (IShi): New.
106         (ISqi): New.
107         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
108         (movhi): Likewise.
109         (movsi): Change from expand to insn-and-split.
110         (ashrsi3): Clobber AX.
111         (lshrsi3): New.
112         (ashlsi3): New.
113         (cbranchsi4): New.
114         * config/rl78/rl78.md (CC_REG): Fix.
115         (addsi3): Allow memory and immediate operands.
116         (addsi3_internal): Split into...
117         (addsi3_internal_virt): ...new, and ...
118         (addsi3_internal_real): ...new.
119         (subsi): New.
120         (subsi3_internal_virt): New.
121         (subsi3_internal_real): New.
122         (mulsi3): Add memory operand.
123         (mulsi3_rl78): Likewise.
124         (mulsi3_g13): Likewise.
125         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
126         (cbranchqi4_real): Add more constraint options.
127         (cbranchhi4_real): Expand pattern.
128         (cbranchhi4_real_signed): New.
129         (cbranchhi4_real_inverted): New.
130         (cbranchsi4_real_lt): New.
131         (cbranchsi4_real_ge): New.
132         (cbranchsi4_real_signed): New.
133         (cbranchsi4_real): New.
134         (peephole2): New.
135         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
136         constant shifts.
137         (lshrsi3_virt): Likewise.
138         (ashlsi3_virt): Likewise.
139         (cbranchqi4_virt_signed): New.
140         (cbranchhi4_virt_signed): New.
141         (cbranchsi4_virt): New.
142         * config/rl78/rl78.c: Whitespace fixes throughout.
143         (move_elim_pass): New.
144         (pass_data_rl78_move_elim): New.
145         (pass_rl78_move_elim): New.
146         (make_pass_rl78_move_elim): New.
147         (rl78_devirt_info): Run devirt earlier.
148         (rl78_move_elim_info): New.
149         (rl78_asm_file_start): Register it.
150         (rl78_split_movsi): New.
151         (rl78_as_legitimate_address): Allow virtual base registers when
152         appropriate.
153         (rl78_addr_space_convert): Remove spurious debug stuff.
154         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
155         (rl78_print_operand): More cases for not printing '#'.
156         (rl78_expand_compare): Remove most of the logic.
157         (content_memory): New.
158         (clear_content_memory): New.
159         (get_content_index): New.
160         (get_content_name): New.
161         (display_content_memory): New.
162         (update_content): New.
163         (record_content): New.
164         (already_contains): New.
165         (insn_ok_now): Re-recog insns with virtual registers.
166         (add_postponed_content_update): New.
167         (process_postponed_content_update): New.
168         (gen_and_emit_move): New.
169         (transcode_memory_rtx): Record new location content.
170         Use gen_and_emit_move.
171         (force_into_acc): New.
172         (move_to_acc): Use gen_and_emit_move.
173         (move_from_acc): Likewise.
174         (move_acc_to_reg): Likewise.
175         (move_to_x): Likewise.
176         (move_to_hl): Likewise.
177         (move_to_de): Likewise.
178         (rl78_alloc_physical_registers_op1): Record location content.
179         (has_constraint): New.
180         (rl78_alloc_physical_registers_op2): Record location content.
181         Optimize use of HL.
182         (rl78_alloc_physical_registers_ro1): Likewise.
183         (rl78_alloc_physical_registers_cmp): Likewise.
184         (rl78_alloc_physical_registers_umul): Likewise.
185         (rl78_alloc_address_registers_macax): New.
186         (rl78_alloc_physical_registers): Initialize and set location
187         content memory as needed.
188         (rl78_reorg): Make sure split2 is called.
189         (rl78_rtx_costs): New.
191 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
193         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
194         for (not (neg ...)) and (neg (not ...)) cases.
196 2013-09-11  Richard Biener  <rguenther@suse.de>
198         PR middle-end/58377
199         * passes.def: Split critical edges before late uninit warning passes.
200         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
202 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
204         PR tree-optimization/58385
205         * fold-const.c (build_range_check): If both low and high are NULL,
206         use omit_one_operand_loc to preserve exp side-effects.
208 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
210         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
212 2013-09-11  Richard Biener  <rguenther@suse.de>
214         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
215         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
216         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
217         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
218         stmts_from_loop, known_dependences_p, build_empty_rdg,
219         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
220         * tree-loop-distribution.c: ... here.
221         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
222         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
223         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
224         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
225         * tree-loop-distribution.c: ... here.
226         * tree-loop-distribution.c: Include gimple-pretty-print.h.
227         (struct partition_s): Add loops member.
228         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
229         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
230         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
232 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
233             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
234             Sergey Lega  <sergey.s.lega@intel.com>
235             Anna Tikhonova  <anna.tikhonova@intel.com>
236             Ilya Tocar  <ilya.tocar@intel.com>
237             Andrey Turetskiy  <andrey.turetskiy@intel.com>
238             Ilya Verbin  <ilya.verbin@intel.com>
239             Kirill Yukhin  <kirill.yukhin@intel.com>
240             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
242         * config/i386/constraints.md (k): New.
243         (Yk): Ditto.
244         * config/i386/i386.c (const regclass_map): Add new mask registers.
245         (dbx_register_map): Ditto.
246         (dbx64_register_map): Ditto.
247         (svr4_dbx_register_map): Ditto.
248         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
249         disabled.
250         (ix86_preferred_reload_class): Disable constants for mask registers.
251         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
252         (ix86_hard_regno_mode_ok): Support new mask registers.
253         (x86_order_regs_for_local_alloc): Ditto.
254         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
255         (FIXED_REGISTERS): Add new mask registers.
256         (CALL_USED_REGISTERS): Ditto.
257         (REG_ALLOC_ORDER): Ditto.
258         (VALID_MASK_REG_MODE): New.
259         (FIRST_MASK_REG): Ditto.
260         (LAST_MASK_REG): Ditto.
261         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
262         (MAYBE_MASK_CLASS_P): New.
263         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
264         (REG_CLASS_CONTENTS): Ditto.
265         (MASK_REGNO_P): New.
266         (ANY_MASK_REG_P): Ditto.
267         (HI_REGISTER_NAMES): Add new mask registers.
268         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
269         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
270         mask registers.
271         (attribute "type"): Add mskmov, msklog.
272         (attribute "length_immediate"): Support them.
273         (attribute "memory"): Ditto.
274         (attribute "prefix_0f"): Ditto.
275         (*movhi_internal): Support new mask registers.
276         (*movqi_internal): Ditto.
277         (define_split): Split out clobber pattern is a logic
278         insn on mask registers.
279         (*k<logic><mode>): New.
280         (*andhi_1): Extend to support mask regs.
281         (*andqi_1): Extend to support mask regs.
282         (kandn<mode>): New.
283         (define_split): Split and-not to and and not if operands
284         are not mask regs.
285         (*<code><mode>_1): Separate HI mode to new pattern...
286         (*<code>hi_1): This.
287         (*<code>qi_1): Extend to support mask regs.
288         (kxnor<mode>): New.
289         (kortestzhi): Ditto.
290         (kortestchi): Ditto.
291         (kunpckhi): Ditto.
292         (*one_cmpl<mode>2_1): Remove HImode and handle it...
293         (*one_cmplhi2_1): ...Here, now with mask registers support.
294         (*one_cmplqi2_1): Support new mask registers.
295         (HI/QImode arithmetics splitter): Don't split if mask registers
296         are used.
297         (HI/QImode not splitter): Ditto.
298         * config/i386/predicated.md (mask_reg_operand): New.
299         (general_reg_operand): Ditto.
301 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
303         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
304         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
306 2013-09-10  Jeff Law  <law@redhat.com>
308         PR tree-optimization/58380
309         * tree-ssa-threadupdate.c (thread_block): Recognize another case
310         of threading through a buried loop header.
312         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
313         return value for single successor case.
315 2013-09-10  Jan Hubicka  <jh@suse.cz>
317         * ipa-devirt.c (ipa_devirt): Enable with LTO.
319 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
321         PR target/58361
322         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
323         support conditional execution.
324         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
326 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
328         * lra.c (lra): Clear lra_optional_reload_pseudos before every
329         constraint pass.
330         * lra-constraints.c (curr_insn_transform): Switch on optional
331         reloads.  Check destination too to check move insn.
332         (undo_optional_reloads): Add check that the original peudo did not
333         changed its allocation and the optional reload was inherited on
334         last inheritance pass.  Break loop after deciding to keep optional
335         reload.
336         (lra_undo_inheritance): Add check that inherited pseudo still in
337         memory.
339 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
341         * config/aarch64/aarch64.md (generic_sched): New.
342         * config/aarch64/aarch64-generic.md (load): Make conditional
343         on generic_sched attribute.
344         (nonload): Likewise.
346 2013-09-10  Jan Hubicka  <jh@suse.cz>
348         * lto-cgraph.c: Include ipa-utils.h.
349         (compute_ltrans_boundary): Also add possible targets into the boundary.
351 2013-09-10  Jan Hubicka  <jh@suse.cz>
353         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
354         VAR_DECL of vtable rather than full expression.
356 2013-09-10  Jan Hubicka  <jh@suse.cz>
357             Paolo Carlini  <paolo.carlini@oracle.com>
359         * cgraphunit.c (analyze_functions): Save input_location, set it
360         to UNKNOWN_LOCATION and restore it at the end.
362 2013-09-10  Martin Jambor  <mjambor@suse.cz>
364         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
365         represented by a thunk.
367 2013-09-10  Jeff Law  <law@redhat.com>
369         PR tree-optimization/58343
370         * tree-ssa-threadupdate.c (thread_block): Identify and disable
371         jump threading requests through loop headers buried in the middle
372         of a jump threading path.
374         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
375         in return value/type.
377 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
379         PR rtl-optimization/58365
380         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
381         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
382         it differs.
384 2013-09-10  Richard Biener  <rguenther@suse.de>
386         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
387         * tree-data-ref.c (create_rdg_vertices): Collect all data
388         references, signal failure to the caller, use data-ref API.
389         (build_rdg): Compute data references only once.  Maintain lifetime
390         of data references and data dependences from within RDG.
391         (free_rdg): Free dependence relations.
392         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
393         inventing extra dependences.
394         (distribute_loop): Update for RDG API changes.
396 2013-09-10  Kai Tietz  <ktietz@redhat.com>
398         * doc/invoke.texi (fms-extensions): Document changed
399         behavior for ms-abi targets.
400         * config/i386/i386.c (ix86_option_override_internal):
401         Set default value of option -fms-extension for ms-abi targets.
403 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
405         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
407 2013-09-10  Alan Modra  <amodra@gmail.com>
409         PR target/58330
410         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
412 2013-09-10  Alan Modra  <amodra@gmail.com>
414         * config/rs6000/predicates.md (add_cint_operand): New.
415         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
416         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
417         using add_cint_operand.
418         (largetoc_high_plus_aix): Likewise.
420 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
422         PR tree-optimization/58364
423         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
424         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
425         the current range can't be an unconditional true or false.
427 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
429         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
431 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
433         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
435 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
437         PR c++/43452
438         * doc/invoke.texi (-Wdelete-incomplete): Document it.
440 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
442         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
443         NO_REGS for immediate that can't be moved directly into FP_REGS.
445 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
447         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
448         comparison with negated operand.
449         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
450         RTL form.
452 2013-09-09  Richard Biener  <rguenther@suse.de>
454         PR middle-end/58326
455         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
456         of a subloop record all its block as affecting loop-closed SSA form.
458 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
460         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
461         of an rtx/bitpos pair.
462         (store_fixed_bit_field): Update accordingly.
464 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
466         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
467         GEN_INT.
468         * builtins.c (expand_errno_check): Likewise.
469         * dwarf2cfi.c (init_return_column_size): Likewise.
470         * except.c (sjlj_mark_call_sites): Likewise.
471         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
472         * lra-constraints.c (emit_inc): Likewise.
473         * ree.c (combine_set_extension): Likewise.
474         * regmove.c (fixup_match_2): Likewise.
475         * reload1.c (inc_for_reload): Likewise.
477 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
479         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
480         (force_to_mode, simplify_shift_const_1, simplify_comparison):
481         Use gen_int_mode with the mode of the associated simplify_* call.
482         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
483         * expmed.c (expand_shift_1): Likewise.
484         * function.c (instantiate_virtual_regs_in_insn): Likewise.
485         * loop-iv.c (iv_number_of_iterations): Likewise.
486         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
487         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
489 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
491         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
492         of the associated expand_* call.
493         (asan_emit_stack_protection): Likewise.
494         * builtins.c (round_trampoline_addr): Likewise.
495         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
496         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
497         (emit_store_flag): Likewise.
498         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
499         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
500         Likewise.
501         * function.c (instantiate_virtual_regs_in_insn): Likewise.
502         * ifcvt.c (noce_try_store_flag_constants): Likewise.
503         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
504         * modulo-sched.c (generate_prolog_epilog): Likewise.
505         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
506         (expand_ctz, expand_ffs, expand_unop): Likewise.
508 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
510         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
511         of the associated gen_rtx_* call.
512         * caller-save.c (init_caller_save): Likewise.
513         * combine.c (find_split_point, make_extraction): Likewise.
514         (make_compound_operation): Likewise.
515         * dwarf2out.c (mem_loc_descriptor): Likewise.
516         * explow.c (plus_constant, probe_stack_range): Likewise.
517         * expmed.c (expand_mult_const): Likewise.
518         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
519         * reload1.c (init_reload): Likewise.
520         * valtrack.c (cleanup_auto_inc_dec): Likewise.
521         * var-tracking.c (adjust_mems): Likewise.
522         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
523         rather than gen_rtx_fmt_ee.
525 2013-09-09  Jan Hubicka  <jh@suse.cz>
527         PR middle-end/58294
528         * value-prof.c (gimple_ic): Copy also abnormal edges.
530 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
532         * asan.c (asan_shadow_cst): Use gen_int_mode.
534 2013-09-08  Jan Hubicka  <jh@suse.cz>
536         * ipa-profile.c: Add toplevel comment.
537         (ipa_propagate_frequency_1): Be more conservative when profile is read.
538         (contains_hot_call_p): New function.
539         (ipa_propagate_frequency): Set frequencies based on counts when
540         profile is read.
541         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
542         profile; do not tamper with profile after inlining if it is read.
544 2013-09-08  Jan Hubicka  <jh@suse.cz>
546         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
547         speculative edges.
549 2013-09-08  Jan Hubicka  <jh@suse.cz>
551         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
552         summary generation.
554 2013-09-08  Jeff Law  <law@redhat.com>
556         PR bootstrap/58340
557         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
558         of 'found'.
560 2013-09-08  Andi Kleen  <ak@linux.intel.com>
562         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
564 2013-09-08  Jan Hubicka  <jh@suse.cz>
566         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
568 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
570         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
571         for non-debug insns.
572         * lra.c (new_insn_reg): Take the containing insn as a parameter.
573         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
574         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
575         accordingly.
577 2013-09-08  Jan Hubicka  <jh@suse.cz>
579         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
580         targets and devirtualize to BUILT_IN_UNREACHABLE.
581         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
582         * ipa.c (walk_polymorphic_call_targets): New function.
583         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
584         functions; use the new timevar.
585         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
586         was removed from the program.
587         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
588         not consider it in the walk when its vtable is dead.
589         (possible_polymorphic_call_targets_1): Pass anonymous flag to
590         record_binfo.
591         (devirt_variable_node_removal_hook): New function.
592         (possible_polymorphic_call_targets): Also register
593         devirt_variable_node_removal_hook.
594         (ipa_devirt): Do not do non-speculative devirtualization.
595         (gate_ipa_devirt): One execute if devirtualizing speculatively.
597 2013-09-08  Jan Hubicka  <jh@suse.cz>
599         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
600         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
601         varpool_remove_variable_insertion_hook): Declare.
602         * varpool.c (varpool_node_hook_list): New structure.
603         (first_varpool_node_removal_hook,
604         first_varpool_variable_insertion_hook): New variables.
605         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
606         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
607         varpool_remove_variable_insertion_hook,
608         varpool_call_variable_insertion_hooks): New functions.
609         (varpool_remove_node): Use it.
611 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
613         PR c++/54941
614         * diagnostic.c (diagnostic_build_prefix): When s.file is
615         "<built-in>" don't output line and column numbers.
617 2013-09-06  Jan Hubicka  <jh@suse.cz>
619         * cgraphunit.c (expand_thunk): Get body before touching arguments.
620         * lto-streamer-out.c: Stream thunks, too.
621         * lto-streamer-in.c (input_function): Pop cfun here
622         (lto_read_body): Instead of here.
624 2013-09-06  Caroline Tice  <cmtice@google.com>
626         * doc/install.texi: Add documentation for the --enable-vtable-verify
627         and the --disable-libvtv configure options.
629 2013-09-06  Jeff Law  <law@redhat.com>
631         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
632         edge implied equivalences into successor phis.
634 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
636         * resource.c (mark_referenced_resources): Handle COND_EXEC.
638 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
640         * resource.c (mark_target_live_regs): Compute resources taking
641         into account if a call is predicated or not.
643 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
645         * toplev.c (output_stack_usage): Be prepared for suffixes created by
646         the compiler in the function names.
648 2013-09-06  Jan Hubicka  <jh@suse.cz>
650         PR middle-end/58094
651         * ipa-inline.c (has_caller_p): New function.
652         (want_inline_function_to_all_callers_p): Use it.
653         (sum_callers, inline_to_all_callers): Break out from ...
654         (ipa_inline): ... here.
656 2013-09-06  Jan Hubicka  <jh@suse.cz>
658         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
659         only when AVX is enabled.
661 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
663         * config/aarch64/aarch64.md
664         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
665         is fpsimd_<load/store>2.
666         (load_pair<mode>): Likewise.
667         (store_pair<mode>): Likewise.
669 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
671         * config/arm/types.md (type): Add "mrs" type.
672         * config/aarch64/aarch64.md
673         (aarch64_load_tp_hard): Make type "mrs".
674         * config/arm/arm.md
675         (load_tp_hard): Make type "mrs".
676         * config/arm/cortex-a15.md: Update with new attributes.
677         * config/arm/cortex-a5.md: Update with new attributes.
678         * config/arm/cortex-a53.md: Update with new attributes.
679         * config/arm/cortex-a7.md: Update with new attributes.
680         * config/arm/cortex-a8.md: Update with new attributes.
681         * config/arm/cortex-a9.md: Update with new attributes.
682         * config/arm/cortex-m4.md: Update with new attributes.
683         * config/arm/cortex-r4.md: Update with new attributes.
684         * config/arm/fa526.md: Update with new attributes.
685         * config/arm/fa606te.md: Update with new attributes.
686         * config/arm/fa626te.md: Update with new attributes.
687         * config/arm/fa726te.md: Update with new attributes.
689 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
691         * config/aarch64/aarch64.md
692         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
693         (*movtf_aarch64): Likewise.
694         * config/arm/arm.md
695         (thumb1_movdi_insn): Use "multiple" for type where more than one
696         instruction is used for a move.
697         (*arm32_movhf): Likewise.
698         (*thumb_movdf_insn): Likewise.
700 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
702         * config/arm/types.md (type): Rename fcpys to fmov.
703         * config/arm/vfp.md
704         (*arm_movsi_vfp): Rename type fcpys as fmov.
705         (*thumb2_movsi_vfp): Likewise
706         (*movhf_vfp_neon): Likewise
707         (*movhf_vfp): Likewise
708         (*movsf_vfp): Likewise
709         (*thumb2_movsf_vfp): Likewise
710         (*movsfcc_vfp): Likewise
711         (*thumb2_movsfcc_vfp): Likewise
712         * config/aarch64/aarch64-simd.md
713         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
714         * config/aarch64/aarch64.md
715         (*movsi_aarch64): Replace type mov_reg with fmovs.
716         (*movdi_aarch64): Likewise
717         (*movsf_aarch64): Likewise
718         (*movdf_aarch64): Likewise
719         * config/arm/arm.c
720         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
721         * config/arm/iwmmxt.md
722         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
723         * config/arm/arm1020e.md: Update with new attributes.
724         * config/arm/cortex-a15-neon.md: Update with new attributes.
725         * config/arm/cortex-a5.md: Update with new attributes.
726         * config/arm/cortex-a53.md: Update with new attributes.
727         * config/arm/cortex-a7.md: Update with new attributes.
728         * config/arm/cortex-a8-neon.md: Update with new attributes.
729         * config/arm/cortex-a9.md: Update with new attributes.
730         * config/arm/cortex-m4-fpu.md: Update with new attributes.
731         * config/arm/cortex-r4f.md: Update with new attributes.
732         * config/arm/marvell-pj4.md: Update with new attributes.
733         * config/arm/vfp11.md: Update with new attributes.
735 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
737         * config/aarch64/aarch64.md
738         (*madd<mode>): Fix type attribute.
739         (*maddsi_uxtw): Likewise.
740         (*msub<mode>): Likewise.
741         (*msubsi_uxtw): Likewise.
742         (<su_optab>maddsidi4): Likewise.
743         (<su_optab>msubsidi4): Likewise.
745 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
747         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
748         * config/arm/arm.md (core_cycles): Remove fdiv.
749         * config/arm/vfp.md:
750         (*sqrtsf2_vfp): Update for attribute changes.
751         (*sqrtdf2_vfp): Likewise.
752         * config/aarch64/aarch64.md:
753         (sqrt<mode>2): Update for attribute changes.
754         * config/arm/arm1020e.md: Update with new attributes.
755         * config/arm/cortex-a15-neon.md: Update with new attributes.
756         * config/arm/cortex-a5.md: Update with new attributes.
757         * config/arm/cortex-a53.md: Update with new attributes.
758         * config/arm/cortex-a7.md: Update with new attributes.
759         * config/arm/cortex-a8-neon.md: Update with new attributes.
760         * config/arm/cortex-a9.md: Update with new attributes.
761         * config/arm/cortex-m4-fpu.md: Update with new attributes.
762         * config/arm/cortex-r4f.md: Update with new attributes.
763         * config/arm/marvell-pj4.md: Update with new attributes.
764         * config/arm/vfp11.md: Update with new attributes.
766 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
768         * config/arm/types.md
769         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
770         * config/aarch64/aarch64.md
771         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
772         new attributes.
773         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
774         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
775         (float<GPI:mode><GPF:mode>2): Likewise.
776         * config/arm/vfp.md
777         (*truncsisf2_vfp): Update with new attributes.
778         (*truncsidf2_vfp): Likewise.
779         (fixuns_truncsfsi2): Likewise.
780         (fixuns_truncdfsi2): Likewise.
781         (*floatsisf2_vfp): Likewise.
782         (*floatsidf2_vfp): Likewise.
783         (floatunssisf2): Likewise.
784         (floatunssidf2): Likewise.
785         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
786         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
787         * config/arm/arm1020e.md: Update with new attributes.
788         * config/arm/cortex-a15-neon.md: Update with new attributes.
789         * config/arm/cortex-a5.md: Update with new attributes.
790         * config/arm/cortex-a53.md: Update with new attributes.
791         * config/arm/cortex-a7.md: Update with new attributes.
792         * config/arm/cortex-a8-neon.md: Update with new attributes.
793         * config/arm/cortex-a9.md: Update with new attributes.
794         * config/arm/cortex-m4-fpu.md: Update with new attributes.
795         * config/arm/cortex-r4f.md: Update with new attributes.
796         * config/arm/marvell-pj4.md: Update with new attributes.
797         * config/arm/vfp11.md: Update with new attributes.
799 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
801         * config/aarch64/arm_neon.h
802         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
803         (vqtbx<1,2,3,4><q>_s8): Likewise.
805 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
807         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
808         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
809         (add<mode>3): Add type attribute.
810         (add<mode>3): Likewise.
811         (usadd<mode>3): Likewise.
812         (ssadd<mode>3): Likewise.
813         (sub<mode>3): Likewise.
814         (sub<mode>3): Likewise.
815         (ussub<mode>3): Likewise.
816         (sssub<mode>3): Likewise.
817         (ssmulsa3): Likewise.
818         (usmulusa3): Likewise.
819         (arm_usatsihi): Likewise.
820         * config/arm/vfp.md
821         (*movdi_vfp): Add types for all instructions.
822         (*movdi_vfp_cortexa8): Likewise.
823         (*movhf_vfp_neon): Likewise.
824         (*movhf_vfp): Likewise.
825         (*movdf_vfp): Likewise.
826         (*thumb2_movdf_vfp): Likewise.
827         (*thumb2_movdfcc_vfp): Likewise.
828         * config/arm/arm.md: Add type attribute to all insn patterns.
829         (*thumb1_adddi3): Add type attribute.
830         (*arm_adddi3): Likewise.
831         (*adddi_sesidi_di): Likewise.
832         (*adddi_zesidi_di): Likewise.
833         (*thumb1_addsi3): Likewise.
834         (addsi3_compare0): Likewise.
835         (*addsi3_compare0_scratch): Likewise.
836         (*compare_negsi_si): Likewise.
837         (cmpsi2_addneg): Likewise.
838         (*addsi3_carryin_<optab>): Likewise.
839         (*addsi3_carryin_alt2_<optab>): Likewise.
840         (*addsi3_carryin_clobercc_<optab>): Likewise.
841         (*subsi3_carryin): Likewise.
842         (*subsi3_carryin_const): Likewise.
843         (*subsi3_carryin_compare): Likewise.
844         (*subsi3_carryin_compare_const): Likewise.
845         (*arm_subdi3): Likewise.
846         (*thumb_subdi3): Likewise.
847         (*subdi_di_zesidi): Likewise.
848         (*subdi_di_sesidi): Likewise.
849         (*subdi_zesidi_di): Likewise.
850         (*subdi_sesidi_di): Likewise.
851         (*subdi_zesidi_ze): Likewise.
852         (thumb1_subsi3_insn): Likewise.
853         (*arm_subsi3_insn): Likewise.
854         (*anddi3_insn): Likewise.
855         (*anddi_zesidi_di): Likewise.
856         (*anddi_sesdi_di): Likewise.
857         (*ne_zeroextracts): Likewise.
858         (*ne_zeroextracts): Likewise.
859         (*ite_ne_zeroextr): Likewise.
860         (*ite_ne_zeroextr): Likewise.
861         (*anddi_notdi_di): Likewise.
862         (*anddi_notzesidi): Likewise.
863         (*anddi_notsesidi): Likewise.
864         (andsi_notsi_si): Likewise.
865         (thumb1_bicsi3): Likewise.
866         (*iordi3_insn): Likewise.
867         (*iordi_zesidi_di): Likewise.
868         (*iordi_sesidi_di): Likewise.
869         (*thumb1_iorsi3_insn): Likewise.
870         (*xordi3_insn): Likewise.
871         (*xordi_zesidi_di): Likewise.
872         (*xordi_sesidi_di): Likewise.
873         (*arm_xorsi3): Likewise.
874         (*andsi_iorsi3_no): Likewise.
875         (*smax_0): Likewise.
876         (*smax_m1): Likewise.
877         (*arm_smax_insn): Likewise.
878         (*smin_0): Likewise.
879         (*arm_smin_insn): Likewise.
880         (*arm_umaxsi3): Likewise.
881         (*arm_uminsi3): Likewise.
882         (*minmax_arithsi): Likewise.
883         (*minmax_arithsi_): Likewise.
884         (*satsi_<SAT:code>): Likewise.
885         (arm_ashldi3_1bit): Likewise.
886         (arm_ashrdi3_1bit): Likewise.
887         (arm_lshrdi3_1bit): Likewise.
888         (*arm_negdi2): Likewise.
889         (*thumb1_negdi2): Likewise.
890         (*arm_negsi2): Likewise.
891         (*thumb1_negsi2): Likewise.
892         (*negdi_extendsid): Likewise.
893         (*negdi_zero_extend): Likewise.
894         (*arm_abssi2): Likewise.
895         (*thumb1_abssi2): Likewise.
896         (*arm_neg_abssi2): Likewise.
897         (*thumb1_neg_abss): Likewise.
898         (one_cmpldi2): Likewise.
899         (extend<mode>di2): Likewise.
900         (*compareqi_eq0): Likewise.
901         (*arm_extendhisi2addsi): Likewise.
902         (*arm_movdi): Likewise.
903         (*thumb1_movdi_insn): Likewise.
904         (*arm_movt): Likewise.
905         (*thumb1_movsi_insn): Likewise.
906         (pic_add_dot_plus_four): Likewise.
907         (pic_add_dot_plus_eight): Likewise.
908         (tls_load_dot_plus_eight): Likewise.
909         (*thumb1_movhi_insn): Likewise.
910         (*thumb1_movsf_insn): Likewise.
911         (*movdf_soft_insn): Likewise.
912         (*thumb_movdf_insn): Likewise.
913         (cbranchsi4_insn): Likewise.
914         (cbranchsi4_scratch): Likewise.
915         (*negated_cbranchsi4): Likewise.
916         (*tbit_cbranch): Likewise.
917         (*tlobits_cbranch): Likewise.
918         (*tstsi3_cbranch): Likewise.
919         (*cbranchne_decr1): Likewise.
920         (*addsi3_cbranch): Likewise.
921         (*addsi3_cbranch_scratch): Likewise.
922         (*arm_cmpdi_insn): Likewise.
923         (*arm_cmpdi_unsig): Likewise.
924         (*arm_cmpdi_zero): Likewise.
925         (*thumb_cmpdi_zero): Likewise.
926         (*deleted_compare): Likewise.
927         (*mov_scc): Likewise.
928         (*mov_negscc): Likewise.
929         (*mov_notscc): Likewise.
930         (*cstoresi_eq0_thumb1_insn): Likewise.
931         (cstoresi_nltu_thumb1): Likewise.
932         (cstoresi_ltu_thu): Likewise.
933         (thumb1_addsi3_addgeu): Likewise.
934         (*arm_jump): Likewise.
935         (*thumb_jump): Likewise.
936         (*check_arch2): Likewise.
937         (arm_casesi_internal): Likewise.
938         (thumb1_casesi_dispatch): Likewise.
939         (*arm_indirect_jump): Likewise.
940         (*thumb1_indirect_jump): Likewise.
941         (nop): Likewise.
942         (*and_scc): Likewise.
943         (*ior_scc): Likewise.
944         (*compare_scc): Likewise.
945         (*cond_move): Likewise.
946         (*cond_arith): Likewise.
947         (*cond_sub): Likewise.
948         (*cmp_ite0): Likewise.
949         (*cmp_ite1): Likewise.
950         (*cmp_and): Likewise.
951         (*cmp_ior): Likewise.
952         (*ior_scc_scc): Likewise.
953         (*ior_scc_scc_cmp): Likewise.
954         (*and_scc_scc): Likewise.
955         (*and_scc_scc_cmp): Likewise.
956         (*and_scc_scc_nod): Likewise.
957         (*negscc): Likewise.
958         (movcond_addsi): Likewise.
959         (movcond): Likewise.
960         (*ifcompare_plus_move): Likewise.
961         (*if_plus_move): Likewise.
962         (*ifcompare_move_plus): Likewise.
963         (*if_move_plus): Likewise.
964         (*ifcompare_arith_arith): Likewise.
965         (*if_arith_arith): Likewise.
966         (*ifcompare_arith_move): Likewise.
967         (*if_arith_move): Likewise.
968         (*ifcompare_move_arith): Likewise.
969         (*if_move_arith): Likewise.
970         (*ifcompare_move_not): Likewise.
971         (*if_move_not): Likewise.
972         (*ifcompare_not_move): Likewise.
973         (*if_not_move): Likewise.
974         (*ifcompare_shift_move): Likewise.
975         (*if_shift_move): Likewise.
976         (*ifcompare_move_shift): Likewise.
977         (*if_move_shift): Likewise.
978         (*ifcompare_shift_shift): Likewise.
979         (*ifcompare_not_arith): Likewise.
980         (*ifcompare_arith_not): Likewise.
981         (*if_arith_not): Likewise.
982         (*ifcompare_neg_move): Likewise.
983         (*if_neg_move): Likewise.
984         (*ifcompare_move_neg): Likewise.
985         (*if_move_neg): Likewise.
986         (prologue_thumb1_interwork): Likewise.
987         (*cond_move_not): Likewise.
988         (*sign_extract_onebit): Likewise.
989         (*not_signextract_onebit): Likewise.
990         (stack_tie): Likewise.
991         (align_4): Likewise.
992         (align_8): Likewise.
993         (consttable_end): Likewise.
994         (consttable_1): Likewise.
995         (consttable_2): Likewise.
996         (consttable_4): Likewise.
997         (consttable_8): Likewise.
998         (consttable_16): Likewise.
999         (*thumb1_tablejump): Likewise.
1000         (prefetch): Likewise.
1001         (force_register_use): Likewise.
1002         (thumb_eh_return): Likewise.
1003         (load_tp_hard): Likewise.
1004         (load_tp_soft): Likewise.
1005         (tlscall): Likewise.
1006         (*arm_movtas_ze): Likewise.
1007         (*arm_rev): Likewise.
1008         (*arm_revsh): Likewise.
1009         (*arm_rev16): Likewise.
1010         * config/arm/thumb2.md
1011         (*thumb2_smaxsi3): Likewise.
1012         (*thumb2_sminsi3): Likewise.
1013         (*thumb32_umaxsi3): Likewise.
1014         (*thumb2_uminsi3): Likewise.
1015         (*thumb2_negdi2): Likewise.
1016         (*thumb2_abssi2): Likewise.
1017         (*thumb2_neg_abss): Likewise.
1018         (*thumb2_movsi_insn): Likewise.
1019         (tls_load_dot_plus_four): Likewise.
1020         (*thumb2_movhi_insn): Likewise.
1021         (*thumb2_mov_scc): Likewise.
1022         (*thumb2_mov_negs): Likewise.
1023         (*thumb2_mov_negs): Likewise.
1024         (*thumb2_mov_nots): Likewise.
1025         (*thumb2_mov_nots): Likewise.
1026         (*thumb2_movsicc_): Likewise.
1027         (*thumb2_movsfcc_soft_insn): Likewise.
1028         (*thumb2_indirect_jump): Likewise.
1029         (*thumb2_and_scc): Likewise.
1030         (*thumb2_ior_scc): Likewise.
1031         (*thumb2_ior_scc_strict_it): Likewise.
1032         (*thumb2_cond_move): Likewise.
1033         (*thumb2_cond_arith): Likewise.
1034         (*thumb2_cond_ari): Likewise.
1035         (*thumb2_cond_sub): Likewise.
1036         (*thumb2_negscc): Likewise.
1037         (*thumb2_movcond): Likewise.
1038         (thumb2_casesi_internal): Likewise.
1039         (thumb2_casesi_internal_pic): Likewise.
1040         (*thumb2_alusi3_short): Likewise.
1041         (*thumb2_mov<mode>_shortim): Likewise.
1042         (*thumb2_addsi_short): Likewise.
1043         (*thumb2_subsi_short): Likewise.
1044         (thumb2_addsi3_compare0): Likewise.
1045         (*thumb2_cbz): Likewise.
1046         (*thumb2_cbnz): Likewise.
1047         (*thumb2_one_cmplsi2_short): Likewise.
1048         (*thumb2_negsi2_short): Likewise.
1049         (*orsi_notsi_si): Likewise.
1050         * config/arm/arm1020e.md: Update with new attributes.
1051         * config/arm/arm1026ejs.md: Update with new attributes.
1052         * config/arm/arm1136jfs.md: Update with new attributes.
1053         * config/arm/arm926ejs.md: Update with new attributes.
1054         * config/arm/cortex-a15.md: Update with new attributes.
1055         * config/arm/cortex-a5.md: Update with new attributes.
1056         * config/arm/cortex-a53.md: Update with new attributes.
1057         * config/arm/cortex-a7.md: Update with new attributes.
1058         * config/arm/cortex-a8.md: Update with new attributes.
1059         * config/arm/cortex-a9.md: Update with new attributes.
1060         * config/arm/cortex-m4.md: Update with new attributes.
1061         * config/arm/cortex-r4.md: Update with new attributes.
1062         * config/arm/fa526.md: Update with new attributes.
1063         * config/arm/fa606te.md: Update with new attributes.
1064         * config/arm/fa626te.md: Update with new attributes.
1065         * config/arm/fa726te.md: Update with new attributes.
1067 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1069         * config/aarch64/aarch64-simd.md
1070         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1071         <vwx> iterator to ensure correct register choice.
1072         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1073         (aarch64_sqdmull_n<mode>): Likewise.
1074         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1075         * config/aarch64/arm_neon.h
1076         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1077         (vml<as><q>_n_<su>16): Likewise.
1078         (vml<as>l_high_lane<q>_<su>16): Likewise.
1079         (vml<as>l_high_n_<su>16): Likewise.
1080         (vml<as>l_lane<q>_<su>16): Likewise.
1081         (vml<as>l_n_<su>16): Likewise.
1082         (vmul<q>_lane<q>_<su>16): Likewise.
1083         (vmul<q>_n_<su>16): Likewise.
1084         (vmull_lane<q>_<su>16): Likewise.
1085         (vmull_n_<su>16): Likewise.
1086         (vmull_high_lane<q>_<su>16): Likewise.
1087         (vmull_high_n_<su>16): Likewise.
1088         (vqrdmulh<q>_n_s16): Likewise.
1090 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1092         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1093         have the correct lane parameter.
1095 2013-09-06  Richard Biener <rguenther@suse.de>
1097         * cfganal.c (control_dependences::~control_dependences):
1098         Properly free all of the vector.
1100 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1102         PR target/58269
1103         * config/i386/i386.c (ix86_conditional_register_usage):
1104         Proper initialize extended SSE registers.
1106 2013-09-06  Jan Hubicka  <jh@suse.cz>
1108         PR tree-optimization/58311
1109         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1111 2013-09-06  Jan Hubicka  <jh@suse.cz>
1113         * Makefile.in (tree-sra.o): Update dependencies.
1114         * tree-sra.c: Include ipa-utils.h
1115         (scan_function): Use recursive_call_p.
1116         (has_caller_p): New function.
1117         (cgraph_for_node_and_aliases): Count also callers of aliases.
1119 2013-09-06  Jan Hubicka  <jh@suse.cz>
1121         PR middle-end/58094
1122         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1123         * tree-tailcall.c: Include ipa-utils.h.
1124         (find_tail_calls): Use it.
1125         * ipa-pure-const.c (check_call): Likewise.
1126         * ipa-utils.c (recursive_call_p): New function.
1127         * ipa-utils.h (recursive_call_p): Dclare.
1128         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1129         (symtab_semantically_equivalent_p): New function.
1130         * Makefile.in (tree-tailcall.o): Update dependencies.
1132 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1134         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1135         non-inlinable part.
1137 2013-09-06  Richard Biener  <rguenther@suse.de>
1139         * lto-streamer.h (lto_global_var_decls): Remove.
1140         * Makefile.in (OBJS): Remove lto-symtab.o.
1141         (lto-symtab.o): Remove.
1142         (GTFILES): Remove lto-symtab.c
1143         * lto-symtab.c: Move to lto/
1145 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1147         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1148         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1149         (UNSPEC_FPINT_RINT): New constant definitions.
1150         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1151         definition with 2 attributes.
1152         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1153         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1154         definitions.
1156 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1158         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1159         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1160         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1161         attribute to "z196_alone".
1162         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1163         "zEC12_simple".
1165 2013-09-06  Richard Biener  <rguenther@suse.de>
1167         * basic-block.h (class control_dependences): New.
1168         * tree-ssa-dce.c (control_dependence_map): Remove.
1169         (cd): New global.
1170         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1171         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1172         find_pdom, find_control_dependence, find_all_control_dependences):
1173         Move to cfganal.c.
1174         (mark_control_dependent_edges_necessary,
1175         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1176         tree_dce_done, perform_tree_ssa_dce): Adjust.
1177         * cfganal.c (set_control_dependence_map_bit,
1178         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1179         find_all_control_dependences): Move from tree-ssa-dce.c and
1180         implement as methods of control_dependences class.
1181         (control_dependences::control_dependences): New.
1182         (control_dependences::~control_dependences): Likewise.
1183         (control_dependences::get_edges_dependent_on): Likewise.
1184         (control_dependences::get_edge): Likewise.
1186 2013-09-04  Jan Hubicka  <jh@suse.cz>
1188         * tree.c (types_same_for_odr): Drop overactive check.
1189         * ipa-devirt.c (hash_type_name): Likewise.
1191 2013-09-04  Jan Hubicka  <jh@suse.cz>
1193         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1194         (analyze_functions): ... here.
1196 2013-09-04  Jan Hubicka  <jh@suse.cz>
1198         PR middle-end/58201
1199         * cgraphunit.c (analyze_functions): Clear AUX fields
1200         after processing; initialize assembler name has.
1202 2013-09-05  Jeff Law  <law@redhat.com>
1204         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1205         from thread_around_empty_block.  Record threading path into PATH.
1206         Recurse if threading through the initial block is successful.
1207         (thread_across_edge): Corresponding changes to slightly simplify.
1209 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1211         * config/aarch64/aarch64.md
1212         (type): Remove frecpe, frecps, frecpx.
1213         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1214         fix to be a TARGET_SIMD instruction.
1215         (aarch64_frecps): Remove.
1216         * config/aarch64/aarch64-simd.md
1217         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1218         (aarch64_frecps<mode>): Handle all float/vector of float modes.
1220 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1221             Sofiane Naci  <sofiane.naci@arm.com>
1223         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1224         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1225         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1226         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1227         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
1228         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1229         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
1230         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
1231         into "clz, "rbit".  Rename "shift" to "shift_imm".
1232         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1233         changes.  Update for attribute changes all occurrences of arlo_* and
1234         shift* types.
1235         * config/arm/arm-fixed.md: Update for attribute changes
1236         all occurrences of arlo_* types.
1237         * config/arm/thumb2.md: Update for attribute changes all occurrences
1238         of arlo_* types.
1239         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
1240         (cortexa7_older_only): Likewise.
1241         (cortexa7_younger):  Likewise.
1242         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1243         (1020alu_shift_op): Likewise.
1244         (1020alu_shift_reg_op): Likewise.
1245         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1246         (alu_shift_op): Likewise.
1247         (alu_shift_reg_op): Likewise.
1248         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1249         (11_alu_shift_op): Likewise.
1250         (11_alu_shift_reg_op): Likewise.
1251         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1252         (9_alu_shift_reg_op): Likewise.
1253         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1254         attribute changes.
1255         (cortex_a15_alu_shift): Likewise.
1256         (cortex_a15_alu_shift_reg): Likewise.
1257         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1258         attribute changes.
1259         (cortex_a5_alu_shift): Likewise.
1260         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1261         attribute changes.
1262         (cortex_a53_alu_shift): Likewise.
1263         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1264         attribute changes.
1265         (cortex_a7_alu_reg): Likewise.
1266         (cortex_a7_alu_shift): Likewise.
1267         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1268         attribute changes.
1269         (cortex_a8_alu_shift): Likewise.
1270         (cortex_a8_alu_shift_reg): Likewise.
1271         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1272         (cortex_a9_dp_shift): Likewise.
1273         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1274         attribute changes.
1275         * config/arm/cortex-r4.md
1276         (cortex_r4_alu): Update for attribute changes.
1277         (cortex_r4_mov): Likewise.
1278         (cortex_r4_alu_shift_reg): Likewise.
1279         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1280         (526_alu_shift_op): Likewise.
1281         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1282         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1283         (626te_alu_shift_op): Likewise.
1284         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1285         (726te_alu_shift_op): Likewise.
1286         (726te_alu_shift_reg_op): Likewise.
1287         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1288         (mp626_alu_shift_op): Likewise.
1289         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1290         (pj4_alu_conds): Likewise.
1291         (pj4_shift): Likewise.
1292         (pj4_shift_conds): Likewise.
1293         (pj4_alu_shift): Likewise.
1294         (pj4_alu_shift_conds): Likewise.
1295         * config/aarch64/aarch64.md: Update for attribute change
1296         all occurrences of arlo_* and shift* types.
1298 2013-09-05  Mike Stump  <mikestump@comcast.net>
1300         * tree.h: Move documentation for tree_function_decl to tree-core.h
1301         with the declaration.
1303 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
1305         PR target/58139
1306         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1307         looking for widest mode.
1309 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
1311         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1313 2013-09-05  Richard Biener  <rguenther@suse.de>
1315         PR tree-optimization/58137
1316         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1317         Do not create vectors of pointers.
1318         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1319         types for the components of the vector initializer.
1320         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1321         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1323 2013-09-05  Martin Jambor  <mjambor@suse.cz>
1325         * ipa-prop.c (remove_described_reference): Accept missing references,
1326         return false if that hppens, otherwise return true.
1327         (cgraph_node_for_jfunc): New function.
1328         (try_decrement_rdesc_refcount): Likewise.
1329         (try_make_edge_direct_simple_call): Use them.
1330         (ipa_edge_removal_hook): Remove references from rdescs.
1331         (ipa_edge_duplication_hook): Clone rdescs and their references
1332         when the new edge has the same caller as the old one.
1333         * cgraph.c (cgraph_resolve_speculation): Remove speculative
1334         reference before removing any edges.
1336 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
1338         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1339         initial store.
1340         * thumb2.md (thumb2_storewb_parisi): New pattern.
1342 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
1344         * config/aarch64/aarch64-option-extensions.def: Add
1345         AARCH64_OPT_EXTENSION of 'crc'.
1346         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1347         (AARCH64_ISA_CRC): Ditto.
1348         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1349         description of the CRC extension.
1351 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1353         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1354         * config/rs6000/linux.h: Ditto.
1355         * alpha/linux.h: Ditto.
1356         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1357         no_c99_libc_has_function.
1358         * config/c6x/uclinux-elf.h: Ditto.
1359         * config/lm32/uclinux-elf.h: Ditto.
1360         * config/m68k/uclinux.h: Ditto.
1361         * config/moxie/uclinux.h: Ditto.
1362         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1363         (crisv32-*-linux*, cris-*-linux*): Ditto.
1364         * config/bfin/bfin.c: Include "tm_p.h".
1366 2013-09-05  Richard Biener  <rguenther@suse.de>
1368         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1369         check for a definition without a basic-block.
1371 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1372             Sofiane Naci <sofiane.naci@arm.com>
1374         * config/aarch64/aarch64.md
1375         (*movti_aarch64): Rename r_2_f and f_2_r.
1376         (*movsf_aarch64): Likewise.
1377         (*movdf_aarch64): Likewise.
1378         (*movtf_aarch64): Likewise.
1379         (aarch64_movdi_<mode>low): Likewise.
1380         (aarch64_movdi_<mode>high): Likewise.
1381         (aarch64_mov<mode>high_di): Likewise.
1382         (aarch64_mov<mode>low_di): Likewise.
1383         (aarch64_movtilow_tilow): Likewise.
1384         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
1385         values to config/arm/types.md
1386         (attribute "conds"): Update for attribute change.
1387         (anddi3_insn): Likewise.
1388         (iordi3_insn): Likewise.
1389         (xordi3_insn): Likewise.
1390         (one_cmpldi2): Likewise.
1391         * config/arm/types.md (type): Add Neon types.
1392         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1393         use "type" attribute.
1394         (movmisalign<mode>_neon_store): Likewise.
1395         (movmisalign<mode>_neon_load): Likewise.
1396         (vec_set<mode>_internal): Likewise.
1397         (vec_setv2di_internal): Likewise.
1398         (vec_extract<mode>): Likewise.
1399         (vec_extractv2di): Likewise.
1400         (add<mode>3_neon): Likewise.
1401         (adddi3_neon): Likewise.
1402         (sub<mode>3_neon): Likewise.
1403         (subdi3_neon): Likewise.
1404         (mul<mode>3_neon): Likewise.
1405         (mul<mode>3add<mode>_neon): Likewise.
1406         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1407         (fma<VCVTF:mode>4)): Likewise.
1408         (fma<VCVTF:mode>4_intrinsic): Likewise.
1409         (fmsub<VCVTF:mode>4)): Likewise.
1410         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1411         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1412         (ior<mode>3): Likewise.
1413         (and<mode>3): Likewise.
1414         (anddi3_neon): Likewise.
1415         (orn<mode>3_neon): Likewise.
1416         (orndi3_neon): Likewise.
1417         (bic<mode>3_neon): Likewise.
1418         (bicdi3_neon): Likewise.
1419         (xor<mode>3): Likewise.
1420         (one_cmpl<mode>2): Likewise.
1421         (abs<mode>2): Likewise.
1422         (neg<mode>2): Likewise.
1423         (umin<mode>3_neon): Likewise.
1424         (umax<mode>3_neon): Likewise.
1425         (smin<mode>3_neon): Likewise.
1426         (smax<mode>3_neon): Likewise.
1427         (vashl<mode>3): Likewise.
1428         (vashr<mode>3_imm): Likewise.
1429         (vlshr<mode>3_imm): Likewise.
1430         (ashl<mode>3_signed): Likewise.
1431         (ashl<mode>3_unsigned): Likewise.
1432         (neon_load_count): Likewise.
1433         (ashldi3_neon_noclobber): Likewise.
1434         (signed_shift_di3_neon): Likewise.
1435         (unsigned_shift_di3_neon): Likewise.
1436         (ashrdi3_neon_imm_noclobber): Likewise.
1437         (lshrdi3_neon_imm_noclobber): Likewise.
1438         (widen_ssum<mode>3): Likewise.
1439         (widen_usum<mode>3): Likewise.
1440         (quad_halves_<code>v4si): Likewise.
1441         (quad_halves_<code>v4sf): Likewise.
1442         (quad_halves_<code>v8hi): Likewise.
1443         (quad_halves_<code>v16qi): Likewise.
1444         (reduc_splus_v2di): Likewise.
1445         (neon_vpadd_internal<mode>): Likewise.
1446         (neon_vpsmin<mode>): Likewise.
1447         (neon_vpsmax<mode>): Likewise.
1448         (neon_vpumin<mode>): Likewise.
1449         (neon_vpumax<mode>): Likewise.
1450         (ss_add<mode>_neon): Likewise.
1451         (us_add<mode>_neon): Likewise.
1452         (ss_sub<mode>_neon): Likewise.
1453         (us_sub<mode>_neon): Likewise.
1454         (neon_vadd<mode>_unspec): Likewise.
1455         (neon_vaddl<mode>): Likewise.
1456         (neon_vaddw<mode>): Likewise.
1457         (neon_vhadd<mode>): Likewise.
1458         (neon_vqadd<mode>): Likewise.
1459         (neon_vaddhn<mode>): Likewise.
1460         (neon_vmul<mode>): Likewise.
1461         (neon_vmla<mode>): Likewise.
1462         (neon_vmlal<mode>): Likewise.
1463         (neon_vmls<mode>): Likewise.
1464         (neon_vmlsl<mode>): Likewise.
1465         (neon_vqdmulh<mode>): Likewise.
1466         (neon_vqdmlal<mode>): Likewise.
1467         (neon_vqdmlsl<mode>): Likewise.
1468         (neon_vmull<mode>): Likewise.
1469         (neon_vqdmull<mode>): Likewise.
1470         (neon_vsub<mode>_unspec): Likewise.
1471         (neon_vsubl<mode>): Likewise.
1472         (neon_vsubw<mode>): Likewise.
1473         (neon_vqsub<mode>): Likewise.
1474         (neon_vhsub<mode>): Likewise.
1475         (neon_vsubhn<mode>): Likewise.
1476         (neon_vceq<mode>): Likewise.
1477         (neon_vcge<mode>): Likewise.
1478         (neon_vcgeu<mode>): Likewise.
1479         (neon_vcgt<mode>): Likewise.
1480         (neon_vcgtu<mode>): Likewise.
1481         (neon_vcle<mode>): Likewise.
1482         (neon_vclt<mode>): Likewise.
1483         (neon_vcage<mode>): Likewise.
1484         (neon_vcagt<mode>): Likewise.
1485         (neon_vtst<mode>): Likewise.
1486         (neon_vabd<mode>): Likewise.
1487         (neon_vabdl<mode>): Likewise.
1488         (neon_vaba<mode>): Likewise.
1489         (neon_vabal<mode>): Likewise.
1490         (neon_vmax<mode>): Likewise.
1491         (neon_vmin<mode>): Likewise.
1492         (neon_vpaddl<mode>): Likewise.
1493         (neon_vpadal<mode>): Likewise.
1494         (neon_vpmax<mode>): Likewise.
1495         (neon_vpmin<mode>): Likewise.
1496         (neon_vrecps<mode>): Likewise.
1497         (neon_vrsqrts<mode>): Likewise.
1498         (neon_vqabs<mode>): Likewise.
1499         (neon_vqneg<mode>): Likewise.
1500         (neon_vcls<mode>): Likewise.
1501         (clz<mode>2): Likewise.
1502         (popcount<mode>2): Likewise.
1503         (neon_vrecpe): Likewise.
1504         (neon_vrsqrte): Likewise.
1505         (neon_vget_lane<mode>_sext_internal): Likewise.
1506         (neon_vget_lane<mode>_zext_internal): Likewise.
1507         (neon_vdup_n<mode>): Likewise.
1508         (neon_vdup_nv2di): Likewise.
1509         (neon_vdpu_lane<mode>_internal): Likewise.
1510         (neon_vswp<mode>): Likewise.
1511         (float<mode><V_cvtto>2): Likewise.
1512         (floatuns<mode><V_cvtto>2): Likewise.
1513         (fix_trunc<mode><V_cvtto>)2): Likewise
1514         (fixuns_trunc<mode><V_cvtto)2): Likewise.
1515         (neon_vcvt<mode>): Likewise.
1516         (neon_vcvtv4sfv4hf): Likewise.
1517         (neon_vcvtv4hfv4sf): Likewise.
1518         (neon_vcvt_n<mode>): Likewise.
1519         (neon_vmovn<mode>): Likewise.
1520         (neon_vqmovn<mode>): Likewise.
1521         (neon_vqmovun<mode>): Likewise.
1522         (neon_vmovl<mode>): Likewise.
1523         (neon_vmul_lane<mode>): Likewise.
1524         (neon_vmull_lane<mode>): Likewise.
1525         (neon_vqdmull_lane<mode>): Likewise.
1526         (neon_vqdmulh_lane<mode>): Likewise.
1527         (neon_vmla_lane<mode>): Likewise.
1528         (neon_vmlal_lane<mode>): Likewise.
1529         (neon_vqdmlal_lane<mode>): Likewise.
1530         (neon_vmls_lane<mode>): Likewise.
1531         (neon_vmlsl_lane<mode>): Likewise.
1532         (neon_vqdmlsl_lane<mode>): Likewise.
1533         (neon_vext<mode>): Likewise.
1534         (neon_vrev64<mode>): Likewise.
1535         (neon_vrev32<mode>): Likewise.
1536         (neon_vrev16<mode>): Likewise.
1537         (neon_vbsl<mode>_internal): Likewise.
1538         (neon_vshl<mode>): Likewise.
1539         (neon_vqshl<mode>): Likewise.
1540         (neon_vshr_n<mode>): Likewise.
1541         (neon_vshrn_n<mode>): Likewise.
1542         (neon_vqshrn_n<mode>): Likewise.
1543         (neon_vqshrun_n<mode>): Likewise.
1544         (neon_vshl_n<mode>): Likewise.
1545         (neon_vqshl_n<mode>): Likewise.
1546         (neon_vqshlu_n<mode>): Likewise.
1547         (neon_vshll_n<mode>): Likewise.
1548         (neon_vsra_n<mode>): Likewise.
1549         (neon_vsri_n<mode>): Likewise.
1550         (neon_vsli_n<mode>): Likewise.
1551         (neon_vtbl1v8qi): Likewise.
1552         (neon_vtbl2v8qi): Likewise.
1553         (neon_vtbl3v8qi): Likewise.
1554         (neon_vtbl4v8qi): Likewise.
1555         (neon_vtbx1v8qi): Likewise.
1556         (neon_vtbx2v8qi): Likewise.
1557         (neon_vtbx3v8qi): Likewise.
1558         (neon_vtbx4v8qi): Likewise.
1559         (neon_vtrn<mode>_internal): Likewise.
1560         (neon_vzip<mode>_internal): Likewise.
1561         (neon_vuzp<mode>_internal): Likewise.
1562         (neon_vld1<mode>): Likewise.
1563         (neon_vld1_lane<mode>): Likewise.
1564         (neon_vld1_dup<mode>): Likewise.
1565         (neon_vld1_dupv2di): Likewise.
1566         (neon_vst1<mode>): Likewise.
1567         (neon_vst1_lane<mode>): Likewise.
1568         (neon_vld2<mode>): Likewise.
1569         (neon_vld2_lane<mode>): Likewise.
1570         (neon_vld2_dup<mode>): Likewise.
1571         (neon_vst2<mode>): Likewise.
1572         (neon_vst2_lane<mode>): Likewise.
1573         (neon_vld3<mode>): Likewise.
1574         (neon_vld3qa<mode>): Likewise.
1575         (neon_vld3qb<mode>): Likewise.
1576         (neon_vld3_lane<mode>): Likewise.
1577         (neon_vld3_dup<mode>): Likewise.
1578         (neon_vst3<mode>): Likewise.
1579         (neon_vst3qa<mode>): Likewise.
1580         (neon_vst3qb<mode>): Likewise.
1581         (neon_vst3_lane<mode>): Likewise.
1582         (neon_vld4<mode>): Likewise.
1583         (neon_vld4qa<mode>): Likewise.
1584         (neon_vld4qb<mode>): Likewise.
1585         (neon_vld4_lane<mode>): Likewise.
1586         (neon_vld4_dup<mode>): Likewise.
1587         (neon_vst4<mode>): Likewise.
1588         (neon_vst4qa<mode>): Likewise.
1589         (neon_vst4qb<mode>): Likewise.
1590         (neon_vst4_lane<mode>): Likewise.
1591         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1592         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1593         (neon_vec_<US>mult_lo_<mode>): Likewise.
1594         (neon_vec_<US>mult_hi_<mode>): Likewise.
1595         (neon_vec_<US>shiftl_<mode>): Likewise.
1596         (neon_unpack<US>_<mode>): Likewise.
1597         (neon_vec_<US>mult_<mode>): Likewise.
1598         (vec_pack_trunc_<mode>): Likewise.
1599         (neon_vec_pack_trunk_<mode>): Likewise.
1600         (neon_vabd<mode>_2): Likewise.
1601         (neon_vabd<mode>_3): Likewise.
1602         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
1603         (thumb2_movsi_vfp): Likewise.
1604         (movdi_vfp): Likewise.
1605         (movdi_vfp_cortexa8): Likewise.
1606         (movhf_vfp_neon): Likewise.
1607         (movhf_vfp): Likewiwse.
1608         (movsf_vfp): Likewiwse.
1609         (thumb2_movsf_vfp): Likewiwse.
1610         (movdf_vfp): Likewise.
1611         (thumb2_movdf_vfp): Likewise.
1612         (movsfcc_vfp): Likewise.
1613         (thumb2_movsfcc_vfp): Likewise.
1614         (movdfcc_vfp): Likewise.
1615         (thumb2_movdfcc_vfp): Likewise.
1616         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
1617         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
1618         (v10_v2c): Likewise.
1619         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
1620         attribute change.
1621         (cortex_a15_neon_int_2): Likewise.
1622         (cortex_a15_neon_int_3): Likewise.
1623         (cortex_a15_neon_int_4): Likewise.
1624         (cortex_a15_neon_int_5): Likewise.
1625         (cortex_a15_neon_vqneg_vqabs): Likewise.
1626         (cortex_a15_neon_vmov): Likewise.
1627         (cortex_a15_neon_vaba): Likewise.
1628         (cortex_a15_neon_vaba_qqq): Likewise.
1629         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1630         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1631         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
1632         scalar_64_32_long_scalar): Likewise.
1633         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1634         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1635         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1636         lotype_qdd_64_32_long): Likewise.
1637         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1638         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1639         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1640         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1641         (cortex_a15_neon_shift_1): Likewise.
1642         (cortex_a15_neon_shift_2): Likewise.
1643         (cortex_a15_neon_shift_3): Likewise.
1644         (cortex_a15_neon_vshl_ddd): Likewise.
1645         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1646         (cortex_a15_neon_vsra_vrsra): Likewise.
1647         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1648         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1649         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1650         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1651         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1652         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1653         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1654         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1655         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1656         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1657         (cortex_a15_neon_bp_simple): Likewise.
1658         (cortex_a15_neon_bp_2cycle): Likewise.
1659         (cortex_a15_neon_bp_3cycle): Likewise.
1660         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1661         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1662         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1663         (cortex_a15_neon_vld2_4_regs): Likewise.
1664         (cortex_a15_neon_vld3_vld4): Likewise.
1665         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1666         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1667         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
1668         (cortex_a15_neon_vst3_vst4): Likewise.
1669         (cortex_a15_neon_vld1_vld2_lane): Likewise.
1670         (cortex_a15_neon_vld3_vld4_lane" 10
1671         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1672         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1673         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1674         (cortex_a15_neon_ldm_2): Likewise.0
1675         (cortex_a15_neon_stm_2): Likewise.
1676         (cortex_a15_neon_mcr): Likewise.
1677         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1678         (cortex_a15_neon_mrc): Likewise.
1679         (cortex_a15_neon_mrrc): Likewise.
1680         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
1681         change.
1682         (cortex_a15_alu_shift): Likewise.
1683         (cortex_a15_alu_shift_reg): Likewise.
1684         (cortex_a15_mult32): Likewise.
1685         (cortex_a15_mult64): Likewise.
1686         (cortex_a15_block): Likewise.
1687         (cortex_a15_branch): Likewise.
1688         (cortex_a15_load1): Likewise.
1689         (cortex_a15_load3): Likewise.
1690         (cortex_a15_store1): Likewise.
1691         (cortex_a15_store3): Likewise.
1692         (cortex_a15_call): Likewise.
1693         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
1694         (cortex_a5_f2r): Likewise.
1695         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
1696         change.
1697         (cortex_a53_f2r): Likewise.
1698         * config/arm/cortex-a7.md
1699         (cortex_a7_branch): Update for attribute change.
1700         (cortex_a7_call): Likewise.
1701         (cortex_a7_alu_imm): Likewise.
1702         (cortex_a7_alu_reg): Likewise.
1703         (cortex_a7_alu_shift): Likewise.
1704         (cortex_a7_mul): Likewise.
1705         (cortex_a7_load1): Likewise.
1706         (cortex_a7_store1): Likewise.
1707         (cortex_a7_load2): Likewise.
1708         (cortex_a7_store2): Likewise.
1709         (cortex_a7_load3): Likewise.
1710         (cortex_a7_store3): Likewise.
1711         (cortex_a7_load4): Likewise.
1712         (cortex_a7_store4): Likewise.
1713         (cortex_a7_fpalu): Likewise.
1714         (cortex_a7_fconst): Likewise.
1715         (cortex_a7_fpmuls): Likewise.
1716         (cortex_a7_neon_mul): Likewise.
1717         (cortex_a7_fpmacs): Likewise.
1718         (cortex_a7_neon_mla: Likewise.
1719         (cortex_a7_fpmuld: Likewise.
1720         (cortex_a7_fpmacd: Likewise.
1721         (cortex_a7_fpfmad: Likewise.
1722         (cortex_a7_fdivs: Likewise.
1723         (cortex_a7_fdivd: Likewise.
1724         (cortex_a7_r2f: Likewise.
1725         (cortex_a7_f2r: Likewise.
1726         (cortex_a7_f_flags: Likewise.
1727         (cortex_a7_f_loads: Likewise.
1728         (cortex_a7_f_loadd: Likewise.
1729         (cortex_a7_f_stores: Likewise.
1730         (cortex_a7_f_stored: Likewise.
1731         (cortex_a7_neon): Likewise.
1732         * config/arm/cortex-a8-neon.md
1733         (cortex_a8_neon_mrc): Update for attribute change.
1734         (cortex_a8_neon_mrrc): Likewise.
1735         (cortex_a8_neon_int_1): Likewise.
1736         (cortex_a8_neon_int_2): Likewise.
1737         (cortex_a8_neon_int_3): Likewise.
1738         (cortex_a8_neon_int_4): Likewise.
1739         (cortex_a8_neon_int_5): Likewise.
1740         (cortex_a8_neon_vqneg_vqabs): Likewise.
1741         (cortex_a8_neon_vmov): Likewise.
1742         (cortex_a8_neon_vaba): Likewise.
1743         (cortex_a8_neon_vaba_qqq): Likewise.
1744         (cortex_a8_neon_vsma): Likewise.
1745         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1746         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1747         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1748         long_scalar): Likewise.
1749         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1750         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1751         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1752         long_scalar_qdd_64_32_long): Likewise.
1753         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1754         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1755         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1756         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1757         (cortex_a8_neon_shift_1): Likewise.
1758         (cortex_a8_neon_shift_2): Likewise.
1759         (cortex_a8_neon_shift_3): Likewise.
1760         (cortex_a8_neon_vshl_ddd): Likewise.
1761         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1762         (cortex_a8_neon_vsra_vrsra): Likewise.
1763         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1764         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1765         (cortex_a8_neon_fp_vsum): Likewise.
1766         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1767         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1768         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1769         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1770         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1771         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1772         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1773         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1774         (cortex_a8_neon_bp_simple): Likewise.
1775         (cortex_a8_neon_bp_2cycle): Likewise.
1776         (cortex_a8_neon_bp_3cycle): Likewise.
1777         (cortex_a8_neon_ldr): Likewise.
1778         (cortex_a8_neon_str): Likewise.
1779         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1780         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1781         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1782         (cortex_a8_neon_vld2_4_regs): Likewise.
1783         (cortex_a8_neon_vld3_vld4): Likewise.
1784         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1785         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1786         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1787         (cortex_a8_neon_vst3_vst4): Likewise.
1788         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1789         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1790         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1791         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1792         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1793         (cortex_a8_neon_mcr): Likewise.
1794         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1795         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
1796         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
1797         change.
1798         (ca9_neon_mrrc): Likewise.
1799         (cortex_a9_neon_int_1): Likewise.
1800         (cortex_a9_neon_int_2): Likewise.
1801         (cortex_a9_neon_int_3): Likewise.
1802         (cortex_a9_neon_int_4): Likewise.
1803         (cortex_a9_neon_int_5): Likewise.
1804         (cortex_a9_neon_vqneg_vqabs): Likewise.
1805         (cortex_a9_neon_vmov): Likewise.
1806         (cortex_a9_neon_vaba): Likewise.
1807         (cortex_a9_neon_vaba_qqq): Likewise.
1808         (cortex_a9_neon_vsma): Likewise.
1809         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1810         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1811         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1812         long_scalar): Likewise.
1813         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1814         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1815         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1816         long_scalar_qdd_64_32_long): Likewise.
1817         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1818         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1819         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1820         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1821         (cortex_a9_neon_shift_1): Likewise.
1822         (cortex_a9_neon_shift_2): Likewise.
1823         (cortex_a9_neon_shift_3): Likewise.
1824         (cortex_a9_neon_vshl_ddd): Likewise.
1825         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1826         (cortex_a9_neon_vsra_vrsra): Likewise.
1827         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1828         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1829         (cortex_a9_neon_fp_vsum): Likewise.
1830         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1831         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1832         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1833         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1834         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1835         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1836         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1837         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1838         (cortex_a9_neon_bp_simple): Likewise.
1839         (cortex_a9_neon_bp_2cycle): Likewise.
1840         (cortex_a9_neon_bp_3cycle): Likewise.
1841         (cortex_a9_neon_ldr): Likewise.
1842         (cortex_a9_neon_str): Likewise.
1843         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1844         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1845         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1846         (cortex_a9_neon_vld2_4_regs): Likewise.
1847         (cortex_a9_neon_vld3_vld4): Likewise.
1848         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1849         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1850         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1851         (cortex_a9_neon_vst3_vst4): Likewise.
1852         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1853         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1854         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1855         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1856         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1857         (cortex_a9_neon_mcr): Likewise.
1858         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1859         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
1860         (cortex_a9_fps): Likewise.
1861         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
1862         change.
1863         (cortex_m4_fmuls): Likewise.
1864         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
1865         change.
1866         (cortex_r4_mrc): Likewise.
1867         * config/arm/iterators.md: Update comment referring to neon_type.
1868         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
1869         (iwmmxt_movsi_insn): Likewise.
1870         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
1871         attribute change.
1872         (pj4_core_to_vfp): Likewise.
1873         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
1874         attribute change.
1875         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
1876         (vfp_fstore): Likewise.
1877         * doc/md.texi: Change references to neon_type to refer to type.
1879 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1881         * tree.h (DECL_BUILT_IN): Fix typo in comment.
1883 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
1885         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
1886         lglobl if not weak.
1888 2013-09-04  Easwaran Raman  <eraman@google.com>
1890         PR middle-end/57370
1891         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
1892         (build_and_add_sum): Use it.
1893         (appears_later_in_bb): Simplify code.
1895 2013-09-04  Teresa Johnson  <tejohnson@google.com>
1897         * dumpfile.c (dump_finish): Don't close stderr/stdout.
1899 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
1901         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
1903 2013-09-04  Jan Hubicka  <jh@suse.cz>
1905         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
1906         * ipa-devirt.c: Include diganostic.h
1907         (odr_type_d): Add types and types_set.
1908         (hash_type_name): Work for types with vtables during LTO.
1909         (odr_hasher::remove): Fix comment; destroy types_set.
1910         (add_type_duplicate): New function,
1911         (get_odr_type): Use it.
1912         (dump_type_inheritance_graph): Dump type duplicates.
1913         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
1914         graph.
1915         * tree.c (types_same_for_odr): Give exact answers on types with
1916         virtual tables.
1918 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1920         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
1921         explaining their differences.
1923 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
1925         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
1927 2013-09-03  Jeff Law  <law@redhat.com>
1929         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
1930         when not threading through a joiner block.  Pass joiner/no joiner
1931         state to register_jump_thread.
1932         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
1933         state from argument rather than implying on path length.
1934         Dump the entire jump thread path into debugging dump.
1935         * tree-flow.h (register_jump_thread): Update prototype.
1937 2013-08-29  Xinliang David Li  <davidxl@google.com>
1939         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1940         Remove a trivial gcc_assert.
1942 2013-08-29  Xinliang David Li  <davidxl@google.com>
1944         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
1945         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
1946         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1947         Delay base decl alignment adjustment.
1948         * tree-vectorizer.c (vect_destroy_datarefs): New function.
1949         * tree-vectorizer.h: New data structure.
1950         (set_dr_misalignment): New function.
1951         (dr_misalignment): Ditto.
1952         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
1953         (vectorizable_load): Ditto.
1954         (ensure_base_align): New function.
1955         (vectorize_loops): Add dbg_cnt support.
1956         (execute_vect_slp): Ditto.
1957         * dbgcnt.def: New debug counter.
1958         * Makefile: New dependency.
1960 2013-09-03  Meador Inge  <meadori@codesourcery.com>
1962         Revert:
1964         2013-08-30  Meador Inge  <meadori@codesourcery.com>
1966         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
1968 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
1970         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
1971         function descriptor.
1973 2013-09-03  Richard Biener  <rguenther@suse.de>
1975         * tree-affine.c (add_elt_to_tree): Fix association issue,
1976         avoid useless converts and make sure to always return a
1977         properly typed result.
1979 2013-09-03  Richard Biener  <rguenther@suse.de>
1981         PR middle-end/57656
1982         * fold-const.c (negate_expr_p): Fix division case.
1983         (negate_expr): Likewise.
1985 2013-09-03  Richard Biener  <rguenther@suse.de>
1987         PR lto/58285
1988         * tree-streamer-out.c: Include tm.h.
1989         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
1991 2013-09-03  Jan Hubicka  <jh@suse.cz>
1993         * tree-profile.c (tree_profiling): Cleanup CFG when done.
1995 2013-09-03  Alan Modra  <amodra@gmail.com>
1997         * config.gcc (powerpc*-*-linux*): Add support for little-endian
1998         multilibs to big-endian target and vice versa.
1999         * config/rs6000/t-linux64: Use := assignment on all vars.
2000         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2001         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2002         * config/rs6000/t-linux64le: New file.
2003         * config/rs6000/t-linux64bele: New file.
2004         * config/rs6000/t-linux64lebe: New file.
2006 2013-09-02  Jan Hubicka  <jh@suse.cz>
2008         * ipa-inline-transform.c (inline_transform): Do not
2009         optimize_inline_calls when not optimizing.
2011 2013-09-02  Jan Hubicka  <jh@suse.cz>
2013         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2014         duplicated nodes for assembler names.
2015         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2016         hard registers from assembler name hash.
2018 2013-09-02  Jan Hubicka  <jh@suse.cz>
2020         * ipa-split.c (execute_split_functions): Split externally visible
2021         functions called once.
2023 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2025         PR ipa/58106
2026         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2027         linked list.  When finding the correct duplicate, also consider also
2028         the caller in additon to its inlined_to node.
2030 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2032         * config/aarch64/aarch64-simd-builtins.def
2033         (dup_lane_scalar): Remove.
2034         * config/aarch64/aarch64-simd.md
2035         (aarch64_simd_dup): Add 'w->w' alternative.
2036         (aarch64_dup_lane<mode>): Allow for VALL.
2037         (aarch64_dup_lane_scalar<mode>): Remove.
2038         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2039         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2040         (aarch64_get_lane_unsigned<mode>): Likewise.
2041         (aarch64_get_lane<mode>): Likewise.
2042         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2043         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2044         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2045         (VCON): Change container of V2SF.
2046         (vswap_width_name): Likewise.
2047         * config/aarch64/arm_neon.h
2048         (__aarch64_vdup_lane_any): New.
2049         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2050         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2051         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2053 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2055         PR middle-end/56382
2056         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2057         the source or the destination is a single hard register.
2059 2013-09-02  Richard Biener  <rguenther@suse.de>
2061         PR middle-end/57511
2062         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2063         non-linear SCEVs.
2065 2013-09-02  Richard Biener  <rguenther@suse.de>
2067         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2068         arithmetic to sizetype.
2070 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2072         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2073         Find auto-increment use both before and after candidate.
2075 2013-09-02  Marek Polacek  <polacek@redhat.com>
2077         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2079 2013-09-01  Jan Hubicka  <jh@suse.cz>
2081         * Makefile.in: Add ipa-profile.o
2082         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2083         * cgraph.c (struct cgraph_propagate_frequency_data,
2084         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2085         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2086         * cgraph.h (cgraph_propagate_frequency): Remove.
2087         * ipa-inline-analysis.c: Include ipa-utils.h;
2088         drop duplicated cfgloop.h.
2089         (inline_update_callee_summaries): Update.
2090         * ipa-profile.c: New file.
2091         * ipa-utils.h (ipa_propagate_frequency): Declare.
2092         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2093         data-streamer.h, value-prof.h.
2094         (symtab_remove_unreachable_nodes): Update profile.
2095         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2096         account_time_size, cmp_counts, dump_histogram,
2097         ipa_profile_generate_summary, ipa_profile_write_summary,
2098         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2099         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2100         Move to ipa-profile.c.
2102 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2104         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2106 2013-09-01  Jan Hubicka  <jh@suse.cz>
2108         * common.opt (fdevirtualize-speculatively): New function.
2109         * invoke.texi (fdevirtualize-speculatively): Document.
2110         * ipa-devirt.c: Include ipa-inline.h
2111         (likely_target_p): New function.
2112         (ipa_devirt): New function.
2113         (gate_ipa_devirt): New function.
2114         (pass_data_ipa_devirt): New static var.
2115         (pass_ipa_devirt): Likewise.
2116         (make_pass_ipa_devirt): New function.
2117         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2118         (common_handle_option): Disable devirtualization when
2119         value range profiling is available.
2120         * passes.def (pass_ipa_devirt): Add.
2121         * timever.def (TV_IPA_DEVIRT): New timevar.
2122         * tree-pass.h (make_pass_ipa_devirt):
2124 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2126         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2127         include sanitize(undefined).
2129 2013-08-31  Diego Novillo  <dnovillo@google.com>
2131         * Makefile.in (TREE_CORE_H): Define.
2132         (TREE_H): Use.
2133         (GTFILES): Add tree-core.h.
2134         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2135         size the array.
2136         * tree-core.h: New file.
2137         Move all data structures, enum, typedefs, global
2138         declarations and constants from ...
2139         * tree.h: ... here.
2141 2013-08-31  Jan Hubicka  <jh@suse.cz>
2143         * bulitins.c (expand_builtin): Do not early exit for gcov
2144         instrumented functions.
2146 2013-08-31  Marek Polacek  <polacek@redhat.com>
2148         * ubsan.c: Include tm_p.h.
2150 2013-08-31  Jan Hubicka  <jh@suse.cz>
2152         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2153         warning.
2155         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2156         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2157         * tree-cfg.c (verify_gimple_label): ... here.
2158         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2159         (ipa_merge_profiles): New function.
2160         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2161         (lto_input_function_body): Likewise.
2162         * ipa-utils.h (ipa_merge_profiles): Declare.
2163         * lto-streamer.h (lto_input_function_body): Update prototype.
2164         (emit_label_in_global_context_p): Remove.
2165         * lto-symtab.c: Include ipa-utils.h
2166         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2168 2013-08-31  Jan Hubicka  <jh@suse.cz>
2170         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2172 2013-08-31  Jan Hubicka  <jh@suse.cz>
2174         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2176 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2178         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2179         "cmp" RTX before signed_comparison_operator check to account
2180         for "code" changes.
2182 2013-08-30  Jan Hubicka  <jh@suse.cz>
2184         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2185         (detect_type_change_1): Rename to ...
2186         (detect_type_change): ... this one; early return on non-polymorphic
2187         types.
2188         (detect_type_change_ssa): Add comp_type parameter; update
2189         use of detect_type_change.
2190         (compute_complex_assign_jump_func): Add param_type parameter;
2191         update use of detect_type_change_ssa.
2192         (compute_complex_ancestor_jump_func): Likewise.
2193         (ipa_get_callee_param_type): New function.
2194         (ipa_compute_jump_functions_for_edge): Compute parameter type;
2195         update calls to the jump function computation functions.
2197 2013-08-30  Teresa Johnson  <tejohnson@google.com>
2198             Steven Bosscher  <steven@gcc.gnu.org>
2200         * cfgrtl.c (fixup_new_cold_bb): New routine.
2201         (commit_edge_insertions): Invoke fixup_partitions.
2202         (find_partition_fixes): New routine.
2203         (fixup_partitions): Ditto.
2204         (verify_hot_cold_block_grouping): Update comments.
2205         (rtl_verify_edges): Invoke find_partition_fixes.
2206         (rtl_verify_bb_pointers): Update comments.
2207         (rtl_verify_bb_layout): Ditto.
2208         * basic-block.h (probably_never_executed_edge_p): Declare.
2209         (fixup_partitions): Ditto.
2210         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2211         * bb-reorder.c (sanitize_hot_paths): New function.
2212         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2213         sanitize_hot_paths.
2214         * predict.c (probably_never_executed_edge_p): New routine.
2215         * cfg.c (check_bb_profile): Add partition insanity warnings.
2217 2013-08-30  Meador Inge  <meadori@codesourcery.com>
2219         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2221 2013-08-30  Marek Polacek  <polacek@redhat.com>
2223         * Makefile.in (ubsan.o): Add.
2224         (c-family/c-ubsan.o): Add.
2225         (builtins.o): Add ubsan.h dependency.
2226         * ubsan.h: New file.
2227         * ubsan.c: New file.
2228         * common.opt: Add -fsanitize=undefined option.
2229         (flag_sanitize): Add variable.
2230         (fsanitize=): Add option.  Add Driver.
2231         (fsanitize=thread): Remove option.
2232         (fsanitize=address): Likewise.
2233         (static-libubsan): New option.
2234         * doc/invoke.texi: Document the new flag and -static-libubsan.
2235         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2236         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2237         * builtin-attrs.def (ATTR_COLD): Define.
2238         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2239         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2240         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2241         * flag-types.h (sanitize_code): New enum.
2242         * opts.c (common_handle_option): Parse command line arguments
2243         of -fsanitize=.  Add -fsanitize=unreachable option.
2244         * varasm.c (get_variable_section): Adjust.
2245         (assemble_noswitch_variable): Likewise.
2246         (assemble_variable): Likewise.
2247         (output_constant_def_contents): Likewise.
2248         (categorize_decl_for_section): Likewise.
2249         (place_block_symbol): Likewise.
2250         (output_object_block): Likewise.
2251         * builtins.def: Likewise.
2252         * toplev.c (compile_file): Likewise.
2253         (process_options): Likewise.
2254         * cppbuiltin.c: Likewise.
2255         * tsan.c (tsan_pass): Likewise.
2256         (tsan_gate): Likewise.
2257         (tsan_gate_O0): Likewise.
2258         * cfgexpand.c (partition_stack_vars): Likewise.
2259         (expand_stack_vars): Likewise.
2260         (defer_stack_allocation): Likewise.
2261         (expand_used_vars): Likewise.
2262         * cfgcleanup.c (old_insns_match_p): Likewise.
2263         * asan.c (asan_finish_file): Likewise.
2264         (asan_instrument): Likewise.
2265         (gate_asan): Likewise.
2266         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2267         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2268         (asan_global_struct): Use pointer_sized_int_node instead
2269         calling build_nonstandard_integer_type.
2270         (initialize_sanitizer_builtins): Likewise.
2271         (asan_finish_file): Likewise.
2272         * gcc.c: Document %{%:function(args):X}.
2273         (static_spec_functions): Add sanitize.
2274         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2275         store funcval != NULL there.
2276         (do_spec_1): Adjust handle_spec_function caller.
2277         (handle_braces): Allow %:function(args) as condition.
2278         (sanitize_spec_function): New function.
2279         (ADD_STATIC_LIBUBSAN_LIBS): Define.
2280         (LIBUBSAN_SPEC): Likewise.
2281         (LIBUBSAN_EARLY_SPEC): Likewise.
2282         (SANITIZER_SPEC): Handle libubsan.
2283         (SANITIZER_EARLY_SPEC): Likewise.
2284         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2285         instead of fsanitize=address.
2286         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2287         instead of fsanitize=address*.
2288         * builtins.c: Include ubsan.h.
2289         (fold_builtin_0): Instrument __builtin_unreachable.
2290         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2291         instead of flag_asan.
2292         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2293         (pointer_sized_int_node): Define.
2294         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2296 2013-08-30  Mike Stump  <mikestump@comcast.net>
2298         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2299         with RE patterns.
2301 2013-08-29  Jan Hubicka  <jh@suse.cz>
2303         * cgraph.c (cgraph_function_body_availability): Handle weakref
2304         correctly.
2305         * passes.def: Remove pass_fixup_cfg.
2306         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2307         track when we need to remove functions.
2308         (gate_ipa_inline): Execute inlining always; add comment why.
2309         (pass_data_ipa_inline): Remove TODO_remove_functions.
2310         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2311         do not produce summaries.
2312         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2313         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2314         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2315         constant pool and vtable.
2317 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
2319         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2320         New arm_neon.h's internal macros to specify 64-bit constants.
2321         Avoid using stdint.h's macros.
2323 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2325         * recog.c (verify_changes): Verify that changes[i].old is non-zero
2326         before applying REG_P.
2328 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
2330         PR tree-optimization/58277
2331         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2332         after seeing too many stmts with vdef in between dombb and current
2333         bb, invalidate everything.
2335 2013-08-30  Richard Biener  <rguenther@suse.de>
2337         * fold-const.c (fold_single_bit_test): Fix overflow test.
2339 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
2341         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2342         and which can live in a register, always retrieve the value on entry.
2343         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2344         passed by invisible reference specially.
2345         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2346         (vt_add_function_parameter): Correctly deal with a parameter passed by
2347         invisible reference.
2349 2013-08-30  Jan Hubicka  <jh@suse.cz>
2351         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2353 2013-08-30  Richard Biener  <rguenther@suse.de>
2355         PR tree-optimization/58228
2356         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2357         allow invariant loads in nested loop vectorization.
2359 2013-08-30  Richard Biener  <rguenther@suse.de>
2361         PR tree-optimization/58223
2362         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2363         (has_anti_or_output_dependence): ... this and adjust to also
2364         look for output dependences.
2365         (mark_nodes_having_upstream_mem_writes): Adjust.
2366         (rdg_flag_uses): Likewise.
2368 2013-08-30  Richard Biener  <rguenther@suse.de>
2370         PR tree-optimization/58010
2371         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2372         assert that we have a loop-closed PHI.
2374 2013-08-29  Jan Hubicka  <jh@suse.cz>
2376         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2377         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2378         * lto-section-in.c (lto_free_function_in_decl_state): New function.
2379         (lto_free_function_in_decl_state_for_node): New function.
2381 2013-08-29  Xinliang David Li  <davidxl@google.com>
2383         * loop-unroll.c (report_unroll_peel): Minor message change.
2384         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2385         Emit alignment peeling message with default -fopt-info.
2386         (vect_loop_versioning): Emit loop version info message.
2387         * tree-vectorizer.c (vectorize_loops): Minor message change.
2388         (execute_vect_slp): Ditto.
2390 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
2392         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2393         of the clone from the DECL_NAME of the original function.
2395 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
2397         * passes.c (register_pass): Add overload.
2398         * tree-pass.h (register_pass): Forward declare it.  Add comment.
2400 2013-08-29  Jan Hubicka  <jh@suse.cz>
2402         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2403         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2404         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2405         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2406         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2407         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2408         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2409         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2411 2013-08-29  Teresa Johnson  <tejohnson@google.com>
2413         * dumpfile.c (dump_loc): Output column number.
2414         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2415         * doc/invoke.texi: Document optall -fopt-info flag.
2416         * profile.c (read_profile_edge_counts): Use new dump framework.
2417         (compute_branch_probabilities): Ditto.
2418         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2419         when pass not in any opt group.
2420         * pass_manager.h (pass_manager::get_pass_profile): New method.
2421         * value-prof.c (check_counter): Use new dump framework.
2422         (check_ic_target): Ditto.
2423         * coverage.c (get_coverage_counts): Ditto.
2424         (coverage_init): Setup new dump framework.
2426 2013-08-29  Richard Biener  <rguenther@suse.de>
2428         PR tree-optimization/58246
2429         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2430         handle the dominance check inside a basic-block.
2432 2013-08-29  Richard Biener  <rguenther@suse.de>
2434         PR middle-end/57287
2435         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2436         of default defs that appear in abnormal PHI nodes.
2438 2013-08-29  Richard Biener  <rguenther@suse.de>
2440         PR tree-optimization/57685
2441         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2442         single-use operands to avoid exponential complexity.
2444 2013-08-28  Dehao Chen  <dehao@google.com>
2446         * ipa-inline.c (edge_badness): Fix integer underflow.
2448 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
2450         * gtm-builtins.def (_ITM_free): Declare leaf.
2452 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2454         PR target/58067
2455         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2456         (*tls_local_dynamic_base_64_largepic): Likewise.
2457         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2458         Remove predicate from call operand.
2459         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2460         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2462 2013-08-28  Jeff Law  <law@redhat.com>
2464         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2465         checks for the number of predecessors and successors allowed.
2466         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2467         which require copying a joiner block if there is a request which
2468         is a subpath that requires no joiner block copying.
2470 2013-08-28  Jan Hubicka  <jh@suse.cz>
2472         * lto-streamer-out.c (DFS_write_tree_body): Drop
2473         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2474         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2475         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2476         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2477         Do not read DECL_ERROR_ISSUED.
2478         (unpack_ts_decl_with_vis_value_fields): Do not read
2479         DECL_DEFER_OUTPUT.
2480         (lto_input_ts_binfo_tree_pointers): Do not read
2481         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2482         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2483         write DECL_ERROR_ISSUED..
2484         (pack_ts_decl_with_vis_value_fields): Do not write
2485         DECL_DEFER_OUTPUT.
2486         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2487         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2488         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2489         * tree.h (tree_decl_common): Update comment.
2490         (DECL_ERROR_ISSUED): Remove.
2492 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2494         PR middle-end/58257
2495         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2497 2013-08-28  Jan Hubicka  <jh@suse.cz>
2499         * builtins.def (free): Declare leaf.
2501 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
2503         * gdbhooks.py: New.
2504         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2505         * configure: Regenerate.
2507 2013-08-27  Martin Jambor  <mjambor@suse.cz>
2509         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2510         (ipa_ancestor_jf_data): Likewise.
2511         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2512         (ipa_get_jf_pass_through_type_preserved): New function.
2513         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2514         (ipa_get_jf_ancestor_type_preserved): New function.
2515         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2516         (ipa_get_jf_ancestor_result): Likewise.
2517         (propagate_vals_accross_pass_through): Use
2518         ipa_get_jf_pass_through_result to do all the value mappings.
2519         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2520         type_preserved flag.
2521         (ipa_set_jf_cst_copy): New function.
2522         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2523         (ipa_set_jf_arith_pass_through): Likewise.
2524         (ipa_set_ancestor_jf): Likewise.
2525         (compute_complex_assign_jump_func): Set type_preserved instead of
2526         punting.
2527         (ipa_compute_jump_functions_for_edge): Likewise.
2528         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2529         (update_jump_functions_after_inlining): Update type_preserved.
2530         Explicitely create jump functions when combining one with pass_through.
2531         (ipa_write_jump_function): Stream the type_preserved flags.
2532         (ipa_read_jump_function): Likewise.
2534 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
2535             Aldy Hernandez  <aldyh@redhat.com>
2537         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
2538         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
2539         * function.h (struct function): Add has_force_vect_loops and
2540         has_simduid_loops.
2541         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
2542         * gimple.c (gimple_build_omp_critical): Add KIND argument and
2543         handle it.
2544         * gimple.def: Update CLAUSES comments.
2545         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
2546         (gimple_build_omp_for): Add argument to prototype.
2547         (gimple_omp_for_kind): New.
2548         (gimple_omp_for_set_kind): New.
2549         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
2550         GOVD_DATA_SHARE_CLASS.
2551         (enum omp_region_type): Add ORT_SIMD.
2552         (gimple_add_tmp_var): Handle ORT_SIMD.
2553         (gimplify_var_or_parm_decl): Same.
2554         (is_gimple_stmt): Same.
2555         (omp_firstprivatize_variable): Same.
2556         (omp_add_variable): Only use splay_tree_insert if lookup failed.
2557         (omp_notice_variable): Handle ORT_SIMD.
2558         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
2559         (omp_check_private): Handle ORT_SIMD.
2560         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2561         OMP_CLAUSE_SAFELEN.
2562         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
2563         Handle OMP_CLAUSE_LASTPRIVATE.
2564         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2565         OMP_CLAUSE_SAFELEN.
2566         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
2567         (gimplify_expr): Handle OMP_SIMD.
2568         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
2569         (expand_GOMP_SIMD_VF): New.
2570         (expand_GOMP_SIMD_LAST_LANE): New.
2571         * internal-fn.def (GOMP_SIMD_LANE): New.
2572         (GOMP_SIMD_VF): New.
2573         (GOMP_SIMD_LAST_LANE): New.
2574         * omp-low.c: Include target.h.
2575         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
2576         OMP_CLAUSE_SAFELEN.
2577         (check_omp_nesting_restrictions): Same.
2578         (omp_max_vf): New.
2579         (lower_rec_simd_input_clauses): New.
2580         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
2581         OMP_CLAUSE_LINEAR.
2582         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
2583         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
2584         (expand_omp_build_assign): New.
2585         (expand_omp_for_init_counts): New.
2586         (expand_omp_for_init_vars): New.
2587         (extract_omp_for_update_vars): New.
2588         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
2589         and rewrite accordingly.
2590         (expand_omp_simd): New.
2591         (expand_omp_for): Use expand_omp_simd.
2592         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
2593         (lower_omp_for): Do not lower the body.
2594         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
2595         in their own loops.
2596         * tree-flow.h (find_omp_clause): Remove prototype.
2597         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
2598         forcing vectorization of the loop, or if flag_tree_vectorize.
2599         (gate_tree_if_conversion): Similarly.
2600         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
2601         gimple_build_omp_for.
2602         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
2603         * tree-parloops (create_parallel_loop): Pass kind argument to
2604         gimple_build_omp_for.
2605         * tree-pretty-print.c (dump_omp_clause): Add cases for
2606         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
2607         OMP_CLAUSE__SIMDUID_.
2608         (dump_generic_node): Handle OMP_SIMD.
2609         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
2610         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
2611         unroll OMP_SIMD loops here.
2612         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
2613         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
2614         loop->safelen.
2615         (vect_analyze_data_refs): Handle simd loops.
2616         * tree-vect-loop.c (vectorizable_live_operation): Handle
2617         IFN_GOMP_SIMD*.
2618         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
2619         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
2620         (vectorizable_load): Same.
2621         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
2622         (struct simduid_to_vf): New.
2623         (simduid_to_vf::hash): New.
2624         (simduid_to-vf::equal): New.
2625         (struct simd_array_to_simduid): New.
2626         (simd_array_to_simduid::hash): New.
2627         (simd_array_to_simduid::equal): New.
2628         (adjust_simduid_builtins): New.
2629         (struct note_simd_array_uses_struct): New.
2630         (note_simd_array_uses_cb): New.
2631         (note_simd_array_uses): New.
2632         (vectorize_loops): Handle simd hints and adjust simd builtins
2633         accordingly.
2634         * tree-vectorizer.h (struct _stmt_vec_info): Add
2635         simd_lane_access_p field.
2636         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
2637         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
2638         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
2639         (omp_clause_code_name): Same.
2640         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
2641         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
2642         * tree.def (OMP_SIMD): New entry.
2643         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
2644         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
2645         (OMP_CLAUSE_DECL): Adjust range for new clauses.
2646         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
2647         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
2648         (OMP_CLAUSE_LINEAR_STEP): New.
2649         (OMP_CLAUSE_SAFELEN_EXPR): New.
2650         (OMP_CLAUSE__SIMDUID__DECL): New.
2651         (find_omp_clause): New prototype.
2653 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
2655         * config/i386/driver-i386.c (host_detect_local_cpu): Update
2656         Haswell processor detection.
2658 2013-08-27  Christian Widmer  <shadow@umbrox.de>
2660         PR target/57927
2661         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
2662         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
2663         AVX2 capable processors.
2665 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
2667         * config/aarch64/arm_neon.h: Replace all inline asm implementations
2668         of vget_low_* with implementations in terms of other intrinsics.
2670 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
2672         PR middle-end/57219
2673         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
2674         values to -1, 0 and 1.
2676 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
2678         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
2679         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
2680         (<optab><mode>3_insn): Remove.
2681         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
2682         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
2683         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
2684         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
2685         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
2686         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
2687         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
2688         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
2689         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
2690         (ror<mode>3_insn): Likewise.
2691         * config/aarch64/predicates.md (aarch64_simd_register): New.
2693 2013-08-27  Richard Biener  <rguenther@suse.de>
2695         PR tree-optimization/57521
2696         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
2697         one edge is non-critical.
2698         (find_phi_replacement_condition): Make sure to use a non-critical
2699         edge.  Cleanup and remove old bug workarounds.
2700         (bb_postdominates_preds): Remove.
2701         (if_convertible_loop_p_1): Do not compute post-dominators.
2702         (combine_blocks): Do not free post-dominators.
2703         (main_tree_if_conversion): Likewise.
2704         (pass_data_if_conversion): Add TODO_verify_ssa.
2706 2013-08-27  DJ Delorie  <dj@redhat.com>
2708         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
2710 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
2712         * function.c (assign_parm_find_data_types): Set passed_mode and
2713         nominal_mode to the TYPE_MODE of nominal_type for the built
2714         pointer type in case of the struct-pass-by-reference.
2716 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2718         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
2719         (UINT16_TYPE): Change default to "unsigned int".
2721         * config/avr/avr.opt (mfract-convert-truncate): New option.
2722         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
2723         is set, round negative fractional integers according to n1169
2724         when converting to integer types.
2726 2013-08-26  Jan Hubicka  <jh@suse.cz>
2728         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
2729         methods can not be called indirectly when their address is not taken.
2731 2013-08-26  Jan Hubicka  <jh@suse.cz>
2733         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
2734         ctor_for_folding.
2736 2013-08-26  Jan Hubicka  <jh@suse.cz>
2738         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
2739         can be unshared.
2741 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2743         * reload.c (find_valid_class): Allow classes that do not include
2744         FIRST_PSEUDO_REGISTER - 1.
2746 2013-08-26  Jan Hubicka  <jh@suse.cz>
2748         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
2749         fix edge count/frequency when speculation failed; fix type check
2750         for the direct call.
2752 2013-08-26  Jan Hubicka  <jh@suse.cz>
2754         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
2756 2013-08-26  Jan Hubicka  <jh@suse.cz>
2758         * ipa-inline-transform.c (inline_transform): Be ready for basic block
2759         to be changed by edge redirection.
2761 2013-08-26  Jan Hubicka  <jh@suse.cz>
2763         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
2764         formating; add sanity check.
2765         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
2766         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
2767         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
2768         (resolve_noninline_speculation): Update callee keys, too.
2770 2013-08-26  Jan Hubicka  <jh@suse.cz>
2772         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
2773         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
2775 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2777         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
2778         into proper place.
2780 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
2782         * config/i386/i386.c (ix86_debug_options): Remove prototype.
2783         (x86_64_elf_select_section): Ditto.
2784         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
2785         arguments.
2786         (ix86_pass_by_reference): Ditto.
2787         (output_set_got): Ditto.
2788         (ix86_unary_operator_ok): Ditto.
2789         (ix86_expand_builtin): Ditto.
2791 2013-08-23  Jan Hubicka  <jh@suse.cz>
2793         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
2795 2013-08-23  Jan Hubicka  <jh@suse.cz>
2797         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
2798         (tree_decl_with_vis): Add FINAL field.
2800 2013-08-23  Jeff Law  <law@redhat.com>
2802         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
2803         the debugging dump when the expression is fully redundant.
2805 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2807         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
2808         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
2809         * pretty-print.c (pp_formatted_text_data): Likewise.
2810         (pp_write_text_to_stream): Likewise.
2811         (pp_write_text_as_dot_label_to_stream): Likewise.
2812         (pp_append_r): Likewise.
2813         (pp_format): Likewise.
2814         (pp_flush): Likewise.
2815         (pp_clear_output_area): Likewise.
2816         (pp_append_text): Likewise.
2817         (pp_formatted_text): Likewise.
2818         (pp_remaining_character_count_for_line): Likewise.
2819         (pp_newline): Likewise.
2820         (pp_character): Likewise.
2821         (output_buffer::~output_buffer): Define.
2822         (pretty_printer::~pretty_printer): Destruct output buffer.
2823         * pretty-print.h (output_buffer::~output_buffer): Declare.
2824         (pretty_printer::~pretty_printer): Declare virtual.
2826 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
2828         PR other/57324
2829         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
2830         HOST_WIDE_INT_M1U): New macros.
2831         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
2832         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
2833         unsigned -1 for lshift.
2834         * cse.c (cse_insn): Likewise.
2835         * double-int.c (rshift_double, lshift_double): Likewise.
2836         * builtins.c (fold_builtin_bitop): Likewise.
2837         * combine.c (force_to_mode): Likewise.
2838         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
2839         * simplify-rtx.c (simplify_const_unary_operation,
2840         simplify_const_binary_operation): Likewise.
2841         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
2842         check_va_list_escapes): Likewise.
2843         * rtlanal.c (nonzero_bits1): Likewise.
2844         * expmed.c (expand_smod_pow2): Likewise.
2845         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
2847 2013-08-23  Jan Hubicka  <jh@suse.cz>
2849         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
2850         as having address taken.
2852 2013-08-23  Jan Hubicka  <jh@suse.cz>
2854         * ipa-utils.h (method_class_type): Declare.
2855         * ipa-devirt.c (method_class_type): Export.
2857         * cgraphunit.c (analyze_functions): Do basic devirtualization;
2858         do not walk base classes of anonymous types.
2860 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
2862         PR rtl-optimization/58220
2863         PR regression/58221
2864         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
2865         handle SEQUENCE insns properly.
2867 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2869         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
2870         definition.
2871         (pp_newline_and_indent): Likewise.
2872         (pp_separate_with): Likewise.
2873         * pretty-print.c (pp_newline_and_flush): Define.
2874         (pp_newline_and_indent): Likewise.
2875         (pp_separate_with): Likewise.
2877 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2879         PR target/58218
2880         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
2881         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
2883 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2885         * config/i386/predicates.md (ext_sse_reg_operand): New.
2886         * config/i386/i386.md (*movti_internal): Use
2887         predicate to determine if EVEX is needed.
2888         (*movsi_internal): Ditto.
2889         (*movdf_internal): Ditto.
2890         (*movsf_internal): Ditto.
2891         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
2893 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2895         PR tree-optimization/58209
2896         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
2897         (find_tail_calls): Give up for pointer result types if m is non-NULL.
2898         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
2899         emit POINTER_PLUS_EXPR.
2900         (create_tailcall_accumulator): For pointer result type accumulate in
2901         sizetype type.
2903 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
2905         * configure.ac: Add backslashes missing from the last change.
2906         * configure: Regenerate.
2908 2013-08-22  Jan Hubicka  <jh@susue.cz>
2910         * ipa.c (function_and_variable_visibility): First remember function
2911         was global and then make it local.
2913 2013-08-22  Julian Brown  <julian@codesourcery.com>
2915         * configure.ac: Add aarch64 to list of arches which use "nop" in
2916         debug_line test.
2917         * configure: Regenerate.
2919 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2921         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
2922         gnu_libc_has_function.
2923         * config/s390/tpf.h: Likewise.
2925 2013-08-22  Jan Hubicka  <jh@susue.cz>
2927         * timevar.c (validate_phases): Add cast.
2929 2013-08-22  Jan Hubicka  <jh@susue.cz>
2931         * timevar.c (validate_phases): Use size_t for memory.
2932         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
2934 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2936         * pretty-print.h (output_buffer::output_buffer): Declare.
2937         (pretty_printer::pretty_printer): Likewise.
2938         (pp_construct): Remove.
2939         * pretty-print.c (output_buffer::output_buffer): Define.
2940         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
2941         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
2942         (print_gimple_expr): Likewise.
2943         (print_gimple_seq): Likewise.
2944         (gimple_dump_bb): Likewise.
2945         * sched-vis.c (dump_value_slim): Likewise.
2946         (dump_insn_slim): Likewise.
2947         (dump_rtl_slim): Likewise.
2948         (str_pattern_slim): Likewise.
2949         * tree-mudflap.c (mf_varname_tree): Likewise.
2950         * graph.c (print_graph_cfg): Likewise.
2951         (start_graph_dump): Likewise.
2952         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
2953         placement-new.
2954         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
2955         pretty printer initialization.
2956         * coretypes.h (diagnostic_context): Remove superflous type alias
2957         declaration.
2958         (pretty_printer): Likewise.  Declare directly as a class.
2959         (pretty_print_info): Remove declaration as class.
2960         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
2961         and pp_clear_output_area.
2962         (asan_add_global): Likewise.
2964 2013-08-22  Jan Hubicka  <jh@suse.cz>
2966         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
2967         * ipa-utils.h (update_type_inheritance_graph): Declare.
2968         (possible_polymorphic_call_target_p): Declare.
2969         (possible_polymorphic_call_target_p): New.
2970         * ipa-devirt.c: Update toplevel comments.
2971         (cached_polymorphic_call_targets): Move up.
2972         (odr_type_d): Move ID down.
2973         (polymorphic_type_binfo_p): Update comment.
2974         (odr_hasher::remove): Likewise;
2975         (get_odr_type): Set anonymous_namespace.
2976         (dump_odr_type): Dump it.
2977         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
2978         (maybe_record_node): Record node in cached_polymorphic_call_targets.
2979         (record_binfo): Add comment.
2980         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
2981         built.
2982         (devirt_node_removal_hook): Do not iCE when cache is freed.
2983         (possible_polymorphic_call_target_p): New predicate.
2984         (update_type_inheritance_graph): New function.
2986 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2987             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2988             Sergey Lega  <sergey.s.lega@intel.com>
2989             Anna Tikhonova  <anna.tikhonova@intel.com>
2990             Ilya Tocar  <ilya.tocar@intel.com>
2991             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2992             Ilya Verbin  <ilya.verbin@intel.com>
2993             Kirill Yukhin  <kirill.yukhin@intel.com>
2994             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2996         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
2997         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
2998         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
2999         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3000         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3001         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3002         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3003         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3004         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3005         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3006         OPT_mavx512pf, OPT_mavx512er cases.
3007         * config/i386/constraints.md (v): New constraint.
3008         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3009         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3010         (bit_AVX512CD): New.
3011         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3012         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3013         * config/i386/i386-c.c (ix86_target_macros_internal):
3014         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3015         __AVX512PF__.
3016         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3017         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3018         * config/i386/i386.c (regclass_map, dbx_register_map)
3019         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3020         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3021         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3022         -mavx512pf options.
3023         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3024         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3025         -mavx512pf options.  Fix formatting.
3026         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3027         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3028         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3029         -mavx512cd, -mavx512pf options.
3030         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3031         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3032         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3033         ALL_SSE_REGS.
3034         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3035         (ix86_set_reg_reg_cost): Ditto.
3036         (x86_order_regs_for_local_alloc): Ditto.
3037         (MAX_VECT_LEN): Extend to 64-byte.
3038         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3039         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3040         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3041         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3042         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3043         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3044         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3045         (SSE_REG_MODE_P): Support new modes.
3046         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3047         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3048         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3049         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3050         (REG_CLASS_CONTENTS): Add new registers.
3051         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3052         (EXT_REX_SSE_REGNO_P): New.
3053         (HI_REGISTER_NAMES): Add new registers.
3054         * config/i386/i386.md: Define constants for new registers.
3055         (mode): Add new 512-bit modes.
3056         (prefix): Support evex prefix.
3057         (isa): Support avx512f, noavx512f, fma_avx512f.
3058         (ssemodesuffix): Add new 512-bit modes.
3059         (movxi): New.
3060         (*movxi_internal_avx512f): Ditto.
3061         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3062         Support MODE_XI.
3063         (*movsi_internal): Likewise.
3064         (*movdf_internal): Likewise.
3065         (*movsf_internal): Likewise.
3066         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3067         (<code><mode>3): Likewise.
3068         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3069         New.
3070         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3071         with the new constraint "v".
3072         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3073         modes.
3074         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3075         with the new constraint "v".
3076         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3077         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3078         (*<plusminus_insn><mode>3): Likewise.
3079         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3080         (*mul<mode>3): Likewise.
3081         (<sse>_vmmul<mode>3): Likewise.
3082         (<sse>_div<mode>3): Likewise.
3083         (<sse>_vmdiv<mode>3): Likewise.
3084         (<sse>_sqrt<mode>2): Likewise.
3085         (<sse>_vmsqrt<mode>2): Likewise.
3086         (*<code><mode>3_finite): Likewise.
3087         (*<code><mode>3) <smaxmin>: Likewise.
3088         (<sse>_vm<code><mode>3): Likewise.
3089         (*<code><mode>3) <any_logic>: Likewise.
3090         (*fma_fmadd_<mode>): Likewise.
3091         (*fma_fmsub_<mode>): Likewise.
3092         (*fma_fnmadd_<mode>): Likewise.
3093         (*fma_fnmsub_<mode>): Likewise.
3094         (*fma_fmaddsub_<mode>): Likewise.
3095         (*fma_fmsubadd_<mode>): Likewise.
3096         (*fmai_fmadd_<mode>): Likewise.
3097         (*fmai_fmsub_<mode>): Likewise.
3098         (*fmai_fnmadd_<mode>): Likewise.
3099         (*fmai_fnmsub_<mode>): Likewise.
3100         (sse_cvtsi2ss): Likewise.
3101         (sse_cvtsi2ssq): Likewise.
3102         (sse_cvtss2si): Likewise.
3103         (sse_cvtss2si_2): Likewise.
3104         (sse_cvtss2siq): Likewise.
3105         (sse_cvtss2siq_2): Likewise.
3106         (sse_cvttss2si): Likewise.
3107         (sse_cvtss2siq_2): Likewise.
3108         (float<sseintvecmodelower><mode>2): Likewise.
3109         (sse2_cvtsd2si_2): Likewise.
3110         (sse2_cvtsd2siq_2): Likewise.
3111         (*<plusminus_insn><mode>3): Likewise.
3112         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3113         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3114         (ashr<mode>3): Likewise.
3115         (<shift_insn><mode>3): Likewise.
3116         (avx2_<code><mode>3): Likewise.
3117         (*avx2_<code><mode>3): Likewise.
3118         (*andnot<mode>3): Likewise.
3119         (*<code><mode>3) <any_logic>: Likewise.
3120         (abs<mode>2): Likewise.
3121         (avx2_permvar<mode>): Likewise.
3122         (avx2_perm<mode>_1): Likewise.
3123         (*avx_vpermilp<mode>): Likewise.
3124         (avx_vpermilvar<mode>3): Likewise.
3125         (avx2_ashrv<mode>): Likewise.
3126         (avx2_<shift_insn>v<mode>): Likewise.
3127         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3128         -mavx512cd.
3129         * doc/rtl.texi: Document XImode.
3131 2013-08-21  Jeff Law  <law@redhat.com>
3133         * tree-flow.h (register_jump_thread): Pass vector of edges
3134         instead of each important edge.
3135         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3136         thread path into a vector and pass that to register_jump_thread.
3137         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3138         passed in edge vector to the current 3-edge form.
3140         Revert:
3141         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3143         * dce.c (fini_dce): Call df_analyze again just in case
3144         delete_unmarked_insns removed anything.
3146 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3148         * reload.h (struct reg_equivs): Rename to ..
3149         (struct reg_equivs_s): .. this.
3151 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3153         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3155 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3157         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3159 2013-08-21  Jeff Law  <law@redhat.com>
3161         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3162         simplify assignments too.  If the RHS collapses to a singleton
3163         range, then return the value for the range.
3165 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3167         * config/i386/sse.md (V16): Rename to...
3168         (VMOVE): this.
3169         (mov<mode>): Update iterator name.
3170         (*mov<mode>_internal): Ditto.
3171         (push<mode>1): Ditto.
3172         (movmisalign<mode>): Ditto.
3174 2013-08-20  Jan Hubicka  <jh@suse.cz>
3176         PR bootstrap/58186
3177         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3178         entry for direct edges.
3179         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3181 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3183         Revert my last two changes, r201865 and r201864:
3185         Revert r201865:
3186         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3188         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3189         instances can own GC refs.
3191         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3192         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3193         (gcc::context::gt_pch_nx): Likewise.
3194         (gcc::context::gt_pch_nx):  Likewise.
3195         * ggc.h (gt_ggc_mx <T>): New.
3196         (gt_pch_nx_with_op <T>): New.
3197         (gt_pch_nx <T>): New.
3198         * passes.c (opt_pass::gt_ggc_mx): New.
3199         (opt_pass::gt_pch_nx): New.
3200         (opt_pass::gt_pch_nx_with_op): New.
3201         (pass_manager::gt_ggc_mx): New.
3202         (pass_manager::gt_pch_nx): New.
3203         (pass_manager::gt_pch_nx_with_op): New.
3204         (pass_manager::operator new): Use
3205         ggc_internal_cleared_alloc_stat rather than xcalloc.
3206         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3207         (pass_manager::gt_ggc_mx): New.
3208         (pass_manager::gt_pch_nx): New.
3209         (pass_manager::gt_pch_nx_with_op): New.
3210         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3211         (opt_pass::operator new): New.
3212         (opt_pass::gt_ggc_mx): New.
3213         (opt_pass::gt_pch_nx): New.
3214         (opt_pass::gt_pch_nx_with_op): New.
3216         Revert r201864:
3217         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3219         * Makefile.in (GTFILES): Add context.h.
3220         * context.c (gcc::context::operator new): New.
3221         (gcc::context::gt_ggc_mx): New.
3222         (gcc::context::gt_pch_nx): New.
3223         (gcc::context::gt_pch_nx): New.
3224         * context.h (gcc::context): Add GTY((user)) marking.
3225         (gcc::context::operator new): New.
3226         (gcc::context::gt_ggc_mx): New.
3227         (gcc::context::gt_pch_nx): New.
3228         (gcc::context::gt_pch_nx): New.
3229         (g): Add GTY marking.
3230         (gt_ggc_mx (gcc::context *)): New.
3231         (gt_pch_nx (gcc::context *)): New.
3232         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3233         void *cookie)): New.
3234         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3236 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3238         * dce.c (fini_dce): Call df_analyze again just in case
3239         delete_unmarked_insns removed anything.
3241 2013-08-20  Teresa Johnson  <tejohnson@google.com>
3243         PR rtl-optimizations/57451
3244         * final.c (reemit_insn_block_notes): Prevent lexical blocks
3245         from crossing split section boundaries.
3247 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3249         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3250         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3251         with MULTLIB_DEFAULTS.
3253 2013-08-20  Nick Clifton  <nickc@redhat.com>
3255         * target.def (narrow_volatile_bitfield): Note that the default
3256         value is false, not !TARGET_STRICT_ALIGN.
3257         * doc/tm.texi: Regenerate.
3259 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
3261         Fix LIB_SPEC for systems without libpthread.
3263         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3264         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3265         for Android.
3266         * config/i386/linux-common.h: Likewise.
3267         * config/mips/linux-common.h: Likewise.
3269 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
3271         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3272         checks.
3274 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3276         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3277         instances can own GC refs.
3279         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3280         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3281         (gcc::context::gt_pch_nx): Likewise.
3282         (gcc::context::gt_pch_nx):  Likewise.
3283         * ggc.h (gt_ggc_mx <T>): New.
3284         (gt_pch_nx_with_op <T>): New.
3285         (gt_pch_nx <T>): New.
3286         * passes.c (opt_pass::gt_ggc_mx): New.
3287         (opt_pass::gt_pch_nx): New.
3288         (opt_pass::gt_pch_nx_with_op): New.
3289         (pass_manager::gt_ggc_mx): New.
3290         (pass_manager::gt_pch_nx): New.
3291         (pass_manager::gt_pch_nx_with_op): New.
3292         (pass_manager::operator new): Use
3293         ggc_internal_cleared_alloc_stat rather than xcalloc.
3294         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3295         (pass_manager::gt_ggc_mx): New.
3296         (pass_manager::gt_pch_nx): New.
3297         (pass_manager::gt_pch_nx_with_op): New.
3298         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3299         (opt_pass::operator new): New.
3300         (opt_pass::gt_ggc_mx): New.
3301         (opt_pass::gt_pch_nx): New.
3302         (opt_pass::gt_pch_nx_with_op): New.
3304 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3306         * Makefile.in (GTFILES): Add context.h.
3307         * context.c (gcc::context::operator new): New.
3308         (gcc::context::gt_ggc_mx): New.
3309         (gcc::context::gt_pch_nx): New.
3310         (gcc::context::gt_pch_nx): New.
3311         * context.h (gcc::context): Add GTY((user)) marking.
3312         (gcc::context::operator new): New.
3313         (gcc::context::gt_ggc_mx): New.
3314         (gcc::context::gt_pch_nx): New.
3315         (gcc::context::gt_pch_nx): New.
3316         (g): Add GTY marking.
3317         (gt_ggc_mx (gcc::context *)): New.
3318         (gt_pch_nx (gcc::context *)): New.
3319         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3320         void *cookie)): New.
3321         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3323 2013-08-20  Alan Modra  <amodra@gmail.com>
3325         PR target/57865
3326         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3327         (rs6000_emit_epilogue): Likewise.
3329 2013-08-19  Dehao Chen  <dehao@google.com>
3331         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3333 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
3334             Jakub Jelinek  <jakub@redhat.com>
3336         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3337         (BUILT_IN_FABSD64): Likewise.
3338         (BUILT_IN_FABSD128): Likewise.
3339         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3340         (fold_builtin_1): Likewise.
3341         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3342         destination and source operands.
3343         (*abstd2_fpr): Likewise.
3344         (*nabstd2_fpr): Likewise.
3346 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
3348         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3349         JUMP_P and INSN_P.
3351 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
3353         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3354         incomplete.
3356 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3358         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3359         * builtins.c (default_libc_has_function): New.
3360         (gnu_libc_has_function): Ditto.
3361         (no_c99_libc_has_function): Ditto.
3362         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3363         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3364         (fold_builtin_sincos): Likewise.
3365         (fold_builtin_cexp): Likewise.
3366         * builtins.def (DEF_C94_BUILTIN): Likewise.
3367         (DEF_C99_BUILTIN): Likewise.
3368         (DEF_C99_C90RES_BUILTIN): Likewise.
3369         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3370         definitions to using this define.
3371         * config/darwin-protos.h (darwin_libc_has_function): New.
3372         * config/darwin.c (darwin_libc_has_function): Ditto.
3373         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3374         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3375         * config/darwin.h: Ditto.
3376         * config/elfos.h: Ditto.
3377         * config/freebsd.h: Ditto.
3378         * config/i386/cygming.h: Ditto.
3379         * config/i386/djgpp.h: Ditto.
3380         * config/i386/i386-interix.h: Ditto.
3381         * config/microblaze/microblaze.h: Ditto.
3382         * config/mmix/mmix.h: Ditto.
3383         * config/gnu-user.h: Ditto.
3384         * config/ia64/hpux.h: Ditto.
3385         * config/pa/pa-hpux.h: Ditto.
3386         * config/pdp11/pdp11.h: Ditto.
3387         * config/picochip/picochip.h: Ditto.
3388         * config/linux.h: Ditto.
3389         * config/netbsd.h: Ditto.
3390         * config/openbsd.h: Ditto.
3391         * config/rs6000/aix43.h: Ditto.
3392         * config/rs6000/aix51.h: Ditto.
3393         * config/rs6000/aix52.h: Ditto.
3394         * config/rs6000/aix53.h: Ditto.
3395         * config/rs6000/aix61.h: Ditto.
3396         * config/rs6000/darwin.h: Ditto.
3397         * config/rs6000/linux.h: Ditto.
3398         * config/rs6000/linux64.h: Ditto.
3399         * config/s390/tpf.h: Ditto.
3400         * config/sol2-10.h: Ditto.
3401         * config/sol2.h: Ditto.
3402         * config/vms/vms.h: Ditto.
3403         * config/vxworks.h: Ditto.
3404         * config/linux-android.c (linux_android_libc_has_function):
3405         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3406         * config/linux-protos.h (linux_android_libc_has_function):
3407         New declaration.
3408         * config/i386/i386.c (ix86_libc_has_function): New.
3409         * config/i386/i386-protos.h
3410         (ix86_libc_has_function): New declaration.
3411         * config/i386/i386.md
3412         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3413         ("isinf<mode>2): Likewise.
3414         * convert.c (convert_to_integer): Using new target hook
3415         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3416         TARGET_C99_FUNCTIONS.
3417         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3418         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3419         * coretypes.h (function_class): New enum for different
3420         classes of functions.
3421         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3422         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3423         (TARGET_HAS_SINCOS): Likewise.
3424         (TARGET_LIBC_HAS_FUNCTION): New.
3425         * doc/tm.texi: Regenerated.
3426         * targhooks.h (default_libc_has_function): New declaration.
3427         (no_c99_libc_has_function): Ditto.
3428         (gnu_libc_has_function): Ditto.
3429         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3430         and TARGET_HAS_SINCOS.
3432 2013-08-18  Jan Hubicka  <jh@suse.cz>
3434         * Makeifle-in (ipa-devirt.o): New.
3435         (GTFILES): Add ipa-utils.h and ipa-devirt.c
3436         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3437         (analyze_functions): Look into possible targets of polymorphic call.
3438         * dumpfile.c (dump_files): Add type-inheritance dump.
3439         * dumpfile.h (TDI_inheritance): New.
3440         * ipa-devirt.c: New file.
3441         * ipa-utils.h (odr_type_d): Forward declare.
3442         (odr_type): New type.
3443         (build_type_inheritance_graph): Declare.
3444         (possible_polymorphic_call_targets): Declare and introduce inline
3445         variant when only edge is pased.
3446         (dump_possible_polymorphic_call_targets): Likewise.
3447         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3448         * tree.c (type_in_anonymous_namespace_p): Break out from ...
3449         (types_same_for_odr): ... here.
3450         * tree.h (type_in_anonymous_namespace_p): Declare.
3452 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
3454         PR tree-optimization/58006
3455         * tree-parloops.c (take_address_of): Don't ICE if get_name
3456         returns NULL.
3457         (eliminate_local_variables_stmt): Remove clobber stmts.
3459 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
3461         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3462         error message is issued for an alias to undefined symbol.
3464 2013-08-18  Jan Hubicka  <jh@suse.cz>
3466         * cgraph.c (cgraph_create_indirect_edge): Discover
3467         polymorphic calls and record basic info into indirect_info.
3468         * gimple-fold.c (gimple_fold_call): When doing BINFO based
3469         devirtualization, ignore objc function calls.
3470         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3471         call with no parm index info.
3472         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3473         * tree.c (virtual_method_call_p): New function.
3474         * tree.h (virtual_method_call_p): Declare.
3476 2013-08-16  Jan Hubicka  <jh@suse.cz>
3478         PR middle-end/58179
3479         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3481 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
3483         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3484         attribute.
3486 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3488         * gengtype.c (type_for_name): Add special-case support for
3489         locating types within the "gcc::" namespace.
3490         (open_base_files): Emit a "using namespace gcc" directive.
3492 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3494         PR target/58160
3495         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3496         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3498         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3499         array instead of each individual operand as a separate argument.
3500         (emit_fusion_gpr_load): Likewise.
3501         (expand_fusion_gpr_load): Add new function declaration.
3503         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3504         signature to have the operands passed as an array, instead of as
3505         separate arguments.  Allow ZERO_EXTEND to be in the memory
3506         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
3507         depend on the register live/dead flags when peepholes are run.
3508         (expand_fusion_gpr_load): New function to be called from the
3509         peephole2 pass, to change the register that addis sets to be the
3510         target register.
3511         (emit_fusion_gpr_load): Change the calling signature to have the
3512         operands passed as an array, instead of as separate arguments.
3513         Allow ZERO_EXTEND to be in the memory address, and also
3514         SIGN_EXTEND if -mpower8-fusion-sign.
3516         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3517         unspec enumeration.
3518         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3519         adjust the register addis loads up in the peephole2 pass.  Do not
3520         depend on the register live/dead state when the peephole pass is done.
3522 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3524         * gengtype.c (create_user_defined_type): Ensure that the kind
3525         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3526         declaration is seen before the GTY((user)) marking.
3528 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3530         PR target/58105
3531         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
3533 2013-08-16  Jan Hubicka  <jh@suse.cz>
3535         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
3536         arugment expected_type.
3537         (gimple_fold_call): Use it.
3538         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
3539         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
3540         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
3541         (try_make_edge_direct_virtual_call): Likewise.
3542         * tree.c (obj_type_ref_class): New.
3543         * tree.h (obj_type_ref_class): Use it.
3545 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3547         * sched-vis.c (rtl_slim_pp_initialized): Remove.
3548         (rtl_slim_pp): Likewise.
3549         (init_rtl_slim_pretty_print): Likewise.
3550         (dump_value_slim):  Don't call it.  Use local pretty printer.
3551         (dump_insn_slim): Likewise.
3552         (dump_rtl_slim): Likewise.
3553         (str_pattern_slim): Likewise.
3554         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
3555         Simplify.
3557 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
3559         PR tree-optimization/58164
3560         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
3561         walk gimple_goto_dest of GIMPLE_GOTO.
3563         PR tree-optimization/58165
3564         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
3565         bi_call must be the last stmt in a bb, don't split_block, instead
3566         use fallthru edge from it and give up if there is none.
3567         Release conds vector when returning early.
3569 2013-08-14  Xinliang David Li  <davidxl@google.com>
3571         * config/i386/i386.c (ix86_option_override_internal):
3572         Remove unused variable and field.
3574 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3576         PR target/57949
3577         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
3578         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
3579         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
3580         and Linux, correct BLKmode alignment when 128-bit alignment is
3581         required and compatibility flag is not set.
3582         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
3583         for zero-size arguments when compatibility flag is not set.
3585 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3587         PR tree-optimization/58145
3588         * tree-sra.c (build_ref_for_offset): If prev_base has
3589         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
3591 2013-08-14  Xinliang David Li  <davidxl@google.com>
3593         * config/i386/i386.c (ix86_option_override_internal):
3594         Fix uninitialized variable error.
3596 2013-08-14  Xinliang David Li  <davidxl@google.com>
3598         * config/i386/i386.opt: Define two new options.
3599         * config/i386/x86-tune.def: Add arch selector field in macros.
3600         * config/i386/i386.h: Adjust macro definition.
3601         * config/i386/i386.c (ix86_option_override_internal):
3602         Refactor the code.
3603         (parse_mtune_ctrl_str): New function.
3604         (set_ix86_tune_features): New function.
3605         (ix86_function_specific_restore): Call the new helper function.
3607 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
3609         PR rtl-optimization/57662
3610         * sel-sched.c (code_motion_process_successors): When the current insn
3611         is removed after the recursive traversal, break from the loop.
3612         Add comments and debug printouts.
3614 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3615             Alexandre Oliva  <aoliva@redhat.com>
3617         PR target/58067
3618         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
3619         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
3620         there also UNSPEC_PLTOFF.
3622 2013-08-14  Marek Polacek  <polacek@redhat.com>
3624         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
3625         NUM_CONDITIONS bit positions.
3627 2013-08-13  Cary Coutant  <ccoutant@google.com>
3629         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
3630         (attr_checksum): Hash vector contents instead of pointer.
3631         (attr_checksum_ordered): Likewise.
3633 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
3635         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
3636         when Pmode != word_mode.  Add length_address attribute.
3637         (sse3_monitor_<mode>): Merge from sse3_monitor and
3638         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
3639         Pmode != word_mode.  Update insn length attribute.
3640         * config/i386/i386.c (ix86_option_override_internal): Update
3641         ix86_gen_monitor selection for merged sse3_monitor insn.
3643 2013-08-13  Julian Brown  <julian@codesourcery.com>
3645         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
3646         perform invalid legitimization on greater-than-word-size modes for
3647         TARGET_E500_DOUBLE.
3649 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
3651         * ira.c (setup_class_translate_array): Use aclass instead of cl
3652         for classes not fully covered by allocno classes.
3654 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
3656         PR tree-optimization/57661
3657         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
3658         * tree-inline.c (tree_function_versioning): Initialize it.
3659         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
3660         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
3661         that is not being copied.
3663         PR sanitizer/56417
3664         * asan.c (instrument_strlen_call): Fix typo in comment.
3665         Use char * type even for the lhs of POINTER_PLUS_EXPR.
3667 2013-08-13  Steve Ellcey  <sellcey@mips.com>
3669         * config/mips/mips.md (prefetch): Use lw instead of ld on
3670         loongson in 32bit mode.
3672 2013-08-13  Nick Clifton  <nickc@redhat.com>
3674         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
3676 2013-08-13  Jan Hubicka  <jh@suse.cz>
3678         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
3679         introduced edge; fix typo in sanity check.
3680         (cgraph_resolve_speculation): Export; improve diagnostic.
3681         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
3682         speculation at type mismatch.
3683         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
3684         (cgraph_resolve_speculation): Declare.
3685         (symtab_can_be_discarded): New function.
3686         * value-prof.c (gimple_ic_transform): Remove actual transform code.
3687         * ipa-inline-transform.c (speculation_removed): New global var.
3688         (clone_inlined_nodes): See if speculation can be removed.
3689         (inline_call): If speculations was removed, we growths may not match.
3690         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
3691         (speculation_useful_p): New function.
3692         (resolve_noninline_speculation): New function.
3693         (inline_small_functions): Resolve useless speculations.
3694         * ipa-inline.h (speculation_useful_p): Declare
3695         * ipa.c (can_replace_by_local_alias): Simplify.
3696         (ipa_profile): Produce speculative calls in non-lto, too;
3697         add simple cost model; produce local aliases.
3699 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
3701         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
3702         PASS_MANAGER_H.
3704 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
3706         * config/i386/i386.c (ix86_function_versions): Use error + inform.
3708 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
3710         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
3711         iterator instead of X87MODEF.
3713 2013-08-12  Perez Read  <netfirewall@gmail.com>
3715         PR target/58132
3716         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
3717         operand 0 for intel asm alternative.
3718         (*movabs<mode>_2): Ditto for operand 1.
3720 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
3722         * config/aarch64/arm_none.h
3723         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
3725 2013-08-12  Nick Clifton  <nickc@redhat.com>
3727         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
3729 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
3731         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
3732         expand for QI/HImode operand to produce more effictive code for
3733         unsigned char(short) --> float(double) conversion.
3735 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
3737         * doc/invoke.texi: Mention that -ftls-model does not force the final
3738         model.
3740 2013-08-12  Marek Polacek  <polacek@redhat.com>
3741             Marc Glisse  <marc.glisse@inria.fr>
3743         PR tree-optimization/57980
3744         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
3745         when creating -1 constant.
3747 2013-08-10  Jan Hubicka  <jh@suse.cz>
3749         Workaround binutils PR14342.
3750         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
3751         (gimple_init_edge_profiler): Likewise.
3752         (gimple_gen_ic_func_profiler): Likewise.
3754 2013-08-09  Jan Hubicka  <jh@suse.cz>
3756         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
3758 2013-08-09  Xinliang David Li  <davidxl@google.com>
3760         * config/i386/stringop.def: New file.
3761         * config/i386/stringop.opt: New file.
3762         * config/i386/i386-opts.h: Include stringopt.def.
3763         * config/i386/i386.opt: Include stringopt.opt.
3764         * config/i386/i386.c (ix86_option_override_internal):
3765         Override default size based stringop inline strategies with options.
3766         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
3767         New function.
3769 2013-08-09  Jan Hubicka  <jh@suse.cz>
3771         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
3773 2013-08-09  Jan Hubicka  <jh@suse.cz>
3775         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
3776         CGRAPH_FREQ_MAX.
3777         (dump_cgraph_node): Dump profile-id.
3778         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
3779         and common_target_probability.
3780         * lto-cgraph.c (lto_output_edge): Stream common targets.
3781         (lto_output_node): Stream profile ids.
3782         (input_node): Stream profile ids.
3783         (input_edge): Stream common targets.
3784         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
3785         * ipa.c: Include value-prof.h
3786         (ipa_profile_generate_summary): Turn indirect call statement histograms
3787         into common targets.
3788         (ipa_profile): Turn common targets into speculative edges.
3790 2013-08-09  Jan Hubicka  <jh@suse.cz>
3792         * cgraph.h (cgraph_node): Add profile_id.
3793         * value-prof.c (cgraph_node_map): Turn into pointer_map.
3794         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
3795         (del_node_map): Update.
3796         (find_func_by_funcdef_no): Replace by ...
3797         (find_func_by_profile_id): ... this one.
3798         (gimple_ic_transform): Do not remove useful histograms when
3799         speculation is not done; dump info when indirect call removal
3800         can happen at LTO.
3801         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
3802         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
3803         (__gcov_indirect_call_profiler_v2): .. this one.
3804         * profile.h (init_node_map): Update.
3805         * coverage.c (coverage_compute_profile_id): New function.
3806         * coverage.h (coverage_compute_profile_id): Declare.
3807         * tree-profile.c (init_ic_make_global_vars): Make
3808         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
3809         (gimple_init_edge_profiler): Update prototype of
3810         __gcov_indirect_call_profiler.
3811         (gimple_gen_ic_func_profiler): Simplify.
3812         (tree_profiling): Use init_node_map
3814 2013-08-09  Jan Hubicka  <jh@suse.cz>
3816         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
3817         non-speculative refs.
3818         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
3819         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
3820         (cgraph_set_call_stmt): Likewise.
3821         (cgraph_create_edge_1): Fix release checking compilatoin;
3822         clear lto_stmt_uid.
3823         (cgraph_free_edge): Free indirect info.
3824         (cgraph_turn_edge_to_speculative): New function.
3825         (cgraph_speculative_call_info): New function.
3826         (cgraph_make_edge_direct): Return direct edge; handle speculation.
3827         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
3828         (dump_cgraph_node): Dump speculation.
3829         (verify_edge_count_and_frequency): Accept speculative edges.
3830         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
3831         (verify_cgraph_node): Handle speculation.
3832         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
3833         (cgraph_set_call_stmt): Update prototype.
3834         (cgraph_make_edge_direct): Update prototype.
3835         (cgraph_speculative_call_info): Declare.
3836         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
3837         to change; update call of ipa_find_references.
3838         * ipa-ref.c (ipa_record_reference): Fix return value; clear
3839         lto_stmt_uid and speculative flags.
3840         (ipa_dump_references): Dump speculation.
3841         (ipa_clone_references): Clone speculative flag.
3842         (ipa_clone_referring): Likewise.
3843         (ipa_clone_ref): New function.
3844         (ipa_find_reference): Look into lto_stmt_uids
3845         (ipa_clear_stmts_in_references): Do not clear speculative calls.
3846         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
3847         (ipa_find_reference): Update declaration.
3848         (ipa_clone_ref): Declare.
3849         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
3850         stream speculative flag.
3851         (lto_output_ref): Stream statements uids and speculation.
3852         (input_ref): Likewise.
3853         (input_edge): Stream speuclation.
3854         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
3855         (cgraph_set_call_stmt_including_clones): Handle speculation.
3856         * ipa-inline.c (heap_edge_removal_hook): New function.
3857         (inline_small_functions): Register it.
3858         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
3859         also initialize refs.
3860         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
3861         edge to change.
3862         (try_make_edge_direct_simple_call): Likewise.
3863         (try_make_edge_direct_simple_call): Likewise.
3864         (update_indirect_edges_after_inlining): Likewise.
3865         (remove_described_reference): Look proper lto_stmt_uid.
3866         (propagate_controlled_uses): Likewise.
3867         (propagate_controlled_uses): Liekwise.
3868         * tree-inline.c (copy_bb): Copy speculative edges.
3869         (redirect_all_calls): New function.
3870         (copy_cfg_body): Do redirection after loop info is updated.
3871         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
3873 2013-08-09  Jan Hubicka  <jh@suse.cz>
3875         * lto-streamer-out.c (output_function): Renumber PHIs.
3876         * lto-streamer-in.c (input_function): Likewise.
3878 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
3880         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
3881         (get_lane_unsigned): Likewise.
3882         (dup_lane_scalar): Likewise.
3883         (get_lane): enable for VALL.
3884         * config/aarch64/aarch64-simd.md
3885         (aarch64_dup_lane_scalar<mode>): Remove.
3886         (aarch64_get_lane_signed<mode>): Likewise.
3887         (aarch64_get_lane_unsigned<mode>): Likewise.
3888         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
3889         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
3890         (aarch64_get_lane<mode>): Enable for all vector modes.
3891         (aarch64_get_lanedi): Remove misleading constraints.
3892         * config/aarch64/arm_neon.h
3893         (__aarch64_vget_lane_any): Define.
3894         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
3895         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
3896         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
3897         * config/aarch64/iterators.md (VDQQH): New.
3898         (VDQQHS): Likewise.
3899         (vwcore): Likewise.
3901 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
3903         * configure.ac: Add GAS check for LEON instructions on SPARC.
3904         * configure: Regenerate.
3905         * config.in: Likewise.
3906         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
3907         sparc*-*-* block.
3908         * config/sparc/sparc.opt (LEON, LEON3): New masks.
3909         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
3910         for LEON or LEON3.
3911         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
3912         (AS_LEON_FLAG): New macro.
3913         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
3914         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
3915         Deal with LEON and LEON3 for the memory model.
3916         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
3917         (atomic_compare_and_swap<mode>_1): Likewise.
3918         (*atomic_compare_and_swap<mode>_1): Likewise.
3920 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3922         * config/arm/neon.md (vcond): Fix floating-point vector
3923         comparisons against 0.
3925 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
3927         * lra-constraints.c (emit_spill_move): Remove assert.
3928         (process_alt_operands): Add more debugging
3929         output.  Increase reject for spilling into memory.  Decrease
3930         reject for reloading scratch.
3931         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
3933 2013-08-08  Steve Ellcey  <sellcey@mips.com>
3935         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
3936         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
3937         micromips incompatible.  Add nan2008.
3938         (MULTILIB_DIRNAMES): Add nan2008.
3939         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3940         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
3941         and micromips incompatible.  Add nan2008.
3942         (MULTILIB_DIRNAMES): Add nan2008.
3943         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3945 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
3947         PR rtl-optimization/58079
3948         * combine.c (combine_simplify_rtx): Avoid using SUBST if
3949         simplify_comparison has widened a comparison with an integer.
3951 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3953         * config/arm/neon.md (movmisalign<mode>): Disable when we
3954         don't allow unaligned accesses.
3955         (*movmisalign<mode>_neon_store): Likewise.
3956         (*movmisalign<mode>_neon_load): Likewise.
3957         (*movmisalign<mode>_neon_store): Likewise.
3958         (*movmisalign<mode>_neon_load): Likewise.
3960 2013-08-08  Jan Hubicka  <jh@suse.cz>
3962         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
3963         (make_pass_rebuild_cgraph_edges): Also clear references.
3964         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
3965         * ipa-inline-transform.c (inline_transform): Remove all references
3966         after inlining.
3967         * cgraphunit.c (expand_function): Remove all references after
3968         expansion.
3969         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
3970         (ipa_find_reference): Rewrite to iterator.
3971         (remove_stmt_references): Likewise.
3972         (ipa_clear_stmts_in_references): New function.
3973         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
3974         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
3975         clear references.
3976         * ipa-split.c (split_function): Remove references in split function.
3978 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
3980         PR target/57431
3981         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
3982         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
3984 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
3986         PR target/56979
3987         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
3988         suggested mode for the assignment isn't compatible with the
3989         registers required.
3991 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3993         PR target/58065
3994         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
3996 2013-08-07  Xinliang David Li  <davidxl@google.com>
3998         * config/i386/i386.opt: New option -mtune-ctrl=.
3999         * config/i386/x86-tune.def: New file.
4000         * config/i386/i386.h: include x86-tune.def.
4001         * config/i386/i386.c (ix86_option_override_internal):
4002         Parsing -mtune-ctrl= option and set tune features.
4004 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4006         PR other/12081
4007         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4008         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4009         to gen_2arg_fn_t.
4011 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4013         * rtl.h (update_alignments): Declare.
4014         * final.c (grow_label_align): New function extracted from...
4015         (shorten_branches): ...here.  Call it.
4016         (update_alignments): New function.
4017         * reorg.c (sibling_labels): New variable.
4018         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4019         the new label along with it onto the sibling_labels vector.
4020         (fill_simple_delay_slots): Adjust call to get_label_before.
4021         (fill_slots_from_thread): Likewise.
4022         (relax_delay_slots): Likewise.
4023         (make_return_insns): Likewise.
4024         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4026 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4028         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4029         document its semantics.
4030         (diagnostic_report_diagnostic): Adjust accordingly.
4032 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4034         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4035         (sparc_option_override): ...and port to new C++ pass API.
4036         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4038 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4040         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4042 2013-08-06  Caroline Tice  <cmtice@google.com>
4044         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4045         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4046         * tree-pass.h: Add pass_vtable_verify.
4047         * varasm.c (assemble_variable): Add code to properly set the comdat
4048         section and name for the .vtable_map_vars section.
4049         (assemble_vtyv_preinit_initializer): New function.
4050         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4051         LINK_ONCE flag.
4052         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4053         * vtable-verify.c: New file.
4054         * vtable-verify.h: New file.
4055         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4056         initialiation levels.
4057         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4058         * passes.def: Insert pass_vtable_verify.
4059         * aclocal.m4: Reorder includes.
4060         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4061         -fvtv-counts options.
4062         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4063         as appropriate, if -fvtable-verify=... is used.
4064         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4065         -fvtable-verify=... is used.
4066         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4067         (vtable-verify.o): Add new build rule.
4068         (GTFILES): Add vtable-verify.c to list.
4069         * common.opt (fvtable-verify=): New flag.
4070         (vtv_priority): Values for fvtable-verify= flag.
4071         (fvtv-counts): New flag.
4072         (fvtv-debug): New flag.
4073         * tree.h (save_vtable_map_decl): New extern function decl.
4075 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4077         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4078         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4079         (pass_data_rl78_devirt): ...new pass_data instance and...
4080         (make_pass_rl78_devirt): ...new function.
4081         (rl78_asm_file_start): Port pass registration to new C++ API.
4083 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4085         * coretypes.h (rtl_opt_pass): Add.
4086         (gcc::context): Add.
4087         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4088         (epiphany_init): Port to new C++ pass API.
4089         (epiphany_optimize_mode_switching): Likewise.
4090         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4091         (pass_manager::get_pass_mode_switching): New.
4092         (pass_manager::get_pass_peephole2): New.
4093         * mode-switching.c (pass_mode_switching): Add clone method.
4094         * recog.c (pass_peephole2): Add clone method.
4095         (pass_split_all_insns): Add clone method.
4097 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4099         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4100         (mips_option_override): ...here, porting to new C++ API for passes.
4102 2013-08-06  Jan Hubicka  <jh@suse.cz>
4104         * cgraph.c (cgraph_get_body): New function based on lto.c
4105         implementation.
4106         * cgraph.h (cgraph_get_body): Declare.
4107         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4108         LTO paths.
4109         * cgraphunit.c (expand_function): Get body prior expanding.
4110         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4111         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4112         really need.
4113         * passes.c (do_per_function_toporder): Get body.
4114         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4115         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4117 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4119         PR fortran/57987
4120         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4121         is not re-finalized.  Rename second parameter to no_collect.
4123 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4125         PR middle-end/58041
4126         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4127         MEM_REF has proper alignment information.
4129 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4131         PR other/12081
4132         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4133         class insn_gen_fn.
4134         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4135         rtx (*) (rtx, ...) with insn_gen_fn.
4136         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4137         insn_gen_fn::stored_funcptr.  Add initializer braces.
4139 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4141         Rewrite how instances of passes are cloned to remove assumptions
4142         about their sizes (thus allowing pass subclasses to have
4143         additional data fields, albeit non-GC-managed ones at this point).
4145         * passes.c (make_pass_instance): Now that passes have clone
4146         methods, rewrite this function to eliminate XNEW and memcpy
4147         calls that used hardcoded sizes.  Since this function no longer
4148         creates pass instances, rename it to...
4149         (add_pass_instance): ...this.  Document the old way that passes were
4150         numbered and flagged, and rework this function to continue using it.
4151         (next_pass_1): Add an initial_pass argument for use by
4152         add_pass_instance.
4153         (position_pass): When adding multiple instances of a pass, use
4154         the pass's clone method, rather than relying on the XNEW/memcpy
4155         within the former make_pass_instance (now add_pass_instance).
4156         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4157         the initial instance of the current pass within the pass manager.
4159 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4161         This is the automated part of the conversion of passes from C
4162         structs to C++ classes.
4164         Patch autogenerated by refactor_passes.py from
4165         https://github.com/davidmalcolm/gcc-refactoring-scripts
4166         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4168         * asan.c (pass_asan): Convert from a global struct to a subclass of
4169         gimple_opt_pass along with...
4170         (pass_data_asan): ...new pass_data instance and...
4171         (make_pass_asan): ...new function.
4172         (pass_asan_O0): Convert from a global struct to a subclass of
4173         gimple_opt_pass along with...
4174         (pass_data_asan_O0): ...new pass_data instance and...
4175         (make_pass_asan_O0): ...new function.
4176         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4177         subclass of rtl_opt_pass along with...
4178         (pass_data_inc_dec): ...new pass_data instance and...
4179         (make_pass_inc_dec): ...new function.
4180         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4181         a subclass of rtl_opt_pass along with...
4182         (pass_data_reorder_blocks): ...new pass_data instance and...
4183         (make_pass_reorder_blocks): ...new function.
4184         (pass_duplicate_computed_gotos): Convert from a global struct to a
4185         subclass of rtl_opt_pass along with...
4186         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4187         (make_pass_duplicate_computed_gotos): ...new function.
4188         (pass_partition_blocks): Convert from a global struct to a subclass of
4189         rtl_opt_pass along with...
4190         (pass_data_partition_blocks): ...new pass_data instance and...
4191         (make_pass_partition_blocks): ...new function.
4192         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4193         struct to a subclass of rtl_opt_pass along with...
4194         (pass_data_branch_target_load_optimize1): ...new pass_data instance
4195         and...
4196         (make_pass_branch_target_load_optimize1): ...new function.
4197         (pass_branch_target_load_optimize2): Convert from a global struct to a
4198         subclass of rtl_opt_pass along with...
4199         (pass_data_branch_target_load_optimize2): ...new pass_data instance
4200         and...
4201         (make_pass_branch_target_load_optimize2): ...new function.
4202         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4203         of rtl_opt_pass along with...
4204         (pass_data_jump): ...new pass_data instance and...
4205         (make_pass_jump): ...new function.
4206         (pass_jump2): Convert from a global struct to a subclass of
4207         rtl_opt_pass along with...
4208         (pass_data_jump2): ...new pass_data instance and...
4209         (make_pass_jump2): ...new function.
4210         * cfgexpand.c (pass_expand): Convert from a global struct to a
4211         subclass of rtl_opt_pass along with...
4212         (pass_data_expand): ...new pass_data instance and...
4213         (make_pass_expand): ...new function.
4214         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4215         of rtl_opt_pass along with...
4216         (pass_data_free_cfg): ...new pass_data instance and...
4217         (make_pass_free_cfg): ...new function.
4218         (pass_into_cfg_layout_mode): Convert from a global struct to a
4219         subclass of rtl_opt_pass along with...
4220         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4221         (make_pass_into_cfg_layout_mode): ...new function.
4222         (pass_outof_cfg_layout_mode): Convert from a global struct to a
4223         subclass of rtl_opt_pass along with...
4224         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4225         (make_pass_outof_cfg_layout_mode): ...new function.
4226         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4227         struct to a subclass of gimple_opt_pass along with...
4228         (pass_data_build_cgraph_edges): ...new pass_data instance and...
4229         (make_pass_build_cgraph_edges): ...new function.
4230         (pass_rebuild_cgraph_edges): Convert from a global struct to a
4231         subclass of gimple_opt_pass along with...
4232         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4233         (make_pass_rebuild_cgraph_edges): ...new function.
4234         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4235         subclass of gimple_opt_pass along with...
4236         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4237         and...
4238         (make_pass_remove_cgraph_callee_edges): ...new function.
4239         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4240         struct to a subclass of rtl_opt_pass along with...
4241         (pass_data_stack_adjustments): ...new pass_data instance and...
4242         (make_pass_stack_adjustments): ...new function.
4243         * combine.c (pass_combine): Convert from a global struct to a subclass
4244         of rtl_opt_pass along with...
4245         (pass_data_combine): ...new pass_data instance and...
4246         (make_pass_combine): ...new function.
4247         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4248         global struct to a subclass of rtl_opt_pass along with...
4249         (pass_data_compare_elim_after_reload): ...new pass_data instance
4250         and...
4251         (make_pass_compare_elim_after_reload): ...new function.
4252         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4253         of rtl_opt_pass along with...
4254         (pass_data_rtl_cprop): ...new pass_data instance and...
4255         (make_pass_rtl_cprop): ...new function.
4256         * cse.c (pass_cse): Convert from a global struct to a subclass of
4257         rtl_opt_pass along with...
4258         (pass_data_cse): ...new pass_data instance and...
4259         (make_pass_cse): ...new function.
4260         (pass_cse2): Convert from a global struct to a subclass of
4261         rtl_opt_pass along with...
4262         (pass_data_cse2): ...new pass_data instance and...
4263         (make_pass_cse2): ...new function.
4264         (pass_cse_after_global_opts): Convert from a global struct to a
4265         subclass of rtl_opt_pass along with...
4266         (pass_data_cse_after_global_opts): ...new pass_data instance and...
4267         (make_pass_cse_after_global_opts): ...new function.
4268         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4269         of rtl_opt_pass along with...
4270         (pass_data_ud_rtl_dce): ...new pass_data instance and...
4271         (make_pass_ud_rtl_dce): ...new function.
4272         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4273         rtl_opt_pass along with...
4274         (pass_data_fast_rtl_dce): ...new pass_data instance and...
4275         (make_pass_fast_rtl_dce): ...new function.
4276         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4277         a subclass of rtl_opt_pass along with...
4278         (pass_data_df_initialize_opt): ...new pass_data instance and...
4279         (make_pass_df_initialize_opt): ...new function.
4280         (pass_df_initialize_no_opt): Convert from a global struct to a
4281         subclass of rtl_opt_pass along with...
4282         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4283         (make_pass_df_initialize_no_opt): ...new function.
4284         (pass_df_finish): Convert from a global struct to a subclass of
4285         rtl_opt_pass along with...
4286         (pass_data_df_finish): ...new pass_data instance and...
4287         (make_pass_df_finish): ...new function.
4288         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4289         rtl_opt_pass along with...
4290         (pass_data_rtl_dse1): ...new pass_data instance and...
4291         (make_pass_rtl_dse1): ...new function.
4292         (pass_rtl_dse2): Convert from a global struct to a subclass of
4293         rtl_opt_pass along with...
4294         (pass_data_rtl_dse2): ...new pass_data instance and...
4295         (make_pass_rtl_dse2): ...new function.
4296         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4297         subclass of rtl_opt_pass along with...
4298         (pass_data_dwarf2_frame): ...new pass_data instance and...
4299         (make_pass_dwarf2_frame): ...new function.
4300         * except.c (pass_set_nothrow_function_flags): Convert from a global
4301         struct to a subclass of rtl_opt_pass along with...
4302         (pass_data_set_nothrow_function_flags): ...new pass_data instance
4303         and...
4304         (make_pass_set_nothrow_function_flags): ...new function.
4305         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4306         subclass of rtl_opt_pass along with...
4307         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4308         and...
4309         (make_pass_convert_to_eh_region_ranges): ...new function.
4310         * final.c (pass_compute_alignments): Convert from a global struct to a
4311         subclass of rtl_opt_pass along with...
4312         (pass_data_compute_alignments): ...new pass_data instance and...
4313         (make_pass_compute_alignments): ...new function.
4314         (pass_final): Convert from a global struct to a subclass of
4315         rtl_opt_pass along with...
4316         (pass_data_final): ...new pass_data instance and...
4317         (make_pass_final): ...new function.
4318         (pass_shorten_branches): Convert from a global struct to a subclass of
4319         rtl_opt_pass along with...
4320         (pass_data_shorten_branches): ...new pass_data instance and...
4321         (make_pass_shorten_branches): ...new function.
4322         (pass_clean_state): Convert from a global struct to a subclass of
4323         rtl_opt_pass along with...
4324         (pass_data_clean_state): ...new pass_data instance and...
4325         (make_pass_clean_state): ...new function.
4326         * function.c (pass_instantiate_virtual_regs): Convert from a global
4327         struct to a subclass of rtl_opt_pass along with...
4328         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4329         (make_pass_instantiate_virtual_regs): ...new function.
4330         (pass_leaf_regs): Convert from a global struct to a subclass of
4331         rtl_opt_pass along with...
4332         (pass_data_leaf_regs): ...new pass_data instance and...
4333         (make_pass_leaf_regs): ...new function.
4334         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4335         subclass of rtl_opt_pass along with...
4336         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4337         and...
4338         (make_pass_thread_prologue_and_epilogue): ...new function.
4339         (pass_match_asm_constraints): Convert from a global struct to a
4340         subclass of rtl_opt_pass along with...
4341         (pass_data_match_asm_constraints): ...new pass_data instance and...
4342         (make_pass_match_asm_constraints): ...new function.
4343         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4344         subclass of rtl_opt_pass along with...
4345         (pass_data_rtl_fwprop): ...new pass_data instance and...
4346         (make_pass_rtl_fwprop): ...new function.
4347         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4348         rtl_opt_pass along with...
4349         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4350         (make_pass_rtl_fwprop_addr): ...new function.
4351         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4352         rtl_opt_pass along with...
4353         (pass_data_rtl_pre): ...new pass_data instance and...
4354         (make_pass_rtl_pre): ...new function.
4355         (pass_rtl_hoist): Convert from a global struct to a subclass of
4356         rtl_opt_pass along with...
4357         (pass_data_rtl_hoist): ...new pass_data instance and...
4358         (make_pass_rtl_hoist): ...new function.
4359         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4360         subclass of gimple_opt_pass along with...
4361         (pass_data_lower_cf): ...new pass_data instance and...
4362         (make_pass_lower_cf): ...new function.
4363         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4364         from a global struct to a subclass of gimple_opt_pass along with...
4365         (pass_data_strength_reduction): ...new pass_data instance and...
4366         (make_pass_strength_reduction): ...new function.
4367         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4368         of rtl_opt_pass along with...
4369         (pass_data_rtl_ifcvt): ...new pass_data instance and...
4370         (make_pass_rtl_ifcvt): ...new function.
4371         (pass_if_after_combine): Convert from a global struct to a subclass of
4372         rtl_opt_pass along with...
4373         (pass_data_if_after_combine): ...new pass_data instance and...
4374         (make_pass_if_after_combine): ...new function.
4375         (pass_if_after_reload): Convert from a global struct to a subclass of
4376         rtl_opt_pass along with...
4377         (pass_data_if_after_reload): ...new pass_data instance and...
4378         (make_pass_if_after_reload): ...new function.
4379         * init-regs.c (pass_initialize_regs): Convert from a global struct to
4380         a subclass of rtl_opt_pass along with...
4381         (pass_data_initialize_regs): ...new pass_data instance and...
4382         (make_pass_initialize_regs): ...new function.
4383         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4384         of ipa_opt_pass_d along with...
4385         (pass_data_ipa_cp): ...new pass_data instance and...
4386         (make_pass_ipa_cp): ...new function.
4387         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4388         global struct to a subclass of gimple_opt_pass along with...
4389         (pass_data_inline_parameters): ...new pass_data instance and...
4390         (make_pass_inline_parameters): ...new function.
4391         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4392         subclass of gimple_opt_pass along with...
4393         (pass_data_early_inline): ...new pass_data instance and...
4394         (make_pass_early_inline): ...new function.
4395         (pass_ipa_inline): Convert from a global struct to a subclass of
4396         ipa_opt_pass_d along with...
4397         (pass_data_ipa_inline): ...new pass_data instance and...
4398         (make_pass_ipa_inline): ...new function.
4399         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4400         struct to a subclass of gimple_opt_pass along with...
4401         (pass_data_local_pure_const): ...new pass_data instance and...
4402         (make_pass_local_pure_const): ...new function.
4403         (pass_ipa_pure_const): Convert from a global struct to a subclass of
4404         ipa_opt_pass_d along with...
4405         (pass_data_ipa_pure_const): ...new pass_data instance and...
4406         (make_pass_ipa_pure_const): ...new function.
4407         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4408         to a subclass of ipa_opt_pass_d along with...
4409         (pass_data_ipa_reference): ...new pass_data instance and...
4410         (make_pass_ipa_reference): ...new function.
4411         * ipa-split.c (pass_split_functions): Convert from a global struct to
4412         a subclass of gimple_opt_pass along with...
4413         (pass_data_split_functions): ...new pass_data instance and...
4414         (make_pass_split_functions): ...new function.
4415         (pass_feedback_split_functions): Convert from a global struct to a
4416         subclass of gimple_opt_pass along with...
4417         (pass_data_feedback_split_functions): ...new pass_data instance and...
4418         (make_pass_feedback_split_functions): ...new function.
4419         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4420         global struct to a subclass of simple_ipa_opt_pass along with...
4421         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4422         instance and...
4423         (make_pass_ipa_function_and_variable_visibility): ...new function.
4424         (pass_ipa_free_inline_summary): Convert from a global struct to a
4425         subclass of simple_ipa_opt_pass along with...
4426         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4427         (make_pass_ipa_free_inline_summary): ...new function.
4428         (pass_ipa_whole_program_visibility): Convert from a global struct to a
4429         subclass of ipa_opt_pass_d along with...
4430         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4431         and...
4432         (make_pass_ipa_whole_program_visibility): ...new function.
4433         (pass_ipa_profile): Convert from a global struct to a subclass of
4434         ipa_opt_pass_d along with...
4435         (pass_data_ipa_profile): ...new pass_data instance and...
4436         (make_pass_ipa_profile): ...new function.
4437         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4438         ipa_opt_pass_d along with...
4439         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4440         (make_pass_ipa_cdtor_merge): ...new function.
4441         * ira.c (pass_ira): Convert from a global struct to a subclass of
4442         rtl_opt_pass along with...
4443         (pass_data_ira): ...new pass_data instance and...
4444         (make_pass_ira): ...new function.
4445         (pass_reload): Convert from a global struct to a subclass of
4446         rtl_opt_pass along with...
4447         (pass_data_reload): ...new pass_data instance and...
4448         (make_pass_reload): ...new function.
4449         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4450         subclass of rtl_opt_pass along with...
4451         (pass_data_cleanup_barriers): ...new pass_data instance and...
4452         (make_pass_cleanup_barriers): ...new function.
4453         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4454         of rtl_opt_pass along with...
4455         (pass_data_loop2): ...new pass_data instance and...
4456         (make_pass_loop2): ...new function.
4457         (pass_rtl_loop_init): Convert from a global struct to a subclass of
4458         rtl_opt_pass along with...
4459         (pass_data_rtl_loop_init): ...new pass_data instance and...
4460         (make_pass_rtl_loop_init): ...new function.
4461         (pass_rtl_loop_done): Convert from a global struct to a subclass of
4462         rtl_opt_pass along with...
4463         (pass_data_rtl_loop_done): ...new pass_data instance and...
4464         (make_pass_rtl_loop_done): ...new function.
4465         (pass_rtl_move_loop_invariants): Convert from a global struct to a
4466         subclass of rtl_opt_pass along with...
4467         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4468         (make_pass_rtl_move_loop_invariants): ...new function.
4469         (pass_rtl_unswitch): Convert from a global struct to a subclass of
4470         rtl_opt_pass along with...
4471         (pass_data_rtl_unswitch): ...new pass_data instance and...
4472         (make_pass_rtl_unswitch): ...new function.
4473         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4474         subclass of rtl_opt_pass along with...
4475         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4476         and...
4477         (make_pass_rtl_unroll_and_peel_loops): ...new function.
4478         (pass_rtl_doloop): Convert from a global struct to a subclass of
4479         rtl_opt_pass along with...
4480         (pass_data_rtl_doloop): ...new pass_data instance and...
4481         (make_pass_rtl_doloop): ...new function.
4482         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4483         a subclass of rtl_opt_pass along with...
4484         (pass_data_lower_subreg): ...new pass_data instance and...
4485         (make_pass_lower_subreg): ...new function.
4486         (pass_lower_subreg2): Convert from a global struct to a subclass of
4487         rtl_opt_pass along with...
4488         (pass_data_lower_subreg2): ...new pass_data instance and...
4489         (make_pass_lower_subreg2): ...new function.
4490         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4491         struct to a subclass of ipa_opt_pass_d along with...
4492         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4493         (make_pass_ipa_lto_gimple_out): ...new function.
4494         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4495         of ipa_opt_pass_d along with...
4496         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4497         (make_pass_ipa_lto_finish_out): ...new function.
4498         * mode-switching.c (pass_mode_switching): Convert from a global struct
4499         to a subclass of rtl_opt_pass along with...
4500         (pass_data_mode_switching): ...new pass_data instance and...
4501         (make_pass_mode_switching): ...new function.
4502         * modulo-sched.c (pass_sms): Convert from a global struct to a
4503         subclass of rtl_opt_pass along with...
4504         (pass_data_sms): ...new pass_data instance and...
4505         (make_pass_sms): ...new function.
4506         * omp-low.c (pass_expand_omp): Convert from a global struct to a
4507         subclass of gimple_opt_pass along with...
4508         (pass_data_expand_omp): ...new pass_data instance and...
4509         (make_pass_expand_omp): ...new function.
4510         (pass_lower_omp): Convert from a global struct to a subclass of
4511         gimple_opt_pass along with...
4512         (pass_data_lower_omp): ...new pass_data instance and...
4513         (make_pass_lower_omp): ...new function.
4514         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4515         of gimple_opt_pass along with...
4516         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4517         (make_pass_diagnose_omp_blocks): ...new function.
4518         * passes.c (pass_early_local_passes): Convert from a global struct to
4519         a subclass of simple_ipa_opt_pass along with...
4520         (pass_data_early_local_passes): ...new pass_data instance and...
4521         (make_pass_early_local_passes): ...new function.
4522         (pass_all_early_optimizations): Convert from a global struct to a
4523         subclass of gimple_opt_pass along with...
4524         (pass_data_all_early_optimizations): ...new pass_data instance and...
4525         (make_pass_all_early_optimizations): ...new function.
4526         (pass_all_optimizations): Convert from a global struct to a subclass
4527         of gimple_opt_pass along with...
4528         (pass_data_all_optimizations): ...new pass_data instance and...
4529         (make_pass_all_optimizations): ...new function.
4530         (pass_all_optimizations_g): Convert from a global struct to a subclass
4531         of gimple_opt_pass along with...
4532         (pass_data_all_optimizations_g): ...new pass_data instance and...
4533         (make_pass_all_optimizations_g): ...new function.
4534         (pass_rest_of_compilation): Convert from a global struct to a subclass
4535         of rtl_opt_pass along with...
4536         (pass_data_rest_of_compilation): ...new pass_data instance and...
4537         (make_pass_rest_of_compilation): ...new function.
4538         (pass_postreload): Convert from a global struct to a subclass of
4539         rtl_opt_pass along with...
4540         (pass_data_postreload): ...new pass_data instance and...
4541         (make_pass_postreload): ...new function.
4542         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
4543         subclass of rtl_opt_pass along with...
4544         (pass_data_gcse2): ...new pass_data instance and...
4545         (make_pass_gcse2): ...new function.
4546         * postreload.c (pass_postreload_cse): Convert from a global struct to
4547         a subclass of rtl_opt_pass along with...
4548         (pass_data_postreload_cse): ...new pass_data instance and...
4549         (make_pass_postreload_cse): ...new function.
4550         * predict.c (pass_profile): Convert from a global struct to a subclass
4551         of gimple_opt_pass along with...
4552         (pass_data_profile): ...new pass_data instance and...
4553         (make_pass_profile): ...new function.
4554         (pass_strip_predict_hints): Convert from a global struct to a subclass
4555         of gimple_opt_pass along with...
4556         (pass_data_strip_predict_hints): ...new pass_data instance and...
4557         (make_pass_strip_predict_hints): ...new function.
4558         * recog.c (pass_peephole2): Convert from a global struct to a subclass
4559         of rtl_opt_pass along with...
4560         (pass_data_peephole2): ...new pass_data instance and...
4561         (make_pass_peephole2): ...new function.
4562         (pass_split_all_insns): Convert from a global struct to a subclass of
4563         rtl_opt_pass along with...
4564         (pass_data_split_all_insns): ...new pass_data instance and...
4565         (make_pass_split_all_insns): ...new function.
4566         (pass_split_after_reload): Convert from a global struct to a subclass
4567         of rtl_opt_pass along with...
4568         (pass_data_split_after_reload): ...new pass_data instance and...
4569         (make_pass_split_after_reload): ...new function.
4570         (pass_split_before_regstack): Convert from a global struct to a
4571         subclass of rtl_opt_pass along with...
4572         (pass_data_split_before_regstack): ...new pass_data instance and...
4573         (make_pass_split_before_regstack): ...new function.
4574         (pass_split_before_sched2): Convert from a global struct to a subclass
4575         of rtl_opt_pass along with...
4576         (pass_data_split_before_sched2): ...new pass_data instance and...
4577         (make_pass_split_before_sched2): ...new function.
4578         (pass_split_for_shorten_branches): Convert from a global struct to a
4579         subclass of rtl_opt_pass along with...
4580         (pass_data_split_for_shorten_branches): ...new pass_data instance
4581         and...
4582         (make_pass_split_for_shorten_branches): ...new function.
4583         * ree.c (pass_ree): Convert from a global struct to a subclass of
4584         rtl_opt_pass along with...
4585         (pass_data_ree): ...new pass_data instance and...
4586         (make_pass_ree): ...new function.
4587         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
4588         subclass of rtl_opt_pass along with...
4589         (pass_data_stack_regs): ...new pass_data instance and...
4590         (make_pass_stack_regs): ...new function.
4591         (pass_stack_regs_run): Convert from a global struct to a subclass of
4592         rtl_opt_pass along with...
4593         (pass_data_stack_regs_run): ...new pass_data instance and...
4594         (make_pass_stack_regs_run): ...new function.
4595         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
4596         subclass of rtl_opt_pass along with...
4597         (pass_data_cprop_hardreg): ...new pass_data instance and...
4598         (make_pass_cprop_hardreg): ...new function.
4599         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
4600         subclass of rtl_opt_pass along with...
4601         (pass_data_reginfo_init): ...new pass_data instance and...
4602         (make_pass_reginfo_init): ...new function.
4603         * regmove.c (pass_regmove): Convert from a global struct to a subclass
4604         of rtl_opt_pass along with...
4605         (pass_data_regmove): ...new pass_data instance and...
4606         (make_pass_regmove): ...new function.
4607         * regrename.c (pass_regrename): Convert from a global struct to a
4608         subclass of rtl_opt_pass along with...
4609         (pass_data_regrename): ...new pass_data instance and...
4610         (make_pass_regrename): ...new function.
4611         * reorg.c (pass_delay_slots): Convert from a global struct to a
4612         subclass of rtl_opt_pass along with...
4613         (pass_data_delay_slots): ...new pass_data instance and...
4614         (make_pass_delay_slots): ...new function.
4615         (pass_machine_reorg): Convert from a global struct to a subclass of
4616         rtl_opt_pass along with...
4617         (pass_data_machine_reorg): ...new pass_data instance and...
4618         (make_pass_machine_reorg): ...new function.
4619         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
4620         of rtl_opt_pass along with...
4621         (pass_data_sched): ...new pass_data instance and...
4622         (make_pass_sched): ...new function.
4623         (pass_sched2): Convert from a global struct to a subclass of
4624         rtl_opt_pass along with...
4625         (pass_data_sched2): ...new pass_data instance and...
4626         (make_pass_sched2): ...new function.
4627         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
4628         to a subclass of rtl_opt_pass along with...
4629         (pass_data_stack_ptr_mod): ...new pass_data instance and...
4630         (make_pass_stack_ptr_mod): ...new function.
4631         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
4632         to a subclass of rtl_opt_pass along with...
4633         (pass_data_rtl_store_motion): ...new pass_data instance and...
4634         (make_pass_rtl_store_motion): ...new function.
4635         * tracer.c (pass_tracer): Convert from a global struct to a subclass
4636         of gimple_opt_pass along with...
4637         (pass_data_tracer): ...new pass_data instance and...
4638         (make_pass_tracer): ...new function.
4639         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
4640         to a subclass of gimple_opt_pass along with...
4641         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
4642         (make_pass_diagnose_tm_blocks): ...new function.
4643         (pass_lower_tm): Convert from a global struct to a subclass of
4644         gimple_opt_pass along with...
4645         (pass_data_lower_tm): ...new pass_data instance and...
4646         (make_pass_lower_tm): ...new function.
4647         (pass_tm_init): Convert from a global struct to a subclass of
4648         gimple_opt_pass along with...
4649         (pass_data_tm_init): ...new pass_data instance and...
4650         (make_pass_tm_init): ...new function.
4651         (pass_tm_mark): Convert from a global struct to a subclass of
4652         gimple_opt_pass along with...
4653         (pass_data_tm_mark): ...new pass_data instance and...
4654         (make_pass_tm_mark): ...new function.
4655         (pass_tm_edges): Convert from a global struct to a subclass of
4656         gimple_opt_pass along with...
4657         (pass_data_tm_edges): ...new pass_data instance and...
4658         (make_pass_tm_edges): ...new function.
4659         (pass_tm_memopt): Convert from a global struct to a subclass of
4660         gimple_opt_pass along with...
4661         (pass_data_tm_memopt): ...new pass_data instance and...
4662         (make_pass_tm_memopt): ...new function.
4663         (pass_ipa_tm): Convert from a global struct to a subclass of
4664         simple_ipa_opt_pass along with...
4665         (pass_data_ipa_tm): ...new pass_data instance and...
4666         (make_pass_ipa_tm): ...new function.
4667         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
4668         subclass of gimple_opt_pass along with...
4669         (pass_data_call_cdce): ...new pass_data instance and...
4670         (make_pass_call_cdce): ...new function.
4671         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
4672         subclass of gimple_opt_pass along with...
4673         (pass_data_build_cfg): ...new pass_data instance and...
4674         (make_pass_build_cfg): ...new function.
4675         (pass_split_crit_edges): Convert from a global struct to a subclass of
4676         gimple_opt_pass along with...
4677         (pass_data_split_crit_edges): ...new pass_data instance and...
4678         (make_pass_split_crit_edges): ...new function.
4679         (pass_warn_function_return): Convert from a global struct to a
4680         subclass of gimple_opt_pass along with...
4681         (pass_data_warn_function_return): ...new pass_data instance and...
4682         (make_pass_warn_function_return): ...new function.
4683         (pass_warn_function_noreturn): Convert from a global struct to a
4684         subclass of gimple_opt_pass along with...
4685         (pass_data_warn_function_noreturn): ...new pass_data instance and...
4686         (make_pass_warn_function_noreturn): ...new function.
4687         (pass_warn_unused_result): Convert from a global struct to a subclass
4688         of gimple_opt_pass along with...
4689         (pass_data_warn_unused_result): ...new pass_data instance and...
4690         (make_pass_warn_unused_result): ...new function.
4691         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
4692         a subclass of gimple_opt_pass along with...
4693         (pass_data_merge_phi): ...new pass_data instance and...
4694         (make_pass_merge_phi): ...new function.
4695         * tree-complex.c (pass_lower_complex): Convert from a global struct to
4696         a subclass of gimple_opt_pass along with...
4697         (pass_data_lower_complex): ...new pass_data instance and...
4698         (make_pass_lower_complex): ...new function.
4699         (pass_lower_complex_O0): Convert from a global struct to a subclass of
4700         gimple_opt_pass along with...
4701         (pass_data_lower_complex_O0): ...new pass_data instance and...
4702         (make_pass_lower_complex_O0): ...new function.
4703         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
4704         subclass of gimple_opt_pass along with...
4705         (pass_data_lower_eh): ...new pass_data instance and...
4706         (make_pass_lower_eh): ...new function.
4707         (pass_refactor_eh): Convert from a global struct to a subclass of
4708         gimple_opt_pass along with...
4709         (pass_data_refactor_eh): ...new pass_data instance and...
4710         (make_pass_refactor_eh): ...new function.
4711         (pass_lower_resx): Convert from a global struct to a subclass of
4712         gimple_opt_pass along with...
4713         (pass_data_lower_resx): ...new pass_data instance and...
4714         (make_pass_lower_resx): ...new function.
4715         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
4716         of gimple_opt_pass along with...
4717         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
4718         (make_pass_lower_eh_dispatch): ...new function.
4719         (pass_cleanup_eh): Convert from a global struct to a subclass of
4720         gimple_opt_pass along with...
4721         (pass_data_cleanup_eh): ...new pass_data instance and...
4722         (make_pass_cleanup_eh): ...new function.
4723         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
4724         to a subclass of simple_ipa_opt_pass along with...
4725         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
4726         (make_pass_ipa_lower_emutls): ...new function.
4727         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
4728         a subclass of gimple_opt_pass along with...
4729         (pass_data_if_conversion): ...new pass_data instance and...
4730         (make_pass_if_conversion): ...new function.
4731         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
4732         subclass of gimple_opt_pass along with...
4733         (pass_data_build_ssa): ...new pass_data instance and...
4734         (make_pass_build_ssa): ...new function.
4735         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
4736         global struct to a subclass of gimple_opt_pass along with...
4737         (pass_data_loop_distribution): ...new pass_data instance and...
4738         (make_pass_loop_distribution): ...new function.
4739         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
4740         subclass of gimple_opt_pass along with...
4741         (pass_data_mudflap_1): ...new pass_data instance and...
4742         (make_pass_mudflap_1): ...new function.
4743         (pass_mudflap_2): Convert from a global struct to a subclass of
4744         gimple_opt_pass along with...
4745         (pass_data_mudflap_2): ...new pass_data instance and...
4746         (make_pass_mudflap_2): ...new function.
4747         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
4748         subclass of gimple_opt_pass along with...
4749         (pass_data_mudflap_1): ...new pass_data instance and...
4750         (make_pass_mudflap_1): ...new function.
4751         (pass_mudflap_2): Convert from a global struct to a subclass of
4752         gimple_opt_pass along with...
4753         (pass_data_mudflap_2): ...new pass_data instance and...
4754         (make_pass_mudflap_2): ...new function.
4755         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
4756         gimple_opt_pass along with...
4757         (pass_data_nrv): ...new pass_data instance and...
4758         (make_pass_nrv): ...new function.
4759         (pass_return_slot): Convert from a global struct to a subclass of
4760         gimple_opt_pass along with...
4761         (pass_data_return_slot): ...new pass_data instance and...
4762         (make_pass_return_slot): ...new function.
4763         * tree-object-size.c (pass_object_sizes): Convert from a global struct
4764         to a subclass of gimple_opt_pass along with...
4765         (pass_data_object_sizes): ...new pass_data instance and...
4766         (make_pass_object_sizes): ...new function.
4767         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
4768         global struct to a subclass of gimple_opt_pass along with...
4769         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
4770         and...
4771         (make_pass_cleanup_cfg_post_optimizing): ...new function.
4772         (pass_fixup_cfg): Convert from a global struct to a subclass of
4773         gimple_opt_pass along with...
4774         (pass_data_fixup_cfg): ...new pass_data instance and...
4775         (make_pass_fixup_cfg): ...new function.
4776         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
4777         (make_pass_mudflap_1): ...new function.
4778         (pass_mudflap_2): Replace declaration with that of...
4779         (make_pass_mudflap_2): ...new function.
4780         (pass_asan): Replace declaration with that of...
4781         (make_pass_asan): ...new function.
4782         (pass_asan_O0): Replace declaration with that of...
4783         (make_pass_asan_O0): ...new function.
4784         (pass_tsan): Replace declaration with that of...
4785         (make_pass_tsan): ...new function.
4786         (pass_tsan_O0): Replace declaration with that of...
4787         (make_pass_tsan_O0): ...new function.
4788         (pass_lower_cf): Replace declaration with that of...
4789         (make_pass_lower_cf): ...new function.
4790         (pass_refactor_eh): Replace declaration with that of...
4791         (make_pass_refactor_eh): ...new function.
4792         (pass_lower_eh): Replace declaration with that of...
4793         (make_pass_lower_eh): ...new function.
4794         (pass_lower_eh_dispatch): Replace declaration with that of...
4795         (make_pass_lower_eh_dispatch): ...new function.
4796         (pass_lower_resx): Replace declaration with that of...
4797         (make_pass_lower_resx): ...new function.
4798         (pass_build_cfg): Replace declaration with that of...
4799         (make_pass_build_cfg): ...new function.
4800         (pass_early_tree_profile): Replace declaration with that of...
4801         (make_pass_early_tree_profile): ...new function.
4802         (pass_cleanup_eh): Replace declaration with that of...
4803         (make_pass_cleanup_eh): ...new function.
4804         (pass_sra): Replace declaration with that of...
4805         (make_pass_sra): ...new function.
4806         (pass_sra_early): Replace declaration with that of...
4807         (make_pass_sra_early): ...new function.
4808         (pass_early_ipa_sra): Replace declaration with that of...
4809         (make_pass_early_ipa_sra): ...new function.
4810         (pass_tail_recursion): Replace declaration with that of...
4811         (make_pass_tail_recursion): ...new function.
4812         (pass_tail_calls): Replace declaration with that of...
4813         (make_pass_tail_calls): ...new function.
4814         (pass_tree_loop): Replace declaration with that of...
4815         (make_pass_tree_loop): ...new function.
4816         (pass_tree_loop_init): Replace declaration with that of...
4817         (make_pass_tree_loop_init): ...new function.
4818         (pass_lim): Replace declaration with that of...
4819         (make_pass_lim): ...new function.
4820         (pass_tree_unswitch): Replace declaration with that of...
4821         (make_pass_tree_unswitch): ...new function.
4822         (pass_predcom): Replace declaration with that of...
4823         (make_pass_predcom): ...new function.
4824         (pass_iv_canon): Replace declaration with that of...
4825         (make_pass_iv_canon): ...new function.
4826         (pass_scev_cprop): Replace declaration with that of...
4827         (make_pass_scev_cprop): ...new function.
4828         (pass_empty_loop): Replace declaration with that of...
4829         (make_pass_empty_loop): ...new function.
4830         (pass_record_bounds): Replace declaration with that of...
4831         (make_pass_record_bounds): ...new function.
4832         (pass_graphite): Replace declaration with that of...
4833         (make_pass_graphite): ...new function.
4834         (pass_graphite_transforms): Replace declaration with that of...
4835         (make_pass_graphite_transforms): ...new function.
4836         (pass_if_conversion): Replace declaration with that of...
4837         (make_pass_if_conversion): ...new function.
4838         (pass_loop_distribution): Replace declaration with that of...
4839         (make_pass_loop_distribution): ...new function.
4840         (pass_vectorize): Replace declaration with that of...
4841         (make_pass_vectorize): ...new function.
4842         (pass_slp_vectorize): Replace declaration with that of...
4843         (make_pass_slp_vectorize): ...new function.
4844         (pass_complete_unroll): Replace declaration with that of...
4845         (make_pass_complete_unroll): ...new function.
4846         (pass_complete_unrolli): Replace declaration with that of...
4847         (make_pass_complete_unrolli): ...new function.
4848         (pass_parallelize_loops): Replace declaration with that of...
4849         (make_pass_parallelize_loops): ...new function.
4850         (pass_loop_prefetch): Replace declaration with that of...
4851         (make_pass_loop_prefetch): ...new function.
4852         (pass_iv_optimize): Replace declaration with that of...
4853         (make_pass_iv_optimize): ...new function.
4854         (pass_tree_loop_done): Replace declaration with that of...
4855         (make_pass_tree_loop_done): ...new function.
4856         (pass_ch): Replace declaration with that of...
4857         (make_pass_ch): ...new function.
4858         (pass_ccp): Replace declaration with that of...
4859         (make_pass_ccp): ...new function.
4860         (pass_phi_only_cprop): Replace declaration with that of...
4861         (make_pass_phi_only_cprop): ...new function.
4862         (pass_build_ssa): Replace declaration with that of...
4863         (make_pass_build_ssa): ...new function.
4864         (pass_build_alias): Replace declaration with that of...
4865         (make_pass_build_alias): ...new function.
4866         (pass_build_ealias): Replace declaration with that of...
4867         (make_pass_build_ealias): ...new function.
4868         (pass_dominator): Replace declaration with that of...
4869         (make_pass_dominator): ...new function.
4870         (pass_dce): Replace declaration with that of...
4871         (make_pass_dce): ...new function.
4872         (pass_dce_loop): Replace declaration with that of...
4873         (make_pass_dce_loop): ...new function.
4874         (pass_cd_dce): Replace declaration with that of...
4875         (make_pass_cd_dce): ...new function.
4876         (pass_call_cdce): Replace declaration with that of...
4877         (make_pass_call_cdce): ...new function.
4878         (pass_merge_phi): Replace declaration with that of...
4879         (make_pass_merge_phi): ...new function.
4880         (pass_split_crit_edges): Replace declaration with that of...
4881         (make_pass_split_crit_edges): ...new function.
4882         (pass_pre): Replace declaration with that of...
4883         (make_pass_pre): ...new function.
4884         (pass_profile): Replace declaration with that of...
4885         (make_pass_profile): ...new function.
4886         (pass_strip_predict_hints): Replace declaration with that of...
4887         (make_pass_strip_predict_hints): ...new function.
4888         (pass_lower_complex_O0): Replace declaration with that of...
4889         (make_pass_lower_complex_O0): ...new function.
4890         (pass_lower_complex): Replace declaration with that of...
4891         (make_pass_lower_complex): ...new function.
4892         (pass_lower_vector): Replace declaration with that of...
4893         (make_pass_lower_vector): ...new function.
4894         (pass_lower_vector_ssa): Replace declaration with that of...
4895         (make_pass_lower_vector_ssa): ...new function.
4896         (pass_lower_omp): Replace declaration with that of...
4897         (make_pass_lower_omp): ...new function.
4898         (pass_diagnose_omp_blocks): Replace declaration with that of...
4899         (make_pass_diagnose_omp_blocks): ...new function.
4900         (pass_expand_omp): Replace declaration with that of...
4901         (make_pass_expand_omp): ...new function.
4902         (pass_expand_omp_ssa): Replace declaration with that of...
4903         (make_pass_expand_omp_ssa): ...new function.
4904         (pass_object_sizes): Replace declaration with that of...
4905         (make_pass_object_sizes): ...new function.
4906         (pass_strlen): Replace declaration with that of...
4907         (make_pass_strlen): ...new function.
4908         (pass_fold_builtins): Replace declaration with that of...
4909         (make_pass_fold_builtins): ...new function.
4910         (pass_stdarg): Replace declaration with that of...
4911         (make_pass_stdarg): ...new function.
4912         (pass_early_warn_uninitialized): Replace declaration with that of...
4913         (make_pass_early_warn_uninitialized): ...new function.
4914         (pass_late_warn_uninitialized): Replace declaration with that of...
4915         (make_pass_late_warn_uninitialized): ...new function.
4916         (pass_cse_reciprocals): Replace declaration with that of...
4917         (make_pass_cse_reciprocals): ...new function.
4918         (pass_cse_sincos): Replace declaration with that of...
4919         (make_pass_cse_sincos): ...new function.
4920         (pass_optimize_bswap): Replace declaration with that of...
4921         (make_pass_optimize_bswap): ...new function.
4922         (pass_optimize_widening_mul): Replace declaration with that of...
4923         (make_pass_optimize_widening_mul): ...new function.
4924         (pass_warn_function_return): Replace declaration with that of...
4925         (make_pass_warn_function_return): ...new function.
4926         (pass_warn_function_noreturn): Replace declaration with that of...
4927         (make_pass_warn_function_noreturn): ...new function.
4928         (pass_cselim): Replace declaration with that of...
4929         (make_pass_cselim): ...new function.
4930         (pass_phiopt): Replace declaration with that of...
4931         (make_pass_phiopt): ...new function.
4932         (pass_forwprop): Replace declaration with that of...
4933         (make_pass_forwprop): ...new function.
4934         (pass_phiprop): Replace declaration with that of...
4935         (make_pass_phiprop): ...new function.
4936         (pass_tree_ifcombine): Replace declaration with that of...
4937         (make_pass_tree_ifcombine): ...new function.
4938         (pass_dse): Replace declaration with that of...
4939         (make_pass_dse): ...new function.
4940         (pass_nrv): Replace declaration with that of...
4941         (make_pass_nrv): ...new function.
4942         (pass_rename_ssa_copies): Replace declaration with that of...
4943         (make_pass_rename_ssa_copies): ...new function.
4944         (pass_sink_code): Replace declaration with that of...
4945         (make_pass_sink_code): ...new function.
4946         (pass_fre): Replace declaration with that of...
4947         (make_pass_fre): ...new function.
4948         (pass_check_data_deps): Replace declaration with that of...
4949         (make_pass_check_data_deps): ...new function.
4950         (pass_copy_prop): Replace declaration with that of...
4951         (make_pass_copy_prop): ...new function.
4952         (pass_vrp): Replace declaration with that of...
4953         (make_pass_vrp): ...new function.
4954         (pass_uncprop): Replace declaration with that of...
4955         (make_pass_uncprop): ...new function.
4956         (pass_return_slot): Replace declaration with that of...
4957         (make_pass_return_slot): ...new function.
4958         (pass_reassoc): Replace declaration with that of...
4959         (make_pass_reassoc): ...new function.
4960         (pass_rebuild_cgraph_edges): Replace declaration with that of...
4961         (make_pass_rebuild_cgraph_edges): ...new function.
4962         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
4963         (make_pass_remove_cgraph_callee_edges): ...new function.
4964         (pass_build_cgraph_edges): Replace declaration with that of...
4965         (make_pass_build_cgraph_edges): ...new function.
4966         (pass_local_pure_const): Replace declaration with that of...
4967         (make_pass_local_pure_const): ...new function.
4968         (pass_tracer): Replace declaration with that of...
4969         (make_pass_tracer): ...new function.
4970         (pass_warn_unused_result): Replace declaration with that of...
4971         (make_pass_warn_unused_result): ...new function.
4972         (pass_diagnose_tm_blocks): Replace declaration with that of...
4973         (make_pass_diagnose_tm_blocks): ...new function.
4974         (pass_lower_tm): Replace declaration with that of...
4975         (make_pass_lower_tm): ...new function.
4976         (pass_tm_init): Replace declaration with that of...
4977         (make_pass_tm_init): ...new function.
4978         (pass_tm_mark): Replace declaration with that of...
4979         (make_pass_tm_mark): ...new function.
4980         (pass_tm_memopt): Replace declaration with that of...
4981         (make_pass_tm_memopt): ...new function.
4982         (pass_tm_edges): Replace declaration with that of...
4983         (make_pass_tm_edges): ...new function.
4984         (pass_split_functions): Replace declaration with that of...
4985         (make_pass_split_functions): ...new function.
4986         (pass_feedback_split_functions): Replace declaration with that of...
4987         (make_pass_feedback_split_functions): ...new function.
4988         (pass_strength_reduction): Replace declaration with that of...
4989         (make_pass_strength_reduction): ...new function.
4990         (pass_ipa_lower_emutls): Replace declaration with that of...
4991         (make_pass_ipa_lower_emutls): ...new function.
4992         (pass_ipa_function_and_variable_visibility): Replace declaration with
4993         that of...
4994         (make_pass_ipa_function_and_variable_visibility): ...new function.
4995         (pass_ipa_tree_profile): Replace declaration with that of...
4996         (make_pass_ipa_tree_profile): ...new function.
4997         (pass_early_local_passes): Replace declaration with that of...
4998         (make_pass_early_local_passes): ...new function.
4999         (pass_ipa_whole_program_visibility): Replace declaration with that
5000         of...
5001         (make_pass_ipa_whole_program_visibility): ...new function.
5002         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5003         (make_pass_ipa_lto_gimple_out): ...new function.
5004         (pass_ipa_increase_alignment): Replace declaration with that of...
5005         (make_pass_ipa_increase_alignment): ...new function.
5006         (pass_ipa_inline): Replace declaration with that of...
5007         (make_pass_ipa_inline): ...new function.
5008         (pass_ipa_free_lang_data): Replace declaration with that of...
5009         (make_pass_ipa_free_lang_data): ...new function.
5010         (pass_ipa_free_inline_summary): Replace declaration with that of...
5011         (make_pass_ipa_free_inline_summary): ...new function.
5012         (pass_ipa_cp): Replace declaration with that of...
5013         (make_pass_ipa_cp): ...new function.
5014         (pass_ipa_reference): Replace declaration with that of...
5015         (make_pass_ipa_reference): ...new function.
5016         (pass_ipa_pure_const): Replace declaration with that of...
5017         (make_pass_ipa_pure_const): ...new function.
5018         (pass_ipa_pta): Replace declaration with that of...
5019         (make_pass_ipa_pta): ...new function.
5020         (pass_ipa_lto_finish_out): Replace declaration with that of...
5021         (make_pass_ipa_lto_finish_out): ...new function.
5022         (pass_ipa_tm): Replace declaration with that of...
5023         (make_pass_ipa_tm): ...new function.
5024         (pass_ipa_profile): Replace declaration with that of...
5025         (make_pass_ipa_profile): ...new function.
5026         (pass_ipa_cdtor_merge): Replace declaration with that of...
5027         (make_pass_ipa_cdtor_merge): ...new function.
5028         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5029         of...
5030         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5031         (pass_init_datastructures): Replace declaration with that of...
5032         (make_pass_init_datastructures): ...new function.
5033         (pass_fixup_cfg): Replace declaration with that of...
5034         (make_pass_fixup_cfg): ...new function.
5035         (pass_expand): Replace declaration with that of...
5036         (make_pass_expand): ...new function.
5037         (pass_instantiate_virtual_regs): Replace declaration with that of...
5038         (make_pass_instantiate_virtual_regs): ...new function.
5039         (pass_rtl_fwprop): Replace declaration with that of...
5040         (make_pass_rtl_fwprop): ...new function.
5041         (pass_rtl_fwprop_addr): Replace declaration with that of...
5042         (make_pass_rtl_fwprop_addr): ...new function.
5043         (pass_jump): Replace declaration with that of...
5044         (make_pass_jump): ...new function.
5045         (pass_jump2): Replace declaration with that of...
5046         (make_pass_jump2): ...new function.
5047         (pass_lower_subreg): Replace declaration with that of...
5048         (make_pass_lower_subreg): ...new function.
5049         (pass_cse): Replace declaration with that of...
5050         (make_pass_cse): ...new function.
5051         (pass_fast_rtl_dce): Replace declaration with that of...
5052         (make_pass_fast_rtl_dce): ...new function.
5053         (pass_ud_rtl_dce): Replace declaration with that of...
5054         (make_pass_ud_rtl_dce): ...new function.
5055         (pass_rtl_dce): Replace declaration with that of...
5056         (make_pass_rtl_dce): ...new function.
5057         (pass_rtl_dse1): Replace declaration with that of...
5058         (make_pass_rtl_dse1): ...new function.
5059         (pass_rtl_dse2): Replace declaration with that of...
5060         (make_pass_rtl_dse2): ...new function.
5061         (pass_rtl_dse3): Replace declaration with that of...
5062         (make_pass_rtl_dse3): ...new function.
5063         (pass_rtl_cprop): Replace declaration with that of...
5064         (make_pass_rtl_cprop): ...new function.
5065         (pass_rtl_pre): Replace declaration with that of...
5066         (make_pass_rtl_pre): ...new function.
5067         (pass_rtl_hoist): Replace declaration with that of...
5068         (make_pass_rtl_hoist): ...new function.
5069         (pass_rtl_store_motion): Replace declaration with that of...
5070         (make_pass_rtl_store_motion): ...new function.
5071         (pass_cse_after_global_opts): Replace declaration with that of...
5072         (make_pass_cse_after_global_opts): ...new function.
5073         (pass_rtl_ifcvt): Replace declaration with that of...
5074         (make_pass_rtl_ifcvt): ...new function.
5075         (pass_into_cfg_layout_mode): Replace declaration with that of...
5076         (make_pass_into_cfg_layout_mode): ...new function.
5077         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5078         (make_pass_outof_cfg_layout_mode): ...new function.
5079         (pass_loop2): Replace declaration with that of...
5080         (make_pass_loop2): ...new function.
5081         (pass_rtl_loop_init): Replace declaration with that of...
5082         (make_pass_rtl_loop_init): ...new function.
5083         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5084         (make_pass_rtl_move_loop_invariants): ...new function.
5085         (pass_rtl_unswitch): Replace declaration with that of...
5086         (make_pass_rtl_unswitch): ...new function.
5087         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5088         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5089         (pass_rtl_doloop): Replace declaration with that of...
5090         (make_pass_rtl_doloop): ...new function.
5091         (pass_rtl_loop_done): Replace declaration with that of...
5092         (make_pass_rtl_loop_done): ...new function.
5093         (pass_web): Replace declaration with that of...
5094         (make_pass_web): ...new function.
5095         (pass_cse2): Replace declaration with that of...
5096         (make_pass_cse2): ...new function.
5097         (pass_df_initialize_opt): Replace declaration with that of...
5098         (make_pass_df_initialize_opt): ...new function.
5099         (pass_df_initialize_no_opt): Replace declaration with that of...
5100         (make_pass_df_initialize_no_opt): ...new function.
5101         (pass_reginfo_init): Replace declaration with that of...
5102         (make_pass_reginfo_init): ...new function.
5103         (pass_inc_dec): Replace declaration with that of...
5104         (make_pass_inc_dec): ...new function.
5105         (pass_stack_ptr_mod): Replace declaration with that of...
5106         (make_pass_stack_ptr_mod): ...new function.
5107         (pass_initialize_regs): Replace declaration with that of...
5108         (make_pass_initialize_regs): ...new function.
5109         (pass_combine): Replace declaration with that of...
5110         (make_pass_combine): ...new function.
5111         (pass_if_after_combine): Replace declaration with that of...
5112         (make_pass_if_after_combine): ...new function.
5113         (pass_ree): Replace declaration with that of...
5114         (make_pass_ree): ...new function.
5115         (pass_partition_blocks): Replace declaration with that of...
5116         (make_pass_partition_blocks): ...new function.
5117         (pass_match_asm_constraints): Replace declaration with that of...
5118         (make_pass_match_asm_constraints): ...new function.
5119         (pass_regmove): Replace declaration with that of...
5120         (make_pass_regmove): ...new function.
5121         (pass_split_all_insns): Replace declaration with that of...
5122         (make_pass_split_all_insns): ...new function.
5123         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5124         (make_pass_fast_rtl_byte_dce): ...new function.
5125         (pass_lower_subreg2): Replace declaration with that of...
5126         (make_pass_lower_subreg2): ...new function.
5127         (pass_mode_switching): Replace declaration with that of...
5128         (make_pass_mode_switching): ...new function.
5129         (pass_sms): Replace declaration with that of...
5130         (make_pass_sms): ...new function.
5131         (pass_sched): Replace declaration with that of...
5132         (make_pass_sched): ...new function.
5133         (pass_ira): Replace declaration with that of...
5134         (make_pass_ira): ...new function.
5135         (pass_reload): Replace declaration with that of...
5136         (make_pass_reload): ...new function.
5137         (pass_clean_state): Replace declaration with that of...
5138         (make_pass_clean_state): ...new function.
5139         (pass_branch_prob): Replace declaration with that of...
5140         (make_pass_branch_prob): ...new function.
5141         (pass_value_profile_transformations): Replace declaration with that
5142         of...
5143         (make_pass_value_profile_transformations): ...new function.
5144         (pass_postreload_cse): Replace declaration with that of...
5145         (make_pass_postreload_cse): ...new function.
5146         (pass_gcse2): Replace declaration with that of...
5147         (make_pass_gcse2): ...new function.
5148         (pass_split_after_reload): Replace declaration with that of...
5149         (make_pass_split_after_reload): ...new function.
5150         (pass_branch_target_load_optimize1): Replace declaration with that
5151         of...
5152         (make_pass_branch_target_load_optimize1): ...new function.
5153         (pass_thread_prologue_and_epilogue): Replace declaration with that
5154         of...
5155         (make_pass_thread_prologue_and_epilogue): ...new function.
5156         (pass_stack_adjustments): Replace declaration with that of...
5157         (make_pass_stack_adjustments): ...new function.
5158         (pass_peephole2): Replace declaration with that of...
5159         (make_pass_peephole2): ...new function.
5160         (pass_if_after_reload): Replace declaration with that of...
5161         (make_pass_if_after_reload): ...new function.
5162         (pass_regrename): Replace declaration with that of...
5163         (make_pass_regrename): ...new function.
5164         (pass_cprop_hardreg): Replace declaration with that of...
5165         (make_pass_cprop_hardreg): ...new function.
5166         (pass_reorder_blocks): Replace declaration with that of...
5167         (make_pass_reorder_blocks): ...new function.
5168         (pass_branch_target_load_optimize2): Replace declaration with that
5169         of...
5170         (make_pass_branch_target_load_optimize2): ...new function.
5171         (pass_leaf_regs): Replace declaration with that of...
5172         (make_pass_leaf_regs): ...new function.
5173         (pass_split_before_sched2): Replace declaration with that of...
5174         (make_pass_split_before_sched2): ...new function.
5175         (pass_compare_elim_after_reload): Replace declaration with that of...
5176         (make_pass_compare_elim_after_reload): ...new function.
5177         (pass_sched2): Replace declaration with that of...
5178         (make_pass_sched2): ...new function.
5179         (pass_stack_regs): Replace declaration with that of...
5180         (make_pass_stack_regs): ...new function.
5181         (pass_stack_regs_run): Replace declaration with that of...
5182         (make_pass_stack_regs_run): ...new function.
5183         (pass_df_finish): Replace declaration with that of...
5184         (make_pass_df_finish): ...new function.
5185         (pass_compute_alignments): Replace declaration with that of...
5186         (make_pass_compute_alignments): ...new function.
5187         (pass_duplicate_computed_gotos): Replace declaration with that of...
5188         (make_pass_duplicate_computed_gotos): ...new function.
5189         (pass_variable_tracking): Replace declaration with that of...
5190         (make_pass_variable_tracking): ...new function.
5191         (pass_free_cfg): Replace declaration with that of...
5192         (make_pass_free_cfg): ...new function.
5193         (pass_machine_reorg): Replace declaration with that of...
5194         (make_pass_machine_reorg): ...new function.
5195         (pass_cleanup_barriers): Replace declaration with that of...
5196         (make_pass_cleanup_barriers): ...new function.
5197         (pass_delay_slots): Replace declaration with that of...
5198         (make_pass_delay_slots): ...new function.
5199         (pass_split_for_shorten_branches): Replace declaration with that of...
5200         (make_pass_split_for_shorten_branches): ...new function.
5201         (pass_split_before_regstack): Replace declaration with that of...
5202         (make_pass_split_before_regstack): ...new function.
5203         (pass_convert_to_eh_region_ranges): Replace declaration with that
5204         of...
5205         (make_pass_convert_to_eh_region_ranges): ...new function.
5206         (pass_shorten_branches): Replace declaration with that of...
5207         (make_pass_shorten_branches): ...new function.
5208         (pass_set_nothrow_function_flags): Replace declaration with that of...
5209         (make_pass_set_nothrow_function_flags): ...new function.
5210         (pass_dwarf2_frame): Replace declaration with that of...
5211         (make_pass_dwarf2_frame): ...new function.
5212         (pass_final): Replace declaration with that of...
5213         (make_pass_final): ...new function.
5214         (pass_rtl_seqabstr): Replace declaration with that of...
5215         (make_pass_rtl_seqabstr): ...new function.
5216         (pass_release_ssa_names): Replace declaration with that of...
5217         (make_pass_release_ssa_names): ...new function.
5218         (pass_early_inline): Replace declaration with that of...
5219         (make_pass_early_inline): ...new function.
5220         (pass_inline_parameters): Replace declaration with that of...
5221         (make_pass_inline_parameters): ...new function.
5222         (pass_update_address_taken): Replace declaration with that of...
5223         (make_pass_update_address_taken): ...new function.
5224         (pass_convert_switch): Replace declaration with that of...
5225         (make_pass_convert_switch): ...new function.
5226         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5227         to a subclass of simple_ipa_opt_pass along with...
5228         (pass_data_ipa_tree_profile): ...new pass_data instance and...
5229         (make_pass_ipa_tree_profile): ...new function.
5230         * tree-sra.c (pass_sra_early): Convert from a global struct to a
5231         subclass of gimple_opt_pass along with...
5232         (pass_data_sra_early): ...new pass_data instance and...
5233         (make_pass_sra_early): ...new function.
5234         (pass_sra): Convert from a global struct to a subclass of
5235         gimple_opt_pass along with...
5236         (pass_data_sra): ...new pass_data instance and...
5237         (make_pass_sra): ...new function.
5238         (pass_early_ipa_sra): Convert from a global struct to a subclass of
5239         gimple_opt_pass along with...
5240         (pass_data_early_ipa_sra): ...new pass_data instance and...
5241         (make_pass_early_ipa_sra): ...new function.
5242         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5243         subclass of gimple_opt_pass along with...
5244         (pass_data_ccp): ...new pass_data instance and...
5245         (make_pass_ccp): ...new function.
5246         (pass_fold_builtins): Convert from a global struct to a subclass of
5247         gimple_opt_pass along with...
5248         (pass_data_fold_builtins): ...new pass_data instance and...
5249         (make_pass_fold_builtins): ...new function.
5250         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5251         subclass of gimple_opt_pass along with...
5252         (pass_data_copy_prop): ...new pass_data instance and...
5253         (make_pass_copy_prop): ...new function.
5254         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5255         global struct to a subclass of gimple_opt_pass along with...
5256         (pass_data_rename_ssa_copies): ...new pass_data instance and...
5257         (make_pass_rename_ssa_copies): ...new function.
5258         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5259         subclass of gimple_opt_pass along with...
5260         (pass_data_dce): ...new pass_data instance and...
5261         (make_pass_dce): ...new function.
5262         (pass_dce_loop): Convert from a global struct to a subclass of
5263         gimple_opt_pass along with...
5264         (pass_data_dce_loop): ...new pass_data instance and...
5265         (make_pass_dce_loop): ...new function.
5266         (pass_cd_dce): Convert from a global struct to a subclass of
5267         gimple_opt_pass along with...
5268         (pass_data_cd_dce): ...new pass_data instance and...
5269         (make_pass_cd_dce): ...new function.
5270         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5271         subclass of gimple_opt_pass along with...
5272         (pass_data_dominator): ...new pass_data instance and...
5273         (make_pass_dominator): ...new function.
5274         (pass_phi_only_cprop): Convert from a global struct to a subclass of
5275         gimple_opt_pass along with...
5276         (pass_data_phi_only_cprop): ...new pass_data instance and...
5277         (make_pass_phi_only_cprop): ...new function.
5278         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5279         subclass of gimple_opt_pass along with...
5280         (pass_data_dse): ...new pass_data instance and...
5281         (make_pass_dse): ...new function.
5282         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5283         a subclass of gimple_opt_pass along with...
5284         (pass_data_forwprop): ...new pass_data instance and...
5285         (make_pass_forwprop): ...new function.
5286         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5287         struct to a subclass of gimple_opt_pass along with...
5288         (pass_data_tree_ifcombine): ...new pass_data instance and...
5289         (make_pass_tree_ifcombine): ...new function.
5290         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5291         subclass of gimple_opt_pass along with...
5292         (pass_data_ch): ...new pass_data instance and...
5293         (make_pass_ch): ...new function.
5294         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5295         subclass of gimple_opt_pass along with...
5296         (pass_data_tree_loop): ...new pass_data instance and...
5297         (make_pass_tree_loop): ...new function.
5298         (pass_tree_loop_init): Convert from a global struct to a subclass of
5299         gimple_opt_pass along with...
5300         (pass_data_tree_loop_init): ...new pass_data instance and...
5301         (make_pass_tree_loop_init): ...new function.
5302         (pass_lim): Convert from a global struct to a subclass of
5303         gimple_opt_pass along with...
5304         (pass_data_lim): ...new pass_data instance and...
5305         (make_pass_lim): ...new function.
5306         (pass_tree_unswitch): Convert from a global struct to a subclass of
5307         gimple_opt_pass along with...
5308         (pass_data_tree_unswitch): ...new pass_data instance and...
5309         (make_pass_tree_unswitch): ...new function.
5310         (pass_predcom): Convert from a global struct to a subclass of
5311         gimple_opt_pass along with...
5312         (pass_data_predcom): ...new pass_data instance and...
5313         (make_pass_predcom): ...new function.
5314         (pass_vectorize): Convert from a global struct to a subclass of
5315         gimple_opt_pass along with...
5316         (pass_data_vectorize): ...new pass_data instance and...
5317         (make_pass_vectorize): ...new function.
5318         (pass_graphite): Convert from a global struct to a subclass of
5319         gimple_opt_pass along with...
5320         (pass_data_graphite): ...new pass_data instance and...
5321         (make_pass_graphite): ...new function.
5322         (pass_graphite_transforms): Convert from a global struct to a subclass
5323         of gimple_opt_pass along with...
5324         (pass_data_graphite_transforms): ...new pass_data instance and...
5325         (make_pass_graphite_transforms): ...new function.
5326         (pass_check_data_deps): Convert from a global struct to a subclass of
5327         gimple_opt_pass along with...
5328         (pass_data_check_data_deps): ...new pass_data instance and...
5329         (make_pass_check_data_deps): ...new function.
5330         (pass_iv_canon): Convert from a global struct to a subclass of
5331         gimple_opt_pass along with...
5332         (pass_data_iv_canon): ...new pass_data instance and...
5333         (make_pass_iv_canon): ...new function.
5334         (pass_scev_cprop): Convert from a global struct to a subclass of
5335         gimple_opt_pass along with...
5336         (pass_data_scev_cprop): ...new pass_data instance and...
5337         (make_pass_scev_cprop): ...new function.
5338         (pass_record_bounds): Convert from a global struct to a subclass of
5339         gimple_opt_pass along with...
5340         (pass_data_record_bounds): ...new pass_data instance and...
5341         (make_pass_record_bounds): ...new function.
5342         (pass_complete_unroll): Convert from a global struct to a subclass of
5343         gimple_opt_pass along with...
5344         (pass_data_complete_unroll): ...new pass_data instance and...
5345         (make_pass_complete_unroll): ...new function.
5346         (pass_complete_unrolli): Convert from a global struct to a subclass of
5347         gimple_opt_pass along with...
5348         (pass_data_complete_unrolli): ...new pass_data instance and...
5349         (make_pass_complete_unrolli): ...new function.
5350         (pass_parallelize_loops): Convert from a global struct to a subclass
5351         of gimple_opt_pass along with...
5352         (pass_data_parallelize_loops): ...new pass_data instance and...
5353         (make_pass_parallelize_loops): ...new function.
5354         (pass_loop_prefetch): Convert from a global struct to a subclass of
5355         gimple_opt_pass along with...
5356         (pass_data_loop_prefetch): ...new pass_data instance and...
5357         (make_pass_loop_prefetch): ...new function.
5358         (pass_iv_optimize): Convert from a global struct to a subclass of
5359         gimple_opt_pass along with...
5360         (pass_data_iv_optimize): ...new pass_data instance and...
5361         (make_pass_iv_optimize): ...new function.
5362         (pass_tree_loop_done): Convert from a global struct to a subclass of
5363         gimple_opt_pass along with...
5364         (pass_data_tree_loop_done): ...new pass_data instance and...
5365         (make_pass_tree_loop_done): ...new function.
5366         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5367         struct to a subclass of gimple_opt_pass along with...
5368         (pass_data_cse_reciprocals): ...new pass_data instance and...
5369         (make_pass_cse_reciprocals): ...new function.
5370         (pass_cse_sincos): Convert from a global struct to a subclass of
5371         gimple_opt_pass along with...
5372         (pass_data_cse_sincos): ...new pass_data instance and...
5373         (make_pass_cse_sincos): ...new function.
5374         (pass_optimize_bswap): Convert from a global struct to a subclass of
5375         gimple_opt_pass along with...
5376         (pass_data_optimize_bswap): ...new pass_data instance and...
5377         (make_pass_optimize_bswap): ...new function.
5378         (pass_optimize_widening_mul): Convert from a global struct to a
5379         subclass of gimple_opt_pass along with...
5380         (pass_data_optimize_widening_mul): ...new pass_data instance and...
5381         (make_pass_optimize_widening_mul): ...new function.
5382         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5383         subclass of gimple_opt_pass along with...
5384         (pass_data_phiopt): ...new pass_data instance and...
5385         (make_pass_phiopt): ...new function.
5386         (pass_cselim): Convert from a global struct to a subclass of
5387         gimple_opt_pass along with...
5388         (pass_data_cselim): ...new pass_data instance and...
5389         (make_pass_cselim): ...new function.
5390         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5391         subclass of gimple_opt_pass along with...
5392         (pass_data_phiprop): ...new pass_data instance and...
5393         (make_pass_phiprop): ...new function.
5394         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5395         subclass of gimple_opt_pass along with...
5396         (pass_data_pre): ...new pass_data instance and...
5397         (make_pass_pre): ...new function.
5398         (pass_fre): Convert from a global struct to a subclass of
5399         gimple_opt_pass along with...
5400         (pass_data_fre): ...new pass_data instance and...
5401         (make_pass_fre): ...new function.
5402         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5403         subclass of gimple_opt_pass along with...
5404         (pass_data_reassoc): ...new pass_data instance and...
5405         (make_pass_reassoc): ...new function.
5406         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5407         subclass of gimple_opt_pass along with...
5408         (pass_data_sink_code): ...new pass_data instance and...
5409         (make_pass_sink_code): ...new function.
5410         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5411         subclass of gimple_opt_pass along with...
5412         (pass_data_strlen): ...new pass_data instance and...
5413         (make_pass_strlen): ...new function.
5414         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5415         struct to a subclass of gimple_opt_pass along with...
5416         (pass_data_build_alias): ...new pass_data instance and...
5417         (make_pass_build_alias): ...new function.
5418         (pass_build_ealias): Convert from a global struct to a subclass of
5419         gimple_opt_pass along with...
5420         (pass_data_build_ealias): ...new pass_data instance and...
5421         (make_pass_build_ealias): ...new function.
5422         (pass_ipa_pta): Convert from a global struct to a subclass of
5423         simple_ipa_opt_pass along with...
5424         (pass_data_ipa_pta): ...new pass_data instance and...
5425         (make_pass_ipa_pta): ...new function.
5426         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5427         subclass of gimple_opt_pass along with...
5428         (pass_data_uncprop): ...new pass_data instance and...
5429         (make_pass_uncprop): ...new function.
5430         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5431         global struct to a subclass of gimple_opt_pass along with...
5432         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5433         (make_pass_late_warn_uninitialized): ...new function.
5434         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5435         to a subclass of gimple_opt_pass along with...
5436         (pass_data_init_datastructures): ...new pass_data instance and...
5437         (make_pass_init_datastructures): ...new function.
5438         (pass_early_warn_uninitialized): Convert from a global struct to a
5439         subclass of gimple_opt_pass along with...
5440         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5441         (make_pass_early_warn_uninitialized): ...new function.
5442         (pass_update_address_taken): Convert from a global struct to a
5443         subclass of gimple_opt_pass along with...
5444         (pass_data_update_address_taken): ...new pass_data instance and...
5445         (make_pass_update_address_taken): ...new function.
5446         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5447         struct to a subclass of gimple_opt_pass along with...
5448         (pass_data_release_ssa_names): ...new pass_data instance and...
5449         (make_pass_release_ssa_names): ...new function.
5450         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5451         subclass of gimple_opt_pass along with...
5452         (pass_data_stdarg): ...new pass_data instance and...
5453         (make_pass_stdarg): ...new function.
5454         * tree-switch-conversion.c (pass_convert_switch): Convert from a
5455         global struct to a subclass of gimple_opt_pass along with...
5456         (pass_data_convert_switch): ...new pass_data instance and...
5457         (make_pass_convert_switch): ...new function.
5458         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5459         to a subclass of gimple_opt_pass along with...
5460         (pass_data_tail_recursion): ...new pass_data instance and...
5461         (make_pass_tail_recursion): ...new function.
5462         (pass_tail_calls): Convert from a global struct to a subclass of
5463         gimple_opt_pass along with...
5464         (pass_data_tail_calls): ...new pass_data instance and...
5465         (make_pass_tail_calls): ...new function.
5466         * tree-vect-generic.c (pass_lower_vector): Convert from a global
5467         struct to a subclass of gimple_opt_pass along with...
5468         (pass_data_lower_vector): ...new pass_data instance and...
5469         (make_pass_lower_vector): ...new function.
5470         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5471         gimple_opt_pass along with...
5472         (pass_data_lower_vector_ssa): ...new pass_data instance and...
5473         (make_pass_lower_vector_ssa): ...new function.
5474         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5475         to a subclass of gimple_opt_pass along with...
5476         (pass_data_slp_vectorize): ...new pass_data instance and...
5477         (make_pass_slp_vectorize): ...new function.
5478         (pass_ipa_increase_alignment): Convert from a global struct to a
5479         subclass of simple_ipa_opt_pass along with...
5480         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5481         (make_pass_ipa_increase_alignment): ...new function.
5482         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5483         gimple_opt_pass along with...
5484         (pass_data_vrp): ...new pass_data instance and...
5485         (make_pass_vrp): ...new function.
5486         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5487         subclass of simple_ipa_opt_pass along with...
5488         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5489         (make_pass_ipa_free_lang_data): ...new function.
5490         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5491         gimple_opt_pass along with...
5492         (pass_data_tsan): ...new pass_data instance and...
5493         (make_pass_tsan): ...new function.
5494         (pass_tsan_O0): Convert from a global struct to a subclass of
5495         gimple_opt_pass along with...
5496         (pass_data_tsan_O0): ...new pass_data instance and...
5497         (make_pass_tsan_O0): ...new function.
5498         * var-tracking.c (pass_variable_tracking): Convert from a global
5499         struct to a subclass of rtl_opt_pass along with...
5500         (pass_data_variable_tracking): ...new pass_data instance and...
5501         (make_pass_variable_tracking): ...new function.
5502         * web.c (pass_web): Convert from a global struct to a subclass of
5503         rtl_opt_pass along with...
5504         (pass_data_web): ...new pass_data instance and...
5505         (make_pass_web): ...new function.
5506         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5507         declaration with that of...
5508         (make_pass_mode_switch_use): ...new function.
5509         (pass_resolve_sw_modes): Replace declaration with that of...
5510         (make_pass_resolve_sw_modes): ...new function.
5511         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5512         from a global struct to a subclass of rtl_opt_pass along with...
5513         (pass_data_mode_switch_use): ...new pass_data instance and...
5514         (make_pass_mode_switch_use): ...new function.
5515         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5516         from a global struct to a subclass of rtl_opt_pass along with...
5517         (pass_data_resolve_sw_modes): ...new pass_data instance and...
5518         (make_pass_resolve_sw_modes): ...new function.
5519         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5520         struct to a subclass of rtl_opt_pass along with...
5521         (pass_data_insert_vzeroupper): ...new pass_data instance and...
5522         (make_pass_insert_vzeroupper): ...new function.
5523         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5524         global struct to a subclass of rtl_opt_pass along with...
5525         (pass_data_work_around_errata): ...new pass_data instance and...
5526         (make_pass_work_around_errata): ...new function.
5527         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5528         struct to a subclass of rtl_opt_pass along with...
5529         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
5530         (make_pass_mips_machine_reorg2): ...new function.
5532 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5534         * passes.c (pass_manager::operator new): New.
5536 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5538         Handwritten part of conversion of passes to C++ classes.
5540         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
5541         (toplev.o): Add dep on PASS_MANAGER_H.
5542         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
5543         of early local pases to reflect this moving from a global to a
5544         member of gcc::pass_manager.
5545         (cgraph_add_new_function): Likewise.
5546         * lto-cgraph.c (lto_output_node): Update for conversion of
5547         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
5548         * passes.c (opt_pass::clone): New.
5549         (opt_pass::gate): New.
5550         (opt_pass::execute): New.
5551         (opt_pass::opt_pass): New.
5552         (pass_manager::execute_early_local_passes): New.
5553         (pass_manager::execute_pass_mode_switching): new.
5554         (finish_optimization_passes): Convert to...
5555         (pass_manager::finish_optimization_passes): ...this.
5556         (finish_optimization_passes): Update for conversion of passes to
5557         C++ classes.
5558         (register_dump_files_1): Use has_gate since we cannot portably
5559         check a vtable entry against NULL.
5560         (dump_one_pass): Likewise.
5561         (ipa_write_summaries_2): Likewise.
5562         (ipa_write_optimization_summaries_1): Likewise.
5563         (ipa_read_summaries_1): Likewise.
5564         (ipa_read_optimization_summaries_1): Likewise.
5565         (execute_ipa_stmt_fixups): Likewise.
5566         (pass_manager::pass_manager): Rewrite pass-creation, invoking
5567         pass-creation functions rather than wiring up globals, and
5568         storing the results in fields of pass_manager generated using
5569         pass-instances.def.
5570         (pass_manager::dump_profile_report): Update for conversion of
5571         passes to C++ classes.
5572         (pass_manager::execute_ipa_summary_passes): Likewise.
5573         (execute_one_ipa_transform_pass): Likewise.
5574         (execute_one_pass): Use has_gate and has_execute since we cannot
5575         portably check a vtable entry against NULL.
5576         * pass_manager.h (pass_manager::finish_optimization_passes): New.
5577         (pass_manager): Use pass-instances.def to add fields for the
5578         various pass instances.
5579         * toplev.c (finalize): Update for move of
5580         finish_optimization_passes to a method of gcc::pass_manager.
5581         * toplev.h (finish_optimization_passes): Move to method of class
5582         pass_manager.
5583         * tree-pass.h (struct pass_data): New.
5584         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
5585         (opt_pass::gate): Convert to virtual function.
5586         (opt_pass::~opt_pass): New.
5587         (opt_pass::clone): New.
5588         (opt_pass::execute): Convert to virtual function.
5589         (opt_pass::opt_pass): New.
5590         (opt_pass::ctxt_): new.
5591         (gimple_opt_pass): Convert to subclass of opt_pass.
5592         (gimple_opt_pass::gimple_opt_pass): New.
5593         (rtl_opt_pass): Convert to subclass of opt_pass.
5594         (rtl_opt_pass::rtl_opt_pass): New.
5595         (ipa_opt_pass_d): Convert to subclass of opt_pass.
5596         (ipa_opt_pass_d::ipa_opt_pass_d): New.
5597         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
5598         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
5599         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
5600         invocation of pass_mode_switching to reflect this moving from a
5601         global to a member of gcc::pass_manager.
5602         (ix86_option_override): Rework how pass_insert_vzeroupper is
5603         added to the pass_manager to reflect autogenerated changes.
5604         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
5606 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
5608         PR rtl-optimization/57708
5609         * recog.c (peep2_find_free_register): Validate all regs in a
5610         multi-reg mode.
5612 2013-08-05  Jan Hubicka  <jh@suse.cz>
5614         PR lto/57602
5615         * cgraph.c (verify_cgraph_node): Accept local flags from other
5616         partitions.
5617         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
5618         (function_and_variable_visibility): Likewise.
5619         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
5621 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5623         * graph.c (init_graph_slim_pretty_print): Remove.
5624         (print_graph_cfg): Do not call it.  Use local pretty printer.
5625         (start_graph_dump): Likewise.
5627 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5629         * gimple-pretty-print.c (buffer): Remove.
5630         (initialized): Likewise.
5631         (maybe_init_pretty_print): Likewise.
5632         (print_gimple_stmt): Do not call it.  Use non-static local
5633         pretty_printer variable.
5634         (print_gimple_expr): Likewise.
5635         (print_gimple_seq): Likewise.
5636         (gimple_dump_bb): Likewise.
5638 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5640         * asan.c (asan_pp): Remove.
5641         (asan_pp_initialized): Likewise.
5642         (asan_pp_initialize): Likewise.
5643         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
5644         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
5645         (asan_add_global): Likewise.
5647 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5649         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
5650         * diagnostic.h (diagnostic_flush_buffer): Adjust.
5651         * pretty-print.c (pp_formatted_text_data): Likewise.
5652         (pp_indent): Rename from pp_base_indent.
5653         (pp_format): Rename from pp_base_format.
5654         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
5655         (pp_format_verbatim): Rename from pp_base_format_verbatim.
5656         (pp_flush): Rename from pp_base_flush.
5657         (pp_set_line_maximum_length): Rename from
5658         pp_base_set_line_maximum_length.
5659         (pp_clear_output_area): Rename from pp_base_clear_output_area.
5660         (pp_set_prefix): Rename from pp_base_set_prefix.
5661         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
5662         (pp_emit_prefix): Rename from pp_base_emit_prefix.
5663         (pp_append_text): Rename from pp_base_append_text.
5664         (pp_formatted_text): Rename from pp_base_formatted_text.
5665         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
5666         (pp_remaining_character_count_for_line): Rename from
5667         pp_base_remaining_character_count_for_line.
5668         (pp_newline): Rename from pp_base_newline.
5669         (pp_character): Rename from pp_base_character.
5670         (pp_string): Rename from pp_base_string.
5671         (pp_maybe_space): Rename from pp_base_maybe_space.
5672         * asan.c (asan_pp_string): Adjust.
5673         (asan_emit_stack_protection): Likewise.
5674         (asan_add_global): Likewise.
5675         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
5676         * tree-mudflap.c (mf_varname_tree): Likewise.
5677         * tree-pretty-print.c (pp_tree_identifier): Rename from
5678         pp_base_tree_identifier.
5679         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
5680         Declare as function.
5682 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5684         * pretty-print.h (pp_bar_bar): New.
5685         (pp_ampersand_ampersand): Likewise.
5686         (pp_less_equal): Likewise.
5687         (pp_greater_equal): Likewise.
5688         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
5689         printer functions instead of pp_string or operators and punctuators.
5690         (dump_gimple_call): Likewise.
5691         (dump_gimple_omp_for): Likewise.
5692         (dump_gimple_transaction): Likewise.
5693         (dump_gimple_phi): Likewise.
5694         (pp_gimple_stmt_1): Likewise.
5695         * sched-vis.c (print_insn): Likewise.
5696         * tree-mudflap.c (mf_varname_tree): Likewise.
5697         * tree-pretty-print.c (dump_block_node): Likewise.
5698         (dump_generic_node): Likewise.
5700 2013-08-02  Jan Hubicka  <jh@suse.cz>
5702         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
5703         boundaries.
5704         * lto-streamer-out.c (tree_is_indexable): Results decls and
5705         parm decls are not indexable.
5706         (DFS_write_tree_body): Do not follow args and results.
5707         (hash_tree): Likewise.
5708         (output_functions): Rearrange so struct function is needed
5709         only when real body is output; be able to also ouptut abstract
5710         functions; output DECL_ARGUMENTS and DECL_RESULT.
5711         (lto_output): When not in WPA, ale store abstract functions.
5712         (write_symbol): Do not care about RESULT_DECL.
5713         (output_symbol_p): Handle correctly sbtract decls.
5714         * lto-streamer-in.c (input_function): Rearrange so struct
5715         function can be NULL at entry; allow streaming of
5716         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
5717         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
5718         sanity check during LTO.
5719         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
5720         RESULT_DECl and DECL_ARGUMENTS.
5721         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
5722         Likewise.
5724 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5726         * pretty-print.h (pp_underscore): New.
5727         (pp_comma): Tidy.
5728         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
5729         printer functions instead of pp_character.
5730         (dump_binary_rhs): Likewise.
5731         (dump_ternary_rhs): Likewise.
5732         (dump_gimple_call_args): Likewise.
5733         (pp_points_to_solution): Likewise.
5734         (dump_gimple_call): Likewise.
5735         (dump_gimple_switch): Likewise.
5736         (dump_gimple_cond): Likewise.
5737         (dump_gimple_bind): Likewise.
5738         (dump_gimple_try): Likewise.
5739         (dump_gimple_omp_for): Likewise.
5740         (dump_gimple_omp_continue): Likewise.
5741         (dump_gimple_omp_single): Likewise.
5742         (dump_gimple_omp_sections): Likewise.
5743         (dump_gimple_omp_block): Likewise.
5744         (dump_gimple_omp_critical): Likewise.
5745         (dump_gimple_transaction): Likewise.
5746         (dump_gimple_asm): Likewise.
5747         (dump_gimple_phi): Likewise.
5748         (dump_gimple_omp_parallel): Likewise.
5749         (dump_gimple_omp_task): Likewise.
5750         (dump_gimple_omp_atomic_load): Likewise.
5751         (dump_gimple_omp_atomic_store): Likewise.
5752         (dump_gimple_mem_ops): Likewise.
5753         (pp_gimple_stmt_1): Likewise.
5754         (pp_cfg_jump): Likewise.
5755         (dump_implicit_edges): Likewise.
5756         (gimple_dump_bb_for_graph): Likewise.
5757         * graph.c (draw_cfg_node): Likewise.
5758         * langhooks.c (lhd_print_error_function): Likewise.
5759         * sched-vis.c (print_exp): Likewise.
5760         (print_value): Likewise.
5761         (print_pattern): Likewise.
5762         (print_insn): Likewise.
5763         (rtl_dump_bb_for_graph): Likewise.
5764         * tree-pretty-print.c (dump_function_declaration): Likewise.
5765         (dump_array_domain): Likewise.
5766         (dump_omp_clause): Likewise.
5767         (dump_location): Likewise.
5768         (dump_generic_node): Likewise.
5769         (print_struct_decl): Likewise.
5770         * diagnostic.c (diagnostic_show_locus): Use pp_space.
5772 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5774         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
5775         candidate table when replacing a candidate statement.
5776         (replace_rhs_if_not_dup): Likewise.
5777         (replace_one_candidate): Likewise.
5779 2013-08-02  Jan Hubicka  <jh@suse.cz>
5780             Martin Liska  <marxin.liska@gmail.com>
5782         * cgraphunit.c (add_new_function): Fix logic when adding from
5783         late IPA pass.
5784         (assemble_thunk): Rename to ...
5785         (expand_thunk); .. this one; export; get it working with
5786         general functions; make produced gimple valid.
5787         * cgraph.h (expand_thunk): Declare.
5789 2013-08-02  Jan Hubicka  <jh@suse.cz>
5791         * ipa-cp.c (gather_context_independent_values): Use
5792         ipa_get_param_move_cost.
5793         (get_replacement_map): Remove PARAM; move parameter folding
5794         into tree-inline.c
5795         (create_specialized_node): Update.
5796         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
5797         assert that we have gimple body; update move_cost.
5798         (count_formal_params): Assert that we have gimple body.
5799         (ipa_dump_param): New function.
5800         (ipa_alloc_node_params): Break out from ...
5801         (ipa_initialize_node_params): ... here.
5802         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
5803         (ipa_write_node_info): Stream move costs.
5804         (ipa_read_node_info): Read move costs.
5805         (ipa_update_after_lto_read): Do not recompute node params.
5806         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
5807         (ipa_get_param): Check we are not in WPA.
5808         (ipa_get_param_move_cost): New.
5809         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
5810         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
5811         parm numbers to be present.
5813 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5815         PR rtl-optimization/58048
5816         * lra-constraints.c (process_alt_operands): Don't check asm
5817         operand on register.
5819 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
5821         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
5822         the implied StoreLoad barrier for atomic operations if before.
5824 2013-08-02  Jan Hubicka  <jh@suse.cz>
5825             Martin Liska  <marxin.liska@gmail.com>
5827         * cgraph.c (cgraph_function_body_availability): Do not check
5828         cgraph flags.
5829         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
5830         symtab_node_availability): Declare.
5831         * ipa.c (can_replace_by_local_alias): New.
5832         (function_and_variable_visibility): Use it.
5833         * symtab.c (symtab_for_node_and_aliases,
5834         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
5836 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5838         PR rtl-optimization/57963
5839         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
5840         (lra_constraints): Use them.
5842 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
5844         * config/arm/types.md (define_attr "type"): Add "load_acq"
5845         and "store_rel".
5846         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
5847         changes.
5848         (cortex_a53_store1): Likewise.
5850 2013-08-01  Jan Hubicka  <jh@suse.cz>
5852         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
5853         partitions are not needed.
5855 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
5857         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
5858         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
5859         MAYBE_NON_Q_CLASS_P where appropriate.
5861 2013-08-01  Jan Hubicka  <jh@suse.cz>
5863         * cgraph.h (release_function_body): Declare.
5864         * tree.c (free_lang_data_in_decl): Free, parameters and return values
5865         of unused delcarations.
5867 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5869         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
5870         RTL form when subtracting a constant.
5872 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5874         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
5875         Generate canonical plus rtx with negated immediate instead of minus
5876         where appropriate.
5877         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
5879 2013-08-01  Jan Hubicka  <jh@suse.cz>
5881         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
5882         (cgraph_release_function_body): Likewise.
5883         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
5884         * cgraph.h (cgrpah_node): Rename abstract_and_needed
5885         to used_as_abstract_origin.
5886         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
5887         symbols used as abstract origins.
5888         * cgraphunit.c (analyze_functions): Update.
5889         * ipa.c (symtab_remove_unreachable_nodes): Recompute
5890         used_as_abstract_origin.
5891         * tree-inline.c (tree_function_versioning): Update
5892         used_as_abstract_origin; be ready for DECL_RESULT and
5893         DECL_ARGUMENTS to be NULL.
5895         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
5896         for abstract functions.
5897         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
5898         real symbols.
5900 2013-08-01  Jan Hubicka  <jh@suse.cz>
5902         * profile.c (compute_value_histograms): Fix thinko.
5904 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5906         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
5907         aarch-common-protos.h to extra_headers.
5908         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5909         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
5910         * config/aarch64/t-aarch64 (aarch-common.o): Define.
5912 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5914         * config/aarch64/aarch64.md (define_attr "type"): Delete.
5915         Include "../arm/types.md".  Define "type" attribute for all patterns.
5916         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
5917         attribute changes.
5919 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
5921         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
5922         to support power8 load fusion.
5923         (fusion_gpr_mem_load): Likewise.
5925         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
5927         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
5928         declarations for power8 load fusion.
5929         (emit_fusion_gpr_load): Likewise.
5931         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
5932         tuning for power8, turn on fusion mode by default.  Turn on sign
5933         extending fusion mode if normal fusion mode is on, and we are at
5934         -O2 or -O3.
5935         (fusion_gpr_load_p): New function, return true if we can fuse an
5936         addis instruction with a dependent load to a GPR.
5937         (emit_fusion_gpr_load): Emit the instructions for power8 load
5938         fusion to GPRs.
5940         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
5941         (VSX load fusion peepholes): New peepholes to fuse together an
5942         addi instruction with a VSX load instruction.
5944         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
5945         peepholes to fuse an addis instruction with a load to a GPR base
5946         register.  If we are supporting sign extending fusions, convert
5947         sign extending loads to zero extending loads and add an explicit
5948         sign extension.
5950 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
5952         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
5953         aarch-common-protos.h to extra_headers.
5954         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5955         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
5956         (arm_early_store_addr_dep): Likewise.
5957         (arm_no_early_alu_shift_dep): Likewise.
5958         (arm_no_early_alu_shift_value_dep): Likewise.
5959         (arm_no_early_mul_dep): Likewise.
5960         (arm_no_early_store_addr_dep): Likewise.
5961         (arm_mac_accumulator_is_mul_result): Likewise.
5962         (arm_mac_accumulator_is_result): Likewise.
5963         * config/arm/aarch-common.c: ... here.  New file.
5964         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
5965         here to ...
5966         (arm_early_store_addr_dep): Likewise.
5967         (arm_no_early_alu_shift_dep): Likewise.
5968         (arm_no_early_alu_shift_value_dep): Likewise.
5969         (arm_no_early_mul_dep): Likewise.
5970         (arm_no_early_store_addr_dep): Likewise.
5971         (arm_mac_accumulator_is_mul_result): Likewise.
5972         (arm_mac_accumulator_is_result): Likewise.
5973         * config/arm/aarch-common-protos.h: ... here.  New file.
5974         * config/arm/t-arm (aarch-common.o): Define.
5976 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
5978         * config/arm/arm.md: Include new file "types.md".
5979         (define_attr "type"): Move from here to ...
5980         (define_attr "mul32"): Likewise.
5981         (define_attr "mul64"): Likewise.
5982         * config/arm/types.md: ... here.  New file.
5984 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5986         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
5987         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
5989 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5991         * gen-pass-instances.awk: Fix offset of substr().
5993 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
5995         * Makefile.in (pass-instances.def): New.
5996         (passes.o): Replace dependency on passes.def with one on
5997         pass-instances.def
5999         * gen-pass-instances.awk: New.
6001         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6002         rather than passes.def, updating local definition of NEXT_PASS
6003         macro to add an extra NUM parameter (currently unused).
6005 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6007         * Makefile.in (PASS_MANAGER_H): New.
6008         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6009         (passes.o): Likewise.
6010         (statistics.o): Likewise.
6011         (cgraphunit.o): Likewise.
6012         (context.o): Depend on PASS_MANAGER_H.
6014         * pass_manager.h: New.
6016         * cgraphunit.c (cgraph_add_new_function): Update for moves
6017         of globals to fields of pass_manager.
6018         (analyze_function): Likewise.
6019         (expand_function): Likewise.
6020         (ipa_passes): Likewise.
6021         (compile): Likewise.
6023         * context.c (context::context): New.
6024         * context.h  (context::context): New.
6025         (context::get_passes): New.
6026         (context::passes_): New.
6028         * lto-cgraph.c (input_node): Update for moves of globals to
6029         fields of pass_manager.
6031         * passes.c (all_passes): Remove, in favor of a field of the
6032         same name within the new class pass_manager.
6033         (all_small_ipa_passes): Likewise.
6034         (all_lowering_passes): Likewise.
6035         (all_regular_ipa_passes): Likewise.
6036         (all_late_ipa_passes): Likewise.
6037         (all_lto_gen_passes): Likewise.
6038         (passes_by_id): Likewise.
6039         (passes_by_id_size): Likewise.
6040         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6041         the new class pass_manager.
6042         (set_pass_for_id): Convert to...
6043         (pass_manager::set_pass_for_id): ...method.
6044         (get_pass_for_id): Convert to...
6045         (pass_manager::get_pass_for_id): ...method.
6046         (register_one_dump_file): Move body of implementation into...
6047         (pass_manager::register_one_dump_file): ...here.
6048         (register_dump_files_1): Convert to...
6049         (pass_manager::register_dump_files_1): ...method.
6050         (register_dump_files): Convert to...
6051         (pass_manager::register_dump_files): ...method.
6052         (create_pass_tab): Update for moves of globals to fields of
6053         pass_manager.
6054         (dump_passes): Move body of implementation into...
6055         (pass_manager::dump_passes): ...here.
6056         (register_pass): Move body of implementation into...
6057         (pass_manager::register_pass): ...here.
6058         (init_optimization_passes): Convert into...
6059         (pass_manager::pass_manager): ...constructor for new
6060         pass_manager class, and initialize the pass_lists array.
6061         (check_profile_consistency): Update for moves of globals to
6062         fields of pass_manager.
6063         (dump_profile_report): Move body of implementation into...
6064         (pass_manager::dump_profile_report): ...here.
6065         (ipa_write_summaries_1): Update for moves of pass lists from
6066         being globals to fields of pass_manager.
6067         (ipa_write_optimization_summaries): Likewise.
6068         (ipa_read_summaries):  Likewise.
6069         (ipa_read_optimization_summaries): Likewise.
6070         (execute_all_ipa_stmt_fixups): Likewise.
6072         * statistics.c (statistics_fini): Update for moves of globals to
6073         fields of pass_manager.
6075         * toplev.c (general_init): Replace call to
6076         init_optimization_passes with construction of the pass_manager
6077         instance.
6079         * tree-pass.h (all_passes): Remove, in favor of a field of the
6080         same name within the new class pass_manager.
6081         (all_small_ipa_passes): Likewise.
6082         (all_lowering_passes): Likewise.
6083         (all_regular_ipa_passes): Likewise.
6084         (all_lto_gen_passes): Likewise.
6085         (all_late_ipa_passes): Likewise.
6086         (passes_by_id): Likewise.
6087         (passes_by_id_size): Likewise.
6088         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6089         the new class pass_manager.
6090         (get_pass_for_id): Remove.
6092 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6094         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6095         configs.
6097 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6099         * arm.md (mulhi3): New expand pattern.
6101 2013-07-30  Jan Hubicka  <jh@suse.cz>
6102             Martin Liska  <marxin.liska@gmail.com>
6104         * profile.c (compute_value_histograms): Do not ICE when
6105         there is mismatch only on some counters.
6107 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6109         PR rtl-optimization/57637
6110         * function.c (move_insn_for_shrink_wrap): Also check the
6111         GEN set of the LIVE problem for the liveness analysis
6112         if it exists, otherwise give up.
6114 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6116         PR tree-optimization/57993
6117         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6118         replaced statement in the candidate table.
6119         (phi_add_costs): Return infinite cost when the hidden basis does
6120         not dominate all phis on which the candidate is dependent.
6121         (replace_one_candidate): Record replaced statement in the
6122         candidate table.
6124 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6126         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6127         (ashlv2si3): New expander.
6128         (*ashlv2si3_i): New define_insn_and_split.
6129         * predicates.md (float_operation): Allow patterns with three
6130         basic sub-patterns.
6132         PR rtl-optimization/58021
6133         * mode-switching.c (create_pre_exit): Always split off preceding
6134         insns if we are not at the basic block head.
6136 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6138         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6139         (UCLIBC_DYNAMIC_LINKER): New macro.
6140         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6141         `-mnan=2008'.
6142         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6143         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6144         `-mnan=2008'.
6145         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6146         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6147         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6148         for SF and DF modes.  Use ieee_quad_format for TF mode.
6149         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6150         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6151         (mips_option_override): Handle `-mnan=legacy'.
6152         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6153         `-mabs=2008' and `-mnan=2008'.
6154         (OPTION_DEFAULT_SPECS): Add "nan" default.
6155         (ASM_SPEC): Handle `-mnan='.
6156         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6157         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6158         comment accordingly.
6159         (neg<mode>2): Likewise.
6160         * config/mips/mips.opt (mabs, mnan): New options.
6161         * doc/install.texi (Configuration): Document `--with-nan=' option.
6162         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6163         `-mnan=' options.
6164         (MIPS Options): Document them.
6165         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6166         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6167         * configure: Regenerate.
6168         * config.in: Regenerate.
6170 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6172         * config/i386/i386.md (float post-reload splitters): Do not check
6173         for subregs of SSE registers.
6175 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6176             H.J. Lu  <hongjiu.lu@intel.com>
6178         PR target/57954
6179         PR target/57988
6180         * config/i386/i386.md (post-reload splitter
6181         to avoid partial SSE reg dependency stalls): New pattern.
6183 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6185         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6187 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6189         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6190         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6192 2013-07-26  Cary Coutant  <ccoutant@google.com>
6194         * dwarf2out.c (die_checksum_ordered): Don't include template
6195         instantiations in signature.
6196         (is_template_parameter): New function.
6197         (is_template_instantiation): New function.
6198         (generate_skeleton_bottom_up): Don't include template instantiations
6199         in type unit DIE.
6200         (generate_skeleton): Likewise.
6201         (break_out_comdat_types): Move recursive call to break out nested
6202         types earlier.
6203         (prune_unused_types_mark_generic_parms_dies): Call
6204         is_template_parameter.
6206 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
6208         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6209         uses vector registers.
6210         * config/aarch64/iterators.md: Add attributes rtn and vas.
6212 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6213             Richard Earnshaw  <richard.earnshaw@arm.com>
6215         * combine.c (simplify_comparison): Re-canonicalize operands
6216         where appropriate.
6217         * config/arm/arm.md (movcond_addsi): New splitter.
6219 2013-07-25  Sterling Augustine  <saugustine@google.com>
6221         * dwarf2out.c (size_of_pubnames): Move code to...
6222         (include_pubname_in_output): ...here.  New.
6223         (want_pubnames): Rearrange.
6224         (output_pubnames): Call include_pubname_in_output.  Move assertion.
6226 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6228         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6230 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6232         PR target/38836
6233         * doc/extend.texi: Remove obsolete builtins. Fix
6234         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6236 2013-07-25  Jan Hubicka  <jh@suse.cz>
6238         * cgraph.c (release_function_body): Break out from ...
6239         (cgraph_release_function_body): ... this one; also release DECL_RESULT
6240         and DECL_ARGUMENTS.
6241         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6242         old_tree in the map.
6243         (create_specialized_node): Update.
6244         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6245         into index.
6246         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6247         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6248         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6249         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6250         DECL_RESULT.
6252 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6254         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6255         addsi3_carryin_alt2_<optab>): Correct output template.
6257 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6259         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6260         Adjust for arm_restrict_it.
6261         Remove trailing whitespace.
6263 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
6265         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6266         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6268         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6270 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
6272         PR rtl-optimization/57960
6273         * lra-constraints.c (process_alt_operands): Use the right mode
6274         when checking strict_low.
6276 2013-07-25  Jan Hubicka  <jh@suse.cz>
6278         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6279         * cgraph.c (cgraph_remove_node): Do not release function body
6280         when in cgraph streaming.
6281         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6282         in other partitions are not considered reachable; fix handling of
6283         clones.
6285 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6287         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6289 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6291         PR target/19599
6292         PR target/57731
6293         PR target/57837
6294         * config/arm/arm.md ("*sibcall_insn): Replace use of
6295         Ss with US. Adjust output for v5 and v4t.
6296         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6298         * config/arm/constraints.md ("Ss"): Rename to US.
6300 2013-07-25  Terry Guo  <terry.guo@arm.com>
6302         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6303         shift_add/shift_sub0/shift_sub1 RTXs.
6305 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
6306             Anton Blanchard  <anton@au1.ibm.com>
6308         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6309         (altivec_vpks<VI_char>ss): Likewise.
6310         (altivec_vpks<VI_char>us): Likewise.
6311         (altivec_vpku<VI_char>us): Likewise.
6312         (altivec_vpku<VI_char>um): Likewise.
6314 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
6316         Introduce context class.
6318         * Makefile.in (CONTEXT_H): New.
6319         (OBJS): Add context.o.
6320         (toplev.o): Add CONTEXT_H to dependencies.
6321         (context.o): New.
6323         * toplev.c (general_init): Create the singleton gcc::context instance.
6325         * context.c: New.
6327         * context.h: New.
6329 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
6331         PR rtl-optimization/57968
6332         * mode-switching.c (create_pre_exit): Allow instructions that
6333         don't set a return register to need a non-exit mode.
6335 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6336             Anton Blanchard  <anton@au1.ibm.com>
6338         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6339         operands to vperm for little endian.
6340         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6341         of lvsl to create the control mask for a vperm for little endian.
6343 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6344             Anton Blanchard  <anton@au1.ibm.com>
6346         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6347         two operands for little-endian.
6349 2013-07-23  Steve Ellcey  <sellcey@mips.com>
6351         * config/mips/mips.c (mips_case_values_threshold): New.
6352         (TARGET_CASE_VALUES_THRESHOLD): Define.
6354 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6355             Anton Blanchard  <anton@au1.ibm.com>
6357         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6358         selection of field for vector splat in little endian mode.
6360 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6362         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6363         expanders to rs6000.md.
6364         (ior<mode>3): Likewise.
6365         (and<mode>3): Likewise.
6366         (one_cmpl<mode>2): Likewise.
6367         (nor<mode>3): Likewise.
6368         (andc<mode>3): Likewise.
6369         (eqv<mode>3): Likewise.
6370         (nand<mode>3): Likewise.
6371         (orc<mode>3): Likewise.
6373         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6374         declaration.
6376         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6377         to split multi-word logical operations.
6378         (rs6000_split_logical_di): Likewise.
6379         (rs6000_split_logical): Likewise.
6381         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6382         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6383         and allow TImode operations in 32-bit.
6384         (vsx_and<mode>3_64bit): Likewise.
6385         (vsx_ior<mode>3_32bit): Likewise.
6386         (vsx_ior<mode>3_64bit): Likewise.
6387         (vsx_xor<mode>3_32bit): Likewise.
6388         (vsx_xor<mode>3_64bit): Likewise.
6389         (vsx_one_cmpl<mode>2_32bit): Likewise.
6390         (vsx_one_cmpl<mode>2_64bit): Likewise.
6391         (vsx_nor<mode>3_32bit): Likewise.
6392         (vsx_nor<mode>3_64bit): Likewise.
6393         (vsx_andc<mode>3_32bit): Likewise.
6394         (vsx_andc<mode>3_64bit): Likewise.
6395         (vsx_eqv<mode>3_32bit): Likewise.
6396         (vsx_eqv<mode>3_64bit): Likewise.
6397         (vsx_nand<mode>3_32bit): Likewise.
6398         (vsx_nand<mode>3_64bit): Likewise.
6399         (vsx_orc<mode>3_32bit): Likewise.
6400         (vsx_orc<mode>3_64bit): Likewise.
6402         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6403         logical types in GPRs.
6405         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6406         logical insns to rs6000.md, and allow TImode operations in
6407         32-bit.
6408         (altivec_ior<mode>3): Likewise.
6409         (altivec_xor<mode>3): Likewise.
6410         (altivec_one_cmpl<mode>2): Likewise.
6411         (altivec_nor<mode>3): Likewise.
6412         (altivec_andc<mode>3): Likewise.
6414         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6415         attributes for moving the 128-bit logical operations into
6416         rs6000.md.
6417         (BOOL_REGS_OUTPUT): Likewise.
6418         (BOOL_REGS_OP1): Likewise.
6419         (BOOL_REGS_OP2): Likewise.
6420         (BOOL_REGS_UNARY): Likewise.
6421         (BOOL_REGS_AND_CR0): Likewise.
6422         (one_cmpl<mode>2): Add support for DI logical operations on
6423         32-bit, splitting the operations to 32-bit.
6424         (anddi3): Likewise.
6425         (iordi3): Likewise.
6426         (xordi3): Likewise.
6427         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6428         changes to combine the 32/64-bit code, allow logical operations on
6429         TI mode in 32-bit, and to use similar match_operator patterns like
6430         scalar mode uses.  Combine the Altivec and VSX code for logical
6431         operations, and move it here.
6432         (ior<mode>3, 128-bit types): Likewise.
6433         (xor<mode>3, 128-bit types): Likewise.
6434         (one_cmpl<mode>3, 128-bit types): Likewise.
6435         (nor<mode>3, 128-bit types): Likewise.
6436         (andc<mode>3, 128-bit types): Likewise.
6437         (eqv<mode>3, 128-bit types): Likewise.
6438         (nand<mode>3, 128-bit types): Likewise.
6439         (orc<mode>3, 128-bit types): Likewise.
6440         (and<mode>3_internal): Likewise.
6441         (bool<mode>3_internal): Likewise.
6442         (boolc<mode>3_internal1): Likewise.
6443         (boolc<mode>3_internal2): Likewise.
6444         (boolcc<mode>3_internal1): Likewise.
6445         (boolcc<mode>3_internal2): Likewise.
6446         (eqv<mode>3_internal1): Likewise.
6447         (eqv<mode>3_internal2): Likewise.
6448         (one_cmpl1<mode>3_internal): Likewise.
6450 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6452         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6453         Rename flag_stack_usage to flag_stack_usage_info.
6455 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6457         * config/microblaze/sync.md: New file.
6458         * config/microblaze/microblaze.md: Include sync.md
6459         * config/microblaze/microblaze.c: Add print_operand 'y'.
6460         * config/microblaze/constraints.md: Add memory_contraint
6461         'Q' which is a single register.
6463 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
6465         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6467 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6469         * reload.c (find_reloads): Exit loop once we find this operand
6470         cannot be reloaded somehow for this alternative.
6472         * reload.c (find_reloads): Exit loop once we find a hard register.
6474         * rtlanal.c (computed_jump_p): Exit loop once we find label
6475         reference is used.
6477         * i386.c (ix86_pad_returns): Exit loop after setting replace.
6479         * cfgloopmanip.c (remove_path): Exit loop after setting
6480         irred_invalidated.
6482         * gensupport.c (subst_dup): Avoid loop if code is not
6483         MATCH_DUP nor MATCH_OP_DUP.
6485 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
6487         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6489 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6491         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6492         true for SP_REGNUM if mode == ptr_mode.
6493         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6494         with value R0_REGNUM + 31.
6496 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6498         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6499         pad pointer-typed argument downward.
6501 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6503         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6504         and __ILP32__ when the ILP32 model is in use.
6506 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6508         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6509         (aarch64_load_symref_appropriately): In the case of
6510         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6511         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6512         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6513         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6514         if 'mode' doesn't equal to 'ptr_mode'.
6515         (aarch64_output_mi_thunk): Add an assertion on the alignment of
6516         'vcall_offset'; change to call aarch64_emit_move differently depending
6517         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6518         to calculate the upper bound of 'vcall_offset'.
6519         (aarch64_cannot_force_const_mem): Change to also return true if
6520         mode != ptr_mode.
6521         (aarch64_legitimize_reload_address): In the case of large
6522         displacements, add new local variable 'xmode' and an assertion
6523         based on it; change to use 'xmode' to generate the new rtx and
6524         reload.
6525         (aarch64_asm_trampoline_template): Change to generate the template
6526         differently depending on TARGET_ILP32 or not; change to use
6527         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6528         (aarch64_trampoline_size): Removed.
6529         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
6530         and replace immediate literals with it.  Change to use 'ptr_mode'
6531         instead of 'DImode' and call convert_memory_address if the mode
6532         of 'fnaddr' doesn't equal to 'ptr_mode'.
6533         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
6534         to output symbol.
6535         (aarch64_elf_asm_destructor): Likewise.
6536         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
6537         on TARGET_ILP32 instead of aarch64_trampoline_size.
6538         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
6539         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
6540         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
6541         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
6542         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
6543         (storewb_pair<GPI:mode>_<P:mode>): ... this.
6544         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
6545         depending on the value of 'mode'.
6546         (add_losym_<mode>): New.
6547         (ldr_got_small_<mode>): New, based on ldr_got_small.
6548         (ldr_got_small): Remove.
6549         (ldr_got_small_sidi): New.
6550         * config/aarch64/iterators.md (P): New.
6551         (PTR): Change to 'ptr_mode' in the condition.
6553 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6555         * config.gcc (aarch64*-*-*): Support --with-abi.
6556         (aarch64*-*-elf): Support --with-multilib-list.
6557         (aarch64*-*-linux*): Likewise.
6558         (supported_defaults): Add abi to aarch64*-*-*.
6559         * configure.ac: Mention AArch64 for --with-multilib-list.
6560         * configure: Re-generated.
6561         * config/aarch64/biarchilp32.h: New file.
6562         * config/aarch64/biarchlp64.h: New file.
6563         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
6564         (ABI_SPEC): Ditto.
6565         (MULTILIB_DEFAULTS): Ditto.
6566         (DRIVER_SELF_SPECS): Ditto.
6567         (ASM_SPEC): Update to also substitute -mabi.
6568         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
6569         file whose name depends on -mabi= and -mbig-endian.
6570         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
6571         TARGET_ILP32.
6572         (POINTER_SIZE): New define.
6573         (POINTERS_EXTEND_UNSIGNED): Ditto.
6574         (enum aarch64_abi_type): New enumeration tag.
6575         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
6576         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
6577         (TARGET_ILP32): New define.
6578         * config/aarch64/aarch64.opt (mabi): New.
6579         (aarch64_abi): New.
6580         (ilp32, lp64): New values for -mabi.
6581         * config/aarch64/t-aarch64 (comma): New define.
6582         (MULTILIB_OPTIONS): Ditto.
6583         (MULTILIB_DIRNAMES): Ditto.
6584         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
6585         * doc/invoke.texi: Document -mabi for AArch64.
6587 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
6589         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
6591 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6592             Anton Blanchard  <anton@au1.ibm.com>
6594         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
6595         endianness when selecting field to splat.
6597 2013-07-22  Eric Christopher  <echristo@gmail.com>
6599         * dwarf2out.c (die_odr_checksum): New function to use
6600         CHECKSUM_ macros and ULEB128 for DIE tag.
6601         (generate_type_signature): Use.
6603 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
6605         * config.gcc (sparc*-*-*): Accept leon3 processor.
6606         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
6607         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
6608         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
6609         * config/sparc/sparc.opt (enum processor_type): Add leon3.
6610         (mfix-ut699): Adjust comment.
6611         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
6612         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
6613         (CPP_CPU_SPEC): Likewise.
6614         (ASM_CPU_SPEC): Likewise.
6615         * config/sparc/sparc.c (leon3_cost): New constant.
6616         (sparc_option_override): Add leon3 support.
6617         (mem_ref): New function.
6618         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
6619         (sparc_do_work_around_errata): Look into the instruction in the delay
6620         slot and adjust accordingly.  Add fix for the data cache nullify issues
6621         of the UT699.  Change insertion position for the NOP.
6622         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
6623         (leon3_load): New reservation.
6624         (leon_store): Bump latency to 2.
6625         (grfpu): New automaton.
6626         (grfpu_alu): New unit.
6627         (grfpu_ds): Likewise.
6628         (leon_fp_alu): Adjust.
6629         (leon_fp_mult): Delete.
6630         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
6631         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
6632         * config/sparc/sparc.md (cpu): Add leon3.
6633         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
6634         (swapsi): Likewise.
6635         (atomic_test_and_set): Likewise.
6636         (ldstub): Likewise.
6638 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
6640         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
6641         default for R5900 targets.
6642         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
6643         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
6644         * config/mips/mips.c (mips_option_override): Report an error for
6645         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
6646         for -march=r5900 -mhard-float.
6648 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6650         * df-problems.c (can_move_insns_across): Exit loop once we
6651         find a non-fixed, non-global register.
6653         * ipa-pure-const.c (propagate_nothrow): Exit loop after
6654         setting can_throw.
6656         * omega.c (omega_eliminate_red): Break after setting red_found.
6657         (omega_problem_has_red_equations): Similarly after setting found.
6658         (omega_query_variable): Similarly after setting coupled.
6660 2013-07-22  Marek Polacek  <polacek@redhat.com>
6662         * gimplify.c: Don't include gimple.h twice.
6664 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6666         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
6667         instead of TARGET_THUMB1.
6668         (Pz): New constraint.
6669         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
6670         encodings.
6671         (compare_negsi_si): Likewise.
6672         (compare_addsi2_op0): Likewise.
6673         (compare_addsi2_op1): Likewise.
6674         (addsi3_carryin_<optab>): Likewise.
6675         (addsi3_carryin_alt2_<optab>): Likewise.
6676         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
6677         for arm_restrict_it.
6678         (subsi3_carryin): Likewise.
6679         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
6680         (minmax_arithsi): Disable for arm_restrict_it.
6681         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
6682         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
6683         (satsi_<SAT:code>_shift): Likewise.
6684         (arm_shiftsi3): Add alternative for 16-bit encoding.
6685         (arm32_movhf): Disable for arm_restrict_it.
6686         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
6687         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
6689 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
6691         * config/arm/arm.md (attribute "insn"): Delete.
6692         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
6693         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
6694         (not_shiftsi): Update for attribute change.
6695         (not_shiftsi_compare0): Likewise.
6696         (not_shiftsi_compare0_scratch): Likewise.
6697         (arm_one_cmplsi2): Likewise.
6698         (thumb1_one_cmplsi2): Likewise.
6699         (notsi_compare0): Likewise.
6700         (notsi_compare0_scratch): Likewise.
6701         (thumb1_movdi_insn): Likewise.
6702         (arm_movsi_insn): Likewise.
6703         (movhi_insn_arch4): Likewise.
6704         (movhi_bytes): Likewise.
6705         (arm_movqi_insn): Likewise.
6706         (thumb1_movqi_insn): Likewise.
6707         (arm32_movhf): Likewise.
6708         (thumb1_movhf): Likewise.
6709         (arm_movsf_soft_insn): Likewise.
6710         (thumb1_movsf_insn): Likewise.
6711         (thumb_movdf_insn): Likewise.
6712         (movsicc_insn): Likewise.
6713         (movsfcc_soft_insn): Likewise.
6714         (and_scc): Likewise.
6715         (cond_move): Likewise.
6716         (if_move_not): Likewise.
6717         (if_not_move): Likewise.
6718         (if_shift_move): Likewise.
6719         (if_move_shift): Likewise.
6720         (if_shift_shift): Likewise.
6721         (if_not_arith): Likewise.
6722         (if_arith_not): Likewise.
6723         (cond_move_not): Likewise.
6724         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
6725         (neon_mov<mode>): Likewise.
6726         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
6727         (thumb2_movsi_vfp): Likewise.
6728         (movsf_vfp): Likewise.
6729         (thumb2_movsf_vfp): Likewise.
6730         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
6731         change.
6732         (cortexa7_older_only): Likewise.
6733         (cortexa7_younger): Likewise.
6734         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
6735         (1020alu_shift_op): Likewise.
6736         (1020alu_shift_reg_op): Likewise.
6737         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
6738         (alu_shift_op): Likewise.
6739         (alu_shift_reg_op): Likewise.
6740         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
6741         (11_alu_shift_op): Likewise.
6742         (11_alu_shift_reg_op): Likewise.
6743         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
6744         (9_alu_shift_reg_op): Likewise.
6745         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6746         change.
6747         (cortex_a15_alu_shift): Likewise.
6748         (cortex_a15_alu_shift_reg): Likewise.
6749         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
6750         (cortex_a5_alu_shift): Likewise.
6751         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
6752         change.
6753         (cortex_a53_alu_shift): Likewise.
6754         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
6755         change.
6756         (cortex_a7_alu_reg): Likewise.
6757         (cortex_a7_alu_shift): Likewise.
6758         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6759         (cortex_a8_alu_shift): Likewise.
6760         (cortex_a8_alu_shift_reg): Likewise.
6761         (cortex_a8_mov): Likewise.
6762         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
6763         (cortex_a9_dp_shift): Likewise.
6764         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
6765         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
6766         (cortex_r4_mov): Likewise.
6767         (cortex_r4_alu_shift): Likewise.
6768         (cortex_r4_alu_shift_reg): Likewise.
6769         * config/arm/fa526.md (526_alu_op): Update for attribute change.
6770         (526_alu_shift_op): Likewise.
6771         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
6772         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
6773         (626te_alu_shift_op): Likewise.
6774         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
6775         (726te_alu_op): Likewise.
6776         (726te_alu_shift_op): Likewise.
6777         (726te_alu_shift_reg_op): Likewise.
6778         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
6779         (mp626_alu_shift_op): Likewise.
6780         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
6781         (pj4_alu_e1_conds): Likewise.
6782         (pj4_alu): Likewise.
6783         (pj4_alu_conds): Likewise.
6784         (pj4_shift): Likewise.
6785         (pj4_shift_conds): Likewise.
6786         (pj4_alu_shift): Likewise.
6787         (pj4_alu_shift_conds): Likewise.
6789 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6791         * config/arm/predicates.md (shiftable_operator_strict_it):
6792         New predicate.
6793         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
6794         Disable cond_exec version for arm_restrict_it.
6795         (thumb2_smaxsi3): Convert to generate cond_exec.
6796         (thumb2_sminsi3): Likewise.
6797         (thumb32_umaxsi3): Likewise.
6798         (thumb2_uminsi3): Likewise.
6799         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
6800         (thumb2_neg_abssi2): Likewise.
6801         (thumb2_mov_scc): Add alternative for 16-bit encoding.
6802         (thumb2_movsicc_insn): Adjust alternatives.
6803         (thumb2_mov_negscc): Disable for arm_restrict_it.
6804         (thumb2_mov_negscc_strict_it): New pattern.
6805         (thumb2_mov_notscc_strict_it): New pattern.
6806         (thumb2_mov_notscc): Disable for arm_restrict_it.
6807         (thumb2_ior_scc): Likewise.
6808         (thumb2_ior_scc_strict_it): New pattern.
6809         (thumb2_cond_move): Adjust for arm_restrict_it.
6810         (thumb2_cond_arith): Disable for arm_restrict_it.
6811         (thumb2_cond_arith_strict_it): New pattern.
6812         (thumb2_cond_sub): Adjust for arm_restrict_it.
6813         (thumb2_movcond): Likewise.
6814         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
6815         (thumb2_zero_extendhisi2_v6): Likewise.
6816         (thumb2_zero_extendqisi2_v6): Likewise.
6817         (orsi_notsi_si): Likewise.
6818         (orsi_not_shiftsi_si): Likewise.
6820 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
6822         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
6823         instruction sequence is 1 byte shorter.
6825 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
6827         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
6828         it is not needed after split.
6830 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
6832         PR target/51784
6833         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
6834         second label for nonlocal goto receivers. Don't output pic base labels
6835         unless we're producing PIC; mark that action unreachable().
6836         (ix86_save_reg): If the function contains a nonlocal label, save the
6837         PIC base reg.
6838         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
6839         * config/darwin.c (emitted_pic_label_num): New GTY.
6840         (update_pic_label_number_if_needed): New.
6841         (machopic_output_function_base_name): Adjust for nonlocal receiver
6842         case.
6843         (machopic_should_output_picbase_label): New.
6844         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
6845         (nonlocal_goto_receiver): New insn and split.
6847 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
6849         * config/aarch64/aarch64-builtins.c
6850         (aarch64_fold_builtin): Fold abs in all modes.
6851         * config/aarch64/aarch64-simd-builtins.def
6852         (abs): Enable for all modes.
6853         * config/aarch64/arm_neon.h
6854         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
6855         (vabs_f64): Add missing intrinsic.
6857 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
6859         * config/aarch64/arm_neon.h (vabs_s64): New function
6861 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
6863         PR target/57516
6864         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
6865         * config/avr/avr.md (adjust_len): Add `round'.
6866         * config/avr/avr-protos.h (avr_out_round): New prototype.
6867         (avr_out_plus): Add `out_label' argument.
6868         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
6869         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
6870         Handle the case where `insn' is just a pattern.
6871         (avr_out_bitop): Handle the case where `insn' is just a pattern.
6872         (avr_out_round): New function.
6873         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
6875 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
6877         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6878         Add check for flag_stack_usage to handle -fstack-usage support
6880 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
6882         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
6883         interaction for new Power8 flags and VSX.
6885 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
6887         PR middle-end/57698
6888         * tree-inline.c (expand_call_inline): Emit errors during
6889         early_inlining only if optimization is not turned on.
6891 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6893         * passes.def: New.
6895         * passes.c (init_optimization_passes): Move the construction of
6896         the pass hierarchy into a new passes.def file.
6898         * Makefile.in (passes.o): Add dependency on passes.def.
6900 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6902         * passes.c (init_optimization_passes): Introduce macros for
6903         constructing the tree of passes (INSERT_PASSES_AFTER,
6904         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
6905         TERMINATE_PASS_LIST).
6907 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
6908             Wei Mi  <wmi@google.com>
6910         PR rtl-optimization/57878
6911         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
6912         top.
6913         (reload_pseudo_compare_func): Check nregs first for reload
6914         pseudos.
6916 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6918         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
6920 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
6922         * read-rtl.c (validate_const_int): Once an invalid character is
6923         seen, quit the loop.
6925         * gengtype.c (write_roots): Similarly once we find the "deletable"
6926         or "if_marked" option.
6928 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6930         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
6931         "xtab" and "sat".  Move value "clz" from here to ...
6932         (attriubte "type"): ... here.
6933         (satsi_<SAT:code>): Delete "insn" attribute.
6934         (satsi_<SAT:code>_shift): Likewise.
6935         (arm_zero_extendqisi2addsi): Likewise.
6936         (arm_extendqisi2addsi): Likewise.
6937         (clzsi2): Update for attribute changes.
6938         (rbitsi2): Likewise.
6939         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
6940         attribute.
6941         (arm_usatsihi): Likewise.
6942         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6944 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6946         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
6947         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
6948         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
6949         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
6950         in alphabetical order.
6951         (attribute "core_cycles"): Update for attribute changes.
6952         (arm_addsi3): Likewise.
6953         (addsi3_compare0): Likewise.
6954         (addsi3_compare0_scratch): Likewise.
6955         (addsi3_compare_op1): Likewise.
6956         (addsi3_compare_op2): Likewise.
6957         (compare_addsi2_op0): Likewise.
6958         (compare_addsi2_op1): Likewise.
6959         (addsi3_carryin_shift_<optab>): Likewise.
6960         (subsi3_carryin_shift): Likewise.
6961         (rsbsi3_carryin_shift): Likewise.
6962         (arm_subsi3_insn): Likewise.
6963         (subsi3_compare0): Likewise.
6964         (subsi3_compare): Likewise.
6965         (arm_andsi3_insn): Likewise.
6966         (thumb1_andsi3_insn): Likewise.
6967         (andsi3_compare0): Likewise.
6968         (andsi3_compare0_scratch): Likewise.
6969         (zeroextractsi_compare0_scratch
6970         (andsi_not_shiftsi_si): Likewise.
6971         (iorsi3_insn): Likewise.
6972         (iorsi3_compare0): Likewise.
6973         (iorsi3_compare0_scratch): Likewise.
6974         (arm_xorsi3): Likewise.
6975         (thumb1_xorsi3_insn): Likewise.
6976         (xorsi3_compare0): Likewise.
6977         (xorsi3_compare0_scratch): Likewise.
6978         (satsi_<SAT:code>_shift): Likewise.
6979         (rrx): Likewise.
6980         (arm_shiftsi3): Likewise.
6981         (shiftsi3_compare0): Likewise.
6982         (not_shiftsi): Likewise.
6983         (not_shiftsi_compare0): Likewise.
6984         (not_shiftsi_compare0_scratch): Likewise.
6985         (arm_one_cmplsi2): Likewise.
6986         (thumb_one_complsi2): Likewise.
6987         (notsi_compare0): Likewise.
6988         (notsi_compare0_scratch): Likewise.
6989         (thumb1_zero_extendhisi2): Likewise.
6990         (arm_zero_extendhisi2): Likewise.
6991         (arm_zero_extendhisi2_v6): Likewise.
6992         (arm_zero_extendhisi2addsi): Likewise.
6993         (thumb1_zero_extendqisi2): Likewise.
6994         (thumb1_zero_extendqisi2_v6): Likewise.
6995         (arm_zero_extendqisi2): Likewise.
6996         (arm_zero_extendqisi2_v6): Likewise.
6997         (arm_zero_extendqisi2addsi): Likewise.
6998         (thumb1_extendhisi2): Likewise.
6999         (arm_extendhisi2): Likewise.
7000         (arm_extendhisi2_v6): Likewise.
7001         (arm_extendqisi): Likewise.
7002         (arm_extendqisi_v6): Likewise.
7003         (arm_extendqisi2addsi): Likewise.
7004         (thumb1_extendqisi2): Likewise.
7005         (thumb1_movdi_insn): Likewise.
7006         (arm_movsi_insn): Likewise.
7007         (movsi_compare0): Likewise.
7008         (movhi_insn_arch4): Likewise.
7009         (movhi_bytes): Likewise.
7010         (arm_movqi_insn): Likewise.
7011         (thumb1_movqi_insn): Likewise.
7012         (arm32_movhf): Likewise.
7013         (thumb1_movhf): Likewise.
7014         (arm_movsf_soft_insn): Likewise.
7015         (thumb1_movsf_insn): Likewise.
7016         (movdf_soft_insn): Likewise.
7017         (thumb_movdf_insn): Likewise.
7018         (arm_cmpsi_insn): Likewise.
7019         (cmpsi_shiftsi): Likewise.
7020         (cmpsi_shiftsi_swp): Likewise.
7021         (arm_cmpsi_negshiftsi_si): Likewise.
7022         (movsicc_insn): Likewise.
7023         (movsfcc_soft_insn): Likewise.
7024         (arith_shiftsi): Likewise.
7025         (arith_shiftsi_compare0
7026         (arith_shiftsi_compare0_scratch
7027         (sub_shiftsi): Likewise.
7028         (sub_shiftsi_compare0
7029         (sub_shiftsi_compare0_scratch
7030         (and_scc): Likewise.
7031         (cond_move): Likewise.
7032         (if_plus_move): Likewise.
7033         (if_move_plus): Likewise.
7034         (if_move_not): Likewise.
7035         (if_not_move): Likewise.
7036         (if_shift_move): Likewise.
7037         (if_move_shift): Likewise.
7038         (if_shift_shift): Likewise.
7039         (if_not_arith): Likewise.
7040         (if_arith_not): Likewise.
7041         (cond_move_not): Likewise.
7042         (thumb1_ashlsi3): Set type attribute.
7043         (thumb1_ashrsi3): Likewise.
7044         (thumb1_lshrsi3): Likewise.
7045         (thumb1_rotrsi3): Likewise.
7046         (shiftsi3_compare0_scratch): Likewise.
7047         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7048         (neon_mov<mode>): Likewise.
7049         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7050         attribute changes.
7051         (thumb2_movsi_insn): Likewise.
7052         (thumb2_cmpsi_neg_shiftsi): Likewise.
7053         (thumb2_extendqisi_v6): Likewise.
7054         (thumb2_zero_extendhisi2_v6): Likewise.
7055         (thumb2_zero_extendqisi2_v6): Likewise.
7056         (thumb2_shiftsi3_short): Likewise.
7057         (thumb2_addsi3_compare0_scratch): Likewise.
7058         (orsi_not_shiftsi_si): Likewise.
7059         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7060         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7061         changes.
7062         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7063         (1020alu_shift_op): Likewise.
7064         (1020alu_shift_reg_op): Likewise.
7065         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7066         (alu_shift_op): Likewise.
7067         (alu_shift_reg_op): Likewise.
7068         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7069         (11_alu_shift_op): Likewise.
7070         (11_alu_shift_reg_op): Likewise.
7071         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7072         (9_alu_shift_reg_op): Likewise.
7073         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7074         changes.
7075         (cortex_a15_alu_shift): Likewise.
7076         (cortex_a15_alu_shift_reg): Likewise.
7077         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7078         changes.
7079         (cortex_a5_alu_shift): Likewise.
7080         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7081         changes.
7082         (cortex_a53_alu_shift): Likewise.
7083         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7084         changes.
7085         (cortex_a7_alu_reg): Likewise.
7086         (cortex_a7_alu_shift): Likewise.
7087         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7088         changes.
7089         (cortex_a8_alu_shift): Likewise.
7090         (cortex_a8_alu_shift_reg): Likewise.
7091         (cortex_a8_mov): Likewise.
7092         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7093         (cortex_a9_dp_shift): Likewise.
7094         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7095         changes.
7096         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7097         changes.
7098         (cortex_r4_mov): Likewise.
7099         (cortex_r4_alu_shift): Likewise.
7100         (cortex_r4_alu_shift_reg): Likewise.
7101         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7102         (526_alu_shift_op): Likewise.
7103         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7104         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7105         (626te_alu_shift_op): Likewise.
7106         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7107         (726te_alu_op): Likewise.
7108         (726te_alu_shift_op): Likewise.
7109         (726te_alu_shift_reg_op): Likewise.
7110         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7111         (mp626_alu_shift_op): Likewise.
7112         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7113         (pj4_alu_e1_conds): Likewise.
7114         (pj4_alu): Likewise.
7115         (pj4_alu_conds): Likewise.
7116         (pj4_shift): Likewise.
7117         (pj4_shift_conds): Likewise.
7118         (pj4_alu_shift): Likewise.
7119         (pj4_alu_shift_conds): Likewise.
7120         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7121         changes.
7122         (cortexa7_older_only): Likewise.
7123         (cortexa7_younger): Likewise.
7125 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7127         * ipa-pure-const.c (generate_summary): Rename to...
7128         (pure_const_generate_summary): ... this.
7130 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7132         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7134 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7136         PR target/57909
7137         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7138         usage in HI mode.
7140 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7142         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7143         enabled without -march=zEC12.
7144         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7145         flags to be set.
7147 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7149         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7150         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7151         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7152         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7153         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7154         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7155         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7156         ISA_HAS_FP4.
7157         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7158         and ISA_HAS_NMADD3_NMSUB3.
7159         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7160         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7161         (nmsub4<mode>, nmsub3<mode>): Likewise.
7162         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7164 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7166         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7167         TARGET_MIPS5400 checking.
7169 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7170             Peter Bergner  <bergner@vnet.ibm.com>
7172         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7173         registers in the comment.
7174         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7175         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7176         rather than FIRST_PSEUDO_REGISTERS.
7178 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7180         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7181         enable extra ISA flags with TARGET_HTM.
7183 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7185         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7186         Fix comment typos.
7188 2013-07-15  Cong Hou  <congh@google.com>
7190         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7191         in compare function for sorting.
7193 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
7195         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7196         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7197         * config/rs6000/rs6000.opt: Add -mhtm option.
7198         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7199         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7200         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7201         __HTM__ if the HTM instructions are available.
7202         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7203         htm_spr_reg_operand): New define_predicates.
7204         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7205         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7206         Include htm.md.
7207         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7208         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7209         HTM builtin functions.
7210         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7211         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7212         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7213         (rs6000_builtin_mask_calculate): Likewise.
7214         (rs6000_option_override_internal): Likewise.
7215         (bdesc_htm): Add new HTM builtin support.
7216         (htm_spr_num): New function.
7217         (htm_spr_regno): Likewise.
7218         (rs6000_htm_spr_icode): Likewise.
7219         (htm_expand_builtin): Likewise.
7220         (htm_init_builtins): Likewise.
7221         (rs6000_expand_builtin): Add support for HTM builtin functions.
7222         (rs6000_init_builtins): Likewise.
7223         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7224         option.
7225         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7226         (TARGET_HTM, MASK_HTM): Define macros.
7227         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7228         (FIXED_REGISTERS): Likewise.
7229         (CALL_USED_REGISTERS): Likewise.
7230         (CALL_REALLY_USED_REGISTERS): Likewise.
7231         (REG_ALLOC_ORDER): Likewise.
7232         (enum reg_class): Likewise.
7233         (REG_CLASS_NAMES): Likewise.
7234         (REG_CLASS_CONTENTS): Likewise.
7235         (REGISTER_NAMES): Likewise.
7236         (ADDITIONAL_REGISTER_NAMES): Likewise.
7237         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7238         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7239         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7240         * config/rs6000/htm.md: New file.
7241         * config/rs6000/htmintrin.h: New file.
7242         * config/rs6000/htmxlintrin.h: New file.
7244 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7246         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7247         Define SYMBOL_TINY_GOT, update comment.
7248         * config/aarch64/aarch64.c
7249         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7250         (aarch64_expand_mov_immediate): Likewise.
7251         (aarch64_print_operand): Likewise.
7252         (aarch64_classify_symbol): Likewise.
7253         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7254         (ldr_got_tiny): Define.
7256 2013-07-13  Tobias Grosser  <tobias@grosser.es>
7258         PR tree-optimization/54094
7259         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7260           scheduling dimension for the parallelism check from the polyhedral
7261           information in the AST.
7262         * graphite-dependences.c (carries_deps): Do not assume the schedule is
7263           in 2D + 1 form.
7265 2013-07-13  Jason Merrill  <jason@redhat.com>
7267         * print-tree.c (debug_vec_tree): Use debug_raw.
7268         (debug_raw (vec<tree, va_gc> &)): New.
7269         (debug_raw (vec<tree, va_gc> *)): New.
7270         * tree.h: Declare them.
7272 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
7274         * ifcvt.c (ifcvt_after_combine): New static variable.
7275         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7276         for size.
7277         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
7278         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7279         rest_of_handle_if_after_reload): Pass new argument for if_convert.
7281 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
7283         * config/mips/mips.c (mips_expand_call): Remove empty statement.
7285 2013-07-12  Michael Matz  <matz@suse.de>
7287         PR middle-end/55771
7288         * convert.c (convert_to_real): Reject non-float inner types.
7290 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
7292         * config/aarch64/aarch64-protos.h
7293         (aarch64_simd_immediate_valid_for_move): Remove.
7294         * config/aarch64/aarch64.c (simd_immediate_info): New member.
7295         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7296         cases.
7297         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7299 2013-07-11  Steve Ellcey  <sellcey@mips.com>
7301         * config/mips/mips.c (mips_conditional_register_usage): Do not
7302         use t[0-7] registers in MIPS16 mode when optimizing for size.
7304 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
7306         * config/i386/i386.c (dispatch_function_versions): Fix array
7307         indexing of function_version_info to match actual_versions.
7309 2013-07-11  Teresa Johnson  <tejohnson@google.com>
7311         * vec.h (struct va_gc): Move release out-of-line.
7312         (va_gc::release): Call ggc_free on released vec.
7314 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7316         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7317         Require GOT register as additional operand in UNSPEC.
7318         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7319         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7320         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7321         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7322         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7323         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7324         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7326 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
7328         PR target/57631
7329         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7330         name seen by assembler/linker rather if available.
7332 2013-07-11  Andreas Schwab  <schwab@suse.de>
7334         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7336 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
7338         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7340 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
7342         * doc/tm.texi.in: Move hook documentation to ....
7343         * target.def: ... here.
7345         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7346         text on @hook line.
7347         * doc/tm.texi: Regenerate.
7349 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
7351         PR c++/57869
7352         * doc/invoke.texi: Document -Wconditionally-supported.
7354 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7356         PR target/57844
7357         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7358         of my_fp.
7360 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7362         PR target/57506
7363         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7364         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7365         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7366         Remove duplicate devices.
7367         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7368         * config/avr/t-multilib: Regenerate.
7369         * config/avr/avr-tables.opt: Regenerate.
7370         * doc/avr-mmcu.texi: Regenerate.
7372 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7374         PR target/56987
7375         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7377 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
7379         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7380         the cost of MULT when optimizing for size.
7382 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7384         * config/cr16/cr16-protos.h: Don't include target.h.
7386 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
7388         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7389         adjust register size for TDmode and TFmode for VSX registers.
7391 2013-07-08  Kai Tietz  <ktietz@redhat.com>
7393         PR target/56892
7394         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7395         hook_bool_const_tree_true.
7397 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7399         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7400         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7401         * config/s390/s390.md: Define FPR*_REGNUM constants.
7402         Fix FPR2_REGNUM constant (18 -> 17).
7403         ("*trunc<BFP:mode><DFP_ALL:mode>2")
7404         ("*trunc<DFP_ALL:mode><BFP:mode>2")
7405         ("trunc<BFP:mode><DFP_ALL:mode>2")
7406         ("trunc<DFP_ALL:mode><BFP:mode>2")
7407         ("*extend<BFP:mode><DFP_ALL:mode>2")
7408         ("*extend<DFP_ALL:mode><BFP:mode>2")
7409         ("extend<BFP:mode><DFP_ALL:mode>2")
7410         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7411         FPR4_REGNUM.
7413 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
7415         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7417 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7419         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7420         and cfun_fpr_bit_p to cfun_fpr_save_p.
7421         (s390_frame_area, s390_register_info, s390_frame_info)
7422         (s390_emit_prologue, s390_emit_epilogue)
7423         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7424         register numbers.
7425         * config/s390/s390.h: Define *_REGNUM macros for floating point
7426         register numbers.
7428 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
7430         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7432 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
7434         PR rtl-optimization/57786
7435         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7436         and break out of the loop when it is set to false.
7438 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
7440         PR target/57819
7441         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7442         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7443         (const_int 63)) 0)).
7444         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7445         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7446         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7448         PR rtl-optimization/57829
7449         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7450         mask bits outside of mode are just sign-extension from mode to HWI.
7452 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7454         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7455         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7456         adjust_address instead of change_address to keep info about alignment.
7457         (emit_strmov): Remove.
7458         (emit_memmov): New function.
7459         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7460         (expand_movmem_epilogue): Likewise and return updated rtx for
7461         destination.
7462         (expand_constant_movmem_prologue): Likewise and return updated rtx for
7463         destination and source.
7464         (decide_alignment): Refactor, handle vector_loop.
7465         (ix86_expand_movmem): Likewise.
7466         (ix86_expand_setmem): Likewise.
7467         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7469 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
7471         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7472         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7474 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7476         * config/i386/sse.md (sse_movlhps): Change alternative 3
7477         of operand 2 to "m".
7479 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7481         PR target/57807
7482         * config/i386/sse.md (iptr): New mode attribute.
7483         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7484         (<sse>_vm<plusminus_insn><mode>3): Ditto.
7485         (<sse>_vmmul<mode>3): Ditto.
7486         (<sse>_vmdiv<mode>3): Ditto.
7487         (sse_vmrcpv4sf2): Ditto.
7488         (<sse>_vmsqrt<mode>2): Ditto.
7489         (sse_vmrsqrtv4sf2): Ditto.
7490         (<sse>_vm<code><mode>3): Ditto.
7491         (avx_vmcmp<mode>3): Ditto.
7492         (<sse>_vmmaskcmp<mode>3): Ditto.
7493         (<sse>_comi): Ditto.
7494         (<sse>_ucomi): Ditto.
7495         (*xop_vmfrcz_<mode>): Ditto.
7496         (*fmai_fmadd_<mode>): Ditto.
7497         (*fmai_fmsub_<mode>): Ditto.
7498         (*fmai_fnmadd_<mode>): Ditto.
7499         (*fmai_fnmsub_<mode>): Ditto.
7500         (*fma4i_vmfmadd_<mode>): Ditto.
7501         (*fma4i_vmfmsub_<mode>): Ditto.
7502         (*fma4i_vmfnmadd_<mode>): Ditto.
7503         (*fma4i_vmfnmsub_<mode>): Ditto.
7504         (*xop_vmfrcz_<mode>): Ditto.
7505         (sse_cvtps2pi): Ditto.
7506         (sse_cvttps2pi): Ditto.
7507         (sse_cvtss2si): Ditto.
7508         (sse_cvtss2si_2): Ditto.
7509         (sse_cvtss2siq_2): Ditto.
7510         (sse_cvttss2si): Ditto.
7511         (sse_cvttss2siq): Ditto.
7512         (sse_cvtsd2si): Ditto.
7513         (sse_cvtsd2si_2): Ditto.
7514         (sse_cvtsd2siq_2): Ditto.
7515         (sse_cvttsd2si): Ditto.
7516         (sse_cvttsd2siq): Ditto.
7517         (sse_cvtsd2ss): Ditto.
7518         (sse_cvtss2sd): Ditto.
7519         (avx2_pbroadcast<mode>): Ditto.
7520         (avx2_pbroadcast<mode>_1): Ditto.
7521         (*avx_vperm_broadcast_v4sf): Ditto.
7523         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7524         (sse_movlhps): Ditto.
7525         (sse_storehps): Ditto.
7526         (sse_loadhps): Ditto.
7527         (sse_storelps): Ditto.
7528         (sse_loadlps): Ditto.
7529         (*vec_concatv4sf): Ditto.
7530         (*vec_interleave_highv2df): Ditto.
7531         (*vec_interleave_lowv2df): Ditto.
7532         (*vec_extractv2df_1_sse): Ditto.
7533         (*vec_extractv2df_0_sse): Ditto.
7534         (sse2_storelpd): Ditto.
7535         (sse2_loadlpd): Ditto.
7536         (sse2_movsd): Ditto.
7537         (*vec_concatv4si): Ditto.
7538         (vec_concatv2di): Ditto.
7540         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
7541         for Intel asm dialect.
7542         (mmx_punpcklwd): Ditto.
7543         (mmx_punpckldq): Ditto.
7545         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
7546         for intel assembler dialect.
7548 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
7550         PR target/29776
7551         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
7552         for BUILT_IN_C{LZ,LRSB}*.
7553         * tree.h (CASE_INT_FN): Add FN##IMAX case.
7554         * tree-vrp.c (extract_range_basic): Handle
7555         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
7556         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
7557         fall thru to code calling set_value*.
7558         * builtins.c (expand_builtin): Remove *IMAX cases.
7559         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
7560         if width is bigger than 2*HWI.
7562 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
7564         PR rtl-optimization/55342
7565         * lra-int.h (lra_subreg_reload_pseudos): New.
7566         * lra.c: Add undoing optional reloads to the block diagram.
7567         (lra_subreg_reload_pseudos): New.
7568         (lra_optional_reload_pseudos): Change comments.
7569         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
7570         lra_optional_reload_pseudos after undo transformations.
7571         * lra-assigns.c (pseudo_prefix_title): New.
7572         (lra_setup_reg_renumber): Use it.
7573         (spill_for): Ditto.  Check subreg reload pseudos too.
7574         (assign_by_spills): Consider subreg reload pseudos too.
7575         * lra-constraints.c (simplify_operand_subreg): Use
7576         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
7577         (curr_insn_transform): Recognize and do optional reloads.
7578         (undo_optional_reloads): New.
7579         (lra_undo_inheritance): Call undo_optional_reloads.
7581 2013-07-05  Thomas Quinot  <quinot@adacore.com>
7583         * tree-complex.c (expand_complex_operations_1): Fix typo.
7585 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
7587         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
7588         (tune_params): New member 'const vec_costs'.
7589         * config/aarch64/aarch64.c (generic_vector_cost): New.
7590         (generic_tunings): New member 'generic_vector_cost'.
7591         (aarch64_builtin_vectorization_cost): New.
7592         (aarch64_add_stmt_cost): New.
7593         (TARGET_VECTORIZE_ADD_STMT_COST): New.
7594         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
7596 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
7598         PR target/57777
7599         * config/i386/predicates.md (vsib_address_operand): Disallow
7600         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
7602 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
7604         PR middle-end/55030
7605         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
7606         expand_builtin_setjmp_receiver.
7607         (expand_label): Adjust, call expand_builtin_setjmp_receiver
7608         with NULL for the label parameter.
7609         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
7610         the frame-pointer.  Adjust comments.
7611         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
7612         only if LABEL is non-NULL.
7614 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
7616         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
7617         (ARM_ABI_AAPCS64): Ditto.
7618         (arm_abi): Ditto.
7619         (ARM_DEFAULT_ABI): Ditto.
7621 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
7623         * config/aarch64/aarch64-builtins.c
7624         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
7625         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
7626         (st1): Likewise.
7627         * config/aarch64/aarch64-simd.md
7628         (aarch64_ld1<VALL:mode>): New.
7629         (aarch64_st1<VALL:mode>): Likewise.
7630         * config/aarch64/arm_neon.h
7631         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
7633 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
7635         * config/i386/i386.c (gate_insert_vzeroupper): Check if
7636         target ISA is AVX.
7637         (ix86_option_override_internal):Turn on all -mavx target flags by
7638         default as they are dependent on AVX anyway.
7640 2013-07-02  Cary Coutant  <ccoutant@google.com>
7642         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
7643         deterministic hash.
7644         (loc_checksum_ordered): Likewise.
7645         (hash_loc_operands): Remove inline keyword.
7647 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
7649         PR tree-optimization/57741
7650         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
7651         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
7652         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
7653         Allow REAL_CST step_exprs if flag_associative_math.
7654         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
7656 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7658         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
7660 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7662         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
7664 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7666         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
7667         encoding.
7668         (iorsi3_insn): Likewise.
7669         (arm_xorsi3): Likewise.
7671 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
7673         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
7674         to ...
7675         (attribute "type"): ... here, and prefix with "wmmx_".
7676         (attribute "core_cycles"): Update for attribute changes.
7677         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
7678         (tbcstv4hi): Likewise.
7679         (tbcstv2si): Likewise.
7680         (iwmmxt_iordi3): Likewise.
7681         (iwmmxt_xordi3): Likewise.
7682         (iwmmxt_anddi3): Likewise.
7683         (iwmmxt_nanddi3): Likewise.
7684         (iwmmxt_arm_movdi): Likewise.
7685         (iwmmxt_movsi_insn): Likewise.
7686         (mov<mode>_internal): Likewise.
7687         (and<mode>3_iwmmxt): Likewise.
7688         (ior<mode>3_iwmmxt): Likewise.
7689         (xor<mode>3_iwmmxt): Likewise.
7690         (add<mode>3_iwmmxt): Likewise.
7691         (ssaddv8qi3): Likewise.
7692         (ssaddv4hi3): Likewise.
7693         (ssaddv2si3): Likewise.
7694         (usaddv8qi3): Likewise.
7695         (usaddv4hi3): Likewise.
7696         (usaddv2si3): Likewise.
7697         (sub<mode>3_iwmmxt): Likewise.
7698         (sssubv8qi3): Likewise.
7699         (sssubv4hi3): Likewise.
7700         (sssubv2si3): Likewise.
7701         (ussubv8qi3): Likewise.
7702         (ussubv4hi3): Likewise.
7703         (ussubv2si3): Likewise.
7704         (mulv4hi3_iwmmxt): Likewise.
7705         (smulv4hi3_highpart): Likewise.
7706         (umulv4hi3_highpart): Likewise.
7707         (iwmmxt_wmacs): Likewise.
7708         (iwmmxt_wmacsz): Likewise.
7709         (iwmmxt_wmacu): Likewise.
7710         (iwmmxt_wmacuz): Likewise.
7711         (iwmmxt_clrdi): Likewise.
7712         (iwmmxt_clrv8qi): Likewise.
7713         (iwmmxt_clr4hi): Likewise.
7714         (iwmmxt_clr2si): Likewise.
7715         (iwmmxt_uavgrndv8qi3): Likewise.
7716         (iwmmxt_uavgrndv4hi3): Likewise.
7717         (iwmmxt_uavgv8qi3): Likewise.
7718         (iwmmxt_uavgv4hi3): Likewise.
7719         (iwmmxt_tinsrb): Likewise.
7720         (iwmmxt_tinsrh): Likewise.
7721         (iwmmxt_tinsrw): Likewise.
7722         (iwmmxt_textrmub): Likewise.
7723         (iwmmxt_textrmsb): Likewise.
7724         (iwmmxt_textrmuh): Likewise.
7725         (iwmmxt_textrmsh): Likewise.
7726         (iwmmxt_textrmw): Likewise.
7727         (iwmxxt_wshufh): Likewise.
7728         (eqv8qi3): Likewise.
7729         (eqv4hi3): Likewise.
7730         (eqv2si3): Likewise.
7731         (gtuv8qi3): Likewise.
7732         (gtuv4hi3): Likewise.
7733         (gtuv2si3): Likewise.
7734         (gtv8qi3): Likewise.
7735         (gtv4hi3): Likewise.
7736         (gtv2si3): Likewise.
7737         (smax<mode>3_iwmmxt): Likewise.
7738         (umax<mode>3_iwmmxt): Likewise.
7739         (smin<mode>3_iwmmxt): Likewise.
7740         (umin<mode>3_iwmmxt): Likewise.
7741         (iwmmxt_wpackhss): Likewise.
7742         (iwmmxt_wpackwss): Likewise.
7743         (iwmmxt_wpackdss): Likewise.
7744         (iwmmxt_wpackhus): Likewise.
7745         (iwmmxt_wpackwus): Likewise.
7746         (iwmmxt_wpackdus): Likewise.
7747         (iwmmxt_wunpckihb): Likewise.
7748         (iwmmxt_wunpckihh): Likewise.
7749         (iwmmxt_wunpckihw): Likewise.
7750         (iwmmxt_wunpckilb): Likewise.
7751         (iwmmxt_wunpckilh): Likewise.
7752         (iwmmxt_wunpckilw): Likewise.
7753         (iwmmxt_wunpckehub): Likewise.
7754         (iwmmxt_wunpckehuh): Likewise.
7755         (iwmmxt_wunpckehuw): Likewise.
7756         (iwmmxt_wunpckehsb): Likewise.
7757         (iwmmxt_wunpckehsh): Likewise.
7758         (iwmmxt_wunpckehsw): Likewise.
7759         (iwmmxt_wunpckelub): Likewise.
7760         (iwmmxt_wunpckeluh): Likewise.
7761         (iwmmxt_wunpckeluw): Likewise.
7762         (iwmmxt_wunpckelsb): Likewise.
7763         (iwmmxt_wunpckelsh): Likewise.
7764         (iwmmxt_wunpckelsw): Likewise.
7765         (ror<mode>3): Likewise.
7766         (ashr<mode>3_iwmmxt): Likewise.
7767         (lshr<mode>3_iwmmxt): Likewise.
7768         (ashl<mode>3_iwmmxt): Likewise.
7769         (ror<mode>3_di): Likewise.
7770         (ashr<mode>3_di): Likewise.
7771         (lshr<mode>3_di): Likewise.
7772         (ashl<mode>3_di): Likewise.
7773         (iwmmxt_wmadds): Likewise.
7774         (iwmmxt_wmaddu): Likewise.
7775         (iwmmxt_tmia): Likewise.
7776         (iwmmxt_tmiaph): Likewise.
7777         (iwmmxt_tmiabb): Likewise.
7778         (iwmmxt_tmiatb): Likewise.
7779         (iwmmxt_tmiabt): Likewise.
7780         (iwmmxt_tmiatt): Likewise.
7781         (iwmmxt_tmovmskb): Likewise.
7782         (iwmmxt_tmovmskh): Likewise.
7783         (iwmmxt_tmovmskw): Likewise.
7784         (iwmmxt_waccb): Likewise.
7785         (iwmmxt_wacch): Likewise.
7786         (iwmmxt_waccw): Likewise.
7787         (iwmmxt_waligni): Likewise.
7788         (iwmmxt_walignr): Likewise.
7789         (iwmmxt_walignr0): Likewise.
7790         (iwmmxt_walignr1): Likewise.
7791         (iwmmxt_walignr2): Likewise.
7792         (iwmmxt_walignr3): Likewise.
7793         (iwmmxt_wsadb): Likewise.
7794         (iwmmxt_wsadh): Likewise.
7795         (iwmmxt_wsadbz): Likewise.
7796         (iwmmxt_wsadhz): Likewise.
7797         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
7798         (iwmmxt_wabsdiffb): Likewise.
7799         (iwmmxt_wabsdiffh): Likewise.
7800         (iwmmxt_wabsdiffw): Likewise.
7801         (iwmmxt_waddsubhx): Likewise
7802         (iwmmxt_wsubaddhx): Likewise.
7803         (addc<mode>3): Likewise.
7804         (iwmmxt_avg4): Likewise.
7805         (iwmmxt_avg4r): Likewise.
7806         (iwmmxt_wmaddsx): Likewise.
7807         (iwmmxt_wmaddux): Likewise.
7808         (iwmmxt_wmaddsn): Likewise.
7809         (iwmmxt_wmaddun): Likewise.
7810         (iwmmxt_wmulwsm): Likewise.
7811         (iwmmxt_wmulwum): Likewise.
7812         (iwmmxt_wmulsmr): Likewise.
7813         (iwmmxt_wmulumr): Likewise.
7814         (iwmmxt_wmulwsmr): Likewise.
7815         (iwmmxt_wmulwumr): Likewise.
7816         (iwmmxt_wmulwl): Likewise.
7817         (iwmmxt_wqmulm): Likewise.
7818         (iwmmxt_wqmulwm): Likewise.
7819         (iwmmxt_wqmulmr): Likewise.
7820         (iwmmxt_wqmulwmr): Likewise.
7821         (iwmmxt_waddbhusm): Likewise.
7822         (iwmmxt_waddbhusl): Likewise.
7823         (iwmmxt_wqmiabb): Likewise.
7824         (iwmmxt_wqmiabt): Likewise.
7825         (iwmmxt_wqmiatb): Likewise.
7826         (iwmmxt_wqmiatt): Likewise.
7827         (iwmmxt_wqmiabbn): Likewise.
7828         (iwmmxt_wqmiabtn): Likewise.
7829         (iwmmxt_wqmiatbn): Likewise.
7830         (iwmmxt_wqmiattn): Likewise.
7831         (iwmmxt_wmiabb): Likewise.
7832         (iwmmxt_wmiabt): Likewise.
7833         (iwmmxt_wmiatb): Likewise.
7834         (iwmmxt_wmiatt): Likewise.
7835         (iwmmxt_wmiabbn): Likewise.
7836         (iwmmxt_wmiabtn): Likewise.
7837         (iwmmxt_wmiatbn): Likewise.
7838         (iwmmxt_wmiattn): Likewise.
7839         (iwmmxt_wmiawbb): Likewise.
7840         (iwmmxt_wmiawbt): Likewise.
7841         (iwmmxt_wmiawtb): Likewise.
7842         (iwmmxt_wmiawtt): Likewise.
7843         (iwmmxt_wmiawbbn): Likewise.
7844         (iwmmxt_wmiawbtn): Likewise.
7845         (iwmmxt_wmiawtbn): Likewise.
7846         (iwmmxt_wmiawttn): Likewise.
7847         (iwmmxt_wmerge): Likewise.
7848         (iwmmxt_tandc<mode>3): Likewise.
7849         (iwmmxt_torc<mode>3): Likewise.
7850         (iwmmxt_torvsc<mode>3): Likewise.
7851         (iwmmxt_textrc<mode>3): Likewise.
7852         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
7853         (wmmxt_pack): Likewise.
7854         (wmmxt_mult_c1): Likewise.
7855         (wmmxt_mult_c2): Likewise.
7856         (wmmxt_alu_c1): Likewise.
7857         (wmmxt_alu_c2): Likewise.
7858         (wmmxt_alu_c3): Likewise.
7859         (wmmxt_transfer_c1): Likewise.
7860         (wmmxt_transfer_c2): Likewise.
7861         (wmmxt_transfer_c3): Likewise.
7862         (marvell_f_iwmmxt_wstr): Likewise.
7863         (marvell_f_iwmmxt_wldr): Likewise.
7865 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
7867         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
7869 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7871         Revert:
7872         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7873         * lra-constraints.c (need_for_split_p): Check call used hard regs
7874         living through calls.
7876         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
7877         call used regs for call insn.
7879 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
7881         PR target/57736
7882         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
7883         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
7885 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7887         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
7889 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7891         * lra-constraints.c (need_for_split_p): Check call used hard regs
7892         living through calls.
7894 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
7896         PR target/57744
7897         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
7898         to tie with any other modes.  Eliminate Altivec vector mode tests,
7899         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
7900         code, to return 0 if testing MODE2 for a condition, if we've
7901         already tested MODE1 for the same condition.
7903 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7905         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
7906         layout.
7908 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7910         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7911         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
7913 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7915         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
7916         Define.
7917         (aarch64_symbolic_constant_p): Remove.
7918         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
7919         static.  Fix line length and white space.
7920         (aarch64_symbolic_constant_p): Remove.
7921         * config/aarch64/predicates.md (aarch64_valid_symref):
7922         Use aarch64_classify_symbol_expression.
7924 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7926         * config/arm/constraints.md (Ts): New constraint.
7927         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
7928         16-bit encodings.
7929         (compare_scc): Use "Ts" constraint for operand 0.
7930         (ior_scc_scc): Likewise.
7931         (and_scc_scc): Likewise.
7932         (and_scc_scc_nodom): Likewise.
7933         (ior_scc_scc_cmp): Likewise for operand 7.
7934         (and_scc_scc_cmp): Likewise.
7935         * config/arm/thumb2.md (thumb2_movsi_insn):
7936         Add alternatives for 16-bit encodings.
7937         (thumb2_movhi_insn): Likewise.
7938         (thumb2_movsicc_insn): Likewise.
7939         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
7940         (thumb2_negscc): Use "Ts" constraint.
7941         Move mvn instruction outside cond_exec block.
7942         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
7943         for 16-bit encodings.
7945 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7947         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
7948         encoding.
7949         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
7950         (mulsi3subsi): Likewise.
7951         (mulsidi3adddi): Likewise.
7952         (mulsidi3_v6): Likewise.
7953         (umulsidi3_v6): Likewise.
7954         (umulsidi3adddi_v6): Likewise.
7955         (smulsi3_highpart_v6): Likewise.
7956         (umulsi3_highpart_v6): Likewise.
7957         (mulhisi3tb): Likewise.
7958         (mulhisi3bt): Likewise.
7959         (mulhisi3tt): Likewise.
7960         (maddhisi4): Likewise.
7961         (maddhisi4tb): Likewise.
7962         (maddhisi4tt): Likewise.
7963         (maddhidi4): Likewise.
7964         (maddhidi4tb): Likewise.
7965         (maddhidi4tt): Likewise.
7966         (zeroextractsi_compare0_scratch): Likewise.
7967         (insv_zero): Likewise.
7968         (insv_t2): Likewise.
7969         (anddi_notzesidi_di): Likewise.
7970         (anddi_notsesidi_di): Likewise.
7971         (andsi_notsi_si): Likewise.
7972         (iordi_zesidi_di): Likewise.
7973         (xordi_zesidi_di): Likewise.
7974         (andsi_iorsi3_notsi): Likewise.
7975         (smax_0): Likewise.
7976         (smax_m1): Likewise.
7977         (smin_0): Likewise.
7978         (not_shiftsi): Likewise.
7979         (unaligned_loadsi): Likewise.
7980         (unaligned_loadhis): Likewise.
7981         (unaligned_loadhiu): Likewise.
7982         (unaligned_storesi): Likewise.
7983         (unaligned_storehi): Likewise.
7984         (extv_reg): Likewise.
7985         (extzv_t2): Likewise.
7986         (divsi3): Likewise.
7987         (udivsi3): Likewise.
7988         (arm_zero_extendhisi2addsi): Likewise.
7989         (arm_zero_extendqisi2addsi): Likewise.
7990         (compareqi_eq0): Likewise.
7991         (arm_extendhisi2_v6): Likewise.
7992         (arm_extendqisi2addsi): Likewise.
7993         (arm_movt): Likewise.
7994         (thumb2_ldrd): Likewise.
7995         (thumb2_ldrd_base): Likewise.
7996         (thumb2_ldrd_base_neg): Likewise.
7997         (thumb2_strd): Likewise.
7998         (thumb2_strd_base): Likewise.
7999         (thumb2_strd_base_neg): Likewise.
8000         (arm_negsi2): Add alternative for 16-bit encoding.
8001         (arm_one_cmplsi2): Likewise.
8003 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8005         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8006         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8007         (movdfcc): Likewise.
8008         * config/arm/vfp.md (*thumb2_movsf_vfp):
8009         Disable predication for arm_restrict_it.
8010         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8011         (*thumb2_movdfcc_vfp): Likewise.
8012         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8013         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8014         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8015         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8016         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8017         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8018         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8019         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8020         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8021         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8022         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8023         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8024         Disable predication for arm_restrict_it.
8026 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8028         * config/i386/bmiintrin.h (_bextr_u32): New.
8029         (_bextr_u64): Ditto.
8031 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8033         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8034         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8035         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8036         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8037         * config/mips/n32-elf.h: ...this new file.
8039 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8041         PR target/57224
8042         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8043         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8045 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8047         * config/mips/mips-tables.opt: Regenerate.
8048         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8049         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8050         * doc/invoke.texi: Add -m14kc.
8052 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8054         PR target/57623
8055         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8056         constraints of operand 1 and 2.
8058         PR target/57623
8059         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8060         to match RTL canonicalization.  Swap predicates and
8061         constraints of operand 1 and 2.
8063 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8065         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8066         Process OP_INOUT regs for splitting too.
8068 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8070         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8071         decl before the loop, initialize to NULL.
8072         (vectorizable_load): Initialize ptr_incr to NULL.
8074 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8076         PR lto/57208
8077         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8078         * ipa-ref.c (ipa_maybe_record_reference): New function.
8079         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8080         * ipa-cp.c (create_specialized_node): Record potential references from
8081         aggvals.
8082         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8084 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8086         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8087         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8088         'mode' to force_reg.
8089         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8090         (aarch64_expand_mov_immediate): Likewise.
8092 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8094         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8095         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8097 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8099         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8100         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8101         (struct machine_function): Add tbegin_p.
8102         (s390_canonicalize_comparison): Fold CC mode compares to
8103         conditional jump if possible.
8104         (s390_emit_jump): Return the emitted jump.
8105         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8106         Handle CCRAWmode compares.
8107         (s390_option_override): Default to -mhtm if available.
8108         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8109         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8110         FPRs instead of df_regs_ever_live_p.
8111         (s390_optimize_nonescaping_tx): New function.
8112         (s390_init_frame_layout): Extend clobbered_regs array to cover
8113         FPRs as well.
8114         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8115         (s390_expand_tbegin): New function.
8116         (enum s390_builtin): New enum definition.
8117         (code_for_builtin): New array definition.
8118         (s390_init_builtins): New function.
8119         (s390_expand_builtin): New function.
8120         (TARGET_INIT_BUILTINS): Define.
8121         (TARGET_EXPAND_BUILTIN): Define.
8122         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8123         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8124         (s390_alc_comparison): Likewise.
8125         * config/s390/s390-modes.def: Add CCRAWmode.
8126         * config/s390/s390.h (processor_flags): Add PF_TX.
8127         (TARGET_CPU_HTM): Define macro.
8128         (TARGET_HTM): Define macro.
8129         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8130         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8131         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8132         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8133         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8134         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8135         ("*cc_to_int"): Move up.
8136         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8137         constants other than 0.
8138         ("*ccraw_to_int"): New insn and splitter definition.
8139         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8140         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8141         ("tx_assist"): New expander.
8142         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8143         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8144         * config/s390/s390.opt: Add -mhtm option.
8145         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8146         * config/s390/htmxlintrin.h: New file.
8147         * config/s390/htmintrin.h: New file.
8148         * config/s390/s390intrin.h: New file.
8149         * doc/extend.texi: Document htm builtins.
8150         * config.gcc: Add the new header files to extra_headers.
8152 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8154         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8155         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8157 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8158             Pat Haugen  <pthaugen@us.ibm.com>
8159             Peter Bergner  <bergner@vnet.ibm.com>
8161         * config/rs6000/power8.md: New.
8162         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8163         setting for power8 entry.
8164         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8165         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8166         test for Power4/Power5 only.
8167         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8168         support.
8169         (force_new_group): Adjust comment.
8170         * config/rs6000/rs6000.md: Include power8.md.
8172 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8174         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8175         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8176         declaration.
8177         (tune_params): Update comment.
8178         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8179         (arm_max_conditional_execute): New function.
8180         (thumb2_final_prescan_insn): Use max_insn_skipped and
8181         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8183 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8185         PR tree-optimization/57705
8186         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8187         SSA_NAME step, provided that it is not defined inside the loop.
8188         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8189         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8191 2013-06-25  Martin Jambor  <mjambor@suse.cz>
8193         PR middle-end/57670
8194         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8195         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8196         calls in the dump.
8197         (ipa_note_param_call): Initialize member_ptr flag.
8198         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8199         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8200         (ipa_write_indirect_edge_info): Stream member_ptr flag.
8201         (ipa_read_indirect_edge_info): Likewise.
8203 2013-06-25  Richard Biener  <rguenther@suse.de>
8205         PR middle-end/56977
8206         * passes.c (init_optimization_passes): Move pass_fold_builtins
8207         and pass_dce earlier with -Og.
8209 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
8211         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8212         <BIT_FIELD_REF>: Remove trailing TAB.
8213         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8214         remove blank line.
8216 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8218         PR tree-optimization/57358
8219         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8220         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8221         (ipa_analyze_params_uses): Generate pessimistic info when true.
8223 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8225         PR tree-optimization/57539
8226         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8227         global.inlined_to of the new node to it.  All callers changed.
8228         * ipa-inline-transform.c (clone_inlined_nodes): New variable
8229         inlining_into, pass it to cgraph_clone_node.
8230         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8231         ipa_free_edge_args_substructures.
8232         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8233         rdesc linked list.  Do not assert rdesc edges have inlined caller.
8234         Assert we have found an rdesc in the rdesc list.
8236 2013-06-24  Richard Biener  <rguenther@suse.de>
8238         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8239         (pointer_set_lookup): Declare.
8240         (class pointer_map): New template class implementing a
8241         generic pointer to T map.
8242         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8243         pointer_map<T>::contains, pointer_map<T>::insert,
8244         pointer_map<T>::traverse): New functions.
8245         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8246         (pointer_set_lookup): New function.
8247         (pointer_set_contains): Use pointer_set_lookup.
8248         (pointer_set_insert): Likewise.
8249         (insert_aux): Remove.
8250         (struct pointer_map_t): Embed a pointer_set_t.
8251         (pointer_map_create): Adjust.
8252         (pointer_map_destroy): Likewise.
8253         (pointer_map_contains): Likewise.
8254         (pointer_map_insert): Likewise.
8255         (pointer_map_traverse): Likewise.
8256         * tree-streamer.h (struct streamer_tree_cache_d): Use a
8257         pointer_map<unsigned> instead of a pointer_map_t.
8258         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8259         (streamer_tree_cache_lookup): Likewise.
8260         (streamer_tree_cache_create): Likewise.
8261         (streamer_tree_cache_delete): Likewise.
8262         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8263         pointer_map<unsigned> instead of a pointer_map_t.
8264         (lto_init_tree_ref_encoder): Adjust.
8265         (lto_destroy_tree_ref_encoder): Likewise.
8266         * lto-section-out.c (lto_output_decl_index): Likewise.
8267         (lto_record_function_out_decl_state): Likewise.
8268         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8270 2013-06-24  Richard Biener  <rguenther@suse.de>
8272         PR tree-optimization/57488
8273         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8275 2013-06-24  Alan Modra  <amodra@gmail.com>
8277         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8278         (gen_easy_altivec_constant): Likewise.
8279         * config/rs6000/predicates.md (easy_vector_constant_add_self,
8280         easy_vector_constant_msb): Likewise.
8282 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
8284         PR target/57688
8285         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8286         add missing return true.
8288 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
8290         PR target/52483
8291         * config/sh/predicates.md (general_extend_operand): Invoke
8292         general_movsrc_operand for memory operands.
8293         (general_movsrc_operand): Allow reg+reg addressing, do not use
8294         general_operand for memory operands.
8296 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
8298         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
8299         when current target options does not apply.
8300         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8301         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8302         * config/i386/bmiintrin.h: Pass appropriate target
8303         attributes to header.
8304         * config/i386/mmintrin.h: Ditto.
8305         * config/i386/nmmintrin.h: Ditto.
8306         * config/i386/avx2intrin.h: Ditto.
8307         * config/i386/fxsrintrin.h: Ditto.
8308         * config/i386/tbmintrin.h: Ditto.
8309         * config/i386/xsaveintrin.h: Ditto.
8310         * config/i386/f16cintrin.h: Ditto.
8311         * config/i386/xtestintrin.h: Ditto.
8312         * config/i386/xsaveoptintrin.h: Ditto.
8313         * config/i386/bmi2intrin.h: Ditto.
8314         * config/i386/lzcntintrin.h: Ditto.
8315         * config/i386/smmintrin.h: Ditto.
8316         * config/i386/wmmintrin.h: Ditto.
8317         * config/i386/x86intrin.h: Remove all header include guards.
8318         * config/i386/prfchwintrin.h: Ditto.
8319         * config/i386/pmmintrin.h: Ditto.
8320         * config/i386/tmmintrin.h: Ditto.
8321         * config/i386/xmmintrin.h: Ditto.
8322         * config/i386/popcntintrin.h: Ditto.
8323         * config/i386/rdseedintrin.h: Ditto.
8324         * config/i386/ammintrin.h: Ditto.
8325         * config/i386/emmintrin.h: Ditto.
8326         * config/i386/immintrin.h: Remove all header include guards.
8327         * config/i386/fma4intrin.h: Ditto.
8328         * config/i386/lwpintrin.h: Ditto.
8329         * config/i386/xopintrin.h: Ditto.
8330         * config/i386/ia32intrin.h: Ditto.
8331         * config/i386/avxintrin.h: Ditto.
8332         * config/i386/rtmintrin.h: Ditto.
8333         * config/i386/fmaintrin.h: Ditto.
8334         * config/i386/mm3dnow.h: Ditto.
8336 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
8338         * common/config/i386/i386-common.c: Handle LZCNT.
8340 2013-06-22  Andi Kleen  <ak@linux.intel.com>
8342         * doc/extend.texi: Use __atomic_store_n instead of
8343         __atomic_store in HLE example.
8345 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
8347         * config/sh/sh.c: Remove <cstdlib> workaround.
8349 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8351         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
8353 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8355         * doc/extend.texi: Document that __atomic_clear and
8356           __atomic_test_and_set should only be used with bool.
8358 2013-06-20  Jan Hubicka  <jh@suse.cz>
8360         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8361         types_same_for_odr.
8362         * tree.c (decls_same_for_odr): New function.
8363         (same_for_edr): New function.
8364         (types_same_for_odr): New function.
8365         (get_binfo_at_offset): Use it.
8366         * tree.h (types_same_for_odr): Declare.
8368 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
8369             Jason Merrill  <jason@redhat.com>
8371         * system.h: Include <cstdlib> as well as <stdlib.h>.
8373 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
8375         PR target/57655
8376         * config/i386/i386.c (construct_container): Report error if
8377         long double is used with disabled x87 float returns.
8379 2013-06-20  Jan Hubicka  <jh@suse.cz>
8381         * lto-cgraph.c (input_symtab): Do not set cgraph state.
8383 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
8385         PR rtl-optimization/57425
8386         PR rtl-optimization/57569
8387         * alias.c (write_dependence_p): Remove parameters mem_mode and
8388         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
8389         Changed all callers.
8390         (canon_anti_dependence): Get comments and semantics in sync.
8391         Add parameter mem_canonicalized.  Changed all callers.
8392         * rtl.h (canon_anti_dependence): Update prototype.
8394 2013-06-20  Richard Biener  <rguenther@suse.de>
8396         * data-streamer-in.c (streamer_read_uhwi): Optimize single
8397         byte case, inline streamer_read_uchar and defer section
8398         overrun check.
8400 2013-06-20  Richard Biener  <rguenther@suse.de>
8402         PR tree-optimization/57584
8403         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8404         SSA names into the expanded expression that take part in
8405         abnormal coalescing.
8407 2013-06-19  Sharad Singhai  <singhai@google.com>
8409         * gcov.c (print_usage): Handle new option.
8410         (process_args): Ditto.
8411         (get_gcov_intermediate_filename): New function.
8412         (output_intermediate_file): New function.
8413         (output_gcov_file): New function
8414         (generate_results): Handle new option.
8415         (release_function): Relase demangled name.
8416         (read_graph_file): Handle demangled name.
8417         (output_lines): Ditto.
8418         * doc/gcov.texi: Document gcov intermediate format.
8420 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
8422         PR bootstrap/57604
8423         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8424         (lra_emit_add): Use the functions.  Add comment about Y as an
8425         address segment.
8427 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
8429         PR driver/57652
8430         * collect2.c (collect_atexit): New.
8431         (collect_exit): Delete.
8432         (main): Register collect_atexit with atexit.
8433         (collect_wait): Change collect_exit to exit.
8434         (do_wait): Same.
8435         * collect2.h (collect_exit): Delete.
8436         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8438 2013-06-19  Wei Mi  <wmi@google.com>
8440         PR rtl-optimization/57518
8441         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8442         if regno is used in paradoxical subreg.
8443         (update_equiv_regs): Check pdx_subregs[regno] before
8444         set a reg to be equivalent with a mem.
8446 2013-06-19  Matthias Klose  <doko@ubuntu.com>
8448         PR driver/57651
8449         * file-find.h (find_a_file): Add a mode parameter.
8450         * file-find.c (find_a_file): Likewise.
8451         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8452         with X_OK for the executables.
8453         * collect2.c (main): Call find_a_file with X_OK.
8455 2013-06-19  Steve Ellcey  <sellcey@mips.com>
8457         PR target/56942
8458         * config/mips/mips.md (casesi_internal_mips16_<mode>):
8459         Use NEXT_INSN instead of next_real_insn.
8461 2013-06-19  Jan Hubicka  <jh@suse.cz>
8463         * cgraph.h (const_value_known_p): Replace by ...
8464         (ctor_for_folding): .. this one.
8465         * cgraphunit.c (process_function_and_variable_attributes): Use it.
8466         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8467         * expr.c (expand_expr_real_1): Likewise.
8468         (string_constant): Likewise.
8469         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8470         * ipa.c (process_references): Likewise.
8471         (symtab_remove_unreachable_nodes): Likewise.
8472         * ipa-inline-analysis.c (param_change_prob): Likewise.
8473         * gimple-fold.c (canonicalize_constructor_val): Likewise.
8474         (get_base_constructor): Likwise.
8475         * varpool.c (varpool_remove_node): Likewise.
8476         (varpool_remove_initializer): LIkewise.
8477         (dump_varpool_node): LIkwise.
8478         (const_value_known_p): Rewrite to ...
8479         (ctor_for_folding): ... this one.
8481 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
8483         PR driver/57651
8484         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8485         PERSONALITY in $PATH derived prefixes.
8487 2013-06-19  Jeff Law  <law@redhat.com>
8489         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8490         in comment.
8492         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8493         (simplify_bitwise_binary): Use it to simpify certain binary ops on
8494         booleans.
8496 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
8498         * config/arm/vfp.md: Move VFP instruction classification documentation
8499         to ...
8500         * config/arm/arm.md: ... here.  Update instruction classification
8501         documentation.
8503 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
8505         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8506         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8507         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
8509 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
8511         PR target/57609
8512         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8513         with NEXT_INSN.  Use tablejump_p to check for jump table data
8514         insns.
8516 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
8518         PR c++/56544
8519         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8520         that now in C++ the value is correct per the C++ standards.
8522 2013-06-19  Richard Biener  <rguenther@suse.de>
8524         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8525         for global context.
8527 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8529         Revert:
8530         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8532         PR target/57609
8533         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8534         with next_active_insn.
8536 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
8538         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
8539         functions are inlined during failures to flag an error.
8540         * tree-inline.c (expand_call_inline): Allow the error to be flagged
8541         in early inline pass.
8543 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
8545         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
8546         in comments.
8548 2013-06-18  Julian Brown  <julian@codesourcery.com>
8550         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
8551         Permit virtual register pre-reload if !strict.
8552         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
8553         change.
8554         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
8555         prototype.
8556         * config/arm/neon.md (movmisalign<mode>): Use
8557         neon_perm_struct_or_reg_operand instead of
8558         neon_struct_or_register_operand.
8559         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
8560         neon_permissive_struct_operand instead of neon_struct_operand.
8561         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
8562         neon_vector_mem_operand.
8563         * config/arm/predicates.md (neon_struct_operand): Adjust call to
8564         neon_vector_mem_operand.
8565         (neon_permissive_struct_operand): New.
8566         (neon_struct_or_register_operand): Rename to...
8567         (neon_perm_struct_or_reg_operand): This. Adjust call to
8568         neon_vector_mem_operand.
8570 2013-06-18  Richard Biener  <rguenther@suse.de>
8572         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
8573         * lto-streamer.h: Include pointer-set.h.
8574         (struct lto_decl_slot): Remove.
8575         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
8576         Remove next_index entry.
8577         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8578         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8579         (lto_init_tree_ref_encoder): Adjust.
8580         (lto_destroy_tree_ref_encoder): Likewise.
8581         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8582         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8583         (lto_output_decl_index): Adjust.
8584         (lto_new_out_decl_state): Likewise.
8585         (lto_record_function_out_decl_state): Likewise.
8586         * lto-streamer-out.c (copy_function): Likewise.
8588 2013-06-18  Richard Biener  <rguenther@suse.de>
8590         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
8591         * cgraphunit.c: Include cfgloop.h.
8592         (init_lowered_empty_function): Initialize the loop tree.
8593         (assemble_thunk): Insert new BBs into loops.
8595 2013-06-18  Richard Biener  <rguenther@suse.de>
8597         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
8598         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
8599         the map from cache entry to cache index optional.
8600         (streamer_tree_cache_replace_tree): Adjust accordingly.
8601         (streamer_tree_cache_append): Likewise.
8602         (streamer_tree_cache_delete): Likewise.
8603         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
8604         streamer cache map from cache entry to cache index.
8605         * lto-streamer-out.c (create_output_block): Adjust.
8607 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
8609         * config/arm/arm.md (attribute "insn"): Move multiplication and
8610         division attributes to...
8611         (attribute "type"): ... here.  Remove mult.
8612         (attribute "mul32"): New attribute.
8613         (attribute "mul64"): Add umaal.
8614         (*arm_mulsi3): Update attributes.
8615         (*arm_mulsi3_v6): Likewise.
8616         (*thumb_mulsi3): Likewise.
8617         (*thumb_mulsi3_v6): Likewise.
8618         (*mulsi3_compare0): Likewise.
8619         (*mulsi3_compare0_v6): Likewise.
8620         (*mulsi_compare0_scratch): Likewise.
8621         (*mulsi_compare0_scratch_v6): Likewise.
8622         (*mulsi3addsi): Likewise.
8623         (*mulsi3addsi_v6): Likewise.
8624         (*mulsi3addsi_compare0): Likewise.
8625         (*mulsi3addsi_compare0_v6): Likewise.
8626         (*mulsi3addsi_compare0_scratch): Likewise.
8627         (*mulsi3addsi_compare0_scratch_v6): Likewise.
8628         (*mulsi3subsi): Likewise.
8629         (*mulsidi3adddi): Likewise.
8630         (*mulsi3addsi_v6): Likewise.
8631         (*mulsidi3adddi_v6): Likewise.
8632         (*mulsidi3_nov6): Likewise.
8633         (*mulsidi3_v6): Likewise.
8634         (*umulsidi3_nov6): Likewise.
8635         (*umulsidi3_v6): Likewise.
8636         (*umulsidi3adddi): Likewise.
8637         (*umulsidi3adddi_v6): Likewise.
8638         (*smulsi3_highpart_nov6): Likewise.
8639         (*smulsi3_highpart_v6): Likewise.
8640         (*umulsi3_highpart_nov6): Likewise.
8641         (*umulsi3_highpart_v6): Likewise.
8642         (mulhisi3): Likewise.
8643         (*mulhisi3tb): Likewise.
8644         (*mulhisi3bt): Likewise.
8645         (*mulhisi3tt): Likewise.
8646         (maddhisi4): Likewise.
8647         (*maddhisi4tb): Likewise.
8648         (*maddhisi4tt): Likewise.
8649         (maddhidi4): Likewise.
8650         (*maddhidi4tb): Likewise.
8651         (*maddhidi4tt): Likewise.
8652         (divsi3): Likewise.
8653         (udivsi3): Likewise.
8654         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
8655         (thumb2_mulsi_short_compare0): Likewise.
8656         (thumb2_mulsi_short_compare0_scratch): Likewise.
8657         * config/arm/arm1020e.md (1020mult1): Update attribute change.
8658         (1020mult2): Likewise.
8659         (1020mult3): Likewise.
8660         (1020mult4): Likewise.
8661         (1020mult5): Likewise.
8662         (1020mult6): Likewise.
8663         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
8664         change.
8665         (cortex_a15_mult64): Likewise.
8666         (cortex_a15_sdiv): Likewise.
8667         (cortex_a15_udiv): Likewise.
8668         * config/arm/arm1026ejs.md (mult1): Update attribute change.
8669         (mult2): Likewise.
8670         (mult3): Likewise.
8671         (mult4): Likewise.
8672         (mult5): Likewise.
8673         (mult6): Likewise.
8674         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
8675         (pj4_ir_div): Likewise.
8676         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
8677         (11_mult2): Likewise.
8678         (11_mult3): Likewise.
8679         (11_mult4): Likewise.
8680         (11_mult5): Likewise.
8681         (11_mult6): Likewise.
8682         (11_mult7): Likewise.
8683         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
8684         (cortex_a8_mla): Likewise.
8685         (cortex_a8_mull): Likewise.
8686         (cortex_a8_smulwy): Likewise.
8687         (cortex_a8_smlald): Likewise.
8688         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
8689         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
8690         (cortex_r4_mul_3): Likewise.
8691         (cortex_r4_mla_4): Likewise.
8692         (cortex_r4_mla_3): Likewise.
8693         (cortex_r4_smlald): Likewise.
8694         (cortex_r4_mull): Likewise.
8695         (cortex_r4_sdiv): Likewise.
8696         (cortex_r4_udiv): Likewise.
8697         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
8698         (cortex_a7_idiv): Likewise.
8699         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
8700         (9_mult2): Likewise.
8701         (9_mult3): Likewise.
8702         (9_mult4): Likewise.
8703         (9_mult5): Likewise.
8704         (9_mult6): Likewise.
8705         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
8706         (cortex_a53_sdiv): Likewise.
8707         (cortex_a53_udiv): Likewise.
8708         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
8709         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
8710         (mp626_mult2): Likewise.
8711         (mp626_mult3): Likewise.
8712         (mp626_mult4): Likewise.
8713         * config/arm/fa526.md (526_mult1): Update attribute change.
8714         (526_mult2): Likewise.
8715         * config/arm/arm-generic.md (mult): Update attribute change.
8716         (mult_ldsched_strongarm): Likewise.
8717         (mult_ldsched): Likewise.
8718         (multi_cycle): Likewise.
8719         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
8720         * config/arm/fa606te.md (606te_mult1): Update attribute change.
8721         (606te_mult2): Likewise.
8722         (606te_mult3): Likewise.
8723         (606te_mult4): Likewise.
8724         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
8725         (cortex_a9_mac16): Likewise.
8726         (cortex_a9_multiply): Likewise.
8727         (cortex_a9_mac): Likewise.
8728         (cortex_a9_multiply_long): Likewise.
8729         * config/arm/fa626te.md (626te_mult1): Update attribute change.
8730         (626te_mult2): Likewise.
8731         (626te_mult3): Likewise.
8732         (626te_mult4): Likewise.
8734 2013-06-18  Richard Biener  <rguenther@suse.de>
8736         PR lto/57334
8737         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
8739 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8741         PR target/57609
8742         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8743         with next_active_insn.
8745 2013-06-18  Alan Modra  <amodra@gmail.com>
8747         * config/rs6000/rs6000.h (enum data_align): New.
8748         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
8749         (DATA_ABI_ALIGNMENT): Define.
8750         (CONSTANT_ALIGNMENT): Correct comment.
8751         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
8752         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
8754 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
8756         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
8757         ATTRIBUTE_UNUSED marking.
8759 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
8761         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
8762         alternative and update.
8763         (aarch64_dup_lanedi): Delete.
8764         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
8765         * config/aarch64/aarch64-simd-builtins.def: Update.
8767 2013-06-17  Richard Biener  <rguenther@suse.de>
8769         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
8770         (lto_input_scc): Declare.
8771         (lto_input_tree_1): Likewise.
8772         (struct lto_stats_d): Add num_tree_bodies_output and
8773         num_pickle_refs_output.
8774         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
8775         (lto_read_tree_1): Split out from ...
8776         (lto_read_tree): ... this.
8777         (lto_input_scc): New function.
8778         (lto_input_tree_1): Split out from ...
8779         (lto_input_tree): ... this.  Handle LTO_tree_scc.
8780         (lto_data_in_create): Create the streamer cache without hashes.
8781         * lto-streamer-out.c (create_output_block): Create the streamer
8782         cache with hashes when not doing WPA.
8783         (lto_write_tree_1): Split out from ...
8784         (lto_write_tree): ... this.
8785         (get_symbol_initial_value): New function.
8786         (lto_output_tree_1): Split out from ...
8787         (lto_output_tree): ... this.  Write trees as series of SCCs
8788         using a DFS walk via DFS_write_tree.
8789         (struct sccs, struct scc_entry): New types.
8790         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
8791         (DFS_write_tree_body): New function.
8792         (DFS_write_tree): Likewise.
8793         (hash_tree): Likewise.
8794         (scc_entry_compare): Likewise.
8795         (hash_scc): Likewise.
8796         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
8797         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
8798         TREE_CHAIN as regular reference.
8799         (streamer_read_integer_cst): Remove.
8800         (streamer_get_pickled_tree): Adjust.
8801         * tree-streamer-out.c (streamer_write_chain): Disable streaming
8802         of DECL_EXTERNALs in BLOCK_VARS for now.
8803         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
8804         reference.
8805         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
8806         Add hash value argument and record that if hashes are recorded
8807         in the cache.
8808         (streamer_tree_cache_insert_1): Adjust.
8809         (streamer_tree_cache_insert): Likewise.
8810         (streamer_tree_cache_insert_at): Rename to ...
8811         (streamer_tree_cache_replace_tree): ... this and adjust.
8812         (streamer_tree_cache_append): Adjust.
8813         (record_common_node): Likewise.
8814         (streamer_tree_cache_create): Add argument whether to
8815         record hash values together with trees.
8816         (streamer_tree_cache_delete): Adjust.
8817         * tree-streamer.h (struct streamer_tree_cache_d): Add
8818         vector of hashes.
8819         (streamer_read_integer_cst): Remove.
8820         (streamer_tree_cache_insert): Adjust.
8821         (streamer_tree_cache_append): Likewise.
8822         (streamer_tree_cache_insert_at): Rename to ...
8823         (streamer_tree_cache_replace_tree): ... this and adjust.
8824         (streamer_tree_cache_create): Add argument whether to record hashes.
8825         (streamer_tree_cache_get): Rename to ...
8826         (streamer_tree_cache_get_tree): ... this.
8827         (streamer_tree_cache_get_hash): New function.
8828         * tree.c (cache_integer_cst): New function.
8829         * tree.h (cache_integer_cst): Declare.
8830         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
8831         * lto-symtab.c (lto_varpool_replace_node): Only release
8832         DECL_INITIAL of non-prevailing decls.
8833         * varpool.c (varpool_remove_initializer): Do not release
8834         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
8836 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8838         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
8839         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
8840         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
8841         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
8842         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
8843         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
8844         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
8845         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
8846         instead of TARGET_64BIT.
8847         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
8848         Require ISA_HAS_<D>DIV.
8850 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
8852         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
8853         (mips*-*-linux*): Move default with_llsc setting to where other
8854         defaults are set.
8855         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
8856         with_arch block.
8857         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
8858         Likewise.  Remove default with_tune setting.  Move default float
8859         setting to its own block.  Handle with_llsc in the same block as above.
8861 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
8863         PR rtl-optimization/57425
8864         PR rtl-optimization/57569
8865         * alias.c (write_dependence_p): Add new parameters mem_mode,
8866         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
8867         Changed all callers.
8868         (canon_anti_dependence): New function.
8869         * cse.c (check_dependence): Use canon_anti_dependence.
8870         * cselib.c (cselib_invalidate_mem): Likewise.
8871         * rtl.h (canon_anti_dependence): Declare.
8873 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8875         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
8876         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
8877         ".set mips3" for 64-bit targets.
8879 2013-06-15  Dehao Chen  <dehao@google.com>
8881         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
8882         * gimple-low.c (gimple_check_call_matching_types): Likewise.
8883         (gimple_check_call_args): Likewise.
8884         * value-prof.c (check_ic_target): Likewise.
8885         * ipa-inline.c (early_inliner): Likewise.
8886         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
8887         * cgraph.c (cgraph_create_edge_1): Likewise.
8888         (cgraph_make_edge_direct): Likewise.
8890 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
8892         PR target/57615
8893         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
8894         rs6000_output_move_128bit to handle emitting quad memory
8895         operations.  Set attribute length to 8 bytes.
8897 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
8899         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
8900         New pattern.
8901         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
8902         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
8903         (aarch64_<su>mlsl<mode>): Likewise.
8905 2013-06-14  Mike Stump  <mikestump@comcast.net>
8907         * Makefile.in (TARGET_H): Add insn-codes.h.
8909 2013-06-14  Alan Modra  <amodra@gmail.com>
8911         PR middle-end/57134
8912         PR middle-end/57586
8913         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
8914         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
8915         bitfield expansion when EXPAND_MEMORY.
8916         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
8918 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
8920         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
8921         test for clearing quad memory on 32-bit later.
8923 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8925         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
8926         (fold_negate_expr): Likewise.
8927         (fold_real_zero_addition_p): Handle vectors.
8928         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
8930 2013-06-14  Alan Modra  <amodra@gmail.com>
8932         * varasm.c (force_const_mem): Revert 2013-06-07 change.
8934 2013-06-13  Jan Hubicka  <jh@suse.cz>
8936         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
8937         Local comdats are not externally visible.
8938         * symtab.c (dump_symtab_base): Dump externally visible.
8939         (verify_symtab_base): Verify back links in the symtab hash.
8941 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8943         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
8944         CONVERT_EXPR as equal nodes.
8946 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8948         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
8950 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8952         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
8953         Generalize to complex and vector.
8954         * tree.c (build_all_ones_cst): New function.
8955         * tree.h (build_all_ones_cst): Declare it.
8957 2013-06-13  Alan Modra  <amodra@gmail.com>
8959         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
8960         * config/rs6000/rs6000.md (signbittf2): New insn.
8961         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
8962         (abstf2_internal, cmptf_internal2): Likewise.
8963         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
8965 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
8966             Pat Haugen  <pthaugen@us.ibm.com>
8967             Peter Bergner  <bergner@vnet.ibm.com>
8969         * config/rs6000/rs6000.c (emit_load_locked): Add support for
8970         power8 byte, half-word, and quad-word atomic instructions.
8971         (emit_store_conditional): Likewise.
8972         (rs6000_expand_atomic_compare_and_swap): Likewise.
8973         (rs6000_expand_atomic_op): Likewise.
8975         * config/rs6000/sync.md (larx): Add new modes for power8.
8976         (stcx): Likewise.
8977         (AINT): New mode iterator to include TImode as well as normal
8978         integer modes on power8.
8979         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
8980         that VSX registers are not considered.  Use AINT mode iterator
8981         instead of INT1 to allow inclusion of quad word atomic operations
8982         on power8.
8983         (load_locked<mode>): Likewise.
8984         (store_conditional<mode>): Likewise.
8985         (atomic_compare_and_swap<mode>): Likewise.
8986         (atomic_exchange<mode>): Likewise.
8987         (atomic_nand<mode>): Likewise.
8988         (atomic_fetch_<fetchop_name><mode>): Likewise.
8989         (atomic_nand_fetch<mode>): Likewise.
8990         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
8991         each type.
8992         (ATOMIC): On power8, add QImode, HImode modes.
8993         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
8994         modes that promote to SImode.
8995         (load_lockedti): Convert TImode arguments to PTImode, so that we
8996         get a guaranteed even/odd register pair.
8997         (load_lockedpti): Likewise.
8998         (store_conditionalti): Likewise.
8999         (store_conditionalpti): Likewise.
9001         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9002         atomic load/store instructions.
9003         (HSI): Likewise.
9005 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9007         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9008         loads.
9009         (insn_count): New attribute, with most cases extracted from...
9010         (length): ...here.  Redefine most cases in terms of insn_count.
9011         (single_insn): Delete.
9012         (can_delay): Use insn_count to check for single instructions.
9013         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9014         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9015         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9016         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9017         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9018         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9019         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9020         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9021         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9022         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9023         rather than "length".
9024         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9025         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9026         Use "insn_count" rather than "length".
9027         * config/mips/mips-dsp.md
9028         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9029         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9030         length attributes.
9032 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9034         PR tree-optimization/57361
9035         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9037 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9039         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9040         to split.
9041         (aarch64_simd_combine<mode>): New instruction expansion.
9042         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9043         function prototype.
9044         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9045         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9047 2013-06-12  Jan Hubicka  <jh@suse.cz>
9049         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9050         decl has when in streaming stage.
9051         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9052         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9054 2013-06-12  Roland Stigge  <stigge@antcom.de>
9056         PR target/57578
9057         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9059 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9061         PR tree-optimization/57537
9062         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9063         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9065 2013-06-12  Richard Biener  <rguenther@suse.de>
9067         * data-streamer.h (streamer_write_char_stream): CSE
9068         obs->current_pointer.
9069         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9070         streamer_write_char_stream manually and optimize the resulting loop.
9071         (streamer_write_hwi_stream): Likewise.
9073 2013-06-12  Jan Hubicka  <jh@suse.cz>
9075         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9076         * cgraph.h (varpool_create_empty_node): Declare.
9077         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9078         duplicated nodes.
9079         * symtab.c (symtab_unregister_node): Be lax about missin entries
9080         in node hash.
9081         (symtab_get_node): Update comment.
9082         * varpool.c (varpool_create_empty_node): Break out from ...
9083         (varpool_node_for_decl): ... here.
9084         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9086 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9088         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9089         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9090         part.  Use straight-line flow at the end.
9091         <COMPONENT_REF>: Remove superfluous else.
9092         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9094 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9096         PR target/56564
9097         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9098         target hook even for !TREE_PUBLIC decls.  If no resolution info
9099         is available, return false for common and external decls.
9101 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9103         * config/rl78/constraints.md (U): New constraint.
9104         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9105         valloc attribute.
9107 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9109         PR target/57589
9110         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9111         to allow returning address to AT_PLATFORM name.
9113 2013-06-11  Jan Hubicka  <jh@suse.cz>
9115         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9116         * cgraph.h (symtab_node_base): Add weakref flag.
9117         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9118         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9119         (output_weakrefs): Use weakref flag.
9120         * fold-const.c (simple_operand_p): Handle WEAK.
9121         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9122         * ipa.c (varpool_externally_visible_p): Drop weakref.
9123         (function_and_variable_visibility): Update comment; fix weakref
9124         sanity checks; do not clear DECL_WEAK on them.
9125         * lto-cgraph.c (lto_output_node): update.
9126         (lto_output_varpool_node): Update.
9127         (input_overwrite_node): Update.
9128         (input_node): Update.
9129         (input_varpool_node): Update.
9130         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9131         (lto_symtab_merge_symbols): Add sanity check.
9132         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9133         * passes.c (rest_of_decl_compilation): Set static flag, too.
9134         * symtab.c (dump_symtab_base): Dump weakref.
9135         (verify_symtab_base): Sanity check weakrefs.
9136         (symtab_make_decl_local): Remove duplicated code.
9137         (symtab_alias_ultimate_target): Simplify.
9138         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9140 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9142         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9143         sequence_vect == NULL.
9145 2013-06-11  DJ Delorie  <dj@redhat.com>
9147         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9148         (rl78_unwind_word_mode): New.
9150 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9152         * final.c (debug_prefix_maps): Make static.
9154 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9156         * function.c (initial_trampoline): Remove stray copy.
9158 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9160         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9162 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9164         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9165         within bounds at the beginning of the function.
9167 2013-06-11  Alan Modra  <amodra@gmail.com>
9169         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9170         reporting.
9171         (get_named_section): Don't NULL !DECL_P decl.
9173 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9175         * doc/invoke.texi (core-avx2): Document.
9176         (slm): Likewise.
9177         (atom): Updated with MOVBE.
9179 2013-06-11  Richard Biener  <rguenther@suse.de>
9181         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9183 2013-06-11  Anton Blanchard  <anton@samba.org>
9185         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9186         correct shift value in little-endian mode.
9188 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9190         PR target/56564
9191         * varasm.c (get_variable_align): Move #endif to the right place.
9193 2013-06-10  Cary Coutant  <ccoutant@google.com>
9195         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9196         for hash so that hash table traversal order is deterministic.
9198 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
9199             Pat Haugen  <pthaugen@us.ibm.com>
9200             Peter Bergner  <bergner@vnet.ibm.com>
9202         * config/rs6000/vector.md (GPR move splitter): Do not split moves
9203         of vectors in GPRS if they are direct moves or quad word load or
9204         store moves.
9206         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9207         declaration.
9208         (direct_move_p): Likewise.
9209         (quad_load_store_p): Likewise.
9211         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9212         classes into bins based on the physical register type.
9213         (reg_class_to_reg_type): Likewise.
9214         (IS_STD_REG_TYPE): Likewise.
9215         (IS_FP_VECT_REG_TYPE): Likewise.
9216         (reload_fpr_gpr): Arrays to determine what insn to use if we can
9217         use direct move instructions.
9218         (reload_gpr_vsx): Likewise.
9219         (reload_vsx_gpr): Likewise.
9220         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9221         information that is a simplification of register classes.  Also
9222         precalculate direct move reload helpers.
9223         (direct_move_p): New function to return true if the operation can
9224         be done as a direct move instruciton.
9225         (quad_load_store_p): New function to return true if the operation
9226         is a quad memory operation.
9227         (rs6000_legitimize_address): If quad memory, only allow register
9228         indirect for TImode addresses.
9229         (rs6000_legitimate_address_p): Likewise.
9230         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9231         (rs6000_reload_register_type): Likewise.
9232         (register_to_reg_type): Return register type.
9233         (rs6000_secondary_reload_simple_move): New helper function for
9234         secondary reload and secondary memory needed to identify anything
9235         that is a simple move, and does not need reloading.
9236         (rs6000_secondary_reload_direct_move): New helper function for
9237         secondary reload to identify cases that can be done with several
9238         instructions via the direct move instructions.
9239         (rs6000_secondary_reload_move): New helper function for secondary
9240         reload to identify moves between register types that can be done.
9241         (rs6000_secondary_reload): Add support for quad memory operations
9242         and for direct move.
9243         (rs6000_secondary_memory_needed): Likewise.
9244         (rs6000_debug_secondary_memory_needed): Change argument names.
9245         (rs6000_output_move_128bit): New function to return the move to
9246         use for 128-bit moves, including knowing about the various
9247         limitations of quad memory operations.
9249         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9250         memory operations.  call rs6000_output_move_128bit for the actual
9251         instruciton(s) to generate.
9252         (vsx_movti_64bit): Likewise.
9254         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9255         (UNSPEC_P8V_MTVSRWZ): Likewise.
9256         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9257         (UNSPEC_P8V_MTVSRD): Likewise.
9258         (UNSPEC_P8V_XXPERMDI): Likewise.
9259         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9260         (UNSPEC_FUSION_GPR): Likewise.
9261         (FMOVE128_GPR): New iterator for direct move.
9262         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9263         (f32_sv): Likewise.
9264         (f32_dm): Likewise.
9265         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9266         loads and direct move instructions.
9267         (zero_extendsidi2_lfiwzx): Likewise.
9268         (extendsidi2_lfiwax): Likewise.
9269         (extendsidi2_nocell): Likewise.
9270         (floatsi<mode>2_lfiwax): Likewise.
9271         (lfiwax): Likewise.
9272         (floatunssi<mode>2_lfiwzx): Likewise.
9273         (lfiwzx): Likewise.
9274         (fix_trunc<mode>_stfiwx): Likewise.
9275         (fixuns_trunc<mode>_stfiwx): Likewise.
9276         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9277         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9278         (parity<mode>2_cmpb): Set length/type attr.
9279         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9280         for 'mr.' to fast_compare.
9281         (bpermd_<mode>): Change type attr to popcnt.
9282         (p8_fmrgow_<mode>): New insns for power8 direct move support.
9283         (p8_mtvsrwz_1): Likewise.
9284         (p8_mtvsrwz_2): Likewise.
9285         (reload_fpr_from_gpr<mode>): Likewise.
9286         (p8_mtvsrd_1): Likewise.
9287         (p8_mtvsrd_2): Likewise.
9288         (p8_xxpermdi_<mode>): Likewise.
9289         (reload_vsx_from_gpr<mode>): Likewise.
9290         (reload_vsx_from_gprsf): Likewise.
9291         (p8_mfvsrd_3_<mode>): LIkewise.
9292         (reload_gpr_from_vsx<mode>): Likewise.
9293         (reload_gpr_from_vsxsf): Likewise.
9294         (p8_mfvsrd_4_disf): Likewise.
9295         (multi-word GPR splits): Do not split direct moves or quad memory
9296         operations.
9298 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
9300         * tree-into-ssa.c (interesting_blocks): Make static.
9302 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
9304         PR target/56564
9305         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9306         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9307         Use DATA_ABI_ALIGNMENT for that case instead if defined.
9308         (get_variable_align): New function.
9309         (get_variable_section, emit_bss, emit_common,
9310         assemble_variable_contents, place_block_symbol): Use
9311         get_variable_align instead of DECL_ALIGN.
9312         (assemble_noswitch_variable): Add align argument, use it
9313         instead of DECL_ALIGN.
9314         (assemble_variable): Adjust caller.  Use get_variable_align
9315         instead of DECL_ALIGN.
9316         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9317         caller.
9318         (DATA_ABI_ALIGNMENT): Define.
9319         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9320         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
9321         opt is false, only return the psABI mandated alignment increase.
9322         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9323         (DATA_ABI_ALIGNMENT): ... this.
9324         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9325         (DATA_ABI_ALIGNMENT): ... this.
9326         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9327         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9328         (DATA_ABI_ALIGNMENT): ... this.
9329         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9330         * doc/tm.texi: Regenerated.
9332 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
9334         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9335         cmp_code to construct REG_EQUAL note.
9337 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
9339         PR target/57568
9340         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9341         that operands[2] doesn't overlap with operands[0].
9343 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
9344             Jan Hubicka  <jh@suse.cz>
9346         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9347         hack to mark symbols as used.
9349 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
9351         PR rtl-optimization/57559
9352         * lra-constraints.c (process_alt_operands): Don't discourage
9353         memory with known offset for offsetable memory constraint.
9354         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9356 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
9358         * varasm.c (struct oc_local_state): Reorder fields.
9359         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9360         and adjust accordingly.
9361         (output_constructor): Reorder initialization code and adjust call to
9362         output_constructor_bitfield.
9364 2013-06-07  Jan Hubicka  <jh@suse.cz>
9366         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9368 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
9370         * tree-object-size.c (unknown): Make const.
9372 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9374         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9375         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9376         for last alternative in the cpu_facility attribute.
9378 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9380         PR target/56315
9381         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9382         (xordi3): Change operand 2 constraint to arm_xordi_operand.
9383         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9384         * config/arm/constraints.md (Dg): New constraint.
9385         * config/arm/neon.md (xordi3_neon): Remove.
9386         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9387         * config/arm/predicates.md (arm_xordi_operand): New predicate.
9389 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9391         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9392         Clean up alternatives.
9394 2013-06-07  Alan Modra  <amodra@gmail.com>
9396         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9397         va_list_gpr_size.
9399 2013-06-07  Alan Modra  <amodra@gmail.com>
9401         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9403 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9405         * config/arm/constraints.md (Df): New constraint.
9406         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9407         Correct length attribute for last two alternatives.
9409 2013-06-07  Alan Modra  <amodra@gmail.com>
9411         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9412         override user -mfp-in-toc.
9413         (offsettable_ok_by_alignment): Consider just the current access
9414         rather than the whole object, unless BLKmode.  Handle
9415         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9416         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9417         for -mcmodel=medium.
9418         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9419         override user -mfp-in-toc or -msum-in-toc.  Default to
9420         -mno-fp-in-toc for -mcmodel=medium.
9422 2013-06-06  DJ Delorie  <dj@redhat.com>
9424         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9425         TARGET_VALID_POINTER_MODE.
9427 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
9428             Pat Haugen  <pthaugen@us.ibm.com>
9429             Peter Bergner  <bergner@vnet.ibm.com>
9431         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9432         Document new power8 builtins.
9434         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9435         condition code register, to allow 128-bit logical operations to be
9436         done in the VSX or GPR registers.
9437         (nor<mode>3): Use the canonical form for nor.
9438         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9439         vclz*, and vpopcnt* vector instructions.
9440         (nand<mode>3): Likewise.
9441         (orc<mode>3): Likewise.
9442         (clz<mode>2): LIkewise.
9443         (popcount<mode>2): Likewise.
9445         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9446         that only the GPRs are recognized.
9448         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9449         support for new power8 builtins.
9451         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9452         builtin functions.
9453         (xscvdpspn): Likewise.
9454         (vclz): Likewise.
9455         (vclzb): Likewise.
9456         (vclzh): Likewise.
9457         (vclzw): Likewise.
9458         (vclzd): Likewise.
9459         (vpopcnt): Likewise.
9460         (vpopcntb): Likewise.
9461         (vpopcnth): Likewise.
9462         (vpopcntw): Likewise.
9463         (vpopcntd): Likewise.
9464         (vgbbd): Likewise.
9465         (vmrgew): Likewise.
9466         (vmrgow): Likewise.
9467         (eqv): Likewise.
9468         (eqv_v16qi3): Likewise.
9469         (eqv_v8hi3): Likewise.
9470         (eqv_v4si3): Likewise.
9471         (eqv_v2di3): Likewise.
9472         (eqv_v4sf3): Likewise.
9473         (eqv_v2df3): Likewise.
9474         (nand): Likewise.
9475         (nand_v16qi3): Likewise.
9476         (nand_v8hi3): Likewise.
9477         (nand_v4si3): Likewise.
9478         (nand_v2di3): Likewise.
9479         (nand_v4sf3): Likewise.
9480         (nand_v2df3): Likewise.
9481         (orc): Likewise.
9482         (orc_v16qi3): Likewise.
9483         (orc_v8hi3): Likewise.
9484         (orc_v4si3): Likewise.
9485         (orc_v2di3): Likewise.
9486         (orc_v4sf3): Likewise.
9487         (orc_v2df3): Likewise.
9489         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9490         allow power8 quad mode in 64-bit.
9491         (rs6000_builtin_vectorized_function): Add support to vectorize
9492         ISA 2.07 count leading zeros, population count builtins.
9493         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9494         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9495         (builtin_function_type): Add vgbbd builtin function which takes an
9496         unsigned argument.
9497         (altivec_expand_vec_perm_const): Add support for new power8 merge
9498         instructions.
9500         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9501         that does not include TImdoe for use with 32-bit.
9502         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9503         instructions.
9504         (UNSPEC_VSX_CVDPSPN): Likewise.
9505         (vsx_xscvdpspn): Likewise.
9506         (vsx_xscvspdpn): Likewise.
9507         (vsx_xscvdpspn_scalar): Likewise.
9508         (vsx_xscvspdpn_directmove): Likewise.
9509         (vsx_and<mode>3): Split logical operations into 32-bit and
9510         64-bit. Add support to do logical operations on TImode as well as
9511         VSX vector types.  Allow logical operations to be done in either
9512         VSX registers or in general purpose registers in 64-bit mode.  Add
9513         splitters if GPRs were used. For AND, add clobber of CCmode to
9514         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
9515         encoding.
9516         (vsx_and<mode>3_32bit): Likewise.
9517         (vsx_and<mode>3_64bit): Likewise.
9518         (vsx_ior<mode>3): Likewise.
9519         (vsx_ior<mode>3_32bit): Likewise.
9520         (vsx_ior<mode>3_64bit): Likewise.
9521         (vsx_xor<mode>3): Likewise.
9522         (vsx_xor<mode>3_32bit): Likewise.
9523         (vsx_xor<mode>3_64bit): Likewise.
9524         (vsx_one_cmpl<mode>2): Likewise.
9525         (vsx_one_cmpl<mode>2_32bit): Likewise.
9526         (vsx_one_cmpl<mode>2_64bit): Likewise.
9527         (vsx_nor<mode>3): Likewise.
9528         (vsx_nor<mode>3_32bit): Likewise.
9529         (vsx_nor<mode>3_64bit): Likewise.
9530         (vsx_andc<mode>3): Likewise.
9531         (vsx_andc<mode>3_32bit): Likewise.
9532         (vsx_andc<mode>3_64bit): Likewise.
9533         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
9534         and xxlorc instructions.
9535         (vsx_eqv<mode>3_64bit): Likewise.
9536         (vsx_nand<mode>3_32bit): Likewise.
9537         (vsx_nand<mode>3_64bit): Likewise.
9538         (vsx_orc<mode>3_32bit): Likewise.
9539         (vsx_orc<mode>3_64bit): Likewise.
9541         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
9543         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
9544         instruction.
9545         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
9546         (p8_vmrgow): Likewise.
9547         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
9548         GPRs to be split under VSX.
9549         (p8v_clz<mode>2): Add power8 count leading zero support.
9550         (p8v_popcount<mode>2): Add power8 population count support.
9551         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
9552         support.
9554         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
9555         instruction.
9557         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
9558         builtin functions.
9559         (vec_nand): Likewise.
9560         (vec_vclz): Likewise.
9561         (vec_vclzb): Likewise.
9562         (vec_vclzd): Likewise.
9563         (vec_vclzh): Likewise.
9564         (vec_vclzw): Likewise.
9565         (vec_vgbbd): Likewise.
9566         (vec_vmrgew): Likewise.
9567         (vec_vmrgow): Likewise.
9568         (vec_vpopcnt): Likewise.
9569         (vec_vpopcntb): Likewise.
9570         (vec_vpopcntd): Likewise.
9571         (vec_vpopcnth): Likewise.
9572         (vec_vpopcntw): Likewise.
9574 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9576         PR rtl-optimization/57468
9577         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
9578         spilled pseudos.
9580 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9582         PR rtl-optimization/57459
9583         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
9584         type when setting live regs.
9586 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9588         * config/s390/s390.opt (mlra): New option.
9589         * config/s390/s390.c (s390_decompose_address): Check displacement
9590         for all registers for LRA.
9591         (s390_secondary_reload): Don't used secondary reloads for LRA.
9592         (s390_lra_p): New function.
9593         (TARGET_LRA_P): Define.
9594         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
9595         of attribute cpu_facility to zarch for the last alternative.
9596         (*cmpmem_short): Ditto.
9598 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
9600         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
9601         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
9602         (arm_expand_prologue): Likewise.
9604 2013-06-06  Teresa Johnson  <tejohnson@google.com>
9606         PR c++/53743
9607         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
9608         as this is now done by redirect_edge_and_branch_force.
9609         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
9610         barriers, and fix interaction with splitting.
9611         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
9612         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
9613         reflect changes made in the routine.
9614         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
9615         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
9616         since this is called in cfglayout mode, and replace partition fixup
9617         with assert as that is now done by force_nonfallthru_and_redirect.
9618         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
9619         already be marked with region crossing note.
9620         (insert_section_boundary_note): Make non-static, gate on flag
9621         has_bb_partition, rewrite to also check for multiple partitions.
9622         (rest_of_handle_reorder_blocks): Remove call to
9623         insert_section_boundary_note, now done later during free_cfg.
9624         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
9625         * bb-reorder.h (insert_section_boundary_note): Declare.
9626         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
9627         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
9628         invoke insert_section_boundary_note.
9629         (try_redirect_by_replacing_jump): Remove unnecessary
9630         check for region crossing note.
9631         (fixup_partition_crossing): New function.
9632         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
9633         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
9634         in non-cfglayout mode.
9635         (force_nonfallthru_and_redirect): Fixup partition boundaries,
9636         remove old code that tried to do this. Emit barrier correctly
9637         when we are in cfglayout mode.
9638         (last_bb_in_partition): New function.
9639         (rtl_split_edge): Correctly fixup partition boundaries.
9640         (commit_one_edge_insertion): Remove old code that tried to
9641         fixup region crossing edge since this is now handled in
9642         split_block, and set up insertion point correctly since
9643         block may now end in a jump.
9644         (verify_hot_cold_block_grouping): Guard against checking when not in
9645         linearized RTL mode.
9646         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
9647         notes.
9648         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
9649         rtl_verify_flow_info, so not called in cfglayout mode.
9650         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
9651         (fixup_reorder_chain): Remove old code that attempted to fixup region
9652         crossing note as this is now handled in force_nonfallthru_and_redirect.
9653         (duplicate_insn_chain): Don't duplicate switch section notes.
9654         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
9655         note.
9656         * basic-block.h (emit_barrier_after_bb): Declare.
9658 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9660         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
9661         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
9662         arm_usatsihi): Adjust alternatives for arm_restrict_it.
9664 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9666         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
9667         where appropriate.
9668         * config/arm/ldmstm.md: Regenerate.
9670 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9672         * config/arm/sync.md (atomic_loaddi_1):
9673         Disable predication for arm_restrict_it.
9674         (arm_load_exclusive<mode>): Likewise.
9675         (arm_load_exclusivesi): Likewise.
9676         (arm_load_exclusivedi): Likewise.
9677         (arm_load_acquire_exclusive<mode>): Likewise.
9678         (arm_load_acquire_exclusivesi): Likewise.
9679         (arm_load_acquire_exclusivedi): Likewise.
9680         (arm_store_exclusive<mode>): Likewise.
9681         (arm_store_exclusive<mode>): Likewise.
9682         (arm_store_release_exclusivedi): Likewise.
9683         (arm_store_release_exclusive<mode>): Likewise.
9685 2013-06-06  Richard Biener  <rguenther@suse.de>
9687         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
9688         after LTO_null.
9689         (lto_tag_is_tree_code_p): Adjust.
9690         (lto_tag_is_gimple_code_p): Likewise.
9691         (lto_gimple_code_to_tag): Likewise.
9692         (lto_tag_to_gimple_code): Likewise.
9693         (lto_tree_code_to_tag): Likewise.
9694         (lto_tag_to_tree_code): Likewise.
9695         * data-streamer.h (streamer_write_hwi_in_range): Use
9696         uhwi streaming to stream the normalized range.
9697         (streamer_read_hwi_in_range): Likewise.
9699 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9701         * config/arm/arm.md (enabled_for_depr_it): New attribute.
9702         (predicable_short_it): Likewise.
9703         (predicated): Likewise.
9704         (enabled): Handle above.
9705         (define_cond_exec): Set predicated attribute to yes.
9707 2013-06-05  Mike Stump  <mikestump@comcast.net>
9709         * gdbinit.in (__FUNCTION__): Add.
9711 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
9713         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
9714         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
9716 2013-06-05  Jan Hubicka  <jh@suse.cz>
9718         * varasm.c (mark_decl_referenced): Revert the removal until targets
9719         are fixed.
9721 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
9723         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
9724         instead of mark_decl_referenced.
9726 2013-06-05  Jan Hubicka  <jh@suse.cz>
9728         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
9729         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
9730         and symtab_used_from_object_file_p.
9731         (cgraph_make_node_local_1): Clear forced_by_abi.
9732         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
9733         * cgraph.h (symtab_node_base): Add forced_by_abi.
9734         (decide_is_variable_needed): Remove.
9735         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
9736         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
9737         (decide_is_symbol_needed): ... this one; handle symbols in general;
9738         always analyze virtuals; honnor forced_by_abi.
9739         (cgraph_finalize_function): Update.
9740         (varpool_finalize_decl): Update.
9741         (symbol_defined_and_needed): Remove.
9742         (analyze_functions): Update.
9743         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9744         output_refs, input_overwrite_node): Handle forced_by_abi.
9745         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
9746         (address_taken_from_non_vtable_p): ... this one.
9747         (comdat_can_be_unshared_p_1): New function.
9748         (cgraph_comdat_can_be_unshared_p): Rename to ...
9749         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
9750         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
9751         (function_and_variable_visibility): Clear forced_by_abi as needed.
9752         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
9753         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
9754         * symtab.c (dump_symtab_base): Dump forced_by_abi.
9755         * varpool.c (decide_is_variable_needed): Remove.
9757 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9759         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
9760         (arm_option_override): Override arm_restrict_it where appropriate.
9761         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
9762         * config/arm/arm.opt (mrestrict-it): New command-line option.
9763         * doc/invoke.texi: Document -mrestrict-it.
9765 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9767         * tsan.c (tsan_atomic_table): Make const.
9769 2013-06-05  Richard Biener  <rguenther@suse.de>
9771         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
9772         index associated with the tree we are supposed to replace.
9773         * tree-streamer-out.c (pack_ts_base_value_fields): Output
9774         TREE_ASM_WRITTEN as zero for everything but SSA names.
9776 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9778         * tree-ssa-structalias.c (call_stmt_vars): Make static.
9780 2013-06-04  Jan Hubicka  <jh@suse.cz>
9782         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
9783         (input_node, input_varpool_node): Handle correctly external same
9784         body aliases.
9785         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
9786         nodes at ltrans stage.
9788 2013-06-04  Jan Hubicka  <jh@suse.cz>
9790         * ipa-inline.c (update_caller_keys): Fix availability test.
9791         (update_callee_keys): Likewise.
9792         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
9793         to follow ELF standard.
9795 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
9797         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
9798         (mips64r5900el-*-elf*): New configurations.
9799         * config/mips/mips-cpus.def (r5900): New processor.
9800         * config/mips/mips-tables.opt: Regenerate.
9801         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
9802         (mips_issue_rate): Handle PROCESSOR_R5900.
9803         (mips_reorg_process_insns): Force reorder mode for the R5900.
9804         * config/mips/mips.h (TARGET_MIPS5900): Define.
9805         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
9806         TARGET_MIPS5900.
9807         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
9808         TARGET_MIPS5900.
9809         * config/mips/mips.md (processor): Add r5900.
9810         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
9812 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9814         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
9815         into function to generate MOVI instruction.
9816         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
9817         (aarch64_preferred_simd_mode): Turn into wrapper.
9818         (aarch64_output_scalar_simd_mov_immediate): New function.
9819         * config/aarch64/aarch64-protos.h: Add prototype for above.
9821 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9823         * config/aarch64/aarch64.c (simd_immediate_info): Remove
9824         element_char member.
9825         (sizetochar): Return signed char.
9826         (aarch64_simd_valid_immediate): Remove elchar and other
9827         unnecessary variables.
9828         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
9829         Calculate element_char as required.
9830         * config/aarch64/aarch64-protos.h: Update and move prototype
9831         for aarch64_output_simd_mov_immediate.
9832         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9833         Update arguments.
9835 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9837         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
9838         information completed by aarch64_simd_valid_immediate.
9839         (aarch64_legitimate_constant_p): Update arguments.
9840         (aarch64_simd_valid_immediate): Work with struct rather than many
9841         pointers.
9842         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
9843         (aarch64_simd_make_constant): Update arguments.
9844         (aarch64_output_simd_mov_immediate): Work with struct rather than
9845         many pointers.  Output immediate directly rather than as operand.
9846         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9847         Update prototype.
9848         * config/aarch64/constraints.md (Dn): Update arguments.
9850 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9852         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
9853         longer static.
9854         (aarch64_simd_immediate_valid_for_move): Remove.
9855         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
9856         (aarch64_simd_make_constant): Update call.
9857         (aarch64_output_simd_mov_immediate): Update call.
9858         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9859         Add prototype.
9860         * config/aarch64/constraints.md (Dn): Update call.
9862 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9864         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
9865         return type to bool for prototype.
9866         (aarch64_legitimate_constant_p): Check for true instead of not -1.
9867         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
9868         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
9870 2013-06-04  Catherine Moore  <clm@codesourcery.com>
9872         * config/mips/mips.opt (meva): New.
9873         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
9874         (ASM_SPEC): Handle -meva.
9875         * doc/invoke.texi (meva):  Document.
9877 2013-06-04  Alan Modra  <amodra@gmail.com>
9879         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
9880         constant output.
9882 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9884         * rtl.def: Add extra fourth optional field to define_cond_exec.
9885         * gensupport.c (process_one_cond_exec): Process attributes from
9886         define_cond_exec.
9887         * doc/md.texi: Document fourth field in define_cond_exec.
9889 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
9891         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
9892         out the processing order as in store_bit_field_1.
9894 2013-06-04  Jan Hubicka  <jh@suse.cz>
9896         PR middle-end/57500
9897         * cgraphunit.c (cgraph_process_same_body_aliases): Create
9898         non-VAR_DECL node if it does not exist yet.
9900 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
9902         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
9903         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
9904         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
9905         target_cpu_default setting.
9907 2013-06-03  Teresa Johnson  <tejohnson@google.com>
9909         * dumpfile.c (opt_info_switch_p): Change -fopt-info
9910         default to -fopt-info=optimized instead of all.
9911         * doc/invoke.texi: Ditto.
9912         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
9913         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
9914         (execute_vect_slp): Emit BB vectorization success under
9915         MSG_OPTIMIZED_LOCATIONS.
9916         * tree-vect-slp.c (vect_slp_transform_bb): Change
9917         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
9918         * tree-vect-loop.c (vect_transform_loop): Ditto.
9920 2013-06-03  Jason Merrill  <jason@redhat.com>
9922         PR c++/57415
9923         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9924         Use TARGET_EXPR for C++.
9926 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
9928         PR rtl-optimization/57268
9929         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
9930         if DEBUG_INSN_P (insn).
9932         Reapply
9933         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
9935         PR rtl-optimization/57268
9936         * sched-deps.c (sched_analyze_2): Flush dependence lists if
9937         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
9939 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
9941         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
9942         (ix86_avoid_lea_for_addr): Likewise.
9943         (exact_dependency_1): Likewise.
9944         (ix86_adjust_cost): Likewise.
9945         (swap_top_of_ready_list): Fix formatting and !reload_completed check
9946         removed.
9947         (do_reorder_for_imul): Fix typo, formatting and
9948         !reload_completed check removed.
9949         (ix86_sched_reorder): Fix typo and formatting.
9950         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
9951         list.
9953 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
9955         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
9957 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
9959         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
9960         <STRING_CST>: Likewise.
9961         <VECTOR_CST>: Likewise.
9963 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
9964             Mikael Morin  <mikael@gcc.gnu.org>
9966         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
9967         * config.in: Regenerated.
9968         * configure: Regenerated.
9970 2013-06-01  Jan Hubicka  <jh@suse.cz>
9972         PR middle-end/57366
9973         * cgraphunit.c (compile): When weakref is not supported,
9974         set up transparent aliases before final output pass.
9975         * varasm.c (assemble_alias): Do not try to do it here.
9977 2013-06-01  Jan Hubicka  <jh@suse.cz>
9979         PR middle-end/57467
9980         * passes.c (for_per_function): Skip unanalyzed functions.
9982 2013-06-01  Jan Hubicka  <jh@suse.cz>
9984         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
9985         (lto_symtab_merge_symbols_1): ... this one.
9986         (lto_symtab_merge_cgraph_nodes): Rename to ...
9987         (lto_symtab_merge_symbols): ... this one; simplify.
9988         * cgraph.c (same_body_aliases_done): Rename to ...
9989         (cpp_implicit_aliases_done): ... this one.
9990         (cgraph_create_function_alias): Update.
9991         (cgraph_same_body_alias): Update.
9992         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
9993         (verify_edge_corresponds_to_fndecl): Simplify.
9994         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
9995         (cgraph_node): Remove same_body_alias.
9996         (varpool_node): Remove alias_of and extra_name_alias.
9997         (same_body_aliases_done): Rename to ..
9998         (cpp_implicit_aliases_done): ... this one.
9999         (symtab_alias_ultimate_target): Add default parameter.
10000         (symtab_resolve_alias): New function.
10001         (fixup_same_cpp_alias_visibility): Declare.
10002         (cgraph_function_node): Add default parameter.
10003         (cgraph_node_asm_name): Likewise.
10004         (cgraph_function_or_thunk_node): Add default parameter; do
10005         not ICE when it is NULL.
10006         (varpool_variable_node): Likewise.
10007         * tree-emutls.c (create_emultls_var): Update.
10008         (ipa_lower_emutls): Update.
10009         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10010         (cgraph_reset_node): Reset alias info.
10011         (cgraph_finalize_function): Update.
10012         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10013         (analyze_function): Simplify.
10014         (cgraph_process_same_body_aliases): Simplify.
10015         (analyze_functions): Fixup same body aliases.
10016         (handle_alias_pairs): Simplify.
10017         (assemble_thunk): Update.
10018         (assemble_thunks_and_aliases): Update.
10019         (output_weakrefs): Rewrite.
10020         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10021         (lto_output_varpool_node): Likewise.
10022         (compute_ltrans_boundary): Remve assert.
10023         (get_alias_symbol): New functoin.
10024         (input_node): Rewrite alias handling.
10025         (input_varpool_node): Likewise.
10026         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10027         * ipa.c (process_references): Handle weakrefs correctly.
10028         (symtab_remove_unreachable_nodes): Likewise.
10029         * trans-mem.c (get_cg_data): Update.
10030         (ipa_tm_create_version_alias): Update.
10031         (ipa_tm_execute): Update.
10032         * symtab.c (dump_symtab_base): Dump aliases.
10033         (verify_symtab_base): Verify aliases.
10034         (symtab_node_availability): New function.
10035         (symtab_alias_ultimate_target): Simplify.
10036         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10037         handle all the fixup cases.
10038         (symtab_resolve_alias): New function.
10039         * passes.c (ipa_write_summaries): Handle weakrefs.
10040         * varpool.c (varpool_analyze_node): Simplify.
10041         (assemble_aliases): Update.
10042         (varpool_create_variable_alias): Simplify.
10043         (varpool_extra_name_alias): Simplify.
10044         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10045         (lto_symtab_merge_symbols): ... this one.
10047 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10049         Revert
10050         PR rtl-optimization/57268
10051         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10052         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10054 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10056         Partially reverted:
10057         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10059         PR middle-end/57073
10060         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10061         further up.
10063 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10065         PR rtl-optimization/57268
10066         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10067         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10069 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10071         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10072         unordered comparison operators when -fno-trapping-math is in effect
10073         on the e500.
10074         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10075         and implement unordered comparison operators properly on the e500.
10077 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10079         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10080         for constant scalar integers.
10081         (simplify_relational_operation_1): Likewise.
10083 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10085         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10086         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10087         Fix comment.
10089 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10090             Igor Zamyatin  <igor.zamyatin@intel.com>
10092         Silvermont (SLM) architecture performance tuning.
10093         * config/i386/i386.h (enum ix86_tune_indices): Add
10094         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10095         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10097         * config/i386/i386.c (initial_ix86_tune_features)
10098         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10099         (ix86_lea_outperforms): Handle Silvermont tuning.
10100         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10101         call.
10102         (ix86_use_lea_for_mov): Likewise.
10103         (ix86_avoid_lea_for_addr): Likewise.
10104         (ix86_lea_for_add_ok): Likewise.
10105         (exact_dependency_1): New function.
10106         (exact_store_load_dependency): Likewise.
10107         (ix86_adjust_cost): Handle Silvermont tuning.
10108         (do_reoder_for_imul): Likewise.
10109         (swap_top_of_ready_list): New function.
10110         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10112         * config/i386/i386.md (peepholes that split memory operand in fp
10113         converts): New.
10115 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10117         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10118         Remove un-necessary braces.
10120 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10122         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10123         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10125 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10127         PR middle-end/57073
10128         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10130 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10132         PR target/56315
10133         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10134         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10135         * config/arm/neon.md (iordi3_neon): Remove.
10136         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10137         * config/arm/predicates.md (imm_for_neon_logic_operand):
10138         Move to earlier in the file.
10139         (neon_logic_op2): Likewise.
10140         (arm_iordi_operand_neon): New predicate.
10142 2013-05-31  Richard Biener  <rguenther@suse.de>
10144         PR tree-optimization/57478
10145         PR tree-optimization/57453
10146         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10147         are life as well.
10149 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10151         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10152         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10154 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10155             Thomas Koenig  <tkoenig@gcc.gnu.org>
10157         PR middle-end/57073
10158         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10159         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10161 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10163         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10165 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10167         * target.def (register_usage_leveling_p): New hook.
10168         * targhooks.c (default_register_usage_leveling_p): New.
10169         * targhooks.h (default_register_usage_leveling_p): New prototype.
10170         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10171         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10172         * doc/tm.texi: Update.
10173         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10175 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10177         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10178         (*insv_reg<mode>): New define_insn.
10180 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10182         PR rtl-optimization/57439
10183         * postreload.c (move2add_valid_value_p): Check that we have
10184         a zero subreg_regno_offset when accessing the register in
10185         the requested mode.
10187 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10188             Igor Zamyatin  <igor.zamyatin@intel.com>
10190         Silvermont (SLM) architecture pipeline model, tuning and
10191         insn selection.
10192         * config.gcc: Add slm config options and target.
10194         * config/i386/slm.md: New.
10196         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10198         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
10199         PROCESSOR_SLM.
10200         (ix86_target_macros_internal): Likewise.
10202         * config/i386/i386.c (slm_cost): New cost.
10203         (m_SLM): New macro flag.
10204         (initial_ix86_tune_features): Set m_SLM.
10205         (x86_accumulate_outgoing_args): Likewise.
10206         (x86_arch_always_fancy_math_387): Likewise.
10207         (processor_target_table): Add slm cost.
10208         (cpu_names): Add slm cpu name.
10209         (x86_option_override_internal): Set SLM ISA.
10210         (ix86_issue_rate): New case PROCESSOR_SLM.
10211         (ia32_multipass_dfa_lookahead): Likewise.
10212         (fold_builtin_cpu): Add slm.
10214         * config/i386/i386.h (TARGET_SLM): New target macro.
10215         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10216         (processor_type): Add PROCESSOR_SLM.
10218         * config/i386/i386.md (cpu): Add new value "slm".
10219         (slm.md): Include slm.md.
10221 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
10222             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10224         * config/arm/arm-protos.h: Add and update function protos.
10225         * config/arm/arm.c (use_simple_return_p): New added.
10226         (thumb2_expand_return): Check simple_return flag.
10227         * config/arm/arm.md: Add simple_return and conditional simple_return.
10228         * config/arm/iterators.md: Add iterator for return and simple_return.
10230 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10232         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10233         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10234         (arm_emit_vfp_multi_reg_pop): Likewise.
10235         (thumb2_emit_ldrd_pop): Likewise.
10236         (arm_expand_epilogue): Add misc REG_CFA notes.
10237         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10239 2013-05-29  Lawrence Crowl  <crowl@google.com>
10241         * config/arm/t-arm: Update for below.
10243         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10244         Change type to hash_table.  Update dependent calls and types.
10246         * config/i386/t-cygming: Update for below.
10248         * config/i386/t-interix: Update for below.
10250         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10251         Change type to hash_table.  Update dependent calls and types.
10252         (i386_find_on_wrapper_list::wrappers): Likewise.
10254         * config/ia64/t-ia64: Update for below.
10256         * config/ia64/ia64.c (bundle_state_table):
10257         Change type to hash_table.  Update dependent calls and types.
10259         * config/mips/mips.c (mips_reorg_process_insns::htab):
10260         Change type to hash_table.  Update dependent calls and types.
10262         * config/sol2.c (solaris_comdat_htab):
10263         Change type to hash_table.  Update dependent calls and types.
10265         * config/t-sol2: Update for above.
10267 2013-05-29  Teresa Johnson  <tejohnson@google.com>
10269         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10270         functions are not yet marked as defined.
10272 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
10273             Pat Haugen  <pthaugen@us.ibm.com>
10274             Peter Bergner  <bergner@vnet.ibm.com>
10276         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10277         instructions.
10278         (VEC_A): Likewise.
10279         (VEC_C): Likewise.
10280         (vrotl<mode>3): Likewise.
10281         (vashl<mode>3): Likewise.
10282         (vlshr<mode>3): Likewise.
10283         (vashr<mode>3): Likewise.
10285         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10286         support for power8 V2DI builtins.
10288         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10289         power8 V2DI builtins.
10290         (vupkhsw): Likewise.
10291         (vupklsw): Likewise.
10292         (vaddudm): Likewise.
10293         (vminsd): Likewise.
10294         (vmaxsd): Likewise.
10295         (vminud): Likewise.
10296         (vmaxud): Likewise.
10297         (vpkudum): Likewise.
10298         (vpksdss): Likewise.
10299         (vpkudus): Likewise.
10300         (vpksdus): Likewise.
10301         (vrld): Likewise.
10302         (vsld): Likewise.
10303         (vsrd): Likewise.
10304         (vsrad): Likewise.
10305         (vsubudm): Likewise.
10306         (vcmpequd): Likewise.
10307         (vcmpgtsd): Likewise.
10308         (vcmpgtud): Likewise.
10309         (vcmpequd_p): Likewise.
10310         (vcmpgtsd_p): Likewise.
10311         (vcmpgtud_p): Likewise.
10312         (vupkhsw): Likewise.
10313         (vupklsw): Likewise.
10314         (vaddudm): Likewise.
10315         (vmaxsd): Likewise.
10316         (vmaxud): Likewise.
10317         (vminsd): Likewise.
10318         (vminud): Likewise.
10319         (vpksdss): Likewise.
10320         (vpksdus): Likewise.
10321         (vpkudum): Likewise.
10322         (vpkudus): Likewise.
10323         (vrld): Likewise.
10324         (vsld): Likewise.
10325         (vsrad): Likewise.
10326         (vsrd): Likewise.
10327         (vsubudm): Likewise.
10329         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10330         support for power8 V2DI instructions.
10332         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10333         power8 V2DI instructions.  Combine pack and unpack insns to use an
10334         iterator for each mode.  Check whether a particular mode supports
10335         Altivec instructions instead of just checking TARGET_ALTIVEC.
10336         (UNSPEC_VPKUWUM): Likewise.
10337         (UNSPEC_VPKSHSS): Likewise.
10338         (UNSPEC_VPKSWSS): Likewise.
10339         (UNSPEC_VPKUHUS): Likewise.
10340         (UNSPEC_VPKSHUS): Likewise.
10341         (UNSPEC_VPKUWUS): Likewise.
10342         (UNSPEC_VPKSWUS): Likewise.
10343         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10344         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10345         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10346         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10347         (UNSPEC_VUPKHSB): Likewise.
10348         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10349         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10350         (UNSPEC_VUPKHSH): Likewise.
10351         (UNSPEC_VUPKLSB): Likewise.
10352         (UNSPEC_VUPKLSH): Likewise.
10353         (VI2): Likewise.
10354         (VI_char): Likewise.
10355         (VI_scalar): Likewise.
10356         (VI_unit): Likewise.
10357         (VP): Likewise.
10358         (VP_small): Likewise.
10359         (VP_small_lc): Likewise.
10360         (VU_char): Likewise.
10361         (add<mode>3): Likewise.
10362         (altivec_vaddcuw): Likewise.
10363         (altivec_vaddu<VI_char>s): Likewise.
10364         (altivec_vadds<VI_char>s): Likewise.
10365         (sub<mode>3): Likewise.
10366         (altivec_vsubcuw): Likewise.
10367         (altivec_vsubu<VI_char>s): Likewise.
10368         (altivec_vsubs<VI_char>s): Likewise.
10369         (altivec_vavgs<VI_char>): Likewise.
10370         (altivec_vcmpbfp): Likewise.
10371         (altivec_eq<mode>): Likewise.
10372         (altivec_gt<mode>): Likewise.
10373         (altivec_gtu<mode>): Likewise.
10374         (umax<mode>3): Likewise.
10375         (smax<mode>3): Likewise.
10376         (umin<mode>3): Likewise.
10377         (smin<mode>3): Likewise.
10378         (altivec_vpkuhum): Likewise.
10379         (altivec_vpkuwum): Likewise.
10380         (altivec_vpkshss): Likewise.
10381         (altivec_vpkswss): Likewise.
10382         (altivec_vpkuhus): Likewise.
10383         (altivec_vpkshus): Likewise.
10384         (altivec_vpkuwus): Likewise.
10385         (altivec_vpkswus): Likewise.
10386         (altivec_vpks<VI_char>ss): Likewise.
10387         (altivec_vpks<VI_char>us): Likewise.
10388         (altivec_vpku<VI_char>us): Likewise.
10389         (altivec_vpku<VI_char>um): Likewise.
10390         (altivec_vrl<VI_char>): Likewise.
10391         (altivec_vsl<VI_char>): Likewise.
10392         (altivec_vsr<VI_char>): Likewise.
10393         (altivec_vsra<VI_char>): Likewise.
10394         (altivec_vsldoi_<mode>): Likewise.
10395         (altivec_vupkhsb): Likewise.
10396         (altivec_vupkhs<VU_char>): Likewise.
10397         (altivec_vupkls<VU_char>): Likewise.
10398         (altivec_vupkhsh): Likewise.
10399         (altivec_vupklsb): Likewise.
10400         (altivec_vupklsh): Likewise.
10401         (altivec_vcmpequ<VI_char>_p): Likewise.
10402         (altivec_vcmpgts<VI_char>_p): Likewise.
10403         (altivec_vcmpgtu<VI_char>_p): Likewise.
10404         (abs<mode>2): Likewise.
10405         (vec_unpacks_hi_v16qi): Likewise.
10406         (vec_unpacks_hi_v8hi): Likewise.
10407         (vec_unpacks_lo_v16qi): Likewise.
10408         (vec_unpacks_hi_<VP_small_lc>): Likewise.
10409         (vec_unpacks_lo_v8hi): Likewise.
10410         (vec_unpacks_lo_<VP_small_lc>): Likewise.
10411         (vec_pack_trunc_v8h): Likewise.
10412         (vec_pack_trunc_v4si): Likewise.
10413         (vec_pack_trunc_<mode>): Likewise.
10415         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10416         V2DI builtins.
10417         (vec_vmaxsd): Likewise.
10418         (vec_vmaxud): Likewise.
10419         (vec_vminsd): Likewise.
10420         (vec_vminud): Likewise.
10421         (vec_vpksdss): Likewise.
10422         (vec_vpksdus): Likewise.
10423         (vec_vpkudum): Likewise.
10424         (vec_vpkudus): Likewise.
10425         (vec_vrld): Likewise.
10426         (vec_vsld): Likewise.
10427         (vec_vsrad): Likewise.
10428         (vec_vsrd): Likewise.
10429         (vec_vsubudm): Likewise.
10430         (vec_vupkhsw): Likewise.
10431         (vec_vupklsw): Likewise.
10433 2013-05-29  Jan Hubicka  <jh@suse.cz>
10435         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10436         flags; reorder rest of fields in more consistent way.
10437         (varpool_node): Remove analyzed, finalized and alias.
10438         (cgraph_ndoe): Likewise.
10439         (symtab_alias_ultimate_target): New function.
10440         (cgraph_function_node): Move offline.
10441         (cgraph_reset_node): Declare.
10442         (cgraph_comdat_can_be_unshared_p): Remove.
10443         (varpool_remove_initializer): Declare.
10444         (varpool_first_defined_variable, varpool_next_defined_variable
10445         cgraph_first_defined_function, cgraph_next_defined_function): Update.
10446         (cgraph_function_with_gimple_body_p): Update.
10447         (varpool_all_refs_explicit_p): Update.
10448         (symtab_alias_target): New function.
10449         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10450         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10451         (cgraph_function_or_thunk_node): Simplify using
10452         symtab_alias_ultimate_target.
10453         (varpool_variable_node): Likewise.
10454         * cgraph.c (cgraph_create_function_alias): Update.
10455         (cgraph_add_thunk): Update.
10456         (cgraph_remove_node): Update.
10457         (dump_cgraph_node): Do not dump removed flags.
10458         (cgraph_function_body_availability): Update.
10459         (cgraph_propagate_frequency): Update.
10460         (verify_cgraph_node): Check sanity of local flag.
10461         (cgraph_function_node): Move here from cgraph.h; revamp for
10462         cgraph_function_or_thunk_node.
10463         * lto-symtab.c (lto_varpool_replace_node): Update.
10464         (lto_symtab_resolve_can_prevail_p): Update.
10465         (lto_symtab_merge_cgraph_nodes): Update.
10466         * ipa-cp.c (determine_versionability, initialize_node_lattices,
10467         propagate_constants_accross_call, devirtualization_time_bonus,
10468         ipcp_propagate_stage): Update.
10469         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10470         * ipa-inline-transform.c (clone_inlined_nodes,
10471         preserve_function_body_p): Update.
10472         * ipa-reference.c (propagate): Update.
10473         (write_node_summary_p): Update.
10474         * toplev.c (wrapup_global_declaration_2): Update.
10475         * cgraphunit.c (cgraph_analyze_function): Rename to ...
10476         (analyze_function) ... this one.
10477         (cgraph_process_new_functions): Update.
10478         (cgraph_reset_node): Export.
10479         (cgraph_finalize_function): Update.
10480         (cgraph_add_new_function): Update.
10481         (process_function_and_variable_attributes): Update.
10482         (varpool_finalize_decl): Update.
10483         (symbol_finalized): Remove.
10484         (symbol_finalized_and_needed): Rename to ...
10485         (symbol_defined_and_needed): ... update.
10486         (cgraph_analyze_functions): Update.
10487         (handle_alias_pairs): Update.
10488         (mark_functions_to_output): Update.
10489         (assemble_thunk): Update.
10490         (output_in_order): Update.
10491         (output_weakrefs): Update.
10492         (finalize_compilation_unit): Update.
10493         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10494         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10495         input_node, input_varpool_node): Update.
10496         * dbxout.c (dbxout_expand_expr): Update.
10497         * cgraphclones.c (cgraph_clone_node): Update.
10498         (cgraph_copy_node_for_versioning): Update.
10499         (cgraph_materialize_clone): Update.
10500         (cgraph_materialize_all_clones): Update.
10501         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10502         propagate_pure_const, propagate_nothrow): Update.
10503         * lto-streamer-out.c (lto_output, write_symbol): Update.
10504         * ipa-utils.c (ipa_reverse_postorder): Update.
10505         * ipa-inline.c (can_inline_edge_p): Update.
10506         (update_caller_keys, ipa_inline): Update.
10507         * dwarf2out.c (reference_to_unused,
10508         premark_types_used_by_global_vars_helper): Update.
10509         * tree-eh.c (tree_could_trap_p): Update.
10510         * ipa-split.c (consider_split, execute_split_functions): Update.
10511         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10512         has_addr_references_p): Update; move ahead in file for better
10513         readability.
10514         (process_references): Simplify.
10515         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10516         bodies are removed.
10517         (cgraph_comdat_can_be_unshared_p): Make static.
10518         (cgraph_externally_visible_p): Update.
10519         (varpool_externally_visible_p): Update.
10520         (function_and_variable_visibility): Update.
10521         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10522         ipa_tm_mark_force_output_node): Update.
10523         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10524         estimate_edge_devirt_benefit, inline_generate_summary,
10525         inline_write_summary): Update.
10526         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10527         * ipa-prop.c (ipa_compute_jump_functions): Update.
10528         (ipa_print_node_params, ipa_prop_read_section,
10529         ipa_update_after_lto_read, read_replacements_section): Update.
10530         * varasm.c (mark_decl_referenced): Update.
10531         (assemble_alias, dump_tm_clone_pairs): Update.
10532         * tree-inline.c (copy_bb): Update.
10533         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
10534         Update.
10535         * symtab.c (dump_symtab_base): Print new flags.
10536         (verify_symtab_base): Verify new flags.
10537         (symtab_alias_ultimate_target): New function.
10538         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
10539         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
10540         Update.
10541         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
10542         Update.
10543         * i386.c (ix86_get_function_versions_dispatcher,
10544         ix86_generate_version_dispatcher_body): Update.
10545         (fold_builtin_cpu): Use varpool_add_new_variable.
10546         * varpool.c (varpool_remove_initializer): Break out from ...
10547         (varpool_remove_node): ... this one.
10548         (dump_varpool_node, varpool_node_for_asm,
10549         cgraph_variable_initializer_availability, varpool_analyze_node,
10550         varpool_assemble_decl, varpool_remove_unreferenced_decls,
10551         varpool_finalize_named_section_flags, varpool_create_variable_alias):
10552         Update.
10554 2013-05-29  Jan Hubicka  <jh@suse.cz>
10556         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
10558 2013-05-29  Easwaran Raman  <eraman@google.com>
10560         PR tree-optimization/57442
10561         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
10562         when control exits the main loop.
10564 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
10566         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
10567         and RX600.
10568         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
10569         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
10570         * rx/t-rx: Add rx100 under multi library matches option for nofpu
10571         option.
10573 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10575         PR tree-optimization/57441
10576         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
10577         Don't limit size of incr_vec to number of candidates.
10579 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
10581         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
10582         and mips16 directories.
10583         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
10584         (MULTILIB_DIRNAMES): Ditto.
10585         (MULTILIB_EXCEPTIONS): Add new exceptions.
10586         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
10587         (MULTILIB_DIRNAMES): Ditto.
10588         (MULTILIB_EXCEPTIONS): Add new exceptions.
10590 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10591             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10593         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
10594         SYMBOL_TINY_ABSOLUTE.
10595         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
10596         SYMBOL_TINY_ABSOLUTE.
10597         (aarch64_expand_mov_immediate): Likewise.
10598         (aarch64_classify_symbol): Likewise.
10599         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
10600         Permit SYMBOL_TINY_ABSOLUTE.
10601         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
10603 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10604             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10606         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
10607         Refactor if/switch.  Replace gcc_assert with if.
10609 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10611         * config/i386/i386.c (initial_ix86_tune_features): Enable
10612         FP Reassociation for AMD bdver1 and bdver2.
10614 2013-05-29  Martin Jambor  <mjambor@suse.cz>
10616         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
10617         and IMAGPART_EXPR do not occur within other handled_components.
10619 2013-05-29  Richard Biener  <rguenther@suse.de>
10621         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
10622         access on whether the use is in the BB we currently try to
10623         vectorize.
10624         (vect_bb_vectorization_profitable_p): Pass the BB we currently
10625         vectorize to vect_bb_slp_scalar_cost.
10627 2013-05-29  Richard Biener  <rguenther@suse.de>
10629         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
10630         computing scalar cost offsetted by stmts that are kept live
10631         by scalar uses.
10632         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
10633         for computation of scalar cost.
10635 2013-05-28  Steve Ellcey  <sellcey@mips.com>
10637         * config/mips/mips-cpus.def (mips32r2): Change processor type.
10639 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
10641         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
10642         array notation built-in reduction functions.
10643         * doc/passes.texi (Passes): Added documentation about changes done
10644         for Cilk Plus.
10645         * doc/invoke.texi (C Dialect Options): Added documentation about
10646         the -fcilkplus flag.
10647         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
10648         (BUILTINS_DEF): Depend on cilkplus.def.
10649         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
10650         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
10651         * cilkplus.def: New file.
10653 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
10655         PR rtl-optimization/57439
10656         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
10658 2013-05-28  Easwaran Raman  <eraman@google.com>
10660         PR tree-optimization/57337
10661         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
10662         (find_insert_point): Correctly identify the insertion point
10663         when two statements with the same UID is compared.
10665 2013-05-28  Richard Biener  <rguenther@suse.de>
10667         PR tree-optimization/56787
10668         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
10669         from the list of data references.
10670         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
10671         clobbers.
10672         (vect_analyze_loop_operations): Likewise.
10673         (vect_transform_loop): Remove clobbers.
10675 2013-05-28  Martin Jambor  <mjambor@suse.cz>
10677         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
10678         and REALPART_EXPRs have scalar type.
10680 2013-05-28  Richard Biener  <rguenther@suse.de>
10682         PR tree-optimization/57411
10683         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
10684         virtual operands.
10685         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
10686         virtual operand propagation.
10688 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10690         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
10691         destination register for bmasksi_vis.
10692         (vector_init_bshuffle): Likewise.
10693         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
10695 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10697         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
10698         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
10699         mode if the instruction isn't available in the original mode.
10700         * config/sparc/sparc.opt (mfix-ut699): New option.
10701         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
10702         (divdf3): Turn into expander.
10703         (divdf3_nofix): New insn.
10704         (divdf3_fix): Likewise.
10705         (divsf3): Disable if -mfix-ut699.
10706         (sqrtdf2): Turn into expander.
10707         (sqrtdf2_nofix): New insn.
10708         (sqrtdf2_fix): Likewise.
10709         (sqrtsf2): Disable if -mfix-ut699.
10711 2013-05-27  Richard Biener  <rguenther@suse.de>
10713         PR middle-end/57412
10714         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
10715         block for the new loop.
10717 2013-05-27  Richard Biener  <rguenther@suse.de>
10719         PR tree-optimization/57343
10720         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
10721         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
10722         (number_of_iterations_cond): Do not build the folded tree.
10724 2013-05-27  Richard Biener  <rguenther@suse.de>
10726         Revert
10727         PR middle-end/57381
10728         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10729         OEP_CONSTANT_ADDRESS_OF retained.
10731         PR tree-optimization/57417
10732         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
10733         for unchanged base.
10734         (set_ssa_val_to): Compare addresses using
10735         get_addr_base_and_unit_offset.
10737 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
10739         PR rtl-optimization/56833
10740         * postreload.c (move2add_record_mode): New function.
10741         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
10742         (move2add_use_add2_insn): Use move2add_record_sym_value.
10743         (move2add_use_add3_insn): Likewise.
10744         (reload_cse_move2add): Use move2add_valid_value_p and
10745         move2add_record_mode.  Invalidate call-clobbered and REG_INC
10746         affected regs by setting reg_mode to VOIDmode.
10747         (move2add_note_store): Don't pretend the inside of a SUBREG is
10748         the actual destination.  Invalidate single/leading registers by
10749         setting reg_mode to VOIDmode.
10750         Use move2add_record_sym_value, move2add_valid_value_p and
10751         move2add_record_mode.
10753 2013-05-27  Richard Biener  <rguenther@suse.de>
10755         PR tree-optimization/57396
10756         * tree-affine.c (double_int_constant_multiple_p): Properly
10757         return false for val == 0 and div != 0.
10759 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10761         * config/mips/mips.h: Use #elif in preprocessor conditions.
10763 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10765         PR target/53916
10766         * config/mips/constraints.md (kl): New constraint.
10767         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
10768         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
10769         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
10770         from LO for MIPS16.
10771         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
10773 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10775         PR target/55777
10776         * config/mips/mips.c (mips_can_inline_p): New function.
10777         (TARGET_CAN_INLINE_P): Define.
10779 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
10781         * sched-int.h (ds_t, dw_t): Make unsigned int.
10782         Fix documentation that describes how all the ds_t bits are used.
10783         Reserve the last bit for delayed-branch scheduling.
10784         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
10785         (BITS_PER_DEP_WEAK): Fix definition and documentation.
10786         (gen_dep_weak_1): Remove prototype.
10787         * sched-deps.c (get_dep_weak_1): Make static.
10788         * target.def (speculate_insn, needs_block_p, gen_spec_check,
10789         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
10790         * doc/tm.texi: Regenerate.
10791         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
10793 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
10795         PR debug/56950
10796         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
10798 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
10799             Sandra Loosemore  <sandra@codesourcery.com>
10801         * config.gcc (powerpc-*): Allow native for with-cpu.
10803 2013-05-24  Jeff Law  <law@redhat.com>
10805         PR tree-optimization/57124
10806         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
10807         conversion feeding a condition if the range has an overflow
10808         if -fstrict-overflow.  Add warnings for when we do make the
10809         transformation.
10811 2013-05-24  Dehao Chen  <dehao@google.com>
10813         * tree-cfg.c (locus_discrim_map): Fix the typo.
10814         (locus_discrim_hasher): Likewise.
10815         (locus_discrim_hasher::hash): Likewise.
10816         (locus_discrim_hasher::equal): Likewise.
10818 2013-05-24  Martin Jambor  <mjambor@suse.cz>
10820         PR tree-optimization/57294
10821         * cgraph.h (ipa_record_stmt_references): Declare.
10822         * cgraphbuild.c (ipa_record_stmt_references): New function.
10823         (build_cgraph_edges): Use ipa_record_stmt_references.
10824         (rebuild_cgraph_edges): Likewise.
10825         (cgraph_rebuild_references): Likewise.
10826         * ipa-prop.c (ipa_modify_call_arguments): Discard references
10827         associated with the old statement and build references from the
10828         newly built statements.
10829         * ipa-ref.c (ipa_remove_stmt_references): New function.
10830         * ipa-ref.h (ipa_remove_stmt_references): Declare.
10832 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
10834         * lra-constraints.c (emit_spill_move): Use smaller mode for
10835         mem-mem moves.
10836         (check_and_process_move): Consider mem-reg moves for secondary
10837         too.
10838         (curr_insn_transform): Don't lose insns emitted before for
10839         secondary memory moves.
10840         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
10841         reg set up in the current insn.
10843 2013-05-24  Dehao Chen  <dehao@google.com>
10845         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
10846         hash function.
10847         (locus_descrim_hasher::equal): Likewise.
10848         (build_gimple_cfg): New discriminator assignment algorithm.
10849         (make_edges): Likewise.
10850         (next_discriminator_for_locus): Likewise.
10851         (same_line_p): Likewise.
10852         (assign_discriminators): Likewise.
10853         (make_cond_expr_edges): Likewise.
10854         (make_gimple_switch_edges): Likewise.
10855         (make_goto_expr_edges): Likewise.
10856         (make_gimple_asm_edges): Likewise.
10858 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
10860         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
10861         X format specifier to only display bottom 16 bits.
10862         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
10863         immediate to match for operand 2, since it will be masked.
10865 2013-05-24  Richard Biener  <rguenther@suse.de>
10867         PR tree-optimization/57287
10868         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
10869         all SSA names that occur in abnormal PHIs.
10871 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10873         PR tree-ssa/57385
10874         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
10875         that index is not negative.
10877 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
10879         PR rtl-optimization/55177
10880         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
10881         (simplify_byte_swapping_operation): New.
10882         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
10883         (simplify_relational_operation_1): Deal with BSWAP.
10885 2013-05-23  Richard Henderson  <rth@redhat.com>
10887         PR target/56742
10888         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
10889         (ix86_reorg): Call it.
10891 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
10893         PR target/57379
10894         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
10895         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
10896         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
10898 2013-05-23  Christian Bruel  <christian.bruel@st.com>
10900         PR debug/57351
10901         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
10903 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10904             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10906         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
10907         * config/aarch64/constraints.md (Usa): Remove.
10908         * doc/md.texi (AArch64 Usa): Remove.
10910 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10911             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10913         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
10914         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
10915         * config/aarch64/predicates.md (aarch64_const_address): Remove.
10916         (aarch64_mov_operand): Use aarch64_mov_operand_p.
10918 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
10920         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
10921         instruction (AdvSIMD).
10922         * config/aarch64/aarch64-builtins.c
10923         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
10924         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
10926 2013-05-23  Martin Jambor  <mjambor@suse.cz>
10928         PR middle-end/57347
10929         * tree.h (contains_bitfld_component_ref_p): Declare.
10930         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
10931         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
10932         caller.
10933         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
10934         not access a bit-field.  Assert all final offsets are byte-aligned.
10936 2013-05-23  Richard Biener  <rguenther@suse.de>
10938         PR tree-optimization/57380
10939         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
10940         least one invariant or re-used load.
10941         * passes.c (init_optimization_passes): Move pass_phiprop before
10942         pass_forwprop.
10944 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
10946         * config/aarch64/aarch64-simd.md
10947         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
10949 2013-05-23  Richard Biener  <rguenther@suse.de>
10951         PR middle-end/57381
10952         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10953         OEP_CONSTANT_ADDRESS_OF retained.
10955 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
10957         PR middle-end/57344
10958         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
10959         don't lower unit.  Handle unit not being always BITS_PER_WORD.
10961 2013-05-23  Richard Biener  <rguenther@suse.de>
10963         PR rtl-optimization/57341
10964         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
10965         instead of true_dependence.
10967 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
10969         * bb-reorder.c (branch_threshold): Make const.
10970         (exec_threshold): Ditto.
10972 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
10973             Pat Haugen  <pthaugen@us.ibm.com>
10974             Peter Bergner  <bergner@vnet.ibm.com>
10976         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
10977         documentation for the power8 crypto builtins.
10979         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
10981         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
10982         macros for defining power8 builtin functions.
10983         (BU_P8V_AV_2): Likewise.
10984         (BU_P8V_AV_P): Likewise.
10985         (BU_P8V_VSX_1): Likewise.
10986         (BU_P8V_OVERLOAD_1): Likewise.
10987         (BU_P8V_OVERLOAD_2): Likewise.
10988         (BU_CRYPTO_1): Likewise.
10989         (BU_CRYPTO_2): Likewise.
10990         (BU_CRYPTO_3): Likewise.
10991         (BU_CRYPTO_OVERLOAD_1): Likewise.
10992         (BU_CRYPTO_OVERLOAD_2): Likewise.
10993         (XSCVSPDP): Fix typo, point to the correct instruction.
10994         (VCIPHER): Add power8 crypto builtins.
10995         (VCIPHERLAST): Likewise.
10996         (VNCIPHER): Likewise.
10997         (VNCIPHERLAST): Likewise.
10998         (VPMSUMB): Likewise.
10999         (VPMSUMH): Likewise.
11000         (VPMSUMW): Likewise.
11001         (VPERMXOR_V2DI): Likewise.
11002         (VPERMXOR_V4SI: Likewise.
11003         (VPERMXOR_V8HI: Likewise.
11004         (VPERMXOR_V16QI: Likewise.
11005         (VSHASIGMAW): Likewise.
11006         (VSHASIGMAD): Likewise.
11007         (VPMSUM): Likewise.
11008         (VPERMXOR): Likewise.
11009         (VSHASIGMA): Likewise.
11011         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11012         __CRYPTO__ if the crypto instructions are available.
11013         (altivec_overloaded_builtins): Add support for overloaded power8
11014         builtins.
11016         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11017         support for power8 crypto builtins.
11018         (builtin_function_type): Likewise.
11019         (altivec_init_builtins): Add support for builtins that take vector
11020         long long (V2DI) arguments.
11022         * config/rs6000/crypto.md: New file, define power8 crypto
11023         instructions.
11025 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11026             Pat Haugen  <pthaugen@us.ibm.com>
11027             Peter Bergner  <bergner@vnet.ibm.com>
11029         * doc/invoke.texi (Option Summary): Add power8 options.
11030         (RS/6000 and PowerPC Options): Likewise.
11032         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11033         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11034         wm, wn, wr documentation.
11036         * config/rs6000/constraints.md (wm): New constraint for VSX
11037         registers if direct move instructions are enabled.
11038         (wn): New constraint for no registers.
11039         (wq): New constraint for quad word even GPR registers.
11040         (wr): New constraint if 64-bit instructions are enabled.
11041         (wv): New constraint if power8 vector instructions are enabled.
11042         (wQ): New constraint for quad word memory locations.
11044         * config/rs6000/predicates.md (const_0_to_15_operand): New
11045         constraint for 0..15 for crypto instructions.
11046         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11047         as GPR and floating point registers.
11048         (int_reg_operand): New predicate to match only GPR registers.
11049         (base_reg_operand): New predicate to match base registers.
11050         (quad_int_reg_operand): New predicate to match even GPR registers
11051         for quad memory operations.
11052         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11053         operations in both GPR and VSX registers.
11054         (quad_memory_operand): New predicate for quad memory operations.
11055         (reg_or_indexed_operand): New predicate for direct move support.
11057         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11058         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11059         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11060         (POWERPC_MASKS): Add power8 options.
11061         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11062         various options.
11064         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11065         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11067         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11068         (-mpower8-fusion): New power8 options.
11069         (-mpower8-fusion-sign): Likewise.
11070         (-mpower8-vector): Likewise.
11071         (-mcrypto): Likewise.
11072         (-mdirect-move): Likewise.
11073         (-mquad-memory): Likewise.
11075         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11076         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11077         registers.
11078         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11079         (rs6000_debug_vector_unit): Add p8_vector.
11080         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11081         definitions.  Also print fusion state.
11082         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11083         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11084         (rs6000_option_override_internal): Add support for power8.
11085         (rs6000_common_init_builtins): Add debugging for skipped builtins
11086         if -mdebug=builtin.
11087         (rs6000_adjust_cost): Add power8 support.
11088         (rs6000_issue_rate): Likewise.
11089         (insn_must_be_first_in_group): Likewise.
11090         (insn_must_be_last_in_group): Likewise.
11091         (force_new_group): Likewise.
11092         (rs6000_register_move_cost): Likewise.
11093         (rs6000_opt_masks): Likewise.
11095         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11096         power8 capable assembler, default to power7 options.
11097         (TARGET_DIRECT_MOVE): Likewise.
11098         (TARGET_CRYPTO): Likewise.
11099         (TARGET_P8_VECTOR): Likewise.
11100         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11101         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11102         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11103         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11104         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11105         (TARGET_XSCVDPSPN): Likewise.
11106         (TARGET_XSCVSPDPN): Likewsie.
11107         (TARGET_SYNC_HI_QI): Likewise.
11108         (TARGET_SYNC_TI): Likewise.
11109         (MASK_CRYPTO): Likewise.
11110         (MASK_DIRECT_MOVE): Likewise.
11111         (MASK_P8_FUSION): Likewise.
11112         (MASK_P8_VECTOR): Likewise.
11113         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11114         temporary used by some of the direct move instructions to get two FP
11115         temporary registers does not force creation of a stack frame.
11116         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11117         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11118         that any VSX registers are tieable, even if they are also an
11119         Altivec vector mode.
11120         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11121         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11122         (RS6000_BTM_CRYPTO): Likewise.
11123         (RS6000_BTM_COMMON): Likewise.
11125         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11126         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11127         (enum rs6000_vector): Add power8 vector support.
11129 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11131         PR target/19599
11132         PR target/57340
11133         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11134         (any_sibcall_could_use_r3): this and handle indirect calls.
11135         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11137 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11139         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11141 2013-05-22  Richard Biener  <rguenther@suse.de>
11143         PR middle-end/57349
11144         * profile.c (branch_prob): Do not split blocks that are
11145         abnormally receiving from ECF_RETURNS_TWICE functions.
11147 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11149         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11150         address mode.  Move pointer mode initialization to the same place.
11152 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11154         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11155         while it has any effect.
11157 2013-05-21  Easwaran Raman  <eraman@google.com>
11159         PR tree-optimization/57322
11160         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11161         UID of the statement added to the BB to be 1.
11163 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11165         PR tree-optimization/57331
11166         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11167         of conversion from pointer type to integral type with integer.
11169 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11171         PR lto/57289
11172         * ipa-prop.c (ipa_read_node_info): Process param_used and
11173         controlled_uses in the same order as when writing.
11175 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11177         PR plugins/56754
11178         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11180 2013-05-21  Richard Biener  <rguenther@suse.de>
11182         PR tree-optimization/57318
11183         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11184         estimate stmts with side-effects as likely eliminated.
11186 2013-05-21  Richard Biener  <rguenther@suse.de>
11188         PR tree-optimization/57330
11189         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11190         preserve the call stmts fntype.
11192 2013-05-21  Richard Biener  <rguenther@suse.de>
11194         PR tree-optimization/57303
11195         * tree-ssa-sink.c (statement_sink_location): Improve killing
11196         stmt detection and properly handle self-assignments.
11198 2013-05-21  Christian Bruel  <christian.bruel@st.com>
11200         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11201         spanning registers. LEAF_REG_REMAP is supported only for contiguous
11202         registers. Set register size out of the PARALLEL loop.
11204 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
11206         PR target/56547
11207         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11208         (*fmasf4, *fmasf4_media): New insns.
11210 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11212         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11213         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11214         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11215         (mips_idiv_insns): Update the comments to say that the returned
11216         instruction counts are in units of BASE_INSN_LENGTH.
11217         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11218         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
11219         using 2 rather than 4 as the length of indirect MIPS16 and
11220         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
11221         length of a NOP.  Don't divide MIPS16 lengths by 2.
11222         (mips16_split_long_branches): Assume a branch is long if the
11223         length is greater than 4 rather than 8.
11224         * config/mips/mips.md (length): Give MIPS16 lengths directly,
11225         rather than multiplying them by 2.  Multiply instruction counts
11226         by BASE_INSN_LENGTH rather than 4.
11227         (*jump_mips16, tls_get_tp_mips16_<mode>)
11228         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11230 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11232         * config/mips/mips.md (extended_mips16): Remove branch case.
11233         (length): Remove duplicated extended_mips16 test.
11235 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11237         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11239 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
11241         * recog.h (Recog_data): Rename to...
11242         (recog_data_d): ...this.
11243         (recog_data): Update accordingly.
11244         * recog.c (recog_data): Likewise.
11245         * reload.c (save_recog_data): Likewise.
11246         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11247         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11249 2013-05-17  Julian Brown  <julian@codesourcery.com>
11251         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11252         found in a REG_EQUAL note, invalidate it.
11254 2013-05-17  Easwaran Raman  <eraman@google.com>
11256         * tree-ssa-reassoc.c (find_insert_point): New function.
11257         (insert_stmt_after): Likewise.
11258         (get_def_stmt): Likewise.
11259         (ensure_ops_are_available): Likewise.
11260         (not_dominated_by): Likewise.
11261         (rewrite_expr_tree): Do not move statements beyond what is
11262         necessary. Remove call to swap_ops_for_binary_stmt...
11263         (reassociate_bb): ... and move it here.
11264         (build_and_add_sum): Assign UIDs for new statements.
11265         (linearize_expr): Likewise.
11266         (do_reassoc): Renumber gimple statement UIDs.
11268 2013-05-17  Jan Hubicka  <jh@suse.cz>
11270         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11271         weakrefs.
11272         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11273         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11274         weakrefs.
11275         (output_weakrefs): Update.
11277 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
11278             Martin Jambor  <mjambor@suse.cz>
11280         PR middle-end/57276
11281         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11282         value that corresponds to the given aggval is found in values vector.
11284 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
11286         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11287         sse, sse2, sse3, ssse3 and sse4a flags to options.
11289 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
11291         * gengtype-state.c: (s_expr_writer): New class, to handle
11292         prettifying of output layout of s-expressions.
11293         (state_writer): New class, to write out gtype.state.
11294         (state_written_type_count): Move this variable into member data of
11295         state_writer.
11296         (s_expr_writer::s_expr_writer): New code: constructor for new class
11297         (state_writer::state_writer(): ditto
11298         (s_expr_writer::write_new_line): New function
11299         (s_expr_writer::write_any_indent): ditto
11300         (s_expr_writer::begin_s_expr): ditto
11301         (s_expr_writer::end_s_expr): ditto
11302         (write_state_fileloc): convert to method of state_writer...
11303         (state_writer:: write_state_fileloc): ...and use methods of
11304         s_expr_writer to write indentation into the gtype.state output file
11305         to visually represent the hierarchical structure of the list
11306         structures
11307         (write_state_fields): ditto, renaming to...
11308         (state_writer::write_state_fields)
11309         (write_state_a_string): ditto, renaming to...
11310         (state_writer::write_state_a_string)
11311         (write_state_string_option): ditto, renaming to...
11312         (state_writer::write_state_string_option)
11313         (write_state_type_option): ditto, renaming to...
11314         (state_writer::write_state_type_option)
11315         (write_state_nested_option): ditto, renaming to...
11316         (state_writer::write_state_nested_option)
11317         (write_state_option): ditto, renaming to...
11318         (state_writer::write_state_option)
11319         (write_state_options): ditto, renaming to...
11320         (state_writer::write_state_options)
11321         (write_state_lang_bitmap): ditto, renaming to...
11322         (state_writer::write_state_lang_bitmap)
11323         (write_state_version): ditto, renaming to...
11324         (state_writer::write_state_version)
11325         (write_state_scalar_type): ditto, renaming to...
11326         (state_writer::write_state_scalar_type)
11327         (write_state_string_type): ditto, renaming to...
11328         (state_writer::write_state_string_type)
11329         (write_state_undefined_type): ditto, renaming to...
11330         (state_writer::write_state_undefined_type)
11331         (write_state_struct_union_type): ditto, renaming to...
11332         (state_writer::write_state_struct_union_type)
11333         (write_state_struct_type): ditto, renaming to...
11334         (state_writer::write_state_struct_type)
11335         (write_state_user_struct_type): ditto, renaming to...
11336         (state_writer::write_state_user_struct_type)
11337         (write_state_lang_struct_type): ditto, renaming to...
11338         (state_writer::write_state_lang_struct_type)
11339         (write_state_param_struct_type): ditto, renaming to...
11340         (state_writer::write_state_param_struct_type)
11341         (write_state_pointer_type): ditto, renaming to...
11342         (state_writer::write_state_pointer_type)
11343         (write_state_array_type): ditto, renaming to...
11344         (state_writer::write_state_array_type)
11345         (write_state_gc_used): ditto, renaming to...
11346         (state_writer::write_state_gc_used)
11347         (write_state_common_type_content): ditto, renaming to...
11348         (state_writer::write_state_common_type_content)
11349         (write_state_type): ditto, renaming to...
11350         (state_writer::write_state_type)
11351         (write_state_pair_list): ditto, renaming to...
11352         (state_writer::write_state_pair_list)
11353         (write_state_pair): ditto, renaming to...
11354         (state_writer::write_state_pair)
11355         (write_state_typedefs): ditto, renaming to...
11356         (state_writer::write_state_typedefs)
11357         (write_state_structures): ditto, renaming to...
11358         (state_writer::write_state_structures)
11359         (write_state_param_structs): ditto, renaming to...
11360         (state_writer::write_state_param_structs)
11361         (write_state_variables): ditto, renaming to...
11362         (state_writer::write_state_variables)
11363         (write_state_srcdir): ditto, renaming to...
11364         (state_writer::write_state_srcdir)
11365         (write_state_files_list): ditto, renaming to...
11366         (state_writer::write_state_files_list)
11367         (write_state_languages): ditto, renaming to...
11368         (state_writer::write_state_languages)
11369         (write_state): create a state_writer instance and use it when
11370         writing out the state file
11372 2013-05-17  Mike Stump  <mikestump@comcast.net>
11374         PR rtl-optimization/57304
11375         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11376         accessing DF_REF_REAL_LOC.
11378 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11380         PR rtl-optimization/57281
11381         PR rtl-optimization/57300
11382         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11383         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11384         what the other splitter did if the registers are dead.
11386 2013-05-17  Richard Biener  <rguenther@suse.de>
11388         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11389         MEM_REF offsets.
11391 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11393         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11394         linking.
11396 2013-05-17  Marek Polacek  <polacek@redhat.com>
11398         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11399         length when doing non-zero store of storing '\0' to '\0'.
11401 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11403         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11404         vect_external_def oprnd1 with loop_vinfo, try to emit
11405         optional cast, negation and and stmts on the loop preheader
11406         edge instead of into the pattern def seq.
11408         PR tree-optimization/57051
11409         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11410         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11412 2013-05-16  Nick Clifton  <nickc@redhat.com>
11414         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11415         (rl78_is_naked_func): New function.
11416         (rl78_expand_prologue): Skip prologue generation for naked functions.
11417         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11418         * doc/extend.texi (naked): Add RL78 to the list of processors
11419         that supports this attribute.
11421 2013-05-16  Jeff Law  <law@redhat.com>
11423         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11425 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11427         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11428         cache parameters using detect_caches_amd also for CYRIX,
11429         NSC and TM2 signatures.
11431 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11432             Dzianis Kahanovich  <mahatma@eu.by>
11434         PR target/45359
11435         PR target/46396
11436         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11437         VIA/Centaur processors and determine their cache parameters
11438         using detect_caches_amd.
11440 2013-05-16  Teresa Johnson  <tejohnson@google.com>
11442         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11443         (rtl_verify_edges): New function.
11444         (rtl_verify_bb_insns): Ditto.
11445         (rtl_verify_bb_pointers): Ditto.
11446         (rtl_verify_bb_insn_chain): Ditto.
11447         (rtl_verify_fallthru): Ditto.
11448         (rtl_verify_bb_layout): Ditto.
11449         (rtl_verify_flow_info_1): Outline checks into new functions.
11450         (rtl_verify_flow_info): Ditto.
11452 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
11454         * cfghooks.c (copy_bbs): Add update_dominance argument.
11455         * cfghooks.h (copy_bbs): Update prototype.
11456         * tree-cfg.c (gimple_duplicate_sese_region):
11457         Add update_dominance argument.
11458         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11459         * tree-ssa-loop-ch.c (copy_loop_headers): Update
11460         gimple_duplicate_sese_region call.
11461         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11462         Update copy_bbs call.
11463         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11464         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11466 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11468         * tree-vectorizer.h (NUM_PATTERNS): Increment.
11469         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11470         vect_recog_rotate_pattern.
11471         (vect_recog_rotate_pattern): New function.
11473 2013-05-16  Jason Merrill  <jason@redhat.com>
11475         * Makefile.in (LLINKER): New variable.
11476         (mostlyclean): Remove link mutex.
11477         * configure.ac: Handle --enable-link-mutex.
11478         * lock-and-run.sh: New script.
11480 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11482         PR target/19599
11483         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11484         for NULL decl.
11486 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11488         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11490 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11492         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11493         * config/arm/arm.c (next_consecutive_mem): New function.
11494         (gen_movmem_ldrd_strd): Likewise.
11495         * config/arm/arm.md (movmemqi): Update condition and code.
11496         (unaligned_loaddi, unaligned_storedi): New patterns.
11498 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11500         * config.gcc: Obsolete *-*-solaris2.9*.
11501         * doc/install.texi (Specific, *-*-solaris2*): Document it.
11503 2013-05-16  Richard Biener  <rguenther@suse.de>
11505         * passes.c (init_optimization_passes): Move pass_parallelize_loops
11506         earlier, after GRAPHITE transforms and IV canonicalization.
11508 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11510         * omp-low.c (extract_omp_for_data): For collapsed loops,
11511         if at least one of the loops is known at compile time to
11512         iterate zero times, set count to 0.
11513         (expand_omp_regimplify_p): New function.
11514         (expand_omp_for_generic): For collapsed loops, if at least
11515         one of the loops isn't known to iterate at least once,
11516         add runtime check with setting count to 0.
11517         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11518         For unsigned types if it isn't known at compile time that
11519         the loop will iterate at least once, add runtime check to bypass
11520         the whole loop if initial condition isn't true.
11522 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
11524         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11526 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
11528         PR middle-end/57286
11529         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
11530         transformations to avoid an infinite loop.
11532 2013-05-16  Marek Polacek  <polacek@redhat.com>
11534         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
11536 2013-05-15  Leif Ekblad  <leif@rdos.net>
11538         * config/i386/i386.c (ix86_decompose_address): Use
11539         DEFAULT_TLS_SEG_REG to access TLS segment register.
11540         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
11541         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
11542         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
11544 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
11546         PR target/57260
11547         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
11548         sibling calls to functions that would normally be lazily bound,
11549         unless $gp is call-clobbered.
11551 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
11553         * config/i386/i386.c (ix86_option_override_internal): Update
11554         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
11555         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
11556         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
11557         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
11558         of TARGET_3DNOW.
11559         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
11561 2013-05-15  Andreas Schwab  <schwab@suse.de>
11563         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
11564         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
11565         third operand.
11567 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11569         * loop-unroll.c (report_unroll_peel): Check decision before
11570         emitting unroll/peel message.
11572 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11574         * function.h (has_bb_partition): New rtl_data flag.
11575         (bb_reorder_complete): Ditto.
11576         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
11577         instead of flag_reorder_blocks_and_partition.
11578         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
11579         with some enhancements.
11580         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
11581         * bb-reorder.c (connect_traces): Check for has_bb_partition
11582         instead of flag_reorder_blocks_and_partition.
11583         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
11584         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
11585         verify_hot_cold_block_grouping.
11586         (partition_hot_cold_basic_blocks): Set has_bb_partition.
11588 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11590         PR target/19599
11591         * config/arm/predicates.md (call_insn_operand): New predicate.
11592         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
11593         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
11594         if insn is not a tail call.
11595         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
11596         registers.
11597         * config/arm/arm.h (enum reg_class): New caller save register class.
11598         (REG_CLASS_NAMES): Likewise.
11599         (REG_CLASS_CONTENTS): Likewise.
11600         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
11601         without decls.
11603 2013-05-15  Richard Biener  <rguenther@suse.de>
11605         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
11606         of MSG_OPTIMIZED_LOCATIONS.
11607         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
11608         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
11609         message.
11610         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
11611         of MSG_OPTIMIZED_LOCATIONS.
11612         (execute_vect_slp): Likewise.
11613         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
11614         (vect_create_cond_for_alias_checks): Likewise.
11615         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11616         (vect_recog_widen_mult_pattern): Likewise.
11617         (vect_recog_widen_sum_pattern): Likewise.
11618         (vect_recog_over_widening_pattern): Likewise.
11619         (vect_recog_widen_shift_pattern): Likewise.
11620         (vect_recog_vector_vector_shift_pattern): Likewise.
11621         (vect_recog_divmod_pattern): Likewise.
11622         (vect_recog_mixed_size_cond_pattern): Likewise.
11623         (vect_recog_bool_pattern): Likewise.
11624         (vect_pattern_recog_1): Likewise.
11626 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11628         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
11629         non-functions to builtin_unreachable.
11630         * ipa-inline-transform.c (inline_call): Do not assert estimates were
11631         correct when new direct edges were discovered.
11633 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11635         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
11636         header, print symbol order instead of node uid, print more information
11637         about indirect edge targets.
11638         (ipa_make_edge_direct_to_target): Print symbol order instead of node
11639         uids.
11640         (ipa_make_edge_direct_to_target): Likewise.
11641         (remove_described_reference): Likewise.
11642         (propagate_controlled_uses): Likewise.
11643         (ipa_print_node_params): Also print symbol order.
11644         (ipcp_transform_function): Print symbol order instead of node uids.
11645         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
11646         (cgraph_get_create_real_symbol_node): Likewise.
11647         * ipa-cp.c (print_lattice): Likewise.
11648         (print_all_lattices): Likewise.
11649         (determine_versionability): Likewise.
11650         (initialize_node_lattices): Likewise.
11651         (estimate_local_effects): Likewise.
11652         (update_profiling_info): Likewise.
11653         (create_specialized_node): Likewise.
11654         (perhaps_add_new_callers): Likewise.
11655         (decide_about_value): Likewise.
11656         (decide_whether_version_node): Likewise.
11657         (identify_dead_nodes): Likewise.
11658         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
11659         (dump_inline_summary): Likewise.
11660         (estimate_node_size_and_time): Likewise.
11661         (inline_analyze_function): Likewise.
11662         * ipa-inline.c (report_inline_failed_reason): Likewise.
11663         (want_early_inline_function_p): Likewise.
11664         (edge_badness): Likewise.
11665         (update_edge_key): Likewise.
11666         (inline_small_functions): Likewise.  Add dumping of order to two other
11667         dumps.
11668         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
11669         instead of node uids.
11670         (propagate_pure_const): Likewise.
11671         (propagate_pure_const): Likewise.
11672         * ipa-utils.c (dump_cgraph_node_set): Likewise.
11673         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
11674         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
11675         of node uids.
11676         * tree-pretty-print.c (dump_function_header): Likewise.
11677         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
11678         Print symbol order instead of node uids.
11680 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11682         * config/s390/s390.c (s390_register_move_cost): Don't impose the
11683         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
11685 2013-05-15  Richard Biener  <rguenther@suse.de>
11687         PR tree-optimization/57275
11688         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
11689         return value for fail to do runtime alias checks for gather loads.
11691 2013-05-15  Jan Hubicka  <jh@suse.cz>
11693         PR lto/57038
11694         PR lto/47375
11695         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
11696         weakrefs are not external.
11697         (lto_symtab_merge_decls): Fix thinko when dealing with
11698         non-lto_symtab decls.
11699         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
11700         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
11701         * varpool.c (dump_varpool_node): Dump more flags.
11703 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11705         * config/i386/i386.c (processor_alias_table): Add instruction
11706         FSGSBASE for AMD bdver3 architecture.
11708 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11710         * tree.c (warn_deprecated_use): Print file:line using locus color.
11711         * diagnostic.c (diagnostic_report_current_module): Print file:line
11712         and file:line:column using locus color.
11714 2013-05-14  Mike Stump  <mikestump@comcast.net>
11716         * gdbinit.in: Add __null.
11718 2013-05-14  Mike Stump  <mikestump@comcast.net>
11720         * recog.h: Rename struct recog_data to Recog_data.
11721         * recog.c: Likewise.
11722         * reload.c (can_reload_into): Likewise.
11723         * config/picochip/picochip.c: Likewise.
11725 2013-05-14  Mike Stump  <mikestump@comcast.net>
11727         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
11729 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
11731         * resource.h (struct resources): Remove unch_memory member.
11732         (CLEAR_RESOURCE): Don't clear unch_memory.
11733         * resource.c (mark_referenced_resources): Don't set it.
11734         (mark_set_resources): Likewise.
11735         (mark_target_live_regs): Don't clear it.
11736         (init_resource_info): Likewise.
11737         * reorg.c (resource_conflicts_p): Don't compare it.
11738         (redundant_insn): Don't set it.
11740         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
11741         Remove prototypes.
11742         * emit-rtl.c (next_label): Remove unused function.
11743         (skip_consecutive_labels, link_cc0_insns): Move to ...
11744         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
11745         only place where these functions are used, and make them static.
11747 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11749         * fold-const.c (fold_negate_expr): Handle vectors.
11750         (fold_truth_not_expr): Make it static.
11751         (fold_invert_truthvalue): New static function.
11752         (invert_truthvalue_loc): Handle vectors. Do not call
11753         fold_truth_not_expr directly.
11754         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
11755         <TRUTH_NOT_EXPR>: Do not cast to boolean.
11756         (fold_comparison): Handle vector constants.
11757         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
11758         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
11759         * tree.h (fold_truth_not_expr): Remove declaration.
11761 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
11763         * config/aarch64/aarch64-simd.md
11764         (aarch64_vcond_internal<mode>): Rename to...
11765         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
11766         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
11767         float modes. Clarify all iterator modes.
11768         (vcond<mode><mode>): Use new name for vcond expanders.
11769         (vcond<v_cmp_result><mode>): Likewise.
11770         (vcondu<mode><mode>: Likewise.
11771         * config/aarch64/iterators.md (VDQF_COND): New.
11773 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11775         PR bootstrap/57266
11776         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
11777         variable for the shift amount. Check that we shift by non-negative
11778         amounts.
11780 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
11782         PR target/42017
11783         * config/arm/arm.h (EPILOGUE_USES): Only return true
11784         for LR_REGNUM after epilogue_completed.
11786 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
11788         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
11789         is error_mark_node.
11791 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11793         PR target/57261
11794         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
11795         and Solaris 11+/x86 with gld.
11796         * configure: Regenerate.
11798 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11800         * expmed.c (expand_shift_1): Canonicalize rotates by
11801         constant bitsize / 2 to bitsize - 1.
11802         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
11803         case ROTATERT>: Likewise.
11805         Revert:
11806         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
11808         * config/i386/i386.md (rotateinv): New code attr.
11809         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
11810         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
11811         roll $31, %eax, etc.
11813 2013-05-14  Richard Biener  <rguenther@suse.de>
11815         PR middle-end/57235
11816         * tree-eh.c (sink_clobbers): Give up for successors with
11817         multiple predecessors and no virtual uses.
11819 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
11821         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
11822         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
11824 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11826         PR middle-end/57251
11827         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
11828         the case when both op0 and op1 have VOIDmode.
11830 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11832         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
11833         in multiply-accumulate mode.
11835 2013-05-13  Guozhi Wei  <carrot@google.com>
11837         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
11839 2013-05-13  Kai Tietz  <ktietz@redhat.com>
11841         PR target/56975
11842         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
11843         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
11844         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
11845         * config/i386/i386.c (ix86_option_override_internal): Likewise.
11846         (ix86_expand_prologue): Likewise.
11847         (ix86_expand_split_stack_prologue): Likewise.
11848         (legitimate_pic_address_disp_p): Likewise.
11849         (legitimize_pic_address): Likewise.
11850         (legitimize_tls_address): Likewise.
11851         (legitimize_pe_coff_symbol): Likewise.
11852         (output_pic_addr_const): Likewise.
11853         (construct_plt_address): Likewise.
11854         (ix86_expand_call): Likewise.
11855         (x86_output_mi_thunk): Likewise.
11856         (x86_function_profiler): Likewise.
11858 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
11860         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
11861         similar switch cases.
11862         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
11863         (aarch64_simd_mov_to_<mode>low): Delete.
11864         (aarch64_simd_mov_to_<mode>high): Delete.
11865         (move_lo_quad_<mode>): Add w<-r alternative.
11866         (aarch64_simd_move_hi_quad_<mode>): Likewise.
11867         (aarch64_simd_mov_from_*): Update type attribute.
11868         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
11869         statement.
11871 2013-05-13  Jan Hubicka  <jh@suse.cz>
11873         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
11874         * config/i386/i386.c (ix86_compute_frame_layout,
11875         ix86_expand_epilogue, emit_i387_cw_initialization,
11876         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
11877         ix86_local_alignment): Fix use of size/speed predicates.
11879 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11881         PR tree-optimization/45216
11882         PR tree-optimization/57157
11883         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
11884         the (-Y) & (B - 1) variant if OP is |.
11885         * expmed.c (expand_shift_1): For rotations by const0_rtx just
11886         return shifted.  Use (-op1) & (prec - 1) as other_amount
11887         instead of prec - op1.
11889 2013-05-13  Martin Jambor  <mjambor@suse.cz>
11891         PR middle-end/42371
11892         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
11893         (ipa_constant_data): New type.
11894         (ipa_jump_func): Use ipa_constant_data to hold information about
11895         constant jump functions.
11896         (ipa_get_jf_constant): Adjust to jump function type changes.
11897         (ipa_get_jf_constant_rdesc): New function.
11898         (ipa_param_descriptor): New field controlled_uses.
11899         (ipa_get_controlled_uses): New function.
11900         (ipa_set_controlled_uses): Likewise.
11901         * ipa-ref.h (ipa_find_reference): Declare.
11902         * ipa-prop.c (ipa_cst_ref_desc): New type.
11903         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
11904         changes.
11905         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
11906         New parameter cs.  Adjust all callers.
11907         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
11908         (remove_described_reference): New function.
11909         (jfunc_rdesc_usable): Likewise.
11910         (try_make_edge_direct_simple_call): Decrement controlled use count,
11911         attempt to remove reference if it hits zero.
11912         (combine_controlled_uses_counters): New function.
11913         (propagate_controlled_uses): Likewise.
11914         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
11915         (ipa_edge_duplication_hook): Duplicate reference descriptions.
11916         (ipa_print_node_params): Print described use counter.
11917         (ipa_write_jump_function): Adjust to jump function type changes.
11918         (ipa_read_jump_function): New parameter CS, pass it to
11919         ipa_set_jf_constant.  Adjust caller.
11920         (ipa_write_node_info): Stream controlled use count
11921         (ipa_read_node_info): Likewise.
11922         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
11923         asserting.
11924         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
11925         count.  Remove cloning-added reference if it reaches zero.
11926         * ipa-ref.c (ipa_find_reference): New function.
11928 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11930         * config/i386/i386.c (processor_target_table): Modified default
11931         alignment values for AMD BD and BT architectures.
11933 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
11935         * tree-vect-generic.c (uniform_vector_p): Move ...
11936         * tree.c (uniform_vector_p): ... here.
11937         * tree.h (uniform_vector_p): Declare it.
11938         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
11939         into a scalar.
11941 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11943         PR tree-optimization/57230
11944         * tree-ssa-strlen.c (handle_char_store): Record length for
11945         array store from STRING_CST.
11947         PR tree-optimization/57230
11948         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
11949         check.
11951 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
11953         * config/epiphany/epiphany.c (epiphany_init): Check size of
11954         NUM_MODES_FOR_MODE_SWITCHING.
11955         (epiphany_expand_prologue):
11956         Remove CONFIG_REGNUM initial value handling code.
11957         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
11958         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
11959         (emit_set_fp_mode, epiphany_mode_after): Likewise.
11960         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
11961         Don't return 1 for FP_MODE_NONE.
11962         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
11963         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
11964         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
11965         * config/epiphany/epiphany.md (save_config): New pattern.
11967 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
11969         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
11971 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
11973         * config/i386/i386.md (memory): Handle sseishft1.
11974         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
11975         (*vec_extractv2di_1): Ditto.
11977 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
11979         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
11980         saved registers.
11982 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11984         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
11985         Add mthumb/march=armv7-a multilib.
11986         Add mthumb/march=armv7-r multilib.
11987         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
11989 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
11991         * config/v850/t-rtems: Add more multilibs.
11993 2013-05-10  Richard Biener  <rguenther@suse.de>
11995         PR tree-optimization/57214
11996         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
11997         not propagate from SSA names that occur in abnormal PHI nodes.
11999 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12001         * stor-layout.c (element_precision): New function.
12002         * machmode.h (element_precision): Declare it.
12003         * tree.c (build_minus_one_cst): New function.
12004         (element_precision): Likewise.
12005         * tree.h (build_minus_one_cst): Declare new function.
12006         (element_precision): Likewise.
12007         * fold-const.c (operand_equal_p): Use element_precision.
12008         (fold_binary_loc): Handle vector types.
12009         * convert.c (convert_to_integer): Use element_precision.
12010         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12011         separately.
12013 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12015         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12016         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12017         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12018         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12019         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12020         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12021         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12022         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12023         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12024         (Uuw8): New constraints.
12025         (Usb4): Move into alphabetical order.
12026         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12027         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12028         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12029         previously unnamed patterns.
12030         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12031         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12032         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12033         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12034         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12035         of specific lengths.
12037 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12039         * config/i386/i386.md (rotateinv): New code attr.
12040         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12041         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12042         roll $31, %eax, etc.
12044         PR tree-optimization/45216
12045         PR tree-optimization/57157
12046         * tree-ssa-forwprop.c (simplify_rotate): New function.
12047         (ssa_forward_propagate_and_combine): Call it.
12049 2013-05-10  Richard Biener  <rguenther@suse.de>
12051         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12052         disable peeling when we version for aliasing.
12053         (vector_alignment_reachable_p): Honor explicit user alignment.
12054         (vect_supportable_dr_alignment): Likewise.
12055         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12056         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12057         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12058         then peeling to arrange for the cost-model check to come first.
12060 2013-05-10  Alan Modra  <amodra@gmail.com>
12062         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12063         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12064         * configure: Regenerate.
12066 2013-05-10  Alan Modra  <amodra@gmail.com>
12068         PR target/55033
12069         * varasm.c (default_elf_select_section): Move !DECL_P check..
12070         (get_named_section): ..to here before calling get_section_name.
12071         Adjust assertion.
12072         (default_section_type_flags): Add DECL_P check.
12073         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12074         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12076 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12078         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12079         When using gen_stack_adjust_str with a register offset, add a
12080         REG_FRAME_RELATED_EXPR note.
12082 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12084         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12085         (*vec_extractv4si_zext_mem): Ditto.
12086         (*vec_extractv2di): Add 0->x and x->x alternatives.
12087         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12088         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12090 2013-05-09  Jason Merrill  <jason@redhat.com>
12092         N3639 C++1y VLA support
12093         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12094         DECL_VALUE_EXPR.
12096         * tree.c (build_constructor_va): New.
12097         * tree.h: Declare it.
12099 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12101         PR lto/57084
12102         * gimple-fold.c (canonicalize_constructor_val): Call
12103         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12105 2013-05-09  Jan Hubicka  <jh@suse.cz>
12106             Richard Biener  <rguenther@suse.de>
12108         PR lto/54095
12109         * symtab.c (symtab_make_decl_local): Do not add private names.
12111 2013-05-09  Jan Hubicka  <jh@suse.cz>
12113         PR lto/54095
12114         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12115         (unlink_from_assembler_name_hash): Likewise.
12116         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12117         symtab_unregister_node, symtab_initialize_asm_name_hash,
12118         change_decl_assembler_name): Update.
12120 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12122         * config/aarch64/aarch64.md: New movtf split.
12123         (*movtf_aarch64): Update.
12124         (aarch64_movdi_tilow): Handle TF modes and rename to
12125         aarch64_movdi_<mode>low.
12126         (aarch64_movdi_tihigh): Handle TF modes and rename to
12127         aarch64_movdi_<mode>high
12128         (aarch64_movtihigh_di): Handle TF modes and rename to
12129         aarch64_mov<mode>high_di
12130         (aarch64_movtilow_di): Handle TF modes and rename to
12131         aarch64_mov<mode>low_di
12132         (aarch64_movtilow_tilow): Remove spurious whitespace.
12133         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12134         splits.
12135         (aarch64_print_operand): Update.
12137 2013-05-09  Alan Modra  <amodra@gmail.com>
12139         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12140         powerpc64le.
12141         * configure: Regenerate.
12143 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12145         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12146         splitter preparation statements.
12147         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12148         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12149         adjust_address_nv.
12151 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12153         * gimple-ssa-strength-reduction.c (count_candidates): Change
12154         return value to int.
12155         (analyze_candidates_and_replace): Change type of length to int.
12157 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12159         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12160         (*vec_extract<mode>): Use VI12_128 mode iterator.
12161         (*vec_extract<mode>_mem): Ditto.
12162         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12163         attribute.
12165 2013-05-08  Diego Novillo  <dnovillo@google.com>
12167         PR bootstrap/54659
12169         Revert:
12170         2012-08-17  Diego Novillo  <dnovillo@google.com>
12172         PR bootstrap/54281
12173         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12174         * config.in: Regenerate.
12175         * configure: Regenerate.
12176         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12178 2013-05-08  Jan Hubicka  <jh@suse.cz>
12180         PR lto/54095
12181         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12182         * cgraph.h (symtab_node_base): Add unique_name.
12183         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12184         input_overwrite_node, input_varpool_node): Stream unique_name.
12185         * cgraphclones.c (cgraph_create_virtual_clone,
12186         cgraph_function_versioning): Set unique_name.
12187         * ipa.c (function_and_variable_visibility): Set unique_name.
12189 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12191         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12192         (alloc_cand_and_find_basis): Restrict conditional candidate
12193         processing to CAND_MULTs.
12195 2013-05-08  Jan Hubicka  <jh@suse.cz>
12197         PR lto/54095
12198         lto-symtab.c (lto_symtab_symbol_p): New function.
12199         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12200         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12201         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12202         Skip static symbols.
12204 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
12206         PR tree-optimization/57200
12207         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12208         Only call inform if the preceding warning_at returns true.
12210 2013-05-07  Han Shen  <shenhan@google.com>
12212         * cfgexpand.c (record_or_union_type_has_array_p): New function.
12213         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12214         * common.opt (fstack-protector-strong): New option.
12215         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12216         * doc/invoke.texi (Optimization Options): Document
12217         "-fstack-protector-strong".
12218         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12220 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12222         * config/mips/mips.c (mips_machine_reorg2): Return 0.
12224 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
12226         * ira.c (update_equiv_regs): Add insn having equiv memory even if
12227         it is not lhs of the insn.
12228         (setup_reg_equiv): Remove insn having equiv memory which it is not
12229         lhs of the insn.
12230         * lra-constraints.c (process_address): Try to improve generation
12231         code for address base + disp.
12232         (lra_constraints): Make correct the code for checking insn setting
12233         up backward equivalence.  Remove insn only if it is in the init
12234         insn list.
12235         * lra-eliminations.c (update_reg_eliminate): Change return value.
12236         (lra_eliminate): Use the result.
12238 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
12240         * config/i386/sse.md (ssescalarnummask): New mode attribute.
12241         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12242         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12243         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12244         register target operands.
12245         (*vec_extractv8hi_sse2): New pattern.
12246         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12247         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12248         (*vec_extract<mode>_mem): New insn and split pattern.
12250 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
12252         * config/arm/arm.c (arm_asan_shadow_offset): New function.
12253         (TARGET_ASAN_SHADOW_OFFSET): Define.
12254         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12255         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12257 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12259         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12260         (incr_vec_index): Return -1 if increment not found.
12261         (create_add_on_incoming_edge): Assert if increment not found.
12262         (record_increment): Limit number of increments recorded.
12263         (all_phi_incrs_profitable): Return false if an increment not found.
12264         (replace_profitable_candidates): Don't process increments that were
12265         not recorded.
12266         (analyze_candidates_and_replace): Limit size of incr_vec.
12268 2013-05-07  Richard Biener  <rguenther@suse.de>
12270         * calls.c (special_function_p): setjmp-like functions are leaf.
12271         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12272         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12274 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
12276         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12277         (aarch64_simd_mov<mode>): New expander.
12278         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12279         (aarch64_simd_mov_to_<mode>high): Likewise.
12280         (aarch64_simd_mov_from_<mode>low): Likewise.
12281         (aarch64_simd_mov_from_<mode>high): Likewise.
12282         (aarch64_dup_lane<mode>): Update.
12283         (aarch64_dup_lanedi): New instruction pattern.
12284         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12285         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12287 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12289         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12290         (replace_mult_candidate): Remove unnecessary argument; remove
12291         unnecessary parameter from call to introduce_cast_before_cand.
12292         (replace_unconditional_candidate): Remove unnecessary parameter
12293         from call to replace_mult_candidate.
12294         (replace_conditional_candidate): Likewise.
12295         (insert_initializers): Use make_temp_ssa_name.
12296         (introduce_cast_before_cand): Remove unnecessary argument; use
12297         make_temp_ssa_name.
12298         (replace_one_candidate): Remove unnecessary argument; remove
12299         unnecessary parameter from calls to introduce_cast_before_cand.
12300         (replace_profitable_candidates): Remove unnecessary parameters
12301         from calls to replace_one_candidate.
12303 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12305         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12306         phi def as possibly hiding a basis for a CAND_ADD whose operands
12307         have been commuted in the analysis.
12308         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12310 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12312         * config/aarch64/aarch64.md
12313         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12314         shift value between 0-4.
12316 2013-05-07  Richard Biener  <rguenther@suse.de>
12318         * double-int.h (rshift): New overload.
12319         * double-int.c (rshift): New function.
12320         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12321         (create_reference_ops_from_ref): Remove.
12322         (vn_reference_insert): Use shared ops for constructing the
12323         reference and copy it.
12325 2013-05-07  Richard Biener  <rguenther@suse.de>
12327         PR middle-end/57190
12328         * tree-eh.c (sink_clobbers): Properly propagate
12329         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12331 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
12333         PR tree-optimization/57149
12334         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12335         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12336         collect_phi_def_edges, execute_late_warn_uninitialized): Use
12337         uninit_undefined_value_p instead of ssa_undefined_value_p.
12339         PR debug/57184
12340         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12341         for modifier == EXPAND_INITIALIZER.
12343 2013-05-07  Anton Blanchard  <anton@samba.org>
12345         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12346         for powerpc64 little endian.
12347         * configure: Regenerate.
12349 2013-05-06  Graham Stott  <grahams@btinternet.com>
12351         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12352         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12353         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12354         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12356 2013-05-06  Graham Stott  <grahams@btinternet.com>
12358         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12359         codes which allow non-lvalues.
12361 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
12363         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12364         components are all 1s.
12365         (integer_minus_onep): New function.
12366         * tree.h (integer_minus_onep): Declare it.
12367         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12368         integer_minus_onep instead of integer_all_onesp.
12370 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12372         PR target/52933
12373         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12374         variations of these patterns.
12376 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12378         * config/i386/i386.md (isa): Add x64_sse4 member.
12379         (enabled): Handle x64_sse4.
12380         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12381         instruction for 64bit SSE4_1 targets.  Update insn attributes.
12382         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12383         instruction for SSE4_1 targets.  Update insn attributes.
12384         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12385         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12386         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12387         const_1 selector.
12388         (*vec_extractv4si): Rename from *sse4_1_pextrd.
12389         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12390         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12392 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12394         PR target/57108
12395         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12397 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
12399         * final.c (do_assembler_dialects): Don't handle curly braces and
12400         vertical bar escaped by % as dialect delimiters.
12401         (output_asm_insn): Print curly braces and vertical bar if escaped
12402         by % and ASSEMBLER_DIALECT defined.
12403         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12404         * doc/tm.texi: Regenerated.
12406 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12408         * config/mips/mips.c: Include tree-pass.h.
12409         (mips_reorg): Split in pre- and post-dbr_schedule parts.
12410         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12411         (pass_mips_machine_reorg2): New machine specific pass.
12412         (insert_pass_mips_machine_reorg2): New pass plugin definition.
12413         (mips_option_override): Register the new pass.
12414         * rtl.h (cleanup_barriers): Remove prototype.
12415         (dbr_schedule): Likewise.
12416         * jump.c (cleanup_barriers): Make static.
12417         * reorg.c (dbr_schedule): Likewise.
12419 2013-05-06  Richard Biener  <rguenther@suse.de>
12421         PR tree-optimization/57185
12422         * tree-parloops.c (add_field_for_reduction): Handle anonymous
12423         SSA names properly.
12425 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12427         PR target/57106
12428         * config/i386/i386.c (add_parameter_dependencies): Add dependence
12429         between "first_arg" and "insn", not "last" and "insn".
12431 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12433         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12434         (find_candidates_in_block): Re-enable slsr_process_phi.
12435         (create_phi_basis): Fix double counting of candidate adjustment.
12437 2013-05-06  Richard Biener  <rguenther@suse.de>
12439         PR middle-end/57147
12440         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12441         the edge is also fallthru, preserve it and just clear the
12442         abnormal flag.
12443         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12444         also complex, preserve that and just clear the fallthru flag.
12445         * tree-inline.c (update_ssa_across_abnormal_edges): Also
12446         update virtual operands.
12448 2013-05-06  Alan Modra  <amodra@gmail.com>
12450         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12451         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12452         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12453         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12454         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12455         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12457 2013-05-06  Alan Modra  <amodra@gmail.com>
12459         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12460         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
12461         (DEFAULT_ASM_ENDIAN): Define.
12462         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12463         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12464         Update -K PIC clause from sysv4.h.
12465         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12466         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12468 2013-05-06  Alan Modra  <amodra@gmail.com>
12470         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12471         twice for little-endian.
12472         (ashrdi3_no_power, ashrdi3): Support little-endian.
12474 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12476         PR target/55303
12477         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12478         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12479         related expanders.
12480         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12481         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12482         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12483         New predicates.
12485 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
12486             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12488         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12489         * config/pa/pa.opt: Make mbig-switch a no-op.
12490         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12491         (CASE_VECTOR_MODE): Always return SImode.
12492         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12493         for the !TARGET_BIG_SWITCH case.
12494         * config/pa/pa-linux.h: Likewise.
12495         * config/pa/pa-openbsd.h: Likewise.
12496         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12497         * config/pa/pa.md (short_jump): Remove define_insn.
12498         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12499         (casesi0): Remove define_insn.
12500         (type): Remove btable_branch.
12501         (pa_combine_type): Likewise.
12502         (in_nullified_branch_delay): Likewise.
12503         (in_call_delay): Likewise.
12504         (define_delay): Likewise.
12505         (define_insn_reservation "Z3"): Likewise.
12506         (define_insn_reservation "Z4"): Likewise.
12507         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12508         (pa_adjust_insn_length): Remove adjustment for btable branches.
12509         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12510         and mno-big-switch
12512 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
12514         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12515         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12516         Add m->r,x alternatives.
12517         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12518         splitters using SWI48x mode iterator.
12519         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
12520         TARGET_64BIT.  Add m->x alternative.
12521         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12522         Add o->x alternative.  Enable for TARGET_SSE.
12523         (sse_storeq): Remove expander.
12524         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12525         with memory input operand.
12526         (*vec_extractv2di_1 splitter): New.
12527         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12528         * config/i386/i386.md (ssevecmodelower): New mode attribute.
12530 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12532         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
12533         (INT_LOWPART): Delete.
12534         (extract_MB): Adjust.
12535         (extract_ME): Adjust.
12536         (print_operand): Adjust.
12538 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12540         * config/rs6000/predicates.md (reg_or_add_cint_operand,
12541         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12542         (reg_or_logical_cint_operand, easy_fp_constant,
12543         logical_const_operand): Delete "CONST_DOUBLE" case.
12544         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
12545         "HOST_BITS_PER_WIDE_INT == 64" test.
12546         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
12547         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
12548         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
12549         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
12550         test.
12551         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
12552         CONST_DOUBLE DImode/VOIDmode case.
12553         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
12554         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12555         CONST_DOUBLE VOIDmode case.
12556         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12557         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
12558         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
12559         Delete CONST_DOUBLE case.
12560         (splitters for mov FMOVE64 const_double): Delete
12561         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12562         "HOST_BITS_PER_WIDE_INT >= 64" test.
12563         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
12564         case.
12565         (mov DI const_double): Delete.
12567 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
12569         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
12570         on op shows all bits zero in mode of a lowpart subreg, return zero.
12572 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
12574         PR target/57150
12575         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
12576         to save TFmode registers and DImode to save TImode registers for
12577         caller save operations.
12578         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
12579         mark being partially clobbered since they only use the first
12580         double word.
12582         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
12583         and TDmode only use the upper 64-bits of each VSX register.
12585 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12587         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
12588         (find_candidates_in_block): Disable slsr_process_phi.
12590 2013-05-03  Guozhi Wei  <carrot@google.com>
12592         * coverage.c (coverage_obj_init): Move the construction of gcov
12593         constructor to ...
12594         (build_init_ctor): ... here.
12596 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12598         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
12599         (slsr_cand_d): Redefine def_phi.
12600         (stride_status, phi_adjust_status, count_phis_status): New enums.
12601         (find_phi_def): New.
12602         (find_basis_for_base_expr): New.
12603         (find_basis_for_candidate): Handle hidden bases.
12604         (alloc_cand_and_find_basis): Handle phi candidates.
12605         (slsr_process_phi): New.
12606         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
12607         (create_mul_imm_cand): Likewise.
12608         (create_add_ssa_cand): Exclude phi base candidates.
12609         (create_add_imm_cand): Likewise.
12610         (slsr_process_cast): Likewise.
12611         (slsr_process_copy): Likewise.
12612         (find_candidates_in_block): Handle phi candidates.
12613         (dump_candidate): Likewise.
12614         (unconditional_cands): Delete.
12615         (unconditional_cands_with_known_stride_p): Delete.
12616         (phi_dependent_cand_p): New.
12617         (cand_increment): Handle phi-dependent candidates.
12618         (replace_dependent): Delete.
12619         (replace_mult_candidate): New.
12620         (replace_unconditional_candidate): New.
12621         (incr_vec_index): Move to avoid forward reference.
12622         (create_add_on_incoming_edge): New.
12623         (create_phi_basis): New.
12624         (replace_dependents): Delete.
12625         (replace_conditional_candidate): New.
12626         (phi_add_costs): New.
12627         (replace_uncond_cands_and_profitable_phis): New.
12628         (record_increment): Handle phi adjustments.
12629         (record_phi_increments): New.
12630         (record_increments): Handle phi adjustments.
12631         (phi_incr_cost): New.
12632         (lowest_cost_path): Handle phis.
12633         (total_savings): Likewise.
12634         (analyze_increments): Likewise.
12635         (ncd_with_phi): New.
12636         (ncd_of_cand_and_phis): New.
12637         (nearest_common_dominator_for_cands): Handle phi increments.
12638         (all_phi_incrs_profitable): New.
12639         (replace_profitable_candidates): Handle phi-dependent candidates.
12640         (analyze_candidates_and_replace): Likewise.
12642 2013-05-03  Teresa Johnson  <tejohnson@google.com>
12644         PR bootstrap/57154
12645         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
12646         do not exceed REG_BR_PROB_BASE.
12648 2013-05-03  Jeff Law  <law@redhat.com>
12650         PR tree-optimization/57144
12651         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
12652         operand of the condition will bit into the new type when eliminating
12653         a cast feeding a condition.
12655 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
12657         PR rtl-optimization/57130
12658         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
12659         of COMPARE as in_code to the recursive call if needed.
12661 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
12663         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
12664         (enabled): Handle new members.
12665         * config/i386/sse.md (*vec_concatv2si): Merge from
12666         *vec_concatv2si_sse2 and vec_concatv2si_sse.
12667         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
12669 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
12671         PR tree-optimization/57027
12672         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
12673         for fnms opportunity, check we got the prerequisite kind
12674         of tree / gimple before using accessor functions.
12676 2013-05-03  Richard Biener  <rguenther@suse.de>
12678         * double-int.h (lshift): New overload without precision
12679         and arith argument.
12680         (operator *=, operator +=, operator -=): Move ...
12681         * double-int.c (operator *=, operator +=, operator -=): ... here
12682         and implement more efficiently.
12683         (mul_double_with_sign): Remove.
12684         (lshift_double): Adjust to take unsinged shift argument, push
12685         dispatching code to callers.
12686         (mul_double_wide_with_sign): Add early out for callers that
12687         are not interested in high parts or overflow.
12688         (lshift): New function.
12689         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
12690         dispatch code here.
12691         (lrotate, rrotate): Use logical shifts.
12692         * expr.c (get_inner_reference): Use lshift.
12693         * fixed-value.c (do_fixed_divide): Likewise.
12694         * tree-dfa.c (get_ref_base_and_extent): Likewise.
12695         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
12696         (indirect_refs_may_alias_p): Likewise.
12697         (stmt_kills_ref_p_1): Likewise.
12699 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12701         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
12703 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12705         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
12706         scalar form of FABD instruction.
12708 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
12710         * lra-constraints.c (process_alt_operands): Add checking alt
12711         number to choose the best alternative.
12713 2013-05-02  Richard Biener  <rguenther@suse.de>
12715         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
12716         bitmap and its handling.
12717         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
12719 2013-05-02  Richard Biener  <rguenther@suse.de>
12721         PR middle-end/57140
12722         * tree-inline.c (copy_loops): Properly handle removed loops.
12723         (copy_cfg_body): Mark destination loops for fixup if source
12724         loops needed fixup.
12726 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
12728         PR target/56732
12729         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
12730         generating simple_return for naked functions.
12732 2013-05-02  Martin Jambor  <mjambor@suse.cz>
12734         PR middle-end/56988
12735         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
12736         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
12737         flags match.
12738         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
12739         ipa_agg_replacement_value structures.
12740         (known_aggs_to_agg_replacement_list): Likewise.
12741         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
12742         (read_agg_replacement_chain): Likewise.
12743         (ipcp_transform_function): Also check that by_ref flags match.
12745 2013-05-02  Richard Biener  <rguenther@suse.de>
12747         * graphds.h (struct graph): Add obstack member.
12748         * graphds.c (new_graph): Initialize obstack and allocate
12749         vertices from it.
12750         (add_edge): Allocate edge from the obstack.
12751         (free_graph): Free the obstack instead of all edges and vertices.
12753 2013-05-02  Teresa Johnson  <tejohnson@google.com>
12755         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
12756         divides.
12757         * cfg.c (update_bb_profile_for_threading): Ditto.
12758         * tree-inline.c (copy_bb): Ditto.
12759         (copy_edges_for_bb): Ditto.
12760         (initialize_cfun): Ditto.
12761         (copy_cfg_body): Ditto.
12762         (expand_call_inline): Ditto.
12763         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
12764         (estimate_node_size_and_time): Ditto.
12765         (inline_merge_summary): Ditto.
12766         * cgraphclones.c (cgraph_clone_edge): Ditto.
12767         (cgraph_clone_node): Ditto.
12768         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12769         (compute_trg_info): Ditto.
12771 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12773         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
12774         S reg when fp attribute set.
12775         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
12777 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12779         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
12780         New pattern.
12781         (*and_one_cmplsi3_compare0_uxtw): Likewise.
12782         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
12783         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
12785 2013-05-02  Richard Biener  <rguenther@suse.de>
12787         * tree-scalar-evolution.c (scev_info_hasher): Remove.
12788         (struct instantiate_cache_entry): New type.
12789         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
12790         (struct instantiate_cache_type): New type.
12791         (set_instantiated_value, get_instantiated_value): Remove.
12792         (get_instantiated_value_entry): New function.
12793         (instantiate_scev_name): Use the new cache and adjust.
12794         (instantiate_scev_poly): Adjust.
12795         (instantiate_scev_binary): Likewise.
12796         (instantiate_array_ref): Likewise.
12797         (instantiate_scev_convert): Likewise.
12798         (instantiate_scev_not): Likewise.
12799         (instantiate_scev_3): Likewise.
12800         (instantiate_scev_2): Likewise.
12801         (instantiate_scev_r): Likewise.
12802         (instantiate_scev): Likewise.
12803         (resolve_mixers): Likewise.
12805 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
12807         PR target/57091
12808         * lra-constraints.c (best_small_class_operands_num): Remove.
12809         (process_alt_operands): Remove small_class_operands_num.  Take
12810         small classes operands into losers and only if the operand is not
12811         matched.  Modify debugging output.
12812         (curr_insn_transform): Remove best_small_class_operands_num.
12813         Print insn name.
12815 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12817         * config/aarch64/aarch64-builtins.c
12818         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
12819         * config/aarch64/aarch64-simd-builtins.def
12820         (reduc_splus_): Add new modes.
12821         (reduc_uplus_): New.
12822         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
12823         (reduc_uplus_v4sf): Likewise.
12824         (reduc_splus_v4sf): Likewise.
12825         (aarch64_addv<mode>): Likewise.
12826         (reduc_uplus_<mode>): Likewise.
12827         (reduc_splus_<mode>): Likewise.
12828         (aarch64_addvv2di): Likewise.
12829         (reduc_uplus_v2di): Likewise.
12830         (reduc_splus_v2di): Likewise.
12831         (aarch64_addvv2si): Likewise.
12832         (reduc_uplus_v2si): Likewise.
12833         (reduc_splus_v2si): Likewise.
12834         (reduc_<sur>plus_<mode>): New.
12835         (reduc_<sur>plus_v2di): Likewise.
12836         (reduc_<sur>plus_v2si): Likewise.
12837         (reduc_<sur>plus_v4sf): Likewise.
12838         (aarch64_addpv4sf): Likewise.
12839         * config/aarch64/arm_neon.h
12840         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
12841         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
12842         add UNSPEC_SADDV, UNSPEC_UADDV.
12843         (SUADDV): New.
12844         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
12846 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12848         * config/aarch64/arm_neon.h
12849         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
12851 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12853         * config/aarch64/aarch64-builtins
12854         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
12856 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12858         * config/aarch64/aarch64-simd-builtins.def
12859         (reduc_smax_): New.
12860         (reduc_smin_): Likewise.
12861         (reduc_umax_): Likewise.
12862         (reduc_umin_): Likewise.
12863         (reduc_smax_nan_): Likewise.
12864         (reduc_smin_nan_): Likewise.
12865         (fmax): Remove.
12866         (fmin): Likewise.
12867         (smax): Update for V2SF, V4SF and V2DF modes.
12868         (smin): Likewise.
12869         (smax_nan): New.
12870         (smin_nan): Likewise.
12871         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
12872         (<su><maxmin><mode>3): ...This, refactor.
12873         (s<maxmin><mode>3): New.
12874         (<maxmin_uns><mode>3): Likewise.
12875         (reduc_<maxmin_uns>_<mode>): Refactor.
12876         (reduc_<maxmin_uns>_v4sf): Likewise.
12877         (reduc_<maxmin_uns>_v2si): Likewise.
12878         (aarch64_<fmaxmin><mode>: Remove.
12879         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
12880         new builtin names.
12881         (vmin<q>_f<32,64>): Likewise.
12882         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12883         (FMAXMIN): New.
12884         (su): Add mappings for smax, smin, umax, umin.
12885         (maxmin): New.
12886         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12887         (FMAXMIN): Rename as...
12888         (FMAXMIN_UNS): ...This.
12889         (maxminv): Remove.
12890         (fmaxminv): Likewise.
12891         (fmaxmin): Likewise.
12892         (maxmin_uns): New.
12893         (maxmin_uns_op): Likewise.
12895 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12897         * config/aarch64/arm_neon.h
12898         (vac<ge, gt><sd>_f<32, 64>): Rename to...
12899         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
12900         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
12902 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12904         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
12905         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
12907 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12909         * config/aarch64/aarch64-simd.md
12910         (vcond<mode>_internal): Handle special cases for constant masks.
12911         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
12912         (vcondu<mode><mode>): Likewise.
12913         (vcond<v_cmp_result><mode>): New.
12915 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12917         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
12918         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
12919         * config/aarch64/aarch64-simd-builtins.def
12920         (cmeq): Update to BUILTIN_VALLDI.
12921         (cmgt): Likewise.
12922         (cmge): Likewise.
12923         (cmle): Likewise.
12924         (cmlt): Likewise.
12925         * config/aarch64/arm_neon.h
12926         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
12927         to builtins or C as appropriate.
12929 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12931         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
12932         (cmgeu): ...This.
12933         (cmhi): Rename to...
12934         (cmgtu): ...This.
12935         * config/aarch64/aarch64-simd.md
12936         (simd_mode): Add SF.
12937         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
12938         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
12939         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
12940         (cstore<mode>_neg): ...This.
12941         * config/aarch64/iterators.md
12942         (VALLF): new.
12943         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
12944         (COMPARISONS): New.
12945         (UCOMPARISONS): Likewise.
12946         (optab): Add missing comparisons.
12947         (n_optab): New.
12948         (cmp_1): Likewise.
12949         (cmp_2): Likewise.
12950         (CMP): Likewise.
12951         (cmp): Remove.
12952         (VCMP_S): Likewise.
12953         (VCMP_U): Likewise.
12954         (V_cmp_result): Add DF, SF modes.
12955         (v_cmp_result): Likewise.
12956         (v): Likewise.
12957         (vmtype): Likewise.
12958         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
12960 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
12962         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
12963         define_insn to define_insn_and_split.
12964         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
12965         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
12966         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
12967         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
12968         (thumb2_negscc): Likewise.
12970 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
12972         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
12974 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
12976         * config/arm/thumb2.md: Remove trailing whitespaces.
12978 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
12980         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
12981         Use gen_int_mode rather than GEN_INT.
12983 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
12985         * value-prof.c (stream_in_histogram_value): Remove the strayed
12986         debug_gimple_stmt.
12988 2013-04-30  Richard Biener  <rguenther@suse.de>
12990         PR middle-end/57122
12991         * cfghooks.c (split_edge): Properly check for the loop latch edge.
12993 2013-04-30  Richard Biener  <rguenther@suse.de>
12995         PR middle-end/57107
12996         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
12998 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13000         PR rtl-optimization/56957
13001         PR rtl-optimization/57105
13002         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13003         variable.  Use just INSN_UID for determining whether an insn
13004         should be only disconnected from the insn stream.
13005         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13007 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13009         PR tree-optimization/57104
13010         * tsan.c (instrument_expr): Don't instrument accesses to
13011         DECL_HARD_REGISTER VAR_DECLs.
13013 2013-04-30  Richard Biener  <rguenther@suse.de>
13015         * function.h (loops_for_fn): New inline function.
13016         (set_loops_for_fn): Likewise.
13017         * cfgloop.h (place_new_loop): Add struct function parameter.
13018         (get_loop): Likewise.
13019         (get_loops): Likewise.
13020         (number_of_loops): Likewise.
13021         (fel_next): Adjust.
13022         (fel_init): Likewise.
13023         * cfg.c (get_loop_copy): Adjust.
13024         * cfgloop.c (flow_loops_dump): Likewise.
13025         (record_loop_exits): Likewise.
13026         (verify_loop_structure): Likewise.
13027         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13028         (estimate_reg_pressure_cost): Likewise.
13029         (mark_loop_exit_edges): Likewise.
13030         * cfgloopmanip.c (place_new_loop): Likewise.
13031         (add_loop): Likewise.
13032         (duplicate_loop): Likewise.
13033         * graph.c (draw_cfg_nodes): Likewise.
13034         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13035         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13036         (extract_affine_chrec): Likewise.
13037         (build_scop_iteration_domain): Likewise.
13038         * graphite.c (graphite_initialize): Likewise.
13039         * ira-build.c (create_loop_tree_nodes): Likewise.
13040         (more_one_region_p): Likewise.
13041         (rebuild_regno_allocno_maps): Likewise.
13042         (mark_loops_for_removal): Likewise.
13043         (mark_all_loops_for_removal): Likewise.
13044         (remove_unnecessary_regions): Likewise.
13045         (ira_build): Likewise.
13046         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13047         * loop-init.c (fix_loop_structure): Likewise.
13048         (gate_rtl_move_loop_invariants): Likewise.
13049         (gate_rtl_unswitch): Likewise.
13050         (gate_rtl_unroll_and_peel_loops): Likewise.
13051         (rtl_doloop): Likewise.
13052         * lto-streamer-in.c (input_cfg): Likewise.
13053         * lto-streamer-out.c (output_cfg): Likewise.
13054         * modulo-sched.c (sms_schedule): Likewise.
13055         * predict.c (tree_estimate_probability): Likewise.
13056         (tree_estimate_probability_driver): Likewise.
13057         (estimate_loops): Likewise.
13058         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13059         (move_sese_region_to_fn): Likewise.
13060         (debug_loop_num): Likewise.
13061         * tree-chrec.c (chrec_evaluate): Likewise.
13062         (hide_evolution_in_other_loops_than_loop): Likewise.
13063         (chrec_component_in_loop_num): Likewise.
13064         (reset_evolution_in_loop): Likewise.
13065         (evolution_function_is_invariant_rec_p): Likewise.
13066         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13067         * tree-inline.c (copy_loops): Likewise.
13068         (copy_cfg_body): Likewise.
13069         (tree_function_versioning): Likewise.
13070         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13071         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13072         Likewise.
13073         (add_to_evolution_1): Likewise.
13074         (scev_const_prop): Likewise.
13075         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13076         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13077         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13078         (tree_ssa_lim_initialize): Likewise.
13079         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13080         (verify_loop_closed_ssa): Likewise.
13081         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13082         (tree_ssa_loop_im): Likewise.
13083         (tree_ssa_loop_unswitch): Likewise.
13084         (tree_vectorize): Likewise.
13085         (check_data_deps): Likewise.
13086         (tree_ssa_loop_ivcanon): Likewise.
13087         (tree_ssa_loop_bounds): Likewise.
13088         (tree_complete_unroll): Likewise.
13089         (tree_complete_unroll_inner): Likewise.
13090         (tree_parallelize_loops): Likewise.
13091         (tree_ssa_loop_prefetch): Likewise.
13092         (tree_ssa_loop_ivopts): Likewise.
13093         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13094         * tree-vectorizer.c (vectorize_loops): Likewise.
13096 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13098         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13099         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13100         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13101         with EABI_LINK_SPEC.
13103 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13105         PR target/44578
13106         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13107         alternative.
13109 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13111         PR target/57097
13112         * lra-constraints.c (process_alt_operands): Discourage a bit more
13113         using memory for pseudos.  Print cost dump for alternatives.
13114         Modify cost values for conflicts with early clobbers.
13115         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13117 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13119         PR target/57098
13120         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13122 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13124         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13125         from/to S register.
13126         (movdi_aarch64): Support LDR/STR from/to D register.
13128 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13130         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13131         or higher by default.
13133 2013-04-29  Richard Biener  <rguenther@suse.de>
13135         PR middle-end/57075
13136         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13137         even if not adding abnormal edges for calls that can make
13138         abnormal gotos.
13140 2013-04-29  Richard Biener  <rguenther@suse.de>
13142         PR middle-end/57103
13143         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13144         TREE_BLOCK.
13145         (move_stmt_r): Remove redundant checking.
13147 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13149         PR bootstrap/57077
13150         * basic-block.h (apply_scale): New function.
13151         (apply_probability): Use apply_scale.
13152         * gimple-streamer-in.c (input_bb): Ditto.
13153         * lto-streamer-in.c (input_cfg): Ditto.
13154         * lto-cgraph.c (merge_profile_summaries): Ditto.
13155         * tree-optimize.c (execute_fixup_cfg): Ditto.
13156         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13157         (copy_edges_for_bb): Ditto.
13158         (copy_cfg_body): Ditto.
13160 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13162         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13163         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13164         (tail_merge_optimize): Handle current_loops == NULL.
13166 2013-04-26  Jeff Law  <law@redhat.com>
13168         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13169         (simplify_cond_using_ranges): Generalize code to simplify
13170         COND_EXPRs where one argument is a constant and the other
13171         is an SSA_NAME created by an integral type conversion.
13173 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13175         * config/arm/arm.md (store_minmaxsi): Use only when
13176         optimize_insn_for_size_p.
13178 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13180         PR target/57108
13181         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13183 2013-04-29  Richard Biener  <rguenther@suse.de>
13185         PR middle-end/57089
13186         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13187         loop tree make sure to schedule a fixup for the child as well.
13188         (expand_omp_for_generic): Properly add loops.
13189         (expand_omp_for_static_nochunk): Likewise.
13190         (expand_omp_for_static_chunk): Likewise.
13191         (expand_omp_for): For the degenerate case fixup loops.
13192         (expand_omp_sections): Fix default bb placement in loops.
13193         (expand_omp_atomic_pipeline): Properly add loops.
13195 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13197         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13199 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13201         * tree-ssa-tail-merge.c: Update header comment.
13203 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13205         * config/aarch64/arm_neon.h
13206         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13207         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13208         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13209         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13210         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13211         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13212         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13213         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13215 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13217         * config/aarch64/aarch64-simd.md
13218         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13219         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13220         fix_trunc, fixuns_trunc.
13221         (ftrunc<VDQF:mode>2): New.
13222         * config/aarch64/iterators.md (optab): Add fix, fixuns.
13223         (fix_trunc_optab): New.
13225 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13227         * config/aarch64/aarch64-builtins.c
13228         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13229         iceilf, lround, iroundf.
13231 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13233         PR target/54349
13234         * config/i386/i386.h (enum ix86_tune_indices)
13235         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13236         New, split from X86_TUNE_INTER_UNIT_MOVES.
13237         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13238         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13239         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13240         (TARGET_INTER_UNIT_MOVES): Remove.
13241         * config/i386/i386.c (initial_ix86_tune_features): Update.
13242         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13243         (ix86_expand_convert_uns_didf_sse): Use
13244         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13245         (ix86_expand_vector_init_one_nonzero): Ditto.
13246         (ix86_expand_vector_init_interleave): Ditto.
13247         (inline_secondary_memory_needed): Return true for moves from SSE class
13248         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13249         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13250         * config/i386/constraints.md (Yi, Ym): Depend on
13251         TARGET_INTER_UNIT_MOVES_TO_VEC.
13252         (Yj, Yn): New constraints.
13253         * config/i386/i386.md (*movdi_internal): Change constraints of
13254         operand 1 from Yi to Yj and from Ym to Yn.
13255         (*movsi_internal): Ditto.
13256         (*movdf_internal): Ditto.
13257         (*movsf_internal): Ditto.
13258         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13259         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13260         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13261         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13262         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13263         * config/i386/sse.md (movdi_to_sse): Ditto.
13264         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13265         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13266         TARGET_INTER_UNIT_MOVES.
13267         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13268         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13269         instead of TARGET_INTER_UNIT_MOVES.
13270         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13271         operand 1 from Yi to Yj and from Ym to Yn.
13273 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13275         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13276         (float_truncate_hi_): Likewise.
13277         (float_extend_lo_): Likewise.
13278         (float_truncate_lo_): Likewise.
13279         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13280         (aarch64_float_extend_lo_v2df): Likewise.
13281         (vec_unpacks_hi_v4sf): Likewise.
13282         (aarch64_float_truncate_lo_v2sf): Likewise.
13283         (aarch64_float_truncate_hi_v4sf): Likewise.
13284         (vec_pack_trunc_v2df): Likewise.
13285         (vec_pack_trunc_df): Likewise.
13287 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13289         * config/aarch64/aarch64-builtins.c
13290         (aarch64_fold_builtin): Fold float conversions.
13291         * config/aarch64/aarch64-simd-builtins.def
13292         (floatv2si, floatv4si, floatv2di): New.
13293         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13294         * config/aarch64/aarch64-simd.md
13295         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13296         * config/aarch64/iterators.md (FLOATUORS): New.
13297         (optab): Add float, floatuns.
13298         (su_optab): Likewise.
13300 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13302         * config/aarch64/aarch64-builtins.c
13303         (aarch64_builtin_vectorized_function): Use new names for
13304         fcvt builtins.
13305         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13306         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13307         (fcvtzu): Split as...
13308         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13309         (fcvtas): Split as...
13310         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13311         (fcvtau): Split as...
13312         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13313         (fcvtps): Split as...
13314         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13315         (fcvtpu): Split as...
13316         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13317         (fcvtms): Split as...
13318         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13319         (fcvtmu): Split as...
13320         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13321         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13322         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13323         (lfrintnusf, lfrintnudf): Likewise.
13324         * config/aarch64/aarch64-simd.md
13325         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13326         define_insn.
13327         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13328         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13329         (fcvt_pattern): Likewise.
13331 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13333         * config/aarch64/aarch64-simd.md
13334         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13335         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13337 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13339         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13340         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13341         (vrnd<a,m,n,p>_f32): Implement using builtins.
13342         (vrnd<i,x><q>_f<32, 64>): New.
13344 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13346         * config/aarch64/aarch64-builtins.c
13347         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13348         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13349         (frintz): Rename to...
13350         (btrunc): ...this.
13351         (frintp): Rename to...
13352         (ceil): ...this.
13353         (frintm): Rename to...
13354         (floor): ...this.
13355         (frinti): Rename to...
13356         (nearbyint): ...this.
13357         (frintx): Rename to...
13358         (rint): ...this.
13359         (frinta): Rename to...
13360         (round): ...this.
13361         * config/aarch64/aarch64-simd.md
13362         (aarch64_frint<frint_suffix><mode>): Delete.
13363         (<frint_pattern><mode>2): Convert to insn.
13364         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13365         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13366         (frint_pattern): Likewise.
13367         (frint_suffix): Likewise.
13369 2013-04-29  Richard Biener  <rguenther@suse.de>
13371         PR tree-optimization/57081
13372         * loop-init.c: Include tree-flow.h.
13373         (loop_optimizer_finalize): Free number of iteration estimates.
13374         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13376 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
13378         PR tree-optimization/57083
13379         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13380         non-singleton shift count range, zero extend low_bound for uns case.
13382         * config/i386/predicates.md (general_vector_operand): New predicate.
13383         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13384         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13385         if they aren't nonimmediate operands.  If their original values
13386         satisfy const_vector_equal_evenodd_p, don't shift them.
13387         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13388         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
13389         if not nonimmediate_operand.
13390         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13391         instead of register_operand.
13392         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13394 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
13396         * stor-layout.c (finalize_size_functions): Allocate a structure and
13397         reset cfun before dumping the functions.
13399 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
13401         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13403         PR target/56866
13404         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13405         use xop_pmacsdqh if uns_p.
13406         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13407         the immediate rotate count.
13409 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
13411         * rtl.h (struct rtx_def): Add comment for field jump.
13412         (LRA_SUBREG_P): New macro.
13413         * recog.c (register_operand): Check LRA_SUBREG_P.
13414         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13415         stack frame.
13416         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13417         (lra_final_code_change): Skip subreg change for operators.
13418         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13419         if there are no operand changes.
13420         * lra-constraints.c (curr_insn_set): New.
13421         (match_reload): Set LRA_SUBREG_P.
13422         (emit_spill_move): Ditto.
13423         (check_and_process_move): Use curr_insn_set. Process only single
13424         set insns.  Don't initialize sec_mem_p and change_p.
13425         (simplify_operand_subreg): Use LRA_SUBREG_P.
13426         (reg_in_class_p): New function.
13427         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
13428         of #ifdef.  Add code to remove cycling.
13429         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13430         non-null disp.  Reload inner instead of disp when base and index
13431         are null.  Try to put lo_sum into register.
13432         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13433         (check_and_process_move): Move code for move cost check to
13434         simple_move_p.  Remove equiv_substitution.
13435         (simple_move_p): New function.
13436         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
13437         curr_insn_set.  Call check_and_process_move only for single set
13438         insns.  Use the new function.  Move call of check_and_process_move
13439         after operand equiv substitution and address process.
13441 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13443         PR go/57045
13444         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13445         with nonlocal goto receivers or returns twice calls, ignore
13446         unininitialized values from abnormal edges to nl goto receiver
13447         or returns twice call.
13449 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13451         PR tree-optimization/57051
13452         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13453         and VEC_RSHIFT_EXPR if shift count is a multiple of element
13454         bitsize.
13456 2013-04-26  Richard Biener  <rguenther@suse.de>
13458         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13459         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
13460         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13461         (fixup_loop_arrays_after_move): New function.
13462         (move_sese_region_to_fn): Properly outline the loop tree parts
13463         of the SESE region.
13465 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
13467         * config/i386/i386.md (type, unit): Fix long lines.
13469 2013-04-26  Richard Biener  <rguenther@suse.de>
13471         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13472         (lto-streamer-out.o): Likewise.
13473         * cfgloop.c (init_loops_structure): Export, add struct function
13474         argument and adjust.
13475         (flow_loops_find): Adjust.
13476         * cfgloop.h (enum loop_estimation): Add EST_LAST.
13477         (init_loops_structure): Declare.
13478         * lto-streamer-in.c: Include cfgloop.h.
13479         (input_cfg): Input the loop tree.
13480         * lto-streamer-out.c: Include cfgloop.h.
13481         (output_cfg): Output the loop tree.
13482         (output_struct_function_base): Do not drop PROP_loops.
13484 2013-03-26  Richard Biener  <rguenther@suse.de>
13486         * tree-cfg.c (execute_build_cfg): Build the loop tree.
13487         (pass_build_cfg): Provide PROP_loops.
13488         (move_sese_region_to_fn): Remove loops that are outlined into fn
13489         for now.
13490         * tree-inline.c: Include cfgloop.h.
13491         (initialize_cfun): Do not drop PROP_loops.
13492         (copy_loops): New function.
13493         (copy_cfg_body): Copy loop structure.
13494         (tree_function_versioning): Initialize destination loop tree.
13495         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13496         (pass_parallelize_loops): Do IL verification.
13497         * loop-init.c (loop_optimizer_init): Fixup loops if required.
13498         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13499         the CFG make sure we fixup loops as well.
13500         * tree-ssa-tail-merge.c: Include cfgloop.h.
13501         (replace_block_by): When merging loop latches mark loops for fixup.
13502         * lto-streamer-out.c (output_struct_function_base): Drop
13503         PROP_loops for now.
13504         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13505         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13506         * ipa-split.c: Include cfgloop.h.
13507         (split_function): Add the new return block to the loop tree root.
13508         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13509         whether we have removed the forwarder block.
13510         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13511         * cfgloop.h (place_new_loop): Declare.
13512         * cfgloopmanip.c (place_new_loop): Export.
13513         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13514         (tree-switch-conversion.o): Likewise.
13515         (tree-complex.o): Likewise.
13516         (tree-inline.o): Likewise.
13517         (tree-ssa-tailmerge.o): Likewise.
13518         (ipa-split.o): Likewise.
13519         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13520         (tree-ssa-copy.o): Likewise.
13521         * tree-switch-conversion.c: Include cfgloop.h
13522         (process_switch): If we emit a bit-test cascade, schedule loops
13523         for fixup.
13524         * tree-complex.c: Include cfgloop.h.
13525         (expand_complex_div_wide): Properly add new basic-blocks to loops.
13526         * asan.c: Include cfgloop.h.
13527         (create_cond_insert_point): Properly add new basic-blocks to
13528         loops, schedule loop fixup.
13529         * cfgloop.c (verify_loop_structure): Check that looks are not
13530         marked for fixup.
13531         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
13532         to loops.
13533         (expand_omp_for_generic): Likewise.
13534         (expand_omp_sections): Likewise.
13535         (expand_omp_atomic_pipeline): Schedule loops for fixup.
13536         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
13537         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
13538         is initialized, not when loops are present.
13539         * tree-parloops.c (parallelize_loops): Remove checking here.
13540         * passes.c (init_optimization_passes): Schedule a copy-propagation
13541         pass before complete unrolling of inner loops.
13543 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13545         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
13546         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
13547         (colorize_init): Add argument to _WIN32 version.
13548         * toplev.c: Include diagnostic-color.h.
13549         (process_options): Default to -fdiagnostics-color=auto if
13550         GCC_COLORS env var is in the environment.
13551         * common.opt (fdiagnostics-color=): Add Var and Init.
13552         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
13553         env var is in the environment, the default is auto rather than never.
13555         * diagnostic.h (file_name_as_prefix): Add context argument.
13556         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
13557         the string as locus.
13558         * langhooks.c (lhd_print_error_function): Adjust caller.
13560 2013-04-25  Lawrence Crowl  <crowl@google.com>
13562         * var-tracking.c (shared_hash_def::htab):
13563         Change type to hash_table.  Update dependent calls and types.
13565 2013-04-25  Lawrence Crowl  <crowl@google.com>
13567         * Makefile.in: Update as needed below.
13569         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
13570         Move declaration to after the type's method definitons.
13572         * attribs.c (htab_t scoped_attributes::attribute_hash):
13573         Change type to hash_table.  Update dependent calls and types.
13575         * bitmap.c (htab_t bitmap_desc_hash):
13576         Change type to hash_table.  Update dependent calls and types.
13578         * cselib.c (htab_t cselib_hash_table):
13579         Change type to hash_table.  Update dependent calls and types.
13581         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
13582         (hash_string_slot_node): Move implementation into lto-streamer.h
13583         struct string_slot_hasher.
13584         (eq_string_slot_node): Likewise.
13586         * data-streamer-out.c: Update output_block::string_hash_table
13587         dependent calls and types.
13589         * dwarf2cfi.c (htab_t trace_index):
13590         Change type to hash_table.  Update dependent calls and types.
13592         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
13593         Change type to hash_table.  Update dependent calls and types.
13594         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
13595         (htab_t optimize_external_refs::map): Likewise.
13596         (htab_t output_comp_unit::extern_map): Likewise.
13597         (htab_t output_comdat_type_unit::extern_map): Likewise.
13598         (htab_t output_macinfo::macinfo_htab): Likewise.
13599         (htab_t optimize_location_lists::htab): Likewise.
13600         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
13602         * except.c (htab_t ehspec_hash_type):
13603         Change type to hash_table.  Update dependent calls and types.
13604         (assign_filter_values::ttypes): Likewise.
13605         (assign_filter_values::ehspec): Likewise.
13606         (sjlj_assign_call_site_values::ar_hash): Likewise.
13607         (convert_to_eh_region_ranges::ar_hash): Likewise.
13609         * gcse.c (htab_t pre_ldst_table):
13610         Change type to hash_table.  Update dependent calls and types.
13612         * ggc-common.c (htab_t saving_htab):
13613         Change type to hash_table.  Update dependent calls and types.
13614         (htab_t loc_hash): Likewise.
13615         (htab_t ptr_hash): Likewise.
13616         (call_count): Rename ggc_call_count.
13617         (call_alloc): Rename ggc_call_alloc.
13618         (loc_descriptor): Rename make_loc_descriptor.
13619         (add_statistics): Rename ggc_add_statistics.
13621         * ggc-common.c (saving_htab):
13622         Change type to hash_table.  Update dependent calls and types.
13624         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
13625         (push_gimplify_context): Likewise.
13626         (pop_gimplify_context): Likewise.
13627         (struct gimple_temp_hash_elt): Added.
13628         (struct gimplify_hasher): Likewise.
13629         (struct gimplify_ctx.temp_htab):
13630         Change type to hash_table.  Update dependent calls and types.
13632         * gimple-fold.c: Include gimplify-ctx.h.
13634         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
13635         Change type to hash_table.  Update dependent calls and types.
13636         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
13637         avoid potential global name collision.
13639         * gimplify.c: Include gimplify-ctx.h.
13640         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
13641         (htab_t gimplify_ctx::temp_htab):
13642         Update dependent calls and types for new type hash_table.
13643         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
13644         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
13646         * gimplify-ctx.h: New.
13647         (struct gimple_temp_hash_elt): Move from gimplify.c.
13648         (class gimplify_hasher): New.
13649         (struct gimplify_ctx): Move from gimple.h.
13650         (htab_t gimplify_ctx::temp_htab):
13651         Change type to hash_table.  Update dependent calls and types.
13653         * graphite-clast-to-gimple.c: Include graphite-htab.h.
13654         (htab_t ivs_params::newivs_index):
13655         Change type to hash_table.  Update dependent calls and types.
13656         (htab_t ivs_params::params_index): Likewise.
13657         (htab_t print_generated_program::params_index): Likewise.
13658         (htab_t gloog::newivs_index): Likewise.
13659         (htab_t gloog::params_index): Likewise.
13661         * graphite.c: Include graphite-htab.h.
13662         4htab_t graphite_transform_loops::bb_pbb_mapping):
13663         Change type to hash_table.  Update dependent calls and types.
13665         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
13666         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
13667         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
13669         * graphite-dependences.c: Include graphite-htab.h.
13670         (loop_is_parallel_p): Change hash table type of parameter.
13672         * graphite-htab.h: New.
13673         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
13674         (extern find_pbb_via_hash): Move from graphite-poly.h.
13675         (extern loop_is_parallel_p): Move from graphite-poly.h.
13676         (extern get_loop_body_pbbs): Move from graphite-poly.h.
13678         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
13679         (extern loop_is_parallel_p): Move to graphite-htab.h.
13680         (extern get_loop_body_pbbs): Move to graphite-htab.h.
13682         * haifa-sched.c (htab_t delay_htab):
13683         Change type to hash_table.  Update dependent calls and types.
13684         (htab_t delay_htab_i2): Likewise.
13686         * ira-color.c (htab_t allocno_hard_regs_htab):
13687         Change type to hash_table.  Update dependent calls and types.
13689         * ira-costs.c (htab_t cost_classes_htab):
13690         Change type to hash_table.  Update dependent calls and types.
13692         * loop-invariant.c (htab_t merge_identical_invariants::eq):
13693         Change type to hash_table.  Update dependent calls and types.
13695         * loop-iv.c (htab_t bivs):
13696         Change type to hash_table.  Update dependent calls and types.
13698         * loop-unroll.c (htab_t opt_info::insns_to_split):
13699         Change type to hash_table.  Update dependent calls and types.
13700         (htab_t opt_info::insns_with_var_to_expand): Likewise.
13702         * lto-streamer.h (struct string_slot): Move from data-streamer.h
13703         (struct string_slot_hasher): New.
13704         (htab_t output_block::string_hash_table):
13705         Change type to hash_table.  Update dependent calls and types.
13707         * lto-streamer-in.c (freeing_string_slot_hasher): New.
13708         (htab_t file_name_hash_table):
13709         Change type to hash_table.  Update dependent calls and types.
13711         * lto-streamer-out.c: Update output_block::string_hash_table dependent
13712         calls and types.
13714         * lto-streamer.c (htab_t tree_htab):
13715         Change type to hash_table.  Update dependent calls and types.
13717         * omp-low.c: Include gimplify-ctx.h.
13719         * passes.c (htab_t name_to_pass_map):
13720         Change type to hash_table.  Update dependent calls and types.
13721         (pass_traverse): Rename to passes_pass_traverse.
13723         * plugin.c (htab_t event_tab):
13724         Change type to hash_table.  Update dependent calls and types.
13726         * postreload-gcse.c (htab_t expr_table):
13727         Change type to hash_table.  Update dependent calls and types.
13728         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
13730         * sese.c (debug_rename_map_1): Make extern.
13731         (htab_t copy_bb_and_scalar_dependences::rename_map):
13732         Change type to hash_table.  Update dependent calls and types.
13734         * sese.h (extern debug_rename_map): Move to .c file.
13736         * store-motion.c (htab_t store_motion_mems_table):
13737         Change type to hash_table.  Update dependent calls and types.
13739         * trans-mem.c (htab_t tm_new_mem_hash):
13740         Change type to hash_table.  Update dependent calls and types.
13742         * tree-browser.c (htab_t TB_up_ht):
13743         Change type to hash_table.  Update dependent calls and types.
13745         * tree-cfg.c (htab_t discriminator_per_locus):
13746         Change type to hash_table.  Update dependent calls and types.
13748         * tree-complex.c: Include tree-hasher.h
13749         (htab_t complex_variable_components):
13750         Change type to hash_table.  Update dependent calls and types.
13752         * tree-eh.c (htab_t finally_tree):
13753         Change type to hash_table.  Update dependent calls and types.
13755         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
13756         struct int_tree_hasher.
13757         (extern int_tree_map_eq): Likewise.
13758         (uid_decl_map_hash): Removed.
13759         (extern decl_tree_map_eq): Likewise.
13761         * tree-hasher.h: New.
13762         (struct int_tree_hasher): New.
13763         (typedef int_tree_htab_type): New.
13765         * tree-inline.c: Include gimplify-ctx.h.
13767         * tree-mudflap.c: Include gimplify-ctx.h.
13769         * tree-parloops.c: Include tree-hasher.h.
13770         (htab_t eliminate_local_variables_stmt::decl_address):
13771         Change type to hash_table.  Update dependent calls and types.
13772         (htab_t separate_decls_in_region::decl_copies): Likewise.
13774         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
13775         Change type to hash_table.  Update dependent calls and types.
13777         * tree-sra.c (candidates):
13778         Change type to hash_table.  Update dependent calls and types.
13780         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
13781         in tree-flow.h.
13782         (int_tree_map_hash): Likewise.
13784         * tree-ssa-dom.c (htab_t avail_exprs):
13785         Change type to hash_table.  Update dependent calls and types.
13787         * tree-ssa-live.c (var_map_base_init::tree_to_index):
13788         Change type to hash_table.  Update dependent calls and types.
13790         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
13791         Change type to hash_table.  Update dependent calls and types.
13793         * tree-ssa-phiopt.c (seen_ssa_names):
13794         Change type to hash_table.  Update dependent calls and types.
13796         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
13797         Change type to hash_table.  Update dependent calls and types.
13799         * tree-ssa-uncprop.c (equiv):
13800         Change type to hash_table.  Update dependent calls and types.
13802 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
13804         PR rtl-optimization/57003
13805         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
13806         call note_stores with kill_clobbered_value callback again after
13807         killing regs_invalidated_by_call.
13809 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13811         * config/aarch64/aarch64-simd.md
13812         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
13813         (aarch64_simd_bsl<mode>): Likewise.
13814         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
13816 2013-04-25  Marek Polacek  <polacek@redhat.com>
13818         PR tree-optimization/57066
13819         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
13821 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
13823         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
13825 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13827         * config/aarch64/aarch64-builtins.c
13828         (aarch64_fold_builtin): New.
13829         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
13830         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
13831         * config/aarch64/aarch64-simd-builtins.def (abs): New.
13832         * config/aarch64/arm_neon.h
13833         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
13835 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13836             Tejas Belagod  <tejas.belagod@arm.com>
13838         * config/aarch64/aarch64-builtins.c
13839         (aarch64_gimple_fold_builtin): New.
13840         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
13841         * config/aarch64/aarch64-simd-builtins.def (addv): New.
13842         * config/aarch64/aarch64-simd.md (addpv4sf): New.
13843         (addvv4sf): Update.
13844         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
13846 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13848         * config/aarch64/aarch64.md
13849         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
13851 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13853         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
13854         (*ngcsi_uxtw): New pattern.
13856 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13857             Julian Brown  <julian@codesourcery.com>
13859         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
13860         (TB_DREG): Add T_V4HF.
13861         (v4hf_UP): New macro.
13862         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13863         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13864         Handle initialisation of V4HF. Adjust initialisation of reinterpret
13865         built-ins.
13866         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13867         (arm_vector_mode_supported_p): Handle V4HF.
13868         (arm_mangle_map): Handle V4HFmode.
13869         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
13870         * config/arm/arm_neon_builtins.def: Add entries for
13871         vcvtv4hfv4sf, vcvtv4sfv4hf.
13872         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
13873         (neon_vcvtv4hfv4sf): Likewise.
13874         * config/arm/neon-gen.ml: Handle half-precision floating point
13875         features.
13876         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
13877         * config/arm/arm_neon.h: Regenerate.
13878         * config/arm/neon.ml (type elts): Add F16.
13879         (type vectype): Add T_float16x4, T_floatHF.
13880         (type vecmode): Add V4HF.
13881         (type features): Add Requires_FP_bit feature.
13882         (elt_width): Handle F16.
13883         (elt_class): Likewise.
13884         (elt_of_class_width): Likewise.
13885         (mode_of_elt): Refactor.
13886         (type_for_elt): Handle F16, fix error messages.
13887         (vectype_size): Handle T_float16x4.
13888         (vcvt_sh): New function.
13889         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
13890         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
13891         (string_of_mode): Handle V4HF.
13892         * doc/arm-neon-intrinsics.texi: Regenerate.
13894 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13896         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
13897         format specifier in 'X' case.
13899 2013-04-25  Alan Modra  <amodra@gmail.com>
13901         PR target/57052
13902         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
13903         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
13904         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
13905         Repeat for many other rotate/shift and mask patterns using subregs.
13906         Name lshiftrt insns.
13907         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
13908         on WORDS_BIG_ENDIAN.
13910 2013-04-25  Alan Modra  <amodra@gmail.com>
13912         * config.gcc: Support little-endian powerpc-linux targets.
13913         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
13914         (LINK_OS_LINUX_SPEC): Define.
13915         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
13916         Preserve MASK_LITTLE_ENDIAN.
13917         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
13918         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
13919         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
13920         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
13921         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
13922         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
13923         Correct fp word order for little-endian.  Don't shift toc entries
13924         smaller than a word for little-endian.
13925         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
13926         (bswapdi2 splits): Correct low-part subreg for little-endian.
13927         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
13928         low/high where such is correct only for be.
13929         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
13930         little-endian for -mcall-aixdesc.
13932 2013-04-25  Alan Modra  <amodra@gmail.com>
13934         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
13935         replace_equiv_address_nv.
13937 2013-04-25  Alan Modra  <amodra@gmail.com>
13939         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
13941 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13943         Revert:
13944         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13945         * rtl.h (struct rtx_def): ...
13947 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13949         PR rtl-optimizations/57046
13950         * lra-constraints (split_reg): Set up lra_risky_transformations_p
13951         for multi-reg splits.
13953 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
13955         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
13957 2013-04-24  Sterling Augustine  <saugustine@google.com>
13959         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
13960         (comp_dir_string, debug_str_dwo_section): New.
13961         (DEBUG_STR_DWO_SECTION): Rename to ...
13962         (DEBUG_DWO_STR_SECTION): ... this.
13963         (DEBUG_NORM_STR_SECTION): Delete.
13964         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
13965         (DEBUG_STR_DWO_SECTION_FLAGS): New.
13966         (find_AT_string): Move most logic to ...
13967         (find_AT_string_in_table): ... here.  New.
13968         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
13969         add_skeleton_AT_string.  Delete logic.
13970         (output_skeleton_debug_sections): Remove call to
13971         add_top_level_skeleton_die_attrs.
13972         (add_comp_dir_attribute): Move logic to comp_dir_string.
13973         (dwarf2out_init): Initialize debug_str_dwo_section.
13974         (output_indirect_string): Call find_string_form.
13975         (output_indirect_strings): Rewrite.
13976         (prune_unused_types): Empty skeleton_debug_str_hash.
13977         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
13978         (dwarf2out_finish):  Call output_indirect_strings.
13980 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
13982         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
13984 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13986         * rtl.h (struct rtx_def): Add comment for field jump.
13987         (LRA_SUBREG_P): New macro.
13988         * recog.c (register_operand): Check LRA_SUBREG_P.
13989         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13990         stack frame.
13991         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13992         (lra_final_code_change): Skip subreg change for operators.
13993         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13994         if there are no operand changes.
13995         * lra-constraints.c (curr_insn_set): New.
13996         (match_reload): Set LRA_SUBREG_P.
13997         (emit_spill_move): Ditto.
13998         (check_and_process_move): Use curr_insn_set. Process only single
13999         set insns.  Don't initialize sec_mem_p and change_p.
14000         (simplify_operand_subreg): Use LRA_SUBREG_P.
14001         (reg_in_class_p): New function.
14002         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14003         of #ifdef.  Add code to remove cycling.
14004         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14005         non-null disp.  Reload inner instead of disp when base and index
14006         are null.  Try to put lo_sum into register.
14007         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14008         (check_and_process_move): Move code for move cost check to
14009         simple_move_p.  Remove equiv_substitution.
14010         (simple_move_p): New function.
14011         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14012         curr_insn_set.  Call check_and_process_move only for single set
14013         insns.  Use the new function.  Move call of check_and_process_move
14014         after operand equiv substitution and address process.
14016 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14018         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14019         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14020         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14022 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14024         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14026 2013-04-24  Marek Polacek  <polacek@redhat.com>
14028         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14029         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14030         (select_loops_exit_conditions): Likewise.
14031         (number_of_iterations_for_all_loops): Likewise.
14032         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14033         (scev_analysis): Likewise.
14035 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14036             Chao-ying Fu  <fu@mips.com>
14038         * config/mips/micromips.md (jraddiusp): New pattern.
14039         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14040         instruction if possible.
14042 2013-04-24  Alan Modra  <amodra@gmail.com>
14044         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14046 2013-04-24  Julian Brown  <julian@codesourcery.com>
14047             Chung-Lin Tang  <cltang@codesourcery.com>
14049         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14050         dependency behavior in enumeration type DIE generation. Add TODO note
14051         to comments about future DW_FORM_sdata/udata re-work of related code.
14053 2013-04-23  Lawrence Crowl  <crowl@google.com>
14055         * Makefile.in: Update as needed below.
14057         * hash-table.h (class hash_table):
14058         Correct many methods with parameter types compare_type to the correct
14059         value_type.  (Correct code was unlikely to notice the change.)
14060         (hash_table::elements_with_deleted) New.
14061         (class hashtable::iterator): New.
14062         (hashtable::begin()): New.
14063         (hashtable::end()): New.
14064         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14066         * statistics.c (statistics_hashes):
14067         Change type to hash_table.  Update dependent calls and types.
14069         * tree-into-ssa.c (var_infos):
14070         Change type to hash_table.  Update dependent calls and types.
14072         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14073         Change type to hash_table.  Update dependent calls and types.
14075         * tree-ssa-loop-im.c (struct mem_ref.refs):
14076         Change type to hash_table.  Update dependent calls and types.
14078         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14079         Change type to hash_table.  Update dependent calls and types.
14081         * tree-ssa-sccvn.c (vn_tables_s::nary):
14082         Change type to hash_table.  Update dependent calls and types.
14083         (vn_tables_s::phis): Likewise.
14084         (vn_tables_s::references): Likewise.
14086         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14087         (vn_reference_eq): Update parameter and return types.
14089         * tree-ssa-structalias.c (pointer_equiv_class_table):
14090         Change type to hash_table.  Update dependent calls and types.
14091         (location_equiv_class_table): Likewise.
14093         * tree-vect-data-refs.c: Consequential changes for making
14094         peeling a hash_table.
14096         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14097         (destroy_loop_vec_info): Dependent hash_table update.
14099         * tree-vectorizer.h (peeling_htab):
14100         Change type to hash_table.  Update dependent calls and types.
14102 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14104         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14105         to check the register content is equal or not.
14106         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14107         to assign register content record.
14108         * lra-eliminations.c (update_reg_eliminate): Use
14109         lra_update_reg_val_offset to update register content offset.
14110         * lra-int.h (struct lra_reg): Add offset member.
14111         (lra_reg_val_equal_p): New static inline function.
14112         (lra_update_reg_val_offset): New static inline function.
14113         (lra_assign_reg_val): New static inline function.
14114         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14115         to assign register content record.
14116         (initialize_lra_reg_info_element): Initial offset to zero.
14118 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14120         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14121         operands.  Record compression.
14123 2013-04-23  Xinliang David Li  <davidxl@google.com>
14125         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14127 2013-04-23  Richard Biener  <rguenther@suse.de>
14129         PR middle-end/57036
14130         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14131         parameter, only add abnormal goto edges from the copied body
14132         if the call could perform abnormal gotos.
14133         (copy_cfg_body): Adjust.
14135 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14137         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14139 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14141         * coretypes.h (gimple_stmt_iterator): Add struct to make
14142         compatible with C.
14144 2013-04-23  Richard Biener  <rguenther@suse.de>
14146         PR tree-optimization/57026
14147         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14148         from SSA names occuring in abnormal PHI nodes.
14150 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14152         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14154 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14156         * lto/lto.c (print_lto_report_1): Declare early.
14157         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14159 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14161         * common.opt (-flto-report-wpa): Add.
14162         * doc/invoke.texi (-flto-report-wpa): Add.
14163         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14164         (lto_main): dito.
14166 2013-04-22  Xinliang David Li  <davidxl@google.com>
14168         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14169         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14170         * Makefile.in: New dependency
14172             David Daney <ddaney.cavm@gmail.com>
14174         * configure.ac (gcc_cv_as_micromips_support):  Use the
14175         --fatal-warnings option.
14176         * configure: Regenerate.
14178 2013-04-22  Marek Polacek  <polacek@redhat.com>
14180         PR sanitizer/56990
14181         * tsan.c (instrument_expr): Don't instrument expression
14182         in case its size is zero.
14184 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14186         PR target/57032
14187         Revert:
14188         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14190         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14192 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14194         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14195         (gimple_stmt_iterator): New typedef.
14196         * gimple.h (gimple_stmt_iterator): Rename to...
14197         (gimple_stmt_iterator_d): ... This.
14198         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14199         trees be valid for GIMPLE and GENERIC.
14200         (TARGET_GIMPLE_FOLD_BUILTIN): New.
14201         * gimple-fold.c (gimple_fold_call): Call target hook
14202         gimple_fold_builtin.
14203         * hooks.c (hook_bool_gsiptr_false): New.
14204         * hooks.h (hook_bool_gsiptr_false): New.
14205         * target.def (fold_stmt): New.
14206         * doc/tm.texi: Regenerate.
14208 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
14210         PR target/57018
14211         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14212         a set sp if no stack realignment.
14214 2013-04-22  Nick Clifton  <nickc@redhat.com>
14216         * config.gcc (tilegx-linux): Extend extra_objs rather than
14217         overwriting it.
14218         (tilepro-linux): Likewise.
14220 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14222         * config/aarch64/aarch64-builtins.c
14223         (CF): Remove.
14224         (CF0, CF1, CF2, CF3, CF4, CF10): New.
14225         (VAR<1-12>): Add MAP parameter.
14226         (BUILTIN_*): Likewise.
14227         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14228         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14229         (aarch64_ushl_n<mode>): Likewise.
14230         (aarch64_sshr_n<mode>): Likewise.
14231         (aarch64_ushr_n<mode>): Likewise.
14232         (aarch64_<maxmin><mode>): Likewise.
14233         (aarch64_sqrt<mode>): Likewise.
14234         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14235         (vshr<q>_n_*): Likewise.
14237 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14239         * config/aarch64/aarch64-builtins.c
14240         (aarch64_simd_builtin_type_mode): Handle SF types.
14241         (sf_UP): Define.
14242         (BUILTIN_GPF): Define.
14243         (aarch64_init_simd_builtins): Handle SF types.
14244         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14245         (frecps): Likewise.
14246         (frecpx): Likewise.
14247         * config/aarch64/aarch64-simd.md
14248         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14249         (aarch64_frecpe<mode>): New.
14250         (aarch64_frecps<mode>): Likewise.
14251         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14252         (v8type): Add frecp<esx>.
14253         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14254         (aarch64_frecps<mode>): Likewise.
14255         * config/aarch64/iterators.md (FRECP): New.
14256         (frecp_suffix): Likewise.
14257         * config/aarch64/arm_neon.h
14258         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14260 2013-04-22  Christian Bruel  <christian.bruel@st.com>
14262         PR target/56995
14263         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14264         (REG_CLASS_NAMES): Idem.
14265         (REG_CLASS_CONTENTS): Idem.
14266         (REGCLASS_HAS_FP_REG): Idem.
14267         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14268         (sh_conditional_register_usage): Idem.
14270 2013-04-21  Jeff Law  <law@redhat.com>
14272         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14273         (ssa_forward_propagate_and_combine): Use it.
14275 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14277         * lra.c: Update the flow chart diagram.
14279 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14281         PR rtl-optimization/56847
14282         * lra-constraints.c (process_alt_operands): Discourage alternative
14283         with non-matche doffsettable memory constraint fro memory with
14284         known offset.
14286 2013-04-19  Richard Biener  <rguenther@suse.de>
14288         PR tree-optimization/56982
14289         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14290         function.
14291         * gimplify.c (gimplify_call_expr): Notice special calls.
14292         (gimplify_modify_expr): Likewise.
14293         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14294         abnormal control flow receivers.
14295         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14296         in the same way as cfun->has_nonlocal_labels.
14297         (gimple_purge_dead_abnormal_call_edges): Likewise.
14298         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14299         receivers start a basic-block.
14301 2013-04-19  Richard Biener  <rguenther@suse.de>
14303         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14304         member ...
14305         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
14306         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14307         (SLP_TREE_LOAD_PERMUTATION): Add.
14308         (vect_transform_slp_perm_load): Adjust prototype.
14309         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14310         (vect_free_slp_instance): Likewise.
14311         (vect_create_new_slp_node): Likewise.
14312         (vect_supported_slp_permutation_p): Remove.
14313         (vect_slp_rearrange_stmts): Adjust.
14314         (vect_supported_load_permutation_p): Likewise.  Inline
14315         vect_supported_slp_permutation_p here.
14316         (vect_analyze_slp_instance): Compute load permutations per
14317         slp node instead of per instance.
14318         (vect_get_slp_defs): Adjust.
14319         (vect_transform_slp_perm_load): Likewise.
14320         (vect_schedule_slp_instance): Remove redundant code.
14321         (vect_schedule_slp): Remove hack for PR56270, add it ...
14322         * tree-vect-stmts.c (vectorizable_load): ... here, do not
14323         CSE loads for SLP.  Adjust.
14325 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14327         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14328         spelling in two comments.
14330 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14332         PR target/56797
14333         * config/arm/arm.c (load_multiple_sequence): Require SP
14334         as base register for loads if SP is in the register list.
14336 2013-04-19  Martin Jambor  <mjambor@suse.cz>
14338         PR tree-optimization/56718
14339         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14340         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14341         and made public.  Adjusted all callers.
14342         (ipa_intraprocedural_devirtualization): New function.
14343         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14344         (ipa_intraprocedural_devirtualization): Likewise.
14345         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14347 2013-04-19  Richard Biener  <rguenther@suse.de>
14349         PR tree-optimization/57000
14350         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14352 2013-04-19  Terry Guo  <terry.guo@arm.com>
14354         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14355         Replace with ...
14356         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
14357         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14358         (cortex_m4_fmacs): Use new reservations.
14359         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14361 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
14363         PR rtl-optimization/56999
14364         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14365         related code.
14366         (lra_coalesce): Remove split_origin_bitmap and related code.
14367         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
14368         ranges if necessary.
14370 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
14372         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14373         New array.
14374         (ix86_expand_call): Remove clobbered_registers array and use
14375         x86_64_ms_sysv_extra_clobbered_registers instead.
14376         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14377         Declare here.
14378         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14379         predicate.
14380         * config/i386/i386.md (*call_rex64_ms_sysv): Use
14381         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
14382         (*call_value_rex64_ms_sysv): Ditto.
14384 2013-04-18  Cary Coutant  <ccoutant@google.com>
14386         * dwarf2out.c (output_pubnames): Check die_perennial_p of
14387         parent instead of die_mark.
14389 2013-04-18  Diego Novillo  <dnovillo@google.com>
14391         * gimple.c (create_gimple_tmp): New.
14392         (get_expr_type): New.
14393         (build_assign): New.
14394         (build_type_cast): New.
14395         * gimple.h (enum ssa_mode): Define.
14396         (gimple_seq_set_location): New.
14397         * asan.c (build_check_stmt): Change some gimple_build_* calls
14398         to use build_assign and build_type_cast.
14400 2013-04-18  Richard Biener  <rguenther@suse.de>
14402         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14403         handle negative step.  Remove redundant checks.
14404         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14405         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14406         for negative step and grouped loads fail to vectorize.
14408 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14410         * emit-rtl.c (reset_insn_used_flags): New function.
14411         (reset_all_used_flags): Use it.
14412         (verify_insn_sharing): New function.
14413         (verify_rtl_sharing): Fix verification for SEQUENCEs.
14415 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
14417         PR tree-optimization/56984
14418         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14419         and (x >> M) >= N don't register any assertion if N << M is the
14420         minimum value.
14422 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14424         * lower-subreg.c (resolve_simple_move): If called self-recursive,
14425         do not delete_insn insns that have not yet been emitted, only
14426         unlink them with remove_insn.
14427         * df-scan.c (df_insn_delete): Revert r197492.
14429 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
14431         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14432         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14434 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14436         * config/arm/arm.md (movsicc_insn): Convert define_insn into
14437         define_insn_and_split.
14438         (and_scc,ior_scc,negscc): Likewise.
14439         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14441 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14443         * config/arm/arm.c (use_return_insn): Return 0 for targets that
14444         can benefit from using a sequence of LDRD instructions in epilogue
14445         instead of a single LDM instruction.
14447 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14449         PR 45688
14450         * doc/extend.texi: Fix typo.
14452 2013-04-17  Richard Biener  <rguenther@suse.de>
14454         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14455         (vect_build_slp_tree): ... here.
14456         (vect_build_slp_tree_1): Compute which stmts of the SLP group
14457         match.  Remove special-casing of mismatched complex loads.
14458         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14459         re-try the match with swapped commutative operands.
14460         (vect_supported_load_permutation_p): Remove special-casing of
14461         mismatched complex loads.
14462         (vect_analyze_slp_instance): Adjust.
14464 2013-04-17  Richard Biener  <rguenther@suse.de>
14466         PR rtl-optimization/56921
14467         * cfgloop.h (struct loop): Add simple_loop_desc member.
14468         (struct niter_desc): Mark with GTY(()).
14469         (simple_loop_desc): Do not use aux field but simple_loop_desc.
14470         * loop-iv.c (get_simple_loop_desc): Likewise.
14471         (free_simple_loop_desc): Likewise.
14473         Revert
14474         2013-04-16  Richard Biener  <rguenther@suse.de>
14476         PR rtl-optimization/56921
14477         * loop-init.c (pass_rtl_move_loop_invariants): Add
14478         TODO_do_not_ggc_collect to todo_flags_finish.
14479         (pass_rtl_unswitch): Same.
14480         (pass_rtl_unroll_and_peel_loops): Same.
14481         (pass_rtl_doloop): Same.
14483 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
14485         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14486         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14487         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14488         references.
14489         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14490         * tree-streamer.c (record_common_node): Adjust reference in comment.
14492 2013-04-17  Terry Guo  <terry.guo@arm.com>
14494         * config/arm/cortex-m4.md: Add a new bypass.
14496 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14498         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14499         New pattern.
14500         (*subs_<optab><mode>_multp2): New pattern.
14501         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14502         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14504 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14506         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14507         (*subs_mul_imm_<mode>): New pattern.
14509 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
14511         PR target/56948
14512         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14513         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
14514         (vsx_movti_32bit): Change j->wa to O->wa.
14516 2013-04-16  Richard Biener  <rguenther@suse.de>
14518         PR rtl-optimization/56921
14519         * loop-init.c (pass_rtl_move_loop_invariants): Add
14520         TODO_do_not_ggc_collect to todo_flags_finish.
14521         (pass_rtl_unswitch): Same.
14522         (pass_rtl_unroll_and_peel_loops): Same.
14523         (pass_rtl_doloop): Same.
14525 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
14527         * config/arm/arm.c (emit_multi_reg_push): New declaration
14528         for an existing function.
14529         (arm_emit_strd_push): New function.
14530         (arm_expand_prologue): Used here.
14531         (arm_emit_ldrd_pop): New function.
14532         (arm_expand_epilogue): Used here.
14533         (arm_get_frame_offsets): Update condition.
14534         (arm_emit_multi_reg_pop): Add a special case for load of a single
14535         register with writeback.
14537 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
14539         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
14540         description.
14542 2013-04-16  Richard Biener  <rguenther@suse.de>
14544         PR tree-optimization/56756
14545         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
14546         (first_mem_ref_loc): New.
14547         (execute_sm): Place the load temporarily before a previous
14548         access instead of in the latch edge to ensure its SSA dependencies
14549         are defined at points dominating the load.
14551 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
14553         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
14554         correct fix by moving header and footer insn to the footer of
14555         the merged basic block.  Clear BB_END of the merged-away block.
14557         PR middle-end/43631
14558         * emit-rtl.c (make_note_raw): New function.
14559         (link_insn_into_chain): New static inline function.
14560         (add_insn): Use it.
14561         (add_insn_before, add_insn_after): Factor insn chain linking code...
14562         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
14563         using link_insn_into_chain.
14564         (note_outside_basic_block_p): New helper function for emit_note_after
14565         and emit_note_before.
14566         (emit_note_after): Use nobb variant of add_insn_after if the note
14567         should not be contained in a basic block.
14568         (emit_note_before): Use nobb variant of add_insn_before if the note
14569         should not be contained in a basic block.
14570         (emit_note_copy): Use make_note_raw.
14571         (emit_note): Likewise.
14572         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
14573         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
14574         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
14575         the moved barrier the tail of the basic block it follows.
14576         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
14578 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
14580         PR tree-optimization/56962
14581         * gimple-ssa-strength-reduction.c (record_increment): Only set
14582         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
14583         either rhs1 or rhs2 is equal to c->base_expr.
14585 2013-04-15  Richard Biener  <rguenther@suse.de>
14587         PR tree-optimization/56933
14588         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
14589         member.
14590         (GROUP_READ_WRITE_DEPENDENCE): Remove.
14591         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
14592         * tree-vect-data-refs.c (vect_analyze_group_access): Move
14593         dependence check ...
14594         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
14595         ... here.
14596         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
14597         GROUP_READ_WRITE_DEPENDENCE.
14599 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14601         * emit-rtl.c (reset_all_used_flags): New function.
14602         (verify_rtl_sharing): Call reset_all_used_flags before and after
14603         performing the checks.
14605 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14607         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
14608         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
14609         * config/arm/constraints.md (De): New constraint.
14610         * config/arm/neon.md (anddi3_neon): Delete.
14611         (neon_vand<mode>): Expand to standard anddi3 pattern.
14612         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
14613         Move earlier in the file.
14614         (neon_inv_logic_op2): Likewise.
14615         (arm_anddi_operand_neon): New predicate.
14617 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14619         * configure.ac (gcc_cv_ld_as_needed): Set
14620         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
14621         Use -z ignore, -z record on *-*-solaris2*.
14622         (HAVE_LD_AS_NEEDED): Update comment.
14623         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
14624         * configure: Regenerate.
14625         * config.in: Regenerate.
14626         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
14627         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
14628         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
14629         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
14630         equivalents.  Fix markup.
14631         * doc/tm.texi: Regenerate.
14633 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
14635         * config/i386/i386.opt: New option mstack-protector-guard=.
14636         * config/i386/i386-opts.h: Add enum stack_protector_guard.
14637         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
14638         TARGET_SSP_TLS_GUARD.
14639         * config/i386/i386.c (ix86_option_override_internal): Set
14640         ix86_stack_protector_guard.
14641         * config/i386/i386.md (stack_protect_set): Enable for
14642         TARGET_SSP_TLS_GUARD only.
14643         (stack_protect_set_<mode>): Ditto.
14644         (stack_protect_test): Ditto.
14645         (stack_protect_test_<mode>): Ditto.
14646         * doc/invoke.texi (i386 Option): Document.
14648 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
14650         PR target/56890
14651         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
14652         (S_MODES): Set H_MODE bit.
14653         (SF_MODES): Set only S_MODE and SF_MODE bits.
14654         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
14655         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
14656         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
14657         <MODE_FLOAT>: Likewise.
14659 2013-04-15  Joey Ye  <joey.ye@arm.com>
14661         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
14663 2013-04-15  Joey Ye  <joey.ye@arm.com>
14665         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
14666         for real far jump.
14667         (thumb_far_jump_used_p): Count instruction size and set
14668         far_jump_used.
14670 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
14672         * reorg.c (fill_simple_delay_slots): Reindent block of code.
14673         * resource.c (mark_target_live_regs): Reformat conditional block.
14675 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14677         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
14678         notes, they are emitted only just before final.
14679         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
14681 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14683         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
14684         * cfgrtl.c (delete_insn): Call it here instead.
14685         * lra-spills.c (lra_final_code_change): Use delete_insn.
14686         * haifa-sched.c (sched_remove_insn): Likewise.
14687         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
14688         returning to the nop pool.
14689         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
14690         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
14692 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14694         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
14695         * doc/tm.texi: Regenerated.
14697 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
14699         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
14700         QImode checks.
14702 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14704         * df-core.c (df_find_def): Compare register numbers.
14705         (df_find_use): Likewise.
14707 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
14709         PR target/56903
14710         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
14711         lra_in_progress for return.
14713 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14715         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
14716         define_insn into define_insn_and_split and emit movsicc patterns.
14718 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14720         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
14722 2013-04-12  Richard Biener  <rguenther@suse.de>
14724         * tree-pass.h (TODO_do_not_ggc_collect): New.
14725         * passes.c (execute_one_ipa_transform_pass): Honor
14726         TODO_do_not_ggc_collect.
14727         (execute_one_pass): Likewise.
14729         Revert
14730         2013-04-10  Richard Biener  <rguenther@suse.de>
14732         * passes.c (init_optimization_passes): Remove reload pass.
14733         * ira.c (do_reload): Merge into ...
14734         (ira): ... this.
14735         (rest_of_handle_reload): Remove.
14736         (pass_reload): Likewise.
14737         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14738         of reload for vzeroupper pass placement.
14740 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
14742         PR tree-optimization/56918
14743         PR tree-optimization/56920
14744         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
14745         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
14746         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
14747         use wide_mul_with_sign method.
14749 2013-04-12  Richard Biener  <rguenther@suse.de>
14751         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
14752         not be considered a gimple constant.
14754 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
14756         * fold-const.c (const_binop): Handle vector shifts by a scalar.
14757         (fold_binary_loc): Call const_binop also for mixed vector-scalar
14758         operations.
14760 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14761             Jakub Jelinek  <jakub@redhat.com>
14763         * opts.c: Include diagnostic-color.h.
14764         (common_handle_option): Handle OPT_fdiagnostics_color_.
14765         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
14766         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
14767         (diagnostic-color.o): New.
14768         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
14769         (diagnostic_color_rule): New enum.
14770         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
14771         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
14772         the location string.
14773         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
14774         either NULL, or color kind.
14775         * diagnostic-color.c: New file.
14776         * diagnostic-color.h: New file.
14777         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
14778         arguments.
14779         * doc/invoke.texi (-fdiagnostics-color): Document.
14780         * pretty-print.h (pp_show_color): Define.
14781         (struct pretty_print_info): Add show_color field.
14782         * diagnostic.c: Include diagnostic-color.h.
14783         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
14784         macros.  Colorize error:, warning: etc. strings and also the location
14785         string.
14786         (diagnostic_show_locus): Colorize the caret line.
14787         * pretty-print.c: Include diagnostic-color.h.
14788         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
14789         inside of %< %> quotes or quoted through q format modifier.
14791 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14793         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
14795 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14797         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
14798         code in CC_NZ mode.
14799         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
14800         pattern.
14802 2013-04-11  Marek Polacek  <polacek@redhat.com>
14804         PR tree-optimization/48184
14805         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
14807 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
14809         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
14810         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
14811         (skip_simple_arithmetic): Tidy up.
14812         * tree.h (skip_simple_constant_arithmetic): Declare.
14814 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14816         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
14818 2013-04-11  Richard Biener  <rguenther@suse.de>
14820         * tree-vect-loop.c (get_initial_def_for_induction): Properly
14821         generate vector constants.
14823 2013-04-11  Richard Biener  <rguenther@suse.de>
14825         PR tree-optimization/56878
14826         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
14827         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
14828         New function.
14829         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14830         Prefer to align the DR with the most invariant base address.
14832 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14834         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
14835         comment.
14837 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
14839         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
14840         floating-point vector comparisons against 0.
14842 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
14844         PR tree-optimization/56899
14845         * fold-const.c (extract_muldiv_1): Apply distributive law
14846         only if TYPE_OVERFLOW_WRAPS (ctype).
14848 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
14850         PR target/56124
14851         * ira-costs.c (scan_one_insn): Check whether the source rtx of
14852         loading has side effect.
14854 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
14856         * config/sparc/sparc.c: Include tree-pass.h.
14857         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
14858         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
14859         head of file.  Change return type.  Split off gate function.
14860         (sparc_gate_work_around_errata): New function.
14861         (pass_work_around_errata): New pass definition.
14862         (insert_pass_work_around_errata) New pass insert definition to
14863         insert pass_work_around_errata just after delayed-branch scheduling.
14864         (sparc_option_override): Insert the pass.
14865         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
14867 2013-04-10  David S. Miller  <davem@davemloft.net>
14869         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
14870         or -mcpu=hypersparc.
14872         * target.def (cstore_mode): New hook.
14873         * target.h: Include insn-codes.h
14874         * targhooks.c: Likewise.
14875         (default_cstore_mode): New function.
14876         * targhooks.h: Declare it.
14877         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
14878         * doc/tm.texi: Rebuild.
14879         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
14880         target hook, rather than inspecting the insn_data.
14881         * config/sparc/sparc.c (sparc_cstore_mode): New function.
14882         (TARGET_CSTORE_MODE): Redefine.
14883         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
14884         result patterns.
14885         * config/sparc/predicates.md (cstore_result_operand): New special
14886         predicate.
14887         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
14888         Use it for operand 0.
14889         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
14890         (*snesi_special): Likewise.
14891         (*snesi_zero): Likewise.
14892         (*seqsi_zero): Likewise.
14893         (*sltu_insn): Likewise.
14894         (*sgeu_insn): Likewise.
14895         (*seqdi_special): Make operand 0 and comparison operation be of
14896         DImode.
14897         (*snedi_special): Likewise.
14898         (*snedi_special_vis3): Likewise.
14899         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
14900         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
14901         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
14902         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
14903         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
14904         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
14905         (*sltu_extend_sp64): Likewise.
14906         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
14907         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
14908         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
14909         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
14910         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
14912 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
14914         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
14915         (aarch64_start_file): Use the new function.
14917 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14918             Jason Merrill  <jason@redhat.com>
14920         * common.opt: Add -gdwarf.
14921         * opts.c (common_handle_option): Handle it.
14922         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
14924 2013-04-10  Richard Biener  <rguenther@suse.de>
14926         * passes.c (execute_todo): Do not call ggc_collect conditional here.
14927         (execute_one_ipa_transform_pass): But unconditionally here.
14928         (execute_one_pass): And here.
14929         (init_optimization_passes): Remove reload pass.
14930         * tree-pass.h (TODO_ggc_collect): Remove.
14931         (pass_reload): Likewise.
14932         * ira.c (do_reload): Merge into ...
14933         (ira): ... this.
14934         (rest_of_handle_reload): Remove.
14935         (pass_reload): Likewise.
14936         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14937         of reload for vzeroupper pass placement.
14938         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
14939         and todo_flags_finish of all passes.
14941 2013-04-10  Richard Biener  <rguenther@suse.de>
14943         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
14944         first_const_oprnd field, rename first_def_type to first_op_type.
14945         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
14946         (vect_get_and_check_slp_defs): Always use the type of the
14947         operand.  Allow mixed vect_external_def, vect_constant_def types.
14948         (vect_get_constant_vectors): Handle mixed vect_external_def,
14949         vect_constant_def types.
14951 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
14953         PR tree-optimization/55524
14954         * tree-ssa-math-opts.c
14955         (convert_mult_to_fma): Don't use an fms construct
14956         when we don't have an fms operation, but fnma, and it looks
14957         likely that we'll be able to use the latter.
14959 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
14961         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
14962         function.
14963         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
14964         inline fail caused by overwritable functions.
14966 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
14968         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
14969         unnecessary bits in the constant power of two case.
14971 2013-04-10  Richard Biener  <rguenther@suse.de>
14973         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
14974         broken code swapping operands.
14975         (vect_build_slp_tree): Do not compute load permutations here.
14976         (vect_analyze_slp_instance): Compute load permutations here,
14977         after building the SLP tree.
14979 2013-04-09  Christian Bruel  <christian.bruel@st.com>
14981         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
14982         of next/prev_real_insn.
14984 2013-04-09  Jan Hubicka  <jh@suse.cz>
14986         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14987         Drop aliased parameter.
14988         (function_and_variable_visibility): Do not handle alias pairs.
14989         * cgraph.c (varpool_externally_visible_p): Update prototype.
14990         * varpool.c (varpool_add_new_variable): Update.
14992 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14994         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
14996 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
14998         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15000         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15002 2013-04-09  Marek Polacek  <polacek@redhat.com>
15004         PR tree-optimization/48762
15005         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15007 2013-04-09  Richard Biener  <rguenther@suse.de>
15009         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15010         dealing with cost.
15011         (vect_build_slp_tree): Likewise.
15012         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15013         calculating the cost of a SLP instance.
15014         (vect_analyze_slp_instance): Use it from here, after building
15015         the SLP tree.
15017 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15019         PR middle-end/56883
15020         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15021         expand_omp_for_static_chunk): Use simple_p = true in
15022         force_gimple_operand_gsi calls when assigning to addressable decls.
15024 2013-04-09  Jeff Law  <law@redhat.com>
15026         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15027         when the boolean was created by converting a wider object which
15028         had a boolean range.
15030 2013-04-09  Richard Biener  <rguenther@suse.de>
15032         * tree-vectorizer.h (slp_void_p): Remove.
15033         (slp_tree): Typedef before _slp_tree declaration.
15034         (struct _slp_tree): Use a vector of slp_tree as children.
15035         (vect_get_place_in_interleaving_chain): Remove.
15036         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15037         Move ...
15038         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15039         and make static.
15040         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15041         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15042         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15043         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15044         Use slp_node instead of slp_void_p and adjust.
15046 2013-04-09  Richard Biener  <rguenther@suse.de>
15048         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15049         work that is not necessary.
15051 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15053         PR tree-optimization/56854
15054         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15055         forward into clobber stmts if it would change MEM_REF lhs into
15056         non-MEM_REF.
15058 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15060         * tree.c (type_hash_lookup, type_hash_add): Make static.
15061         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15063 2013-04-09  Richard Biener  <rguenther@suse.de>
15065         * tree.h (unsave_expr_now): Remove.
15066         * tree-inline.c (mark_local_for_remap_r): Remove.
15067         (unsave_expr_1): Likewise.
15068         (unsave_r): Likewise.
15069         (unsave_expr_now): Likewise.
15070         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15071         (propagate_tree_value): Likewise.
15073 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15075         * doc/rtl.texi (sequence): Rewrite documentation to match the
15076         current use of SEQUENCE rtl objects.
15077         * rtl.def (SEQUENCE): Likewise.
15079         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15080         Update documentation.
15081         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15082         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15084         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15086 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15088         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15089         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15090         methods.
15091         (estimate_edge_size_and_time): Add comment to suggest using rounding
15092         methods.
15093         (estimate_node_size_and_time): Ditto.
15094         (remap_edge_change_prob): Use helper rounding divide methods.
15095         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15096         (gimple_mod_pow2_value_transform): Ditto.
15097         (gimple_mod_subtract_transform): Ditto.
15098         (gimple_ic_transform): Ditto.
15099         (gimple_stringops_transform): Ditto.
15100         * stmt.c (conditional_probability): Ditto.
15101         (emit_case_dispatch_table): Ditto.
15102         * lto-cgraph.c (merge_profile_summaries): Ditto.
15103         * tree-optimize.c (execute_fixup_cfg): Ditto.
15104         * cfgcleanup.c (try_forward_edges): Ditto.
15105         * cfgloopmanip.c (scale_loop_profile): Ditto.
15106         (loopify): Ditto.
15107         (duplicate_loop_to_header_edge): Ditto.
15108         (lv_adjust_loop_entry_edge): Ditto.
15109         * tree-vect-loop.c (vect_transform_loop): Ditto.
15110         * profile.c (compute_branch_probabilities): Ditto.
15111         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15112         * lto-streamer-in.c (input_cfg): Ditto.
15113         * gimple-streamer-in.c (input_bb): Ditto.
15114         * ipa-cp.c (update_profiling_info): Ditto.
15115         (update_specialized_profile): Ditto.
15116         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15117         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15118         rounding methods.
15119         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15120         (compute_trg_info): Ditto.
15121         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15122         (purge_dead_edges): Ditto.
15123         * loop-unswitch.c (unswitch_loop): Ditto.
15124         * cgraphclones.c (cgraph_clone_edge): Ditto.
15125         (cgraph_clone_node): Ditto.
15126         * tree-inline.c (copy_bb): Ditto.
15127         (copy_edges_for_bb): Ditto.
15128         (initialize_cfun): Ditto.
15129         (copy_cfg_body): Ditto.
15130         (expand_call_inline): Ditto.
15132 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15134         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15135         TARGET_CYGWIN64 by TARGET_64BIT.
15137 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15139         * config/epiphany/epiphany.md (GPR_1): New constant.
15140         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15141         * config/epiphany/epiphany.c (gen_compare_reg):
15142         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15143         is already in place.
15144         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15145         Don't require being called during rtl expansion; If y operlaps r0,
15146         return 0.
15147         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15148         (epiphany_expand_epilogue): Likewise.
15150         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15151         Don't use CC_FPmode for ORDERED / UNORDERED.
15152         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15154         * config/epiphany/constraints.md (CnL): New constraint.
15155         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15156         * config/epiphany/predicates.md (add_operand): Allow 1024.
15158         * config/epiphany/epiphany.md (logical_op): New code iterator.
15159         (op_mnc): New code attribute.
15160         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15161         (mov_f+1, mov_f+2): New peephole2 patterns.
15163         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15164         (cstoresi4): Also allow re-use of zero result when doing a NE
15165         comparison to a non-zero operand.
15166         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15168         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15169         Use gen_addsi3_i / gen_subsi3_i.
15171 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15173         PR c++/34949
15174         PR c++/50243
15175         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15176         contain anything but clobbers, at most one __builtin_stack_restore,
15177         optionally debug stmts and final resx, and if it has at least one
15178         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15179         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15180         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15181         which isn't defaut definition, remove them.
15182         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15183         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15184         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15185         with MEM_REF LHS with SSA_NAME address.
15187 2013-04-08  Jeff Law  <law@redhat.com>
15189         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15191 2013-04-08  Richard Biener  <rguenther@suse.de>
15193         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15194         extra newline.
15195         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15196         determined vector type.
15197         (vect_analyze_data_refs): Likewise.
15198         (vect_get_new_vect_var): Adjust.
15199         (vect_create_destination_var): Preserve SSA name versions.
15200         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15201         not dump anything here.
15203 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15205         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15206         Add member lr_slot_known.
15207         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15208         if necessary.
15209         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15210         Remove code that sets lr_slot_offset according to what a previous
15211         version of epiphany_emit_save_restore used to do.
15212         (epiphany_emit_save_restore): When doing an lr save or restore,
15213         set/verify lr_slot_known and lr_slot_offset.
15215 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
15217         PR target/54338
15218         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15219         in ALL_REGS.
15221 2013-04-08  Richard Biener  <rguenther@suse.de>
15223         * alias.c (find_base_term): Fix thinko in previous change.
15225 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15227         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15228         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15229         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
15230         if possible to compute val.
15231         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15232         For QImode integers don't require anything about precision.  Use
15233         const_with_all_bytes_same to find out if the constant doesn't have
15234         repeated bytes in it.
15236 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15238         * config/s390/s390.c (s390_expand_insv): Only accept insertions
15239         within mode size.
15241 2013-04-08  Marek Polacek  <polacek@redhat.com>
15243         PR rtl-optimization/48182
15244         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15245         value to 1.
15247 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15249         PR target/55487
15250         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15251         nuses, make sure we have a label.
15253 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15255         PR target/56843
15256         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15257         (rs6000_emit_swdiv_low_precision): Remove.
15258         (rs6000_emit_swdiv): Rewrite to handle between one and four
15259         iterations of Newton-Raphson generally; modify required number of
15260         iterations for some cases.
15261         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15263 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
15265         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15266         set-but-unused variable.
15268         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15269         basic blocks of released function bodies garbage-collectable.
15271         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15272         (struct rtl_opt_pass): Add TODO_df_finish.
15274         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15276 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15278         * config/arm/constraints.md (q): New constraint.
15279         * config/arm/ldrdstrd.md: New file.
15280         * config/arm/arm.md (ldrdstrd.md) New include.
15281         (arm_movdi): Use "q" instead of "r" constraint
15282         for double-word memory access.
15283         (movdf_soft_insn): Likewise.
15284         * config/arm/vfp.md (movdi_vfp): Likewise.
15285         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15286         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15287         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15288         (mem_ok_for_ldrd_strd): Likewise.
15289         (output_move_double): Update assertion.
15291 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15293         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15295 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15297         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15298         define_insn_and_split.
15299         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15301 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15303         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15304         define_insn_and_split.
15305         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15306         (shiftsi3_compare): New pattern.
15307         (rrx): New pattern.
15308         * config/arm/unspecs.md (UNSPEC_RRX): New.
15310 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15312         * config/arm/arm.md (negdi_extendsidi): New pattern.
15313         (negdi_zero_extendsidi): Likewise.
15315 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15317         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15318         define_insn_and_split.
15319         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15320         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15322 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15324         * config/arm/arm.md (arm_subdi3): Convert define_insn into
15325         define_insn_and_split.
15326         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15327         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15329 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15331         * config/arm/arm.md (subsi3_carryin): New pattern.
15332         (subsi3_carryin_const): Likewise.
15333         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15334         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15336 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15338         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15340 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15342         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15343         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15345 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15347         * config/arm/arm.c (arm_expand_builtin): Change fcode
15348         type to unsigned int.
15350 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15352         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15354 2013-04-04  Ian Lance Taylor  <iant@google.com>
15356         * doc/standards.texi (Standards): The Go frontend supports the Go 1
15357         language standard.
15359 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
15361         PR middle-end/56729
15362         * df-scan.c (df_insn_delete): Disable failing assert.
15364 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15366         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15367         New function prototype.
15368         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15369         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15370         (arm_builtin_vectorized_function): New function.
15372 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15374         * config/arm/arm_neon_builtins.def: New file.
15375         * config/arm/arm.c (neon_builtin_data): Move contents to
15376         arm_neon_builtins.def.
15377         (enum arm_builtins): Include neon builtin definitions.
15378         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15379         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15381 2013-04-04  Marek Polacek  <polacek@redhat.com>
15383         PR tree-optimization/48186
15384         * predict.c (maybe_hot_frequency_p): Return false if
15385         HOT_BB_FREQUENCY_FRACTION is 0.
15386         (cgraph_maybe_hot_edge_p): Likewise.
15388 2013-04-04  Richard Biener  <rguenther@suse.de>
15390         PR tree-optimization/56826
15391         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15392         more accurately.
15394 2013-04-04  Richard Biener  <rguenther@suse.de>
15396         PR tree-optimization/56213
15397         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15398         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15400 2013-04-04  Richard Biener  <rguenther@suse.de>
15402         PR tree-optimization/56837
15403         * tree-loop-distribution.c (classify_partition): For non-zero
15404         values require that the value has the same precision as its
15405         mode to be useful as memset value.
15407 2013-04-03  Nick Clifton  <nickc@redhat.com>
15409         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15410         (fmssf4): Use fmsf.s on E3V5 architectures.
15411         (fnmasf4): Use fnmaf.s on E3V5 architectures.
15412         (fnmssf4): Use fnmsf.s on E3V5 architectures.
15414 2013-04-03  Jeff Law  <law@redhat.com>
15416         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15417         (lra-eliminations.o): Likewise.
15419 2013-04-03  Teresa Johnson  <tejohnson@google.com>
15421         * gcov-io.c (compute_working_sets): Moved most of body of old
15422         compute_working_sets here from profile.c.
15423         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15424         (gcov_working_set_t): Moved typedef here from basic-block.h
15425         (compute_working_set): Declare.
15426         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15427         (get_working_sets): Renamed from compute_working_set,
15428         replace most of body with call to new compute_working_sets.
15429         (get_exec_counts): Replace call to compute_working_sets
15430         to get_working_sets.
15431         * profile.h (get_working_sets): Renamed from compute_working_set.
15432         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15433         to get_working_sets.
15434         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15435         * gcov-dump.c (dump_working_sets): New function.
15437 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
15439         * hwint.c (sext_hwi, zext_hwi): New functions.
15440         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15441         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15442         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15443         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15444         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15445         (sext_hwi, zext_hwi): New functions.
15447 2013-04-03  Jeff Law  <law@redhat.com>
15449         PR tree-optimization/56799
15450         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15451         back test for widening conversion erroneously dropped in prior change.
15453 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15455         PR target/56809
15456         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15457         instead of next_real_insn.
15459 2013-04-03  Marek Polacek  <polacek@redhat.com>
15461         PR sanitizer/55702
15462         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15464 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15466         PR target/56809
15467         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15468         next_real_insn.
15469         (thumb1_output_casesi): Likewise.
15470         (thumb2_output_casesi): Likewise.
15472 2013-04-03  Richard Biener  <rguenther@suse.de>
15474         PR tree-optimization/56817
15475         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15476         Split out ...
15477         (tree_unroll_loops_completely_1): ... new function to manually
15478         walk the loop tree, properly defering outer loops of unrolled
15479         loops to later iterations.
15481 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15483         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15484         (vectorizable_load): Likewise.
15485         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15486         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15488 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15490         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15491         BIT_FIELD_REF.
15493 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15495         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15497 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
15499         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15501 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15503         PR tree-optimization/56790
15504         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15505         folding.
15507 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15509         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15510         Handle VEC_MERGE.
15511         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15512         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15513         equal arguments.
15515 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
15517         PR c/19449
15518         * tree.h (force_folding_builtin_constant_p): New decl.
15519         * builtins.c (force_folding_builtin_constant_p): New variable.
15520         (fold_builtin_constant_p): Fold immediately also if
15521         force_folding_builtin_constant_p.
15523 2013-04-03  Richard Biener  <rguenther@suse.de>
15525         PR tree-optimization/56812
15526         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15527         DRs of the same interleaving chain are independent.
15529 2013-04-02  Jason Merrill  <jason@redhat.com>
15531         * gdbinit.in (pbb): Use debug fn.
15533 2013-04-02  Lawrence Crowl  <crowl@google.com>
15535         * sese.h (struct ivtype_map_elt_s): Remove unused.
15536         (extern debug_ivtype_map): Remove unused.
15537         (extern eq_ivtype_map_elts): Remove unused.
15538         * sese.c (debug_ivtype_map): Removed unused.
15539         (debug_ivtype_map_1): Removed unused.
15540         (debug_ivtype_elt): Remove unused.
15541         (eq_ivtype_map_elts): Remove unused.
15544 2013-04-02  Kai Tietz  <ktietz@redhat.com>
15546         PR target/52790
15547         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
15548         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
15549         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
15550         function.
15551         (legitimize_pe_coff_symbol): Likewise.
15552         (is_imported_p): New helper-function.
15553         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
15554         for Windows x64 targets.
15555         (ix86_expand_prologue): Optimize for pe-coff targets.
15556         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
15557         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
15558         medium/large code-model.
15559         (legitimize_pic_address): Likewise.
15560         (legitimize_tls_address): Likewise.
15561         (ix86_expand_call): Likewise.
15562         (x86_output_mi_thunk): Likewise.
15563         (get_dllimport_decl): Add new beimport argument.
15564         (construct_plt_address): Don't assert for x64 pe-coff targets.
15565         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
15566         targets.
15567         (SYMBOL_FLAG_STUBVAR): New macro.
15568         (SYMBOL_REF_STUBVAR_P): Likewise.
15569         * config/i386/winnt.c (stub_list): New structure.
15570         (stub_head): New local variable.
15571         (i386_pe_record_stub): New function.
15572         (i386_pe_file_end): Emit refptr-stubs.
15574 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
15576         PR rtl-optimization/56745
15577         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
15578         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
15580         PR c++/34949
15581         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
15582         and both of them are MEM_REFs, just compare first argument for
15583         equality and attempt to deal even with differing offsets.
15585         PR c++/34949
15586         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
15587         of gimple_clobber_p to be MEM_REF.
15588         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
15589         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
15590         after gimplification.
15591         * asan.c (get_mem_ref_of_assignment): Don't instrument
15592         gimple_clobber_p stmts.
15593         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
15594         gimple_clobber_p stmt if they have MEM_REF lhs and
15595         are dead because of another gimple_clobber_p stmt.
15596         * tree-ssa-live.c (clear_unused_block_pointer): Treat
15597         gimple_clobber_p stmts like debug stmts.
15598         (remove_unused_locals): Remove clobbers with MEM_REF lhs
15599         that refer to unused VAR_DECLs or uninitialized values.
15600         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
15601         gimple_clobber_p stmts if they refer to removed parameters.
15602         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
15603         formatting.
15605 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
15607         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
15608         using SWI48 mode attribute.
15610 2013-04-02  Wei Mi  <wmi@google.com>
15612         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
15613         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
15614         *<rotate_insn><mode>3_mask in i386.md.
15616 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
15618         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
15620 2013-04-02  Richard Biener  <rguenther@suse.de>
15622         PR tree-optimization/56778
15623         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15624         Runtime alias tests are not supported for gather loads.
15625         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
15626         stmts referenced from SSA operands before updating SSA form.
15628 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
15629             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15631         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
15632         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
15633         * config/arm/cortex-a53.md: New file.
15634         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
15635         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
15636         * config/arm/arm.c (arm_issue_rate): Likewise.
15637         * config/arm/arm-tune.md: Regenerate
15638         * config/arm/arm-tables.opt: Regenerate.
15639         * config/arm/arm-cores.def: Add cortex-a53.
15641 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
15643         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
15644         non-static link.
15646 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15648         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
15649         scalar load/store operations using B/H registers.
15650         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
15652 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15654         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
15655         scalar move.
15656         * config/aarch64/aarch64.c
15657         (aarch64_simd_scalar_immediate_valid_for_move): New.
15658         * config/aarch64/aarch64-protos.h
15659         (aarch64_simd_scalar_immediate_valid_for_move): New.
15660         * config/aarch64/constraints.md (Dh, Dq): New.
15661         * config/aarch64/iterators.md (hq): New.
15663 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
15665         * reorg.c (get_branch_condition): Deal with conditional returns.
15666         (fill_simple_delay_slots): Remove dead code dealing with jumps.
15668 2013-04-01  Wei Mi  <wmi@google.com>
15670         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
15671         Truncate operand 2 using %b asm operand modifier.
15672         (*<shift_insn><mode>3_mask): Ditto.
15673         (*<rotate_insn><mode>3_mask): Ditto.
15675 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
15677         PR middle-end/56798
15678         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
15680 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
15682         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
15683         of next_real_insn.
15684         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
15686 2013-03-30  Lawrence Crowl  <crowl@google.com>
15688         * dse.c (clear_alias_sets): Remove never set.
15689         (disqualified_clear_alias_sets): Remove never set.
15690         (clear_alias_mode_pool): Remove never set.
15691         (dse_step0): Remove condition that is never true.
15692         (canon_address): Remove condition that is never true.
15693         (dse_step7): Remove condition that is never true.
15694         (rest_of_handle_dse): Remove condition that is never true.
15695         (rest_of_handle_dse::did_global): Remove never read from above.
15696         (dse_step2_spill): Remove never called from above.
15697         (dse_step5_spill): Remove never called from above.
15699 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
15701         * doc/md.texi (Standard Names) <casesi>: Update documentation for
15702         JUMP_TABLE_DATA changes.
15703         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
15704         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
15705         (Insns) <jump_table_data>: New entry.
15706         * doc/tm.texi: Regenerate.
15708         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
15710         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
15711         for table jump at the end of a basic block using tablejump_p.
15712         * targhooks.c (default_invalid_within_doloop): Likewise.
15713         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
15714         target hook implementation that is identical to the default hook.
15715         (rs6000_invalid_within_doloop): Remove.
15717         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
15718         unused variable from tablejump_p call.
15720         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
15721         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
15722         (INSN_DELETED_P): Likewise.
15723         (emit_jump_table_data): New prototype.
15724         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
15725         after 4th as unused.
15726         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
15727         * sched-vis.c (print_insn): Likewise.
15728         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
15729         insn for compatibility with back ends that use next_active_insn to
15730         identify jump table data.
15731         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
15732         (remove_insn): Likewise.
15733         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
15734         to be emitted.
15735         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
15736         (emit_jump_table_data): New function.
15738         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
15739         basic block, a JUMP_TABLE_DATA never is.
15740         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
15741         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
15742         off from code handling real insns.
15743         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
15744         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
15745         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
15746         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
15747         is not a NONDEBUG_INSN_P.
15748         * ira-costs.c (scan_one_insn): Likewise.
15749         * jump.c (mark_all_labels): Likewise.
15750         (mark_jump_label_1): Likewise.
15751         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15752         * lra.c (get_insn_freq): Expect all insns reaching here to be in
15753         a basic block.
15754         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
15755         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
15756         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
15757         JUMP_TABLE_DATA_P insns.
15758         (calculate_elim_costs_all_insns): Likewise.
15759         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
15760         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
15761         (delete_output_reload): Code style fixups.
15762         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
15763         insn flags on this non-insn.
15764         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
15765         as scheduling barriers, for pre-change compatibility.
15766         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
15767         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
15769         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
15770         redundant JUMP_TABLE_DATA_P test.
15771         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
15772         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15773         (frv_for_each_packet): Likewise.
15774         * config/i386/i386.c (min_insn_size): Likewise.
15775         (ix86_avoid_jump_mispredicts): Likewise.
15776         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15777         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
15778         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15779         (mips16_insn_length): Robustify.
15780         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
15781         (mips16_split_long_branches): Likewise.
15782         * config/pa/pa.c (pa_combine_instructions): Likewise.
15783         * config/rs6000/rs6000.c (get_next_active_insn): Treat
15784         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
15785         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
15786         as contributing to pool range lengths.
15787         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
15788         Remove redundant JUMP_TABLE_DATA_P test.
15789         (sh_loop_align): Likewise.
15790         (split_branches): Likewise.
15791         (sh_insn_length_adjustment): Likewise.
15792         * config/spu/spu.c (get_branch_target): Likewise.
15794 2013-03-29  Jan Hubicka  <jh@suse.cz>
15796         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
15797         gcov streaming; stream hot bb threshold to ltrans.
15798         * predict.c (get_hot_bb_threshold): Break out from ....
15799         (maybe_hot_count_p): ... here.
15800         (set_hot_bb_threshold): New function.
15801         * lto-section-in.c (lto_section_name): Add profile.
15802         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
15803         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
15804         and data-streamer.h
15805         (histogram_entry): New structure.
15806         (histogram, histogram_pool): New global vars.
15807         (histogram_hash): New structure.
15808         (histogram_hash::hash): New method.
15809         (histogram_hash::equal): Likewise.
15810         (account_time_size): New function.
15811         (cmp_counts): New function.
15812         (dump_histogram): New function.
15813         (ipa_profile_generate_summary): New function.
15814         (ipa_profile_write_summary): New function.
15815         (ipa_profile_read_summary): New function.
15816         (ipa_profile): Decide on threshold.
15817         (pass_ipa_profile): Add ipa_profile_write_summary and
15818         ipa_profile_read_summary.
15819         * Makefile.in (ipa.o): Update dependencies.
15820         * lto-streamer.h (LTO_section_ipa_profile): New section.
15822 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
15824         * tree.h (VAR_P): New.
15826 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
15828         PR lto/56777
15829         * doc/invoke.texi ([-fwhole-program]): Fix typo.
15831 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
15833         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
15834         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15835         (control_flow_insn_p): Likewise.
15836         * cfgrtl.c (duplicate_insn_chain): Likewise.
15837         * final.c (get_attr_length_1): Likewise.
15838         (shorten_branches): Likewise.
15839         (final_scan_insn): Likewise.
15840         * function.c (instantiate_virtual_regs): Likewise.
15841         * gcse.c (insert_insn_end_basic_block): Likewise.
15842         * ira-costs.c (scan_one_insn): Likewise.
15843         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15844         * lra.c (check_rtl): Likewise.
15845         * reload1.c (elimination_costs_in_insn): Likewise.
15846         * reorg.c (follow_jumps): Likewise.
15848         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
15849         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15850         (thumb_far_jump_used_p): Likewise.
15851         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
15852         (workaround_speculation): Likewise.
15853         (add_sched_insns_for_speculation): Likewise.
15854         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
15855         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15856         (frv_for_each_packet): Likewise.
15857         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
15858         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
15859         (final_emit_insn_group_barriers): Likewise.
15860         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15861         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15862         (mips16_insn_length): Likewise.
15863         * config/pa/pa.c (pa_reorg): Likewise.
15864         (pa_combine_instructions): Likewise.
15865         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
15866         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
15867         (sh_reorg): Likewise.
15868         (split_branches): Likewise.
15869         * config/spu/spu.c (get_branch_target): Likewise.
15871         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
15872         JUMP_TABLE_DATA_P.
15874 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
15876         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
15877         Fix declaration name.
15879 2013-03-28  Lawrence Crowl  <crowl@google.com>
15881         * graphds.h (struct graph.indicies): Remove unused.
15882         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
15883         (SCOP_ORIGINAL_PDDRS): Remove unused.
15884         * sese.h (extern insert_loop_close_phis): Removed unused.
15885         (extern insert_guard_phis): Removed unused.
15886         (extern ivtype_map_elt_info): Removed unused.
15887         (new_ivtype_map_elt): Removed unused.
15888         * sese.c (ivtype_map_elt_info): Removed unused.
15890 2013-03-28  Lawrence Crowl  <crowl@google.com>
15892         * Makefile.in: Add several missing include dependences.
15893         (DUMPFILE_H): New.
15894         (test-dump.o): New.  This object is not added to any executable,
15895         but is present for ad-hoc testing.
15896         * bitmap.c
15897         (debug (const bitmap_head_def &)): New.
15898         (debug (const bitmap_head_def *)): New.
15899         * bitmap.h
15900         (extern debug (const bitmap_head_def &)): New.
15901         (extern debug (const bitmap_head_def *)): New.
15902         * cfg.c
15903         (debug (edge_def &)): New.
15904         (debug (edge_def *)): New.
15905         * cfghooks.c
15906         (debug (basic_block_def &)): New.
15907         (debug (basic_block_def *)): New.
15908         * dumpfile.h
15909         (dump_node (const_tree, int, FILE *)): Correct source file.
15910         * dwarf2out.c
15911         (debug (die_struct &)): New.
15912         (debug (die_struct *)): New.
15913         * dwarf2out.h
15914         (extern debug (die_struct &)): New.
15915         (extern debug (die_struct *)): New.
15916         * gimple-pretty-print.c
15917         (debug (gimple_statement_d &)): New.
15918         (debug (gimple_statement_d *)): New.
15919         * gimple-pretty-print.h
15920         (extern debug (gimple_statement_d &)): New.
15921         (extern debug (gimple_statement_d *)): New.
15922         * ira-build.c
15923         (debug (ira_allocno_copy &)): New.
15924         (debug (ira_allocno_copy *)): New.
15925         (debug (ira_allocno &)): New.
15926         (debug (ira_allocno *)): New.
15927         * ira-int.h
15928         (extern debug (ira_allocno_copy &)): New.
15929         (extern debug (ira_allocno_copy *)): New.
15930         (extern debug (ira_allocno &)): New.
15931         (extern debug (ira_allocno *)): New.
15932         * ira-lives.c
15933         (debug (live_range &)): New.
15934         (debug (live_range *)): New.
15935         * lra-int.h
15936         (debug (lra_live_range &)): New.
15937         (debug (lra_live_range *)): New.
15938         * lra-lives.c
15939         (debug (lra_live_range &)): New.
15940         (debug (lra_live_range *)): New.
15941         * omega.c
15942         (debug (omega_pb_d &)): New.
15943         (debug (omega_pb_d *)): New.
15944         * omega.h
15945         (extern debug (omega_pb_d &)): New.
15946         (extern debug (omega_pb_d *)): New.
15947         * print-rtl.c
15948         (debug (const rtx_def &)): New.
15949         (debug (const rtx_def *)): New.
15950         * print-tree.c
15951         (debug_tree (tree): Move within file.
15952         (debug_raw (const tree_node &)): New.
15953         (debug_raw (const tree_node *)): New.
15954         (dump_tree_via_hooks (const tree_node *, int)): New.
15955         (debug (const tree_node &)): New.
15956         (debug (const tree_node *)): New.
15957         (debug_verbose (const tree_node &)): New.
15958         (debug_verbose (const tree_node *)): New.
15959         (debug_head (const tree_node &)): New.
15960         (debug_head (const tree_node *)): New.
15961         (debug_body (const tree_node &)): New.
15962         (debug_body (const tree_node *)): New.
15963         (debug_vec_tree (tree): Move and reimplement in terms of dump.
15964         (debug (vec<tree, va_gc> &)): New.
15965         (debug (vec<tree, va_gc> *)): New.
15966         * rtl.h
15967         (extern debug (const rtx_def &)): New.
15968         (extern debug (const rtx_def *)): New.
15969         * sbitmap.c
15970         (debug_raw (simple_bitmap_def &)): New.
15971         (debug_raw (simple_bitmap_def *)): New.
15972         (debug (simple_bitmap_def &)): New.
15973         (debug (simple_bitmap_def *)): New.
15974         * sbitmap.h
15975         (extern debug (simple_bitmap_def &)): New.
15976         (extern debug (simple_bitmap_def *)): New.
15977         (extern debug_raw (simple_bitmap_def &)): New.
15978         (extern debug_raw (simple_bitmap_def *)): New.
15979         * sel-sched-dump.c
15980         (debug (vinsn_def &)): New.
15981         (debug (vinsn_def *)): New.
15982         (debug_verbose (vinsn_def &)): New.
15983         (debug_verbose (vinsn_def *)): New.
15984         (debug (expr_def &)): New.
15985         (debug (expr_def *)): New.
15986         (debug_verbose (expr_def &)): New.
15987         (debug_verbose (expr_def *)): New.
15988         (debug (vec<rtx> &)): New.
15989         (debug (vec<rtx> *)): New.
15990         * sel-sched-dump.h
15991         (extern debug (vinsn_def &)): New.
15992         (extern debug (vinsn_def *)): New.
15993         (extern debug_verbose (vinsn_def &)): New.
15994         (extern debug_verbose (vinsn_def *)): New.
15995         (extern debug (expr_def &)): New.
15996         (extern debug (expr_def *)): New.
15997         (extern debug_verbose (expr_def &)): New.
15998         (extern debug_verbose (expr_def *)): New.
15999         (extern debug (vec<rtx> &)): New.
16000         (extern debug (vec<rtx> *)): New.
16001         * sel-sched-ir.h
16002         (_list_iter_cond_expr): Make inline instead of static.
16003         * sreal.c
16004         (debug (sreal &)): New.
16005         (debug (sreal *)): New.
16006         * sreal.h
16007         (extern debug (sreal &)): New.
16008         (extern debug (sreal *)): New.
16009         * tree.h
16010         (extern debug_raw (const tree_node &)): New.
16011         (extern debug_raw (const tree_node *)): New.
16012         (extern debug (const tree_node &)): New.
16013         (extern debug (const tree_node *)): New.
16014         (extern debug_verbose (const tree_node &)): New.
16015         (extern debug_verbose (const tree_node *)): New.
16016         (extern debug_head (const tree_node &)): New.
16017         (extern debug_head (const tree_node *)): New.
16018         (extern debug_body (const tree_node &)): New.
16019         (extern debug_body (const tree_node *)): New.
16020         (extern debug (vec<tree, va_gc> &)): New.
16021         (extern debug (vec<tree, va_gc> *)): New.
16022         * tree-cfg.c
16023         (debug (struct loop &)): New.
16024         (debug (struct loop *)): New.
16025         (debug_verbose (struct loop &)): New.
16026         (debug_verbose (struct loop *)): New.
16027         * tree-dump.c: Add header dependence.
16028         * tree-flow.h
16029         (extern debug (struct loop &)): New.
16030         (extern debug (struct loop *)): New.
16031         (extern debug_verbose (struct loop &)): New.
16032         (extern debug_verbose (struct loop *)): New.
16033         * tree-data-ref.c
16034         (debug (data_reference &)): New.
16035         (debug (data_reference *)): New.
16036         (debug (vec<data_reference_p> &)): New.
16037         (debug (vec<data_reference_p> *)): New.
16038         (debug (vec<ddr_p> &)): New.
16039         (debug (vec<ddr_p> *)): New.
16040         * tree-data-ref.h
16041         (extern debug (data_reference &)): New.
16042         (extern debug (data_reference *)): New.
16043         (extern debug (vec<data_reference_p> &)): New.
16044         (extern debug (vec<data_reference_p> *)): New.
16045         (extern debug (vec<ddr_p> &)): New.
16046         (extern debug (vec<ddr_p> *)): New.
16047         * tree-ssa-alias.c
16048         (debug (pt_solution &)): New.
16049         (debug (pt_solution *)): New.
16050         * tree-ssa-alias.h
16051         (extern debug (pt_solution &)): New.
16052         (extern debug (pt_solution *)): New.
16053         * tree-ssa-alias.c
16054         (debug (_var_map &)): New.
16055         (debug (_var_map *)): New.
16056         (debug (tree_live_info_d &)): New.
16057         (debug (tree_live_info_d *)): New.
16058         * tree-ssa-alias.h
16059         (extern debug (_var_map &)): New.
16060         (extern debug (_var_map *)): New.
16061         (extern debug (tree_live_info_d &)): New.
16062         (extern debug (tree_live_info_d *)): New.
16064 2013-03-28  Jan Hubicka  <jh@suse.cz>
16066         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16068 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16070         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16071         record only when desired or required.
16073 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16075         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16076         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16078 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16080         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16081         (*andsi3_compare0_uxtw): New pattern.
16082         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16083         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16085 2013-03-28  Jan Hubicka  <jh@suse.cz>
16087         * data-streamer-in.c (streamer_read_gcov_count): New function.
16088         * gimple-streamer-out.c: Include value-prof.h.
16089         (output_gimple_stmt): Output histogram.
16090         (output_bb): Use streamer_write_gcov_count.
16091         * value-prof.c: Include data-streamer.h
16092         (dump_histogram_value): Add HIST_TYPE_MAX.
16093         (stream_out_histogram_value): New function.
16094         (stream_in_histogram_value): New function.
16095         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16096         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16097         * data-streamer-out.c (streamer_write_gcov_count): New function.
16098         (streamer_write_gcov_count_stream): New function.
16099         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16100         (lto_output_node): Likewise.
16101         (input_node, input_edge): Likewise.
16102         * lto-streamer-out.c (output_cfg): Update streaming.
16103         * lto-streamer-in.c (input_cfg): Likewise.
16104         * data-streamer.h (streamer_write_gcov_count,
16105         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16106         * gimple-streamer-in.c: Include value-prof.h
16107         (input_gimple_stmt): Input histograms.
16108         (input_bb): Update profile streaming.
16110 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16112         * genmodes.c (emit_max_int): New function.
16113         (emit_insn_modes_h): Added call to emit_max_function.
16114         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16115         Added doc.
16116         * machmode.def: Fixed comment.
16118 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16120         * combine.c (try_combine): Removed useless assert.
16121         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16123 2013-03-28  Marek Polacek  <polacek@redhat.com>
16124             Richard Biener  <rguenther@suse.de>
16126         PR tree-optimization/56695
16127         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16128         build signed result of a vector comparison.
16129         * tree-cfg.c (verify_gimple_comparison): Check that a result
16130         of a vector comparison has signed type.
16132 2013-03-28  Richard Biener  <rguenther@suse.de>
16134         PR tree-optimization/37021
16135         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16136         do not restrict gaps between groups.
16137         * tree-vect-stmts.c (vectorizable_load): Properly account for
16138         a gap between groups.
16140 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16142         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16143         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16144         is not enabled.
16146 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16148         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16149         * doc/extend.texi (Named Address Spaces): Ditto.
16150         (Variable Attributes): Ditto.
16152 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16154         * config.build: Add support for cygwin x64 target.
16155         * config.gcc: Likewise.
16156         * config.host: Likewise.
16157         * configure.ac: Likewise
16158         * configure: Regenerated.
16160 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16162         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16163         * config/i386/t-cygwin-w64: New file.
16164         * config/i386/cygwin-w64.h: New file.
16165         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16166         and add support for x64-cygwin target.
16167         (CPP_SPEC): Likewise.
16168         (CXX_WRAP_SPEC_LIST): Undefine before define.
16169         (LIBGCJ_SONAME): Use 15 as version.
16171 2013-03-27  Richard Biener  <rguenther@suse.de>
16173         PR tree-optimization/56716
16174         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16175         dumping for ref nodes.
16177 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16179         PR tree-optimization/55334
16180         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16181         restricted pointers to arrays.
16183 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16185         * Makefile.in (.SUFFIXES): Add .cc.
16186         (.c.o): Apply same recipe for implicit rule .cc.o.
16188 2013-03-27  Richard Biener  <rguenther@suse.de>
16190         PR tree-optimization/37021
16191         * tree-vect-data-refs.c (vect_check_strided_load): Allow
16192         REALPART/IMAGPART_EXPRs around the supported refs.
16193         * tree-ssa-structalias.c (find_func_aliases): Assume that
16194         floating-point values are not used to transfer pointers.
16196 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16198         * target.def (TARGET_HAS_IFUNC_P): New target hook.
16199         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16200         * doc/tm.texi: Regenerate.
16201         * targhooks.h (default_has_ifunc_p): New.
16202         * targhooks.c (default_has_ifunc_p): Ditto.
16203         * config/linux-protos.h: New file.
16204         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16205         hook for linux which disables support of indirect functions in android.
16206         * config/linux-android.c: New file.
16207         * config/t-linux-android.c: Ditto.
16208         * config.gcc: Added new object file linux-android.o.
16209         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16210         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16211         * varasm.c (do_assemble_alias): Likewise.
16212         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16213         doesn't support indirect functions.
16214         * configure: Regenerate.
16216 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
16218         PR target/56102
16219         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16220         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16221         mult-word mode.
16223 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16225         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16227 2013-03-27  Terry Guo  <terry.guo@arm.com>
16229         * config/arm/arm-cores.def: Added core cortex-r7.
16230         * config/arm/arm-tune.md: Regenerated.
16231         * config/arm/arm-tables.opt: Regenerated.
16232         * doc/invoke.texi: Added entry for core cortex-r7.
16234 2013-03-27  Walter Lee  <walt@tilera.com>
16236         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16237         double-decrement of next_scratch_regno.
16239 2013-03-27  Walter Lee  <walt@tilera.com>
16241         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16242         input operands.
16243         (insn_v1mulus): Ditto.
16244         (insn_v2muls): Ditto.
16246 2013-03-27  Walter Lee  <walt@tilera.com>
16248         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16249         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16251 2013-03-27  Walter Lee  <walt@tilera.com>
16253         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16254         (*sibcall_value): Ditto.
16256 2013-03-27  Walter Lee  <walt@tilera.com>
16258         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16259         (insn_mnz_v8qi): ... this ...
16260         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
16261         vector equivalent.
16262         (insn_v<n>mnz): Replaced by ...
16263         (insn_v1mnz): ... this ...
16264         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
16265         equivalent.
16266         (insn_mz_<mode>): Replaced by ...
16267         (insn_mz_v8qi): ... this ...
16268         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
16269         vector equivalent.
16270         (insn_v<n>mz): Replaced by ...
16271         (insn_v1mz): ... this ...
16272         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
16273         equivalent.
16275 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
16277         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16279 2013-03-26  Roland McGrath  <mcgrathr@google.com>
16281         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16282         than fprintf with a non-constant, non-format string.
16284 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16286         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16287         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
16288         operand 0 predicate.
16289         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16290         attribute.  Use general_x64nomem_operand as operand 1 predicate.
16291         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16292         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16293         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16294         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16295         (mov<mode>_insv_1): Remove expander.  Merge insn with
16296         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16297         Use general_x64nomem_operand as operand 1 predicate.
16298         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16299         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16300         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16301         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16302         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
16303         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16304         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16305         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16306         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
16307         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16308         (general_x64nomem_operand): Ditto.
16310 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
16312         * config/rtems.opt: Add -pthread option.
16314 2013-03-26  Richard Biener  <rguenther@suse.de>
16316         * alias.c (find_base_term): Avoid redundant and not used recursion.
16317         (base_alias_check): Get the initial base term from the caller.
16318         (true_dependence_1): Compute and pass base terms to base_alias_check.
16319         (write_dependence_p): Likewise.
16320         (may_alias_p): Likewise.
16322 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
16324         * config/aarch64/aarch64.c (aarch64_classify_address): Support
16325         PC-relative load in SI modes and above only.
16327 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
16329         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16330         * config/arm/iwmmxt.md (WCGR0): Update.
16331         (WCGR1, WCGR2, WCGR3): Likewise.
16333 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16335         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16336         Use x64 and nox64 isa attributes.
16338 2013-03-26  Richard Biener  <rguenther@suse.de>
16340         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16341         alignment computations and rely on get_object_alignment_1
16342         for the !TYPE_P case.
16343         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16345 2013-03-26  Walter Lee  <walt@tilera.com>
16347         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16348         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16350 2013-03-25  Jeff Law  <law@redhat.com>
16352         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16353         check for INTEGRAL_TYPE_P that was missing due to checking in
16354         wrong version of prior patch.
16356 2013-03-25  Walter Lee  <walt@tilera.com>
16358         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16359         TILEGX_INSN_SHUFFLEBYTES1.
16360         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16361         shufflebytes1.
16362         (tilegx_builtins): Ditto.
16363         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16365 2013-03-25  Walter Lee  <walt@tilera.com>
16367         * config/tilegx/tilegx.md (floatsisf2): New pattern.
16368         (floatunssisf2): New pattern.
16369         (floatsidf2): New pattern.
16370         (floatunssidf2): New pattern.
16372 2013-03-25  Walter Lee  <walt@tilera.com>
16374         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16375         tests for constraint J, K, N, P.
16377 2013-03-25  Walter Lee  <walt@tilera.com>
16379         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16380         Use indirect/pcrel encoding.
16381         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16382         Ditto.
16384 2013-03-25  Steve Ellcey  <sellcey@mips.com>
16386         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16387         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16388         * config/mips/mips.c (mips_option_override): Set IMADD default.
16389         * config/mips/mips.h (PTF_AVOID_IMADD): New.
16390         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16391         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16392         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16393         * doc/invoke.texi (-mimadd/-mno-imadd): New.
16395 2013-03-25  Jeff Law  <law@redhat.com>
16397         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16398         slightly to avoid creating and folding useless trees.  Simplify
16399         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16401 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
16403         * config/i386/i386.md (*zero_extendsidi2): Merge with
16404         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
16405         * config/i386/predicates.md (x86_64_zext_operand): Rename from
16406         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
16407         targets.  Clarify comment.
16409 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16411         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16412         pass-through jump functions differently.
16413         (ipa_read_jump_function): Likewise.  Also use setter functions to set
16414         up jump functions.
16416 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16418         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16419         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16420         process it.
16421         (ipa_get_indirect_edge_target): New function.
16422         (devirtualization_time_bonus): New parameter known_aggs, pass it to
16423         ipa_get_indirect_edge_target.  Update all callers.
16424         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
16425         ipa_get_indirect_edge_target_1 instead of calling
16426         ipa_get_indirect_edge_target.
16427         (create_specialized_node): Pass aggvlas to
16428         ipcp_discover_new_direct_edges.
16430 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16432         * config/arm/arm.md (f_sels, f_seld): New types.
16433         (*cmov<mode>): New pattern.
16434         * config/arm/predicates.md (arm_vsel_comparison_operator): New
16435         predicate.
16437 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16439         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16440         POSIX-printf for mingw-hosted  builds.
16442 2013-03-25  Richard Biener  <rguenther@suse.de>
16444         PR middle-end/56694
16445         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16446         must-not-throw stmt location.
16448 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16450         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16451         Emit load-acquire versions when acq is true.
16452         (arm_emit_store_exclusive): Add rel parameter.
16453         Emit store-release versions when rel is true.
16454         (arm_split_compare_and_swap): Use acquire-release instructions
16455         instead.
16456         of barriers when appropriate.
16457         (arm_split_atomic_op): Likewise.
16458         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16459         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16460         (VUNSPEC_SLX): Likewise.
16461         (VUNSPEC_LDA): Likewise.
16462         (VUNSPEC_STL): Likewise.
16463         * config/arm/sync.md (atomic_load<mode>): New pattern.
16464         (atomic_store<mode>): Likewise.
16465         (arm_load_acquire_exclusive<mode>): Likewise.
16466         (arm_load_acquire_exclusivesi): Likewise.
16467         (arm_load_acquire_exclusivedi): Likewise.
16468         (arm_store_release_exclusive<mode>): Likewise.
16470 2013-03-25  Catherine Moore  <clm@codesourcery.com>
16472         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16473         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16474         * config/mip/predicates.md (lwsp_swsp_operand,
16475         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16476         sb16_operand, db4_operand, db7_operand, ib3_operand,
16477         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16478         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16479         andi16_operand): New predicates.
16480         * config/mips/mips.md (compression): New attribute.
16481         (enabled): New attribute.
16482         (length): Consider compression in computing length.
16483         (shift_compression): New code attribute.
16484         (*add<mode>3): New operands. Record compression.
16485         (sub<mode>3): Likewise.
16486         (one_cmpl<mode>2): Likewise.
16487         (*and<mode>3): Likewise.
16488         (*ior<mode>3): Likewise.
16489         (unnamed pattern for xor): Likewise.
16490         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16491         (*<optab><mode>3): Likewise.
16492         (*mov<mode>_internal: Likewise.
16493         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16494         (mips_unsigned_immediate_p): New.
16495         (umips_lwsp_swsp_address_p): New.
16496         (m16_based_address_p): New.
16497         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16498         (mips_unsigned_immediate_p): New prototype.
16499         (lwsp_swsp_address_p): New prototype.
16500         (m16_based_address_p): New prototype.
16501         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16502         (mips_signed_immediate_p): New function.
16503         (m16_based_address_p): New function.
16504         (lwsp_swsp_address_p): New function.
16505         (mips_print_operand_punctuation): Recognize short delay slot insns
16506         for microMIPS.add<mode>3"
16508 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16510         PR target/56720
16511         * config/arm/iterators.md (v_cmp_result): New mode attribute.
16512         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16514 2013-03-25  Richard Biener  <rguenther@suse.de>
16516         PR tree-optimization/56689
16517         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16518         any edge.
16520 2013-03-25  Richard Biener  <rguenther@suse.de>
16522         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16523         of bitmap.
16524         (memory_references): Likewise.
16525         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16526         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16527         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16528         (gather_mem_refs_in_loops): Fold into ...
16529         (analyze_memory_references): ... this.  Move initialization
16530         to tree_ssa_lim_initialize.
16531         (fill_always_executed_in): Rename to ...
16532         (fill_always_executed_in_1): ... this.
16533         (fill_always_executed_in): Move contains_call computation to
16534         this new function from ...
16535         (tree_ssa_lim_initialize): ... here.
16536         (tree_ssa_lim): Call fill_always_executed_in.
16538 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16540         * postreload.c (reload_combine): Fix code detecting returns.
16542 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16544         * function.c (emit_use_return_register_into_block): On cc0 targets,
16545         do not emit the sequence between cc0 setter and user.
16547 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16549         * config/i386/predicates.md (local_symbolic_operand): Interpret
16550         dll-imported symbols as none-local.
16552 2013-03-25  Richard Biener  <rguenther@suse.de>
16554         * tree-ssa-loop-im.c (struct depend): Remove.
16555         (struct lim_aux_data): Make depends a vec of gimples.
16556         (free_lim_aux_data): Adjust.
16557         (add_dependency): Likewise.
16558         (set_level): Likewise.
16560 2013-03-25  Richard Biener  <rguenther@suse.de>
16562         PR middle-end/56434
16563         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
16564         the pointer returned by calls with ECF_MALLOC set.
16566 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16568         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
16570 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16572         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
16573         using MMXMODE mode iterator.
16574         (*move<mode>_internal): Merge with *movv2sf_internal and
16575         *movv2sf_internal_rex64 using MMXMODE mode iterator.
16577 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16579         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
16580         (record_last_mem_set_info): Likewise.
16582         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
16583         of XNEWVEC followed by memset.
16584         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
16586 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16588         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
16589         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
16590         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
16591         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
16592         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
16593         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
16594         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
16595         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
16596         BARRIER_P instead of GET_CODE.
16598 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16600         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
16601         inaccuracy in the probing code.
16603         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
16604         (ctrapdi4): Likewise.
16606 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16608         * calls.c (expand_call): Add missing guard to code handling return
16609         of non-BLKmode structures in MSB.
16610         * function.c (expand_function_end): Likewise.
16612 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16614         * combine.c (try_combine): Adjust comment.  Do not add the set of
16615         insn #0 if the destination indirectly is set or dies in insn #2.
16616         Tidy up code to distribute a new note.
16618 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16620         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
16621         also for alternatives 16 and 17.
16623 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16625         * config/i386/sse.md (*mov<mode>_internal): Merge with
16626         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
16627         Emit insn template depending on type attribute.  Use
16628         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16629         movd instead of movq mnemonic for interunit moves.  Rewrite mode
16630         attribute calculation.  Remove unit attribute calculation.
16631         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
16632         Set prefix_data16 attribute for DImode ssemov types.
16633         Use Ym instead of y for SSE-MMX conversion alternatives.
16634         Reorder operand constraints.
16636 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
16638         * df.h (df_insn_delete): Adjust prototype.
16639         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
16640         and let it decide whether mark the basic block dirty.
16641         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
16642         * df-scan.c (df_insn_info_delete): New helper function, split
16643         off from df_insn_delete.
16644         (df_scan_free_bb_info): Use it.
16645         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
16646         Likewise.
16647         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
16648         that the insn is actually an insn and it has a non-NULL basic block.
16649         Do not mark basic block dirty if only deleting a DEBUG_INSN.
16651 2013-03-22  Richard Biener  <rguenther@suse.de>
16653         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
16654         dep_ref members.
16655         (mem_ref_alloc): Do not allocate them.
16656         (refs_independent_p): Do not query or maintain a cache.
16658 2013-03-22  Richard Biener  <rguenther@suse.de>
16660         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
16661         (gather_mem_refs_in_loops): Do not compute it.
16662         (analyze_memory_references): Do not allocate it.
16663         (tree_ssa_lim_finalize): Do not free it.
16664         (for_all_locs_in_loop): Do not query all_refs_in_loop.
16666 2013-03-22  Richard Biener  <rguenther@suse.de>
16668         * is-a.h (as_a): Use gcc_checking_assert.
16670 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
16672         * config/aarch64/aarch64.c (aarch64_print_operand): New
16673         format specifier for printing a constant in hex.
16674         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
16675         format specifier for printing second operand.
16677 2013-03-22  Richard Biener  <rguenther@suse.de>
16679         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
16680         bitmaps.
16681         (gather_mem_refs_in_loops): Perform store accumulation here.
16682         (create_vop_ref_mapping_loop): Remove.
16683         (create_vop_ref_mapping): Likewise.
16684         (analyze_memory_references): Initialize refs_stored_in_loop.
16685         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
16686         (record_indep_loop): Remove.
16687         (record_dep_loop): New function.
16688         (ref_indep_loop_p_1): Adjust to only walk over references
16689         in the loop, not its subloops.
16690         (ref_indep_loop_p): Rename to ...
16691         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
16692         maintaining a more fine-grained cache.
16693         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
16694         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
16696 2013-03-22  Richard Biener  <rguenther@suse.de>
16698         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
16699         (struct mem_ref): Make accesses_in_loop a vec of a vec of
16700         aggregate mem_ref_loc.
16701         (free_mem_ref_locs): Inline into ...
16702         (memref_free): ... this and adjust.
16703         (mem_ref_alloc): Adjust.
16704         (mem_ref_locs_alloc): Remove.
16705         (record_mem_ref_loc): Adjust.
16706         (get_all_locs_in_loop): Rewrite into ...
16707         (for_all_locs_in_loop): ... this iterator.
16708         (rewrite_mem_ref_loc): New functor.
16709         (rewrite_mem_refs): Use for_all_locs_in_loop.
16710         (sm_set_flag_if_changed): New functor.
16711         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
16712         (ref_always_accessed): New functor.
16713         (ref_always_accessed_p): Use for_all_locs_in_loop.
16715 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16717         * tree-pass.h (PROP_gimple_lvec): New.
16718         * passes.c (dump_properties): Handle PROP_gimple_lvec.
16719         (init_optimization_passes): Move pass_lower_vector.
16720         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
16721         PROP_gimple_lvec.
16722         (pass_lower_vector): Provide PROP_gimple_lvec.
16723         (pass_lower_vector_ssa): Likewise.
16724         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
16726 2013-03-21  Mark Wielaard  <mjw@redhat.com>
16728         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
16730 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16732         * config/i386/i386.md (*movdi_internal): Disparage slightly
16733         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
16734         conversion alternatives.
16736 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
16738         PR middle-end/48087
16739         * diagnostic.def (DK_WERROR): New kind.
16740         * diagnostic.h (werrorcount): Define.
16741         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
16742         promoted to DK_ERROR, increment DK_WERROR counter instead of
16743         DK_ERROR counter.
16744         * toplev.c (toplev_main): Call print_ignored_options even if
16745         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
16746         even if just werrorcount is non-zero.
16748         PR debug/55608
16749         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
16750         on failure.
16751         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
16752         (string_cst_pool_decl): New function.
16753         (optimize_one_addr_into_implicit_ptr): New function.
16754         (resolve_addr_in_expr): Optimize DWARF location expression
16755         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
16756         which doesn't live in memory, but has DW_AT_location or
16757         DW_AT_const_value, or refers to a string literal, into
16758         DW_OP_GNU_implicit_pointer.
16759         (optimize_location_into_implicit_ptr): New function.
16760         (resolve_addr): If removing DW_AT_location of a variable because
16761         it was DW_OP_addr of address of the variable, but the variable doesn't
16762         live in memory, try to emit const value attribute for the initializer.
16764 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16766         * tree.h (VECTOR_TYPE_P): New macro.
16767         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
16768         TYPE_MODE): Use it.
16769         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
16770         VEC_COND_EXPR cannot be lvalues.
16771         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
16773 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16775         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
16776         Restrict the transformation to equal modes.
16778 2013-03-21  Richard Biener  <rguenther@suse.de>
16780         PR tree-optimization/39326
16781         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
16782         (MEM_ANALYZABLE): Adjust.
16783         (record_mem_ref_loc): Move bitmap ops ...
16784         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
16785         unanalyzable refs, do not record locations for it.
16786         (analyze_memory_references): Allocate ref zero as shared
16787         unanalyzable ref.
16788         (refs_independent_p): Do not test for unanalyzed mems here.
16789         (ref_indep_loop_p_1): Special-case disambiguation against
16790         the unanalyzed ref.
16791         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
16793 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
16795         * config/arm/arm-protos.h (tune_params): Add
16796         prefer_neon_for_64bits field.
16797         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
16798         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
16799         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
16800         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
16801         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
16802         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
16803         (arm_option_override): Handle -mneon-for-64bits new option.
16804         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
16805         (prefer_neon_for_64bits): Declare new variable.
16806         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
16807         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
16808         (arch_enabled): Handle new arch types. Remove support for onlya8
16809         and nota8.
16810         (one_cmpldi2): Use new arch names.
16811         (zero_extend<mode>di2, extend<mode>di2): Ditto.
16812         * config/arm/arm.opt (mneon-for-64bits): Add option.
16813         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
16814         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
16815         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
16816         of onlya8.
16817         * doc/invoke.texi (-mneon-for-64bits): Document.
16819 2013-03-21  Richard Biener  <rguenther@suse.de>
16821         PR tree-optimization/39326
16822         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
16823         (sort_bbs_in_loop_postorder_cmp): New function.
16824         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
16826 2013-03-21  Richard Biener  <rguenther@suse.de>
16828         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
16829         (vect_insert_into_interleaving_chain): Likewise.
16830         (vect_drs_dependent_in_basic_block): Inline ...
16831         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
16832         split out from ...
16833         (vect_analyze_data_ref_dependence): ... here.  Simplify.
16834         (vect_check_interleaving): Simplify.
16835         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
16836         (vect_slp_analyze_data_ref_dependences): ... this new function.
16837         (dr_group_sort_cmp): New function.
16838         (vect_analyze_data_ref_accesses): Compute data-reference groups
16839         here instead of in vect_analyze_data_ref_dependence.  Use
16840         a more efficient algorithm.
16841         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
16842         vect_slp_analyze_data_ref_dependences.  Call
16843         vect_analyze_data_ref_accesses earlier.
16844         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
16845         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
16846         (vect_slp_analyze_data_ref_dependences): New prototype.
16848 2013-03-21  Richard Biener  <rguenther@suse.de>
16850         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
16851         ref is stored in the loop.
16852         (find_refs_for_sm): Walk only over all stores.
16853         (store_motion_loop): Allocate from lim_bitmap_obstack.
16854         (store_motion): Likewise.
16856 2013-03-21  Richard Biener  <rguenther@suse.de>
16858         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
16859         Update virtual SSA form.
16861 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16863         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
16864         * configure: Regenerate.
16865         * config.in: Regenerate.
16866         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
16867         if !HAVE_LD_EH_FRAME_CIEV3.
16869 2013-03-21  Richard Biener  <rguenther@suse.de>
16871         * tree-cfg.c (verify_expr_no_block): New function.
16872         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
16873         nor DECL_VALUE_EXPR have locations with associated blocks.
16874         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
16875         (clear_unused_block_pointer): Remove code dealing with
16876         blocks in DECL_DEBUG_EXPR locations.
16878 2013-03-21  Richard Biener  <rguenther@suse.de>
16880         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
16881         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
16882         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
16883         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
16884         instead of DECL_DEBUG_EXPR_IS_FROM.
16885         * gimplify.c (gimplify_modify_expr): Likewise.
16886         * tree-cfg.c (verify_expr_location_1): Likewise.
16887         * tree-complex.c (create_one_component_var): Likewise.
16888         * tree-sra.c (create_access_replacement): Likewise.
16889         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
16890         (clear_unused_block_pointer): Likewise.
16891         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
16892         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
16893         * var-tracking.c (var_debug_decl): Likewise.
16894         (track_expr_p): Likewise.
16895         * tree-inline.c (add_local_variables): Likewise.  Set
16896         DECL_HAS_DEBUG_EXPR_P after copying it.
16897         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
16898         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
16900 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16902         PR bootstrap/56656
16903         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
16904         * configure: Regenerate.
16905         * config.in: Regenerate.
16906         * config/i386/i386.md (*movdf_internal): Use
16907         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16908         movd instead of movq mnemonic for interunit moves.
16909         (*movdi_internal): Ditto.
16911 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16913         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
16914         (abd<mode>_3): New pattern.
16915         (aba<mode>_3): New pattern.
16916         (fabd<mode>_3): New pattern.
16918 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16920         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
16921         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
16922         occurrence of REGISTER_PREFIX as its empty string.
16924 2013-03-20  Jeff Law  <law@redhat.com>
16926         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
16927         addititional equivalences for equality comparisons between an SSA_NAME
16928         and a constant where the SSA_NAME was set from a widening conversion.
16930 2013-03-20  Walter Lee  <walt@tilera.com>
16932         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
16934 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16936         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
16937         depending on type attribute.
16938         (*movti_internal): Ditto.
16939         (*movtf_internal): Ditto.
16940         (*movxf_internal): Ditto.
16941         (*movdf_internal): Ditto.
16942         (*movsf_internal): Ditto.
16944 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16946         * config/i386/i386.md (*movti_internal): Set prefix attribute to
16947         maybe_vex for sselog1 and ssemov types.
16948         (*movdi_internal): Reorder operand constraints.
16949         (*movsi_internal): Ditto.  Set prefix attribute to
16950         maybe_vex for sselog1 and ssemov types.
16951         (*movtf_internal): Set prefix attribute to maybe_vex
16952         for sselog1 and ssemov types.
16953         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
16954         DImode ssemov types.  Reorder operand constraints.
16955         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
16956         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
16957         attribute for SImode ssemov types.  Reorder operand constraints.
16959 2013-03-20  Martin Jambor  <mjambor@suse.cz>
16961         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
16962         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
16964 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
16966         * config/rs6000/predicates.md (indexed_address, update_address_mem
16967         update_indexed_address_mem): New predicates.
16968         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
16969         attribute for load/store instructions.
16970         * config/rs6000/dfp.md (movsd_store): Likewise.
16971         (movsd_load): Likewise.
16972         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
16973         (unnamed HI->DI extend define_insn): Likewise.
16974         (unnamed SI->DI extend define_insn): Likewise.
16975         (unnamed QI->SI extend define_insn): Likewise.
16976         (unnamed QI->HI extend define_insn): Likewise.
16977         (unnamed HI->SI extend define_insn): Likewise.
16978         (unnamed HI->SI extend define_insn): Likewise.
16979         (extendsfdf2_fpr): Likewise.
16980         (movsi_internal1): Likewise.
16981         (movsi_internal1_single): Likewise.
16982         (movhi_internal): Likewise.
16983         (movqi_internal): Likewise.
16984         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
16985         attribute for load/store instructions.
16986         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
16987         instructions.
16988         (mov<mode>_softfloat): Likewise.
16989         (mov<mode>_hardfloat32): Likewise.
16990         (mov<mode>_hardfloat64): Likewise.
16991         (mov<mode>_softfloat64): Likewise.
16992         (movdi_internal32): Likewise.
16993         (movdi_internal64): Likewise.
16994         (probe_stack_<mode>): Likewise.
16996 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
16998         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
16999         floating point, and decimal floating point to reload iterator.
17001         * config/rs6000/constraints.md (wl constraint): New constraints to
17002         return FLOAT_REGS if certain options are used to reduce the number
17003         of separate patterns that exist in the file.
17004         (wx constraint): Likewise.
17005         (wz constraint): Likewise.
17007         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17008         -mdebug=reg, print wg, wl, wx, and wz constraints.
17009         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17010         Initialize the reload functions for 64-bit binary/decimal floating
17011         point types.
17012         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17013         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17014         create the buffer on the stack to overcome not having a 32-bit
17015         load and store.
17016         (rs6000_emit_move): Likewise.
17017         (rs6000_secondary_memory_needed_rtx): Likewise.
17018         (rs6000_alloc_sdmode_stack_slot): Likewise.
17019         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17020         via xxlxor, just like DFmode 0.0.
17022         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17023         define as 1 if we are running on a power7 or newer.
17024         (enum r6000_reg_class_enum): Add new constraints.
17026         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17027         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17028         with other moves by using conditional constraits (wg).  Use LFIWZX
17029         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17030         (movsd splitter): Likewise.
17031         (movsd_hardfloat): Likewise.
17032         (movsd_softfloat): Likewise.
17034         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17035         binary and decimal floating point moves.
17036         (fmove_ok): New attributes to combine binary and decimal floating
17037         point moves, and to combine power6x (mfpgpr) moves along normal
17038         floating moves.
17039         (real_value_to_target): Likewise.
17040         (f32_lr): Likewise.
17041         (f32_lm): Likewise.
17042         (f32_li): Likewise.
17043         (f32_sr): Likewise.
17044         (f32_sm): Likewise.
17045         (f32_si): Likewise.
17046         (movsf): Combine binary and decimal floating point moves.  Combine
17047         power6x (mfpgpr) moves with other moves by using conditional
17048         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17049         (mov<mode> for SFmode/SDmode); Likewise.
17050         (SFmode/SDmode splitters): Likewise.
17051         (movsf_hardfloat): Likewise.
17052         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17053         (movsf_softfloat): Likewise.
17054         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17056         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17057         wx and wz constraints.
17059         * config/rs6000/constraints.md (wg constraint): New constraint to
17060         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17062         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17063         constraint.
17065         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17066         -mdebug=reg, print wg, wl, wx, and wz constraints.
17067         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17068         Initialize the reload functions for 64-bit binary/decimal floating
17069         point types.
17070         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17071         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17072         create the buffer on the stack to overcome not having a 32-bit
17073         load and store.
17074         (rs6000_emit_move): Likewise.
17075         (rs6000_secondary_memory_needed_rtx): Likewise.
17076         (rs6000_alloc_sdmode_stack_slot): Likewise.
17077         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17078         via xxlxor, just like DFmode 0.0.
17080         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17081         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17082         with other moves by using conditional constraits (wg).  Use LFIWZX
17083         and STFIWX for loading SDmode on power7.
17084         (movdd splitters): Likewise.
17085         (movdd_hardfloat32): Likewise.
17086         (movdd_softfloat32): Likewise.
17087         (movdd_hardfloat64_mfpgpr): Likewise.
17088         (movdd_hardfloat64): Likewise.
17089         (movdd_softfloat64): Likewise.
17091         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17092         64-bit binary and decimal floating point moves.
17093         (FMOVE64X): Likewise.
17094         (movdf): Combine 64-bit binary and decimal floating point moves.
17095         Combine power6x (mfpgpr) moves with other moves by using
17096         conditional constraits (wg).
17097         (mov<mode> for DFmode/DDmode): Likewise.
17098         (DFmode/DDmode splitters): Likewise.
17099         (movdf_hardfloat32): Likewise.
17100         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17101         (movdf_softfloat32): Likewise.
17102         (movdf_hardfloat64_mfpgpr): Likewise.
17103         (movdf_hardfloat64): Likewise.
17104         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17105         (movdf_softfloat64): Likewise.
17106         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17107         (reload_<mode>_load): Move to later in the file so they aren't in
17108         the middle of the floating point move insns.
17109         (reload_<mode>_store): Likewise.
17111         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17112         constraint.
17114         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17115         constraint if -mdebug=reg.
17116         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17117         Enable using dd reload support if needed.
17119         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17120         binary and decimal floating point moves in rs6000.md.
17121         (movtd_internal): Likewise.
17123         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17124         decimal floating point moves.
17125         (movtf): Likewise.
17126         (movtf_internal): Likewise.
17127         (mov<mode>_internal, TDmode/TFmode): Likewise.
17128         (movtf_softfloat): Likewise.
17129         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17131         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17132         movdi_internal64, using wg constraint for move direct operations.
17133         (movdi_internal64): Likewise.
17135         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17136         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17137         the various virtual registers. Use GPR/FPR first/last values,
17138         instead of hard coding the register numbers.  Print which modes
17139         have reload functions registered.
17140         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17141         settings before/after setting cpu, target and subtarget settings.
17142         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17143         and for secondary reload failures in rs6000_secondary_reload_inner.
17144         (rs6000_secondary_reload_fail): Likewise.
17145         (rs6000_secondary_reload_inner): Likewise.
17147         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17148         macros for first/last GPR and FPR registers.
17149         (LAST_GPR_REGNO): Likewise.
17150         (FIRST_FPR_REGNO): Likewise.
17151         (LAST_FPR_REGNO): Likewise.
17153         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17154         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17155         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17156         (vcond<mode><mode>): Likewise.
17157         (vcondu<mode><mode>): Likewise.
17158         (vector_gtu<mode>): Likewise.
17159         (vector_gte<mode>): Likewise.
17160         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17161         to prevent the compiler from converting DImode operations to TImode.
17162         (ior<mode>3): Likewise.
17163         (and<mode>3): Likewise.
17164         (one_cmpl<mode>2): Likewise.
17165         (nor<mode>3): Likewise.
17166         (andc<mode>3): Likewise.
17168         * config/rs6000/constraints.md (wt constraint): New constraint
17169         that returns VSX_REGS if TImode is allowed in VSX registers.
17171         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17172         constant under VSX.
17174         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17175         similar to TImode, but it is restricted to being in the GPRs.
17177         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17178         TImode to occupy a single VSX register.
17180         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17181         -mvsx-timode for power7/power8.
17182         (power7 cpu): Likewise.
17183         (power8 cpu): Likewise.
17185         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17186         sure that TFmode/TDmode take up two registers if they are ever
17187         allowed in the upper VSX registers.
17188         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17189         registers.
17190         (rs6000_init_hard_regno_mode_ok): Likewise.
17191         (rs6000_debug_reg_global): Add debugging for PTImode and wt
17192         constraint.  Print if LRA is turned on.
17193         (rs6000_option_override_internal): Give an error if -mvsx-timode
17194         and VSX is not enabled.
17195         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
17196         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17197         to reg+offset addressing.  Use PTImode when checking offset
17198         addresses for validity.
17199         (reg_offset_addressing_ok_p): Likewise.
17200         (rs6000_legitimate_offset_address_p): Likewise.
17201         (rs6000_legitimize_address): Likewise.
17202         (rs6000_legitimize_reload_address): Likewise.
17203         (rs6000_legitimate_address_p): Likewise.
17204         (rs6000_eliminate_indexed_memrefs): Likewise.
17205         (rs6000_emit_move): Likewise.
17206         (rs6000_secondary_reload): Likewise.
17207         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
17208         reloads to fpr registers to continue to use reg+offset addressing,
17209         but 64-bit reloads to altivec registers need reg+reg addressing.
17210         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17211         it.  Treat LO_SUM like a PLUS operation.
17212         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17213         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17214         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17215         registers to share a register with a smaller sized type, since VSX
17216         puts scalars in the upper 64-bits.
17217         (print_operand): Add support for PTImode.
17218         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17219         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17220         registers, but don't have arithmetic support.
17221         (rs6000_memory_move_cost): Add test for VSX.
17222         (rs6000_opt_masks): Add -mvsx-timode.
17224         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17225         for TImode.
17226         (VSs): Likewise.
17227         (VSr): Use wt constraint for TImode.
17228         (VSv): Drop TImode support.
17229         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17230         (vsx_movti_64bit): Likewise.
17231         (vsx_movti_32bit): Likewise.
17232         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17233         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17234         one '?' on the appropriate output constraint.  Do not allow TImode
17235         logical operations on 32-bit systems.
17236         (vsx_ior<mode>3): Likewise.
17237         (vsx_xor<mode>3): Likewise.
17238         (vsx_one_cmpl<mode>2): Likewise.
17239         (vsx_nor<mode>3): Likewise.
17240         (vsx_andc<mode>3): Likewise.
17241         (vsx_concat_<mode>): Likewise.
17242         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17244         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17245         OPTION_MASK_VSX_TIMODE.
17246         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17247         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17249         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17250         (TI2 iterator): New iterator for TImode, PTImode.
17251         (wd mode attribute): Add values for vector types.
17252         (movti_string): Replace TI move operations with operations for TImode
17253         and PTImode.  Add support for TImode being allowed in VSX registers.
17254         (mov<mode>_string, TImode/PTImode): Likewise.
17255         (movti_ppc64): Likewise.
17256         (mov<mode>_ppc64, TImode/PTImode): Likewise.
17257         (TI mode splitters): Likewise.
17259         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17260         constraint.
17262 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
17264         PR tree-optimization/56355
17265         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17266         Also handle integers with undefined overflow.
17268 2013-03-20  Catherine Moore  <clm@codesourcery.com>
17269             Maciej W. Rozycki  <macro@codesourcery.com>
17270             Tom de Vries  <tom@codesourcery.com>
17271             Nathan Sidwell  <nathan@codesourcery.com>
17272             Iain Sandoe  <iain@codesourcery.com>
17273             Nathan Froyd  <froydnj@codesourcery.com>
17274             Chao-ying Fu  <fu@mips.com>
17276         * doc/extend.texi: (micromips, nomicromips, nocompression):
17277         Document new function attributes.
17278         * doc/invoke.texi (minterlink-compressed, mmicromips,
17279         m14k, m14ke, m14kec): Document new options.
17280         (minterlink-mips16): Update documentation.
17281         * doc/md.texi (ZC, ZD): Document new constraints.
17282         * configure.ac (gcc_cv_as_micromips): Check if linker
17283         supports the .set micromips directive.
17284         * configure: Regenerate.
17285         * config.in: Regenerate.
17286         * config/mips/mips-tables.opt: Regenerate.
17287         * config/mips/micromips.md: New file.
17288         * constraints.md (ZC, ZD): New constraints.
17289         * config/mips/predicates.md (movep_src_register): New predicate.
17290         (movep_src_operand): New predicate.
17291         (non_volatile_mem_operand): New predicate.
17292         * config/mips/mips.md (multimem): New type.
17293         (length): Differentiate between 17-bit and 18-bit branch offsets.
17294         (MOVEP1, MOVEP2): New mode iterator.
17295         (mov_<load>l): Use ZC constraint.
17296         (mov_<load>r): Likewise.
17297         (mov_<store>l): Likewise.
17298         (mov_<store>r): Likewise.
17299         (*branch_equality<mode>_inverted): Add microMIPS support.
17300         (*branch_equality<mode>): Likewise.
17301         (*jump_absolute): Likewise.
17302         (indirect_jump_<mode>): Likewise.
17303         (tablejump_<mode>): Likewise.
17304         (<optab>_internal): Likewise.
17305         (sibcall_internal): Likewise.
17306         (sibcall_value_internal): Likewise.
17307         (prefetch): Use constraint ZD.
17308         * config/mips/mips.opt (minterlink-compressed): New option.
17309         (minterlink-mips16): Now an alias for minterlink-compressed.
17310         (mmicromips): New option.
17311         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17312         (compare_and_swap_12): Likewise.
17313         (sync_add<mode>): Likewise.
17314         (sync_<optab>_12): Likewise.
17315         (sync_old_<optab>_12): Likewise.
17316         (sync_new_<optab>_12): Likewise.
17317         (sync_nand_12): Likewise.
17318         (sync_old_nand_12): Likewise.
17319         (sync_new_nand_12): Likewise.
17320         (sync_sub<mode>): Likewise.
17321         (sync_old_add<mode>): Likewise.
17322         (sync_old_sub<mode>): Likewise.
17323         (sync_new_add<mode>): Likewise.
17324         (sync_new_sub<mode>): Likewise.
17325         (sync_<optab><mode>): Likewise.
17326         (sync_old_<optab><mode>): Likewise.
17327         (sync_new_<optab><mode>): Likewise.
17328         (sync_nand<mode>): Likewise.
17329         (sync_old_nand<mode>): Likewise.
17330         (sync_new_nand<mode>): Likewise.
17331         (sync_lock_test_and_set<mode>): Likewise.
17332         (test_and_set_12): Likewise.
17333         (atomic_compare_and_swap<mode>): Likewise.
17334         (atomic_exchange<mode>_llsc): Likewise.
17335         (atomic_fetch_add<mode>_llsc): Likewise.
17336         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17337         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17338         (umips_save_restore_pattern_p): Likewise.
17339         (umips_load_store_pair_p): Likewise.
17340         (umips_output_load_store_pair): Likewise.
17341         (umips_movep_target_p): Likewise.
17342         (umips_12bit_offset_address_p): Likewise.
17343         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17344         (mips_base_mips16): Rename this...
17345         (mips_base_compression_flags): ...to this. Update all uses.
17346         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17347         (mips_mips16_decl_p): Delete.
17348         (mips_nomips16_decl_p): Delete.
17349         (mips_get_compress_on_flags): New function.
17350         (mips_get_compress_off_flags): New function.
17351         (mips_get_compress_mode): New function.
17352         (mips_get_compress_on_name): New function.
17353         (mips_get_compress_off_name): New function.
17354         (mips_insert_attributes): Support multiple compression types.
17355         (mips_merge_decl_attributes): Likewise.
17356         (umips_12bit_offset_address_p): New function.
17357         (mips_start_function_definition): Emit .set micromips directive.
17358         (mips_call_may_need_jalx_p): New function.
17359         (mips_function_ok_for_sibcall): Add microMIPS support.
17360         (mips_print_operand_punctuation): Support short delay slots and
17361         compact jumps.
17362         (umips_swm_mask, umips_swm_encoding): New.
17363         (umips_build_save_restore): New function.
17364         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17365         (was_mips16_p): Remove.
17366         (old_compression_mode): New.
17367         (mips_set_compression_mode): New function.
17368         (mips_set_current_function): Add microMIPS support.
17369         (mips_option_override): Likewise.
17370         (umips_save_restore_pattern_p): New function.
17371         (umips_output_save_restore): New function.
17372         (umips_load_store_pair_p_1): New function.
17373         (umips_load_store_pair_p): New function.
17374         (umips_output_load_store_pair_1): New function.
17375         (umips_output_load_store_pair): New function.
17376         (umips_movep_target_p) New function.
17377         (mips_prepare_pch_save): Add microMIPS support.
17378         * config/mips/mips.h (TARGET_COMPRESSION): New.
17379         (TARGET_CPU_CPP_BUILTINS): Update macro
17380         to use new compression flags and to support microMIPS.
17381         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17382         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17383         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17384         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17385         (ASM_SPEC): Support mmicromips and mno-micromips.
17386         (M16STORE_REG_P): New macro.
17387         (MIPS_CALL): Support TARGET_MICROMIPS.
17388         (MICROMIPS_J): New macro.
17389         (mips_base_mips16): Rename this...
17390         (mips_base_compression_flags): ...to this.
17391         (UMIPS_12BIT_OFFSET_P): New macro.
17392         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17393         (MULTILIB_DIRNAMES): Likewise.
17394 2013-03-20  Richard Biener  <rguenther@suse.de>
17396         PR tree-optimization/56661
17397         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17398         the result does not have to be distinct.
17400 2013-03-20  Richard Biener  <rguenther@suse.de>
17402         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17403         remap_gimple_op_r.
17405 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17406             Steven Bosscher  <steven@gcc.gnu.org>
17408         PR rtl-optimization/56605
17409         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17411 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17413         PR bootstrap/56656
17414         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17415         that require movd instead of movq.
17417 2013-03-20  Richard Biener  <rguenther@suse.de>
17419         * tree-ssa-structalias.c (struct variable_info): Add pointer
17420         to the first field of an aggregate with sub-vars.  Make
17421         this and the pointer to the next subfield its ID.
17422         (vi_next): New function.
17423         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17424         storedanything_id, integer_id): Increment by one.
17425         (new_var_info, get_call_vi, lookup_call_clobber_vi,
17426         get_call_clobber_vi): Adjust.
17427         (solution_set_expand): Simplify and speedup.
17428         (solution_set_add): Inline into ...
17429         (set_union_with_increment): ... this.  Adjust accordingly.
17430         (do_sd_constraint): Likewise.
17431         (do_ds_constraint): Likewise.
17432         (do_complex_constraint): Simplify.
17433         (build_pred_graph): Adjust.
17434         (solve_graph): Likewise.  Simplify and speedup.
17435         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17436         get_constraint_for_component_ref, get_constraint_for_1,
17437         first_vi_for_offset, first_or_preceding_vi_for_offset,
17438         create_function_info_for, create_variable_info_for_1,
17439         create_variable_info_for, intra_create_variable_infos): Adjust.
17440         (init_base_vars): Push NULL for ID zero.
17441         (compute_points_to_sets): Adjust.
17443 2013-03-20  Richard Biener  <rguenther@suse.de>
17445         * cfgloop.c (verify_loop_structure): Streamline and avoid
17446         ICEing on corrupt loop tree.
17447         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17448         loop tree.
17450 2013-03-20  Richard Biener  <rguenther@suse.de>
17452         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17453         check whether an SSA update is needed.
17455 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
17457         * config/mips/constraints.md (T): Rename to...
17458         (Yf): ...this.
17459         (U): Rename to...
17460         (Yd): ...this.
17461         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17462         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17464 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17466         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17467         (*subsi3_carryin_uxtw): Likewise.
17469 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17471         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17472         (*rorsi3_insn_uxtw): Likewise.
17474 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17476         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17477         (*extrsi5_insn_uxtw): Likewise.
17479 2013-03-19  Richard Biener  <rguenther@suse.de>
17481         PR tree-optimization/56273
17482         * passes.c (init_optimization_passes): Move second VRP after DOM.
17484 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
17486         * config/i386/i386.md (*movti_internal): Merge from
17487         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
17488         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
17489         nox64 isa attributes.
17491 2013-03-18  Richard Biener  <rguenther@suse.de>
17493         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17494         (unite): Likewise.
17495         (merge_node_constraints): Likewise.
17496         (build_succ_graph): Likewise.
17497         (valid_graph_edge): Inline into single caller.
17498         (unify_nodes): Likewise.  Use bitmap_set_bit return value
17499         and cache varinfo.
17500         (scc_visit): Fix formatting and variable use.
17501         (do_sd_constraint): Use gcc_checking_assert.
17502         (do_ds_constraint): Likewise.
17503         (do_complex_constraint): Likewise.
17504         (condense_visit): Likewise.  Cleanup.
17505         (dump_pred_graph): New function.
17506         (perform_var_substitution): Dump the pred-graph before
17507         variable substitution.
17508         (find_equivalent_node): Use gcc_checking_assert.
17509         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17511 2013-03-18  Richard Biener  <rguenther@suse.de>
17513         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17514         Remove cond_expr_stmt_list argument and do not gimplify the
17515         built expression.
17516         (vect_loop_versioning): Adjust.
17517         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17518         Cleanup to use less temporaries.
17519         (vect_create_data_ref_ptr): Cleanup.
17521 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17523         PR tree-optimization/56635
17524         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17525         require types_compatible_p types.
17527 2013-03-18  Nick Clifton  <nickc@redhat.com>
17529         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
17530         spurious backslash.
17532         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
17533         Add missing line to comment describing function.
17535 2013-03-18  Richard Biener  <rguenther@suse.de>
17537         PR tree-optimization/56210
17538         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17539         Handle string / character search functions.
17540         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
17542 2013-03-18  Richard Biener  <rguenther@suse.de>
17544         PR middle-end/56483
17545         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
17546         and implement properly.
17547         * gimple.h (gimple_cond_single_var_p): Remove.
17549 2013-03-18  Richard Biener  <rguenther@suse.de>
17551         * tree-data-ref.h (find_data_references_in_loop): Declare.
17552         * tree-data-ref.c (get_references_in_stmt): Use a stack
17553         vector pre-allocated in the callers.
17554         (find_data_references_in_stmt): Adjust.
17555         (graphite_find_data_references_in_stmt): Likewise.
17556         (create_rdg_vertices): Likewise.
17557         (find_data_references_in_loop): Export.
17558         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17559         Compute dependences here...
17560         (vect_analyze_data_refs): ...not here.  When we encounter
17561         a non-vectorizable data reference in basic-block vectorization
17562         truncate the data reference vector.  Do not bother to
17563         fixup data-dependence information for gather loads.
17564         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
17565         of data references, as reported.
17567 2013-03-18  Richard Biener  <rguenther@suse.de>
17569         PR tree-optimization/3713
17570         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
17571         has_constants and expr.
17572         (stmt_has_constants): Properly valueize SSA names when deciding
17573         whether the stmt has constants.
17575 2013-03-18  Richard Biener  <rguenther@suse.de>
17577         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
17578         whole function when there is nothing to do.
17579         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
17580         * tree-vectorizer.c (vectorize_loops): Update virtual and
17581         loop-closed SSA once.
17582         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
17584 2013-03-18  Richard Biener  <rguenther@suse.de>
17586         PR middle-end/56113
17587         * domwalk.c (bb_postorder): New global static.
17588         (cmp_bb_postorder): New function.
17589         (walk_dominator_tree): Replace scheme imposing an order for
17590         visiting dominator sons by one sorting them at the time they
17591         are pushed on the stack.
17593 2013-03-18  Richard Biener  <rguenther@suse.de>
17595         PR tree-optimization/39326
17596         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
17597         (struct mem_ref): Replace mem member with ao_ref typed member.
17598         (MEM_ANALYZABLE): Adjust.
17599         (memref_eq): Likewise.
17600         (mem_ref_alloc): Likewise.
17601         (gather_mem_refs_stmt): Likewise.
17602         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
17603         (execute_sm_if_changed_flag_set): Adjust.
17604         (execute_sm): Likewise.
17605         (ref_always_accessed_p): Likewise.
17606         (refs_independent_p): Likewise.
17607         (can_sm_ref_p): Likewise.
17609 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17611         PR c/56566
17612         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
17613         return 1 even for !unsignedp.
17615 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17617         * config/i386/i386.md (isa): Add x64 and nox64.
17618         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
17619         (*pushtf): Enable *roF alternative for x64 isa only.
17620         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
17621         mode attribute of integer alternatives to DImode for TARGET_64BIT.
17622         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
17623         (*movtf_internal): Merge from *movtf_internal_rex64 and
17624         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
17625         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
17626         nox64 isa attributes.
17627         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
17628         nox64 isa attributes.
17629         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
17631 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17633         * config/alpha/alpha.c (TARGET_LRA_P): New define.
17635 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
17637         PR target/56640
17638         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
17639         class names.  Remove trailing comma after "ALL_REGS".
17641 2013-03-16  Jan Hubicka  <jh@suse.cz>
17643         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
17644         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
17645         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
17646         of cgraph_get_create_node.
17647         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17649 2013-03-16  Jason Merrill  <jason@redhat.com>
17651         PR debug/49090
17652         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
17653         with DW_AT_default_value.
17655 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
17657         * BASE-VER: Set to 4.9.0.
17659 2013-03-14  Andi Kleen  <ak@linux.intel.com>
17661         PR target/56619
17662         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
17663         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
17664         Document _x* TSX intrinsics.
17666 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
17667             David Holsgrove  <david.holsgrove@xilinx.com>
17669         * configure.ac: Add MicroBlaze TLS support detection.
17670         * configure: Regenerate.
17671         * config/microblaze/microblaze-protos.h
17672         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
17673         symbol_mentioned_p, label_mentioned_p): Add prototypes.
17674         * config/microblaze/microblaze.c (microblaze_address_type): Add
17675         ADDRESS_TLS and tls_reloc address types.
17676         (microblaze_address_info): Add tls_reloc.
17677         (TARGET_HAVE_TLS): Define.
17678         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
17679          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
17680          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
17681         load_tls_operand,  microblaze_call_tls_get_addr,
17682         microblaze_legitimize_tls_address): New functions.
17683         (microblaze_classify_unspec): Handle UNSPEC_TLS.
17684         (get_base_reg): Use microblaze_tls_symbol_p.
17685         (microblaze_classify_address): Handle TLS.
17686         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
17687         label_mentioned_p and microblaze_tls_referenced_p.
17688         (microblaze_legitimize_address): Handle TLS.
17689         (microblaze_address_insns): Handle ADDRESS_TLS.
17690         (pic_address_needs_scratch): Handle TLS.
17691         (print_operand_address): Handle TLS.
17692         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
17693         (microblaze_expand_move): Handle TLS.
17694         (microblaze_legitimate_constant_p): Check
17695         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
17696         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
17697         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
17698         (PIC_OFFSET_TABLE_REGNUM): Set.
17699         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
17700         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
17701         (addsi3, movsi_internal2, movdf_internal): Update constraints
17702         * config/microblaze/predicates.md (arith_plus_operand): Define
17703         (move_operand): Redefine as move_src_operand,
17704         check microblaze_tls_referenced_p.
17706 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17708         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
17709         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
17711 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17713         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
17714         CC mode for AND.
17716 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
17718         PR tree-optimization/53265
17719         * common.opt (Waggressive-loop-optimizations): New option.
17720         * tree-ssa-loop-niter.c: Include tree-pass.h.
17721         (do_warn_aggressive_loop_optimizations): New function.
17722         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
17723         if number_of_latch_executions returned constant.
17724         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
17725         early.  If number_of_latch_executions returned constant, set
17726         nb_iterations_upper_bound back to it.
17727         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
17728         field.
17729         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
17730         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
17732         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
17733         (MULTILIB_OSDIRNAMES): Set.
17734         * genmultilib: If defaultosdirname doesn't start with :: , set
17735         defaultosdirname2 instead, clear it and emit two . multilib_raw
17736         entries instead of just one.
17738 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
17740         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17741         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
17742         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17743         (SUBTARGET_OVERRIDE_OPTIONS): New.
17745 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
17747         PR target/49880
17748         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
17749         (musermode): Convert to Var(TARGET_USERMODE).
17750         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
17751         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
17752         * config/sh/sh.c (sh_option_override): Use
17753         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
17754         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
17755         condition.
17756         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
17757         TARGET_SH4.
17758         (udivsi3_i4_single, divsi3_i4_single): Use
17759         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
17761 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
17763         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
17764         default setting.
17766 2013-03-13  Richard Biener  <rguenther@suse.de>
17768         PR tree-optimization/56608
17769         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
17770         calls when vectorizing basic-blocks.
17772 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
17774         PR plugins/45078
17775         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
17776         tm_file.
17778 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
17780         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
17782 2013-03-11  Jan Hubicka  <jh@suse.cz>
17784         PR lto/56557
17785         * lto-streamer-out.c (output_symbol_p): Skip references from
17786         constructors of external variables.
17788 2013-03-11  Jan Hubicka  <jh@suse.cz>
17790         PR middle-end/56571
17791         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
17792         from pseudos.
17793         * emit-rtl.c (verify_rtx_sharing): Likewise.
17794         (copy_insn_1): Likewise.
17795         * rtl.c (copy_rtx): Likewise.
17797 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
17799         PR target/56591
17800         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
17801         output_operand_lossage message.
17803 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
17805         PR target/56470
17806         * arm.c (shift_op): Validate RTL pattern on the fly.
17807         (arm_print_operand, case 'S'): Don't use shift_operator to validate
17808         the RTL.
17810 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17812         PR target/56347
17813         * config/pa/pa.md (call_value): Check for calls to powf and direct to
17814         new call patterns that clobber %fr12.
17815         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
17816         split and postreload patterns.
17817         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
17818         registers %fr12 and %fr12R as call used.
17820 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
17822         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
17823         (canon_address, record_store, replace_read, check_mem_read_rtx,
17824         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
17825         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
17826         rest_of_handle_dse): Likewise.
17828 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
17830         PR middle-end/56524
17831         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
17832         Add base_optabs.
17833         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
17834         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
17835         (save_optabs_if_changed): Replace with...
17836         (init_tree_optimization_optabs): ...this.
17837         * optabs.c (save_optabs_if_changed): Rename to...
17838         (init_tree_optimization_optabs): ...this.  Take the optimization node
17839         as argument.  Do nothing if the base optabs are already correct.
17840         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
17841         to recompute optabs.
17842         * function.h (function): Remove optabs field.
17843         * function.c (invoke_set_current_function_hook): Call
17844         init_tree_optimization_optabs.  Use the result to initialize
17845         this_fn_optabs.
17847 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
17849         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
17850         if GTMA_HAS_NO_INSTRUMENTATION.
17851         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
17852         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
17853         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
17854         * gimple-pretty-print.c (dump_gimple_transaction): Handle
17855         GTMA_HAS_NO_INSTRUMENTATION.
17857 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
17859         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
17860         libasan_preinit.o.
17862 2013-03-08  Marek Polacek  <polacek@redhat.com>
17863             Jakub Jelinek  <jakub@redhat.com>
17865         PR tree-optimization/56478
17866         * predict.c (is_comparison_with_loop_invariant_p): Change the
17867         type of loop_step to tree.
17868         (predict_loops): Adjust.
17869         (predict_iv_comparison): Perform the computations on double_ints.
17871 2013-03-08  Richard Biener  <rguenther@suse.de>
17873         PR tree-optimization/56570
17874         * tree-cfg.c (verify_expr_location_1): Verify locations for
17875         DECL_DEBUG_EXPR.
17876         * tree-sra.c (create_access_replacement): Strip locations
17877         from DECL_DEBUG_EXPRs.
17879 2013-03-08  Richard Biener  <rguenther@suse.de>
17881         * tree-inline.c (expand_call_inline): Do not associate
17882         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
17883         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
17885 2013-03-08  Richard Biener  <rguenther@suse.de>
17887         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
17888         or block changes with -Og.  Fix for location / block encoding
17889         changes and PHI arguments with locations.
17891 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
17893         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
17894         for all counters.
17895         (struct output_info): Likewise.
17896         (register_overhead): Remove bad gcc_assert.
17897         (bitmap_find_bit): If there is only a single bitmap element, do not
17898         count a miss as a search.
17899         (print_statistics): Update for counter type changes.
17900         (dump_bitmap_statistics): Likewise.  Print headers such that they
17901         are properly lined up with the printed counters.
17903 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
17905         PR tree-optimization/56559
17906         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
17907         check that it has only a single use.
17909 2013-03-07  Richard Biener  <rguenther@suse.de>
17911         * doc/invoke.texi (fwhole-program): Discourage use in combination
17912         with -flto.
17914 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
17916         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
17918         PR tree-optimization/56539
17919         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
17920         instead of GSI_CONTINUE_LINKING as last argument to
17921         force_gimple_operand_gsi.  Adjust function comment.
17923         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
17924         aarch64-cores.def.
17926         PR middle-end/56548
17927         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
17928         promoted mode, convert the result back to the original mode.
17930 2013-03-06  Richard Biener  <rguenther@suse.de>
17932         PR middle-end/56294
17933         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
17934         (insert_updated_phi_nodes_compare_uids): New function.
17935         (update_ssa): Sort symbols_to_rename after UID before
17936         traversing it to insert PHI nodes.
17938 2013-03-06  Richard Biener  <rguenther@suse.de>
17940         PR middle-end/50494
17941         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
17942         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
17944         Revert
17945         2013-02-13  Richard Biener  <rguenther@suse.de>
17947         PR lto/50494
17948         * varasm.c (output_constant_def_1): Get the decl representing
17949         the constant as argument.
17950         (output_constant_def): Wrap output_constant_def_1.
17951         (make_decl_rtl): Use output_constant_def_1 with the decl
17952         representing the constant.
17953         (build_constant_desc): Optionally re-use a decl already
17954         representing the constant.
17955         (tree_output_constant_def): Adjust.
17957 2013-03-06  Joey Ye  <joey.ye@arm.com>
17959         PR lto/50293
17960         * gcc.c (convert_white_space): New function.
17961         (main): Handles white space in function name.
17963 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
17965         PR target/56529
17966         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
17967         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
17968         to SH_DIV_CALL_TABLE for TARGET_SH2.
17969         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
17970         list.
17971         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
17972         call-table options.
17974 2013-03-05  Sterling Augustine  <saugustine@google.com>
17975             Cary Coutant  <ccoutant@google.com>
17977         PR debug/55364
17978         * dwarf2out.c (resolve_addr): Don't call
17979         remove_loc_list_addr_table_entries a second time for the same
17980         expression.
17982 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
17984         PR debug/56510
17985         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
17986         (avoid_complex_debug_insns): New function.
17987         (expand_debug_locations): Call it.
17989         PR rtl-optimization/56484
17990         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
17991         lifetimes of hard registers on small register class machines.
17993 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
17995         * config/microblaze/microblaze-protos.h: Rename
17996         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
17997         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
17998         fast_interrupt.
17999         (microblaze_fast_interrupt_function_p): New function.
18000         (microblaze_is_interrupt_handler): Rename to
18001         microblaze_is_interrupt_variant and add fast_interrupt check.
18002         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18003         (save_restore_insns): Likewise.
18004         (compute_frame_size): Likewise.
18005         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18006         (microblaze_globalize_label): Likewise.
18007         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18008         * config/microblaze/microblaze.md: Use wrapper
18009         microblaze_is_interrupt_variant.
18011 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18013         * sdbout.c (sdbout_one_type): Switch to current function's section
18014         supporting cold/hot.
18016 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18018         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18019         -mxl-reorder.
18021 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18023         PR middle-end/56461
18024         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18025         if VALGRIND_GET_VBITS is defined, temporarily make object
18026         memory all defined, and restore previous valgrind addressability
18027         and definability afterwards.  Free this_object at the end.
18029         PR middle-end/56461
18030         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18031         right before calling lra_create_live_ranges, also call it
18032         when clearing live_p.  Only call lra_clear_live_ranges
18033         at the end if live_p.
18035         PR middle-end/56461
18036         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18038 2013-03-05  Richard Biener  <rguenther@suse.de>
18040         PR tree-optimization/56521
18041         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18042         value-id.
18044 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18046         PR c++/55135
18047         * except.h (remove_unreachable_eh_regions): New prototype.
18048         * except.c (remove_eh_handler_splicer): New function, split out
18049         of remove_eh_handler.
18050         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18051         warning about running it on many EH regions one at a time.
18052         (remove_unreachable_eh_regions_worker): New function, walk the
18053         EH tree in depth-first order and remove non-marked regions.
18054         (remove_unreachable_eh_regions): New function.
18055         * tree-eh.c (mark_reachable_handlers): New function, split out
18056         from remove_unreachable_handlers.
18057         (remove_unreachable_handlers): Use mark_reachable_handlers and
18058         remove_unreachable_eh_regions.
18059         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18060         and remove_unreachable_eh_regions.
18062 2013-03-05  Richard Biener  <rguenther@suse.de>
18064         PR middle-end/56525
18065         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18066         not freeing them until the end.
18068 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18070         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18072 2013-03-05  Richard Biener  <rguenther@suse.de>
18074         PR tree-optimization/56270
18075         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18076         of loads after scheduling an SLP instance.
18078 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18080         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18081         tic6x.exp.
18082         (check_gcc_parallelize): Run guality.exp as a separate job from
18083         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18084         struct-layout-1.exp with stackalign.exp.
18086         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18088         PR middle-end/56461
18089         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18090         load_index sbitmap even if some bit in it isn't set.
18092         PR middle-end/56461
18093         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18094         (discover_iteration_bound_by_body_walk): Change queues to
18095         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18096         spelling in comment.  Call safe_push on queues[bound_index] directly.
18097         Release queues[queue_index] in every iteration unconditionally.
18098         Release bounds vector.
18100         PR middle-end/56461
18101         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18102         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18103         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18104         inner_phis vector.
18106 2013-03-05  Richard Biener  <rguenther@suse.de>
18108         PR lto/56515
18109         * tree-inline.c (remap_blocks_to_null): New function.
18110         (expand_call_inline): When expanding a call stmt without
18111         an associated block inline remap all callee blocks to NULL.
18113 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18115         PR rtl-optimization/56494
18116         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18117         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18118         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18120         PR middle-end/56461
18121         * sel-sched-ir.c (free_sched_pools): Release
18122         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18123         if succs_info_pool.max_top isn't -1.
18125         PR bootstrap/56509
18126         * opts.c (opts_obstack, opts_concat): Moved to...
18127         * opts-common.c (opts_obstack, opts_concat): ... here.
18129 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18131         PR middle-end/56461
18132         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18134 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18136         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18137         all appropriate places.
18139 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18141         PR tree-optimization/56424
18142         * ipa-split.c (split_function): Do not set the RSO flag if result is
18143         not by reference and its type is a register type.
18145 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18147         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18148         (microblaze_legitimate_pic_operand): Likewise
18149         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18150         new function microblaze_legitimate_pic_operand
18151         * config/microblaze/microblaze-protos.h
18152         (microblaze_legitimate_pic_operand): Declare.
18154 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18156         * config/microblaze/predicates.md (call_insn_simple_operand):
18157         New predicate for supported rtx code types.
18158         * config/microblaze/microblaze.md (call_internal1): Use
18159         call_insn_simple_operand predicate.
18161 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18163         PR middle-end/56461
18164         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18165         partitions.ordered_remove.
18167         PR middle-end/56461
18168         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18169         vec_oprnds0.create (1) for modifier == NONE.
18171         PR middle-end/56461
18172         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18173         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18174         vec_oprnds1 right before pushing anything to it for
18175         scalar_shift_arg.
18177         PR middle-end/56461
18178         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18179         set nbbs to 0 instead of having separate code path.
18180         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18181         instead of false as last argument if returning NULL.
18183 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18185         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18186         the attribute is now called "target" instead of "option".
18187         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18188         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18189         attribute/pragma name for TARGET_OPTION_VALID_P and
18190         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18191         * doc/tm.texi: Regenerated.
18193 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
18195         * config/microblaze/microblaze.c:
18196         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18197         * config/microblaze/microblaze.h: Add -mxl-reorder to
18198         DRIVER_SELF_SPECS.
18199         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18200         instructions emitted if TARGET_REORDER.
18201         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18202         or 0 for -m/-mno case, but initialises as 2 to detect default use case
18203         separately.
18205 2013-03-01  Xinliang David Li  <davidxl@google.com>
18207         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18208         walk length.
18210 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
18212         PR middle-end/56461
18213         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18214         vector even when returning true.  Fix up function comment formatting.
18216         PR middle-end/56461
18217         * ira-build.c (ira_loop_nodes_count): New variable.
18218         (create_loop_tree_nodes): Initialize it.
18219         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18221         PR middle-end/56461
18222         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18223         method on dr_chain and result_chain.
18224         * tree-vect-stmts.c (vectorizable_store): Only call
18225         result_chain.create if j == 0.
18227         PR middle-end/56461
18228         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18229         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18230         before overwriting it.
18232 2013-03-01  Tobias Burnus  <burnus@net-b.de>
18234         * doc/extended.texi (C Extensions): Change order in @menu
18235         to match @node.
18236         (Other MIPS Built-in Functions): Move last MIPS entry before
18237         "picoChip Built-in Functions".
18238         (SH Built-in Functions): Move after RX Built-in Functions.
18239         * doc/gcc.texi (Introduction): Change order in @menu
18240         to match @node.
18241         * doc/md.texi (Constraints): Ditto.
18242         * gty.texi (Type Information): Ditto.
18243         (User-provided marking routines for template types): Make
18244         subsection.
18245         * doc/invoke.texi (AArch64 Options): Move before
18246         "Adapteva Epiphany Options".
18248 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
18249             Jakub Jelinek  <jakub@redhat.com>
18251         PR sanitizer/56454
18252         * asan.c (gate_asan): Lookup no_sanitize_address instead of
18253         no_address_safety_analysis attribute.
18254         * doc/extend.texi (no_address_safety_attribute): Rename to
18255         no_sanitize_address attribute, mention no_address_safety_analysis
18256         attribute as deprecated alias.
18258 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
18260         PR middle-end/56461
18261         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18262         type to vec<vec<tree> > *.
18263         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
18264         to be vec<tree> instead of vec<tree> *, set vec_defs
18265         to vNULL and call vec_defs.create (number_of_vects), adjust other
18266         uses of vec_defs.
18267         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18268         vectorizable_condition): Adjust vect_get_slp_defs callers.
18270 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18272         * config/aarch64/aarch64.c
18273         (aarch64_float_const_representable): Remove unused variable.
18275 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18277         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18279 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18281         * config/aarch64/aarch64-builtins.c
18282         (aarch64_init_simd_builtins): Make static.
18284 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18286         * config/aarch64/aarch64.c
18287         (aarch64_simd_make_constant): Make static.
18289 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18291         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18292         with no initialization to the RHS of debug statements.
18294 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18296         PR tree-optimization/56294
18297         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18298         Adjust dumping.
18299         (get_access_replacement): Do not call create_access_replacement.
18300         Assert a replacement exists.
18301         (get_repl_default_def_ssa_name): Create the replacement declaration
18302         itself.
18304 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18306         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18307         final_end_function.
18309 2013-02-28  Marek Polacek  <polacek@redhat.com>
18311         PR rtl-optimization/56466
18312         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18313         if we're changing a loop.
18314         (peel_loops_completely): Likewise.
18316 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
18318         PR c++/55813
18319         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18321 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18323         PR target/56445
18324         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18325         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18326         INTX_FTYPE_FX, FX_FTYPE_INTX.
18327         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18329 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18331         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18332         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18333         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18334         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18335         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18336         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18337         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18338         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18339         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18340         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18341         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18342         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18343         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18344         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18345         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18346         (avrxmega6): Increase max flash segments from 5 to 6.
18347         * config/avr/t-multilib: Regenerate.
18348         * config/avr/avr-tables.opt: Regenerate.
18349         * doc/avr-mmcu.texi: Regenerate.
18351 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18353         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18354         (avr_device_to_arch): Rename to avr_device_to_ld.
18355         (avr_device_to_as): New prototype.
18356         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18357         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18358         * config/avr/driver-avr.c (avr_device_to_as): New.
18359         (avr_device_to_arch): Rename to avr_device_to_ld.
18361 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18363         PR middle-end/56461
18364         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18365         method on dr_chain and result_chain.
18367         PR middle-end/56461
18368         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18369         pointer_set_destroy on not_executed_last_iteration.
18371         PR middle-end/56461
18372         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18374         PR middle-end/56461
18375         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18376         FOR_EACH_DEFINED_FUNCTION when freeing state.
18378         PR middle-end/56461
18379         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18380         pool_free.
18381         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18382         overwriting it.
18384         PR middle-end/56461
18385         * ipa-cp.c (decide_whether_version_node): Call vec_free on
18386         known_aggs[i].items and release known_aggs vector.
18388         PR middle-end/56461
18389         * ipa-reference.c (propagate): Free node_info even for alias nodes.
18391 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18393         * config/microblaze/microblaze.c (microblaze_emit_compare):
18394         Use xor for EQ/NE comparisions.
18395         * config/microblaze/microblaze.md (cstoresf4): Add constraints
18396         (cbranchsf4): Adjust operator to comparison_operator.
18398 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18400         PR middle-end/56461
18401         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18402         vector.
18403         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18404         vec_safe_push, always update *slot.
18405         (redirect_edge_var_map_clear): Use vec_free.
18406         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18407         (free_var_map_entry): Use vec_free.
18408         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18409         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18411 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
18413         PR middle-end/45472
18414         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18415         when the may_trap_p bit of the exprs being merged differs.
18416         Reorder tests for speculativeness in the logical and operator.
18418 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18420         * incpath.c (add_standard_paths): Use reconcat instead of concat
18421         where appropriate and avoid leaking memory.
18423         * opts.h: Include obstack.h.
18424         (opts_concat): New prototype.
18425         (opts_obstack): New declaration.
18426         * opts.c (opts_concat): New function.
18427         (opts_obstack): New variable.
18428         (init_options_struct): Call gcc_init_obstack on opts_obstack.
18429         (finish_options): Use opts_concat instead of concat
18430         and XOBNEWVEC instead of XNEWVEC.
18431         * opts-common.c (generate_canonical_option, decode_cmdline_option,
18432         generate_option): Likewise.
18433         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18434         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18436         PR target/56455
18437         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18438         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18440 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18442         PR middle-end/56461
18443         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18445 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18447         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18448         (arm_block_move_unaligned_straight): Likewise.
18449         (arm_adjust_block_mem): Likewise.
18451 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18453         PR target/48901
18454         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18455         temp, cond and label.
18456         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18458         PR target/52500
18459         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18460         * config/c6x/c6x.h (dbx_register_map): Update declaration.
18462         PR target/52501
18463         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18464         of prologue/epilogue functions.
18466         PR target/52550
18467         * config/tilegx/tilegx.c (tilegx_expand_prologue):
18468         Remove unused variable cfa_offset.
18469         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18471         PR target/54639
18472         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18473         type promotion to unsigned.
18475         PR target/54640
18476         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18477         for HOST_WIDE_INT of 32 bit / same size as int.
18478         (arm_block_move_unaligned_straight): Likewise.
18479         (arm_adjust_block_mem): Likewise.
18481         PR target/54662
18482         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18483         ALL_CFLAGS.
18485 2013-02-26  Marek Polacek  <polacek@redhat.com>
18487         PR tree-optimization/56426
18488         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18490 2013-02-26  Richard Biener  <rguenther@suse.de>
18492         PR target/56444
18493         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18494         unused variable loops.
18496 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18498         PR tree-optimization/56448
18499         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18500         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18501         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18502         later operands of the references, or even first operand for
18503         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18505         PR tree-optimization/56443
18506         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18507         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18508         to type_for_mode langhook.
18510 2013-02-25  Matt Turner  <mattst88@gmail.com>
18512         * doc/invoke.texi: Document r4700.
18514 2013-02-25  Richard Biener  <rguenther@suse.de>
18516         PR tree-optimization/56175
18517         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18518         split out from ...
18519         (simplify_bitwise_binary): ... here.  Also guard the conversion
18520         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18522 2013-02-25  Catherine Moore  <clm@codesourcery.com>
18524         Revert:
18525         2013-02-24  Catherine Moore  <clm@codesourcery.com>
18526             Maciej W. Rozycki  <macro@codesourcery.com>
18527             Tom de Vries  <tom@codesourcery.com>
18528             Nathan Sidwell  <nathan@codesourcery.com>
18529             Iain Sandoe  <iain@codesourcery.com>
18530             Nathan Froyd  <froydnj@codesourcery.com>
18531             Chao-ying Fu  <fu@mips.com>
18533         * doc/extend.texi: (micromips, nomicromips, nocompression):
18534         Document new function attributes.
18535         * doc/invoke.texi (minterlink-compressed, mmicromips,
18536         m14k, m14ke, m14kec): Document new options.
18537         (minterlink-mips16): Update documentation.
18538         * doc/md.texi (ZC, ZD): Document new constraints.
18539         * configure.ac (gcc_cv_as_micromips): Check if linker
18540         supports the .set micromips directive.
18541         * configure: Regenerate.
18542         * config.in: Regenerate.
18543         * config/mips/mips-tables.opt: Regenerate.
18544         * config/mips/micromips.md: New file.
18545         * constraints.md (ZC, AD): New constraints.
18546         * config/mips/predicates.md (movep_src_register): New predicate.
18547         (movep_src_operand): New predicate.
18548         (non_volatile_mem_operand): New predicate.
18549         * config/mips/mips.md (multimem): New type.
18550         (length): Differentiate between 17-bit and 18-bit branch offsets.
18551         (MOVEP1, MOVEP2): New mode iterator.
18552         (mov_<load>l): Use ZC constraint.
18553         (mov_<load>r): Likewise.
18554         (mov_<store>l): Likewise.
18555         (mov_<store>r): Likewise.
18556         (*branch_equality<mode>_inverted): Add microMIPS support.
18557         (*branch_equality<mode>): Likewise.
18558         (*jump_absolute): Likewise.
18559         (indirect_jump_<mode>): Likewise.
18560         (tablejump_<mode>): Likewise.
18561         (<optab>_internal): Likewise.
18562         (sibcall_internal): Likewise.
18563         (sibcall_value_internal): Likewise.
18564         (prefetch): Use constraint ZD.
18565         * config/mips/mips.opt (minterlink-compressed): New option.
18566         (minterlink-mips16): Now an alias for minterlink-compressed.
18567         (mmicromips): New option.
18568         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18569         (compare_and_swap_12): Likewise.
18570         (sync_add<mode>): Likewise.
18571         (sync_<optab>_12): Likewise.
18572         (sync_old_<optab>_12): Likewise.
18573         (sync_new_<optab>_12): Likewise.
18574         (sync_nand_12): Likewise.
18575         (sync_old_nand_12): Likewise.
18576         (sync_new_nand_12): Likewise.
18577         (sync_sub<mode>): Likewise.
18578         (sync_old_add<mode>): Likewise.
18579         (sync_old_sub<mode>): Likewise.
18580         (sync_new_add<mode>): Likewise.
18581         (sync_new_sub<mode>): Likewise.
18582         (sync_<optab><mode>): Likewise.
18583         (sync_old_<optab><mode>): Likewise.
18584         (sync_new_<optab><mode>): Likewise.
18585         (sync_nand<mode>): Likewise.
18586         (sync_old_nand<mode>): Likewise.
18587         (sync_new_nand<mode>): Likewise.
18588         (sync_lock_test_and_set<mode>): Likewise.
18589         (test_and_set_12): Likewise.
18590         (atomic_compare_and_swap<mode>): Likewise.
18591         (atomic_exchange<mode>_llsc): Likewise.
18592         (atomic_fetch_add<mode>_llsc): Likewise.
18593         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18594         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18595         (umips_save_restore_pattern_p): Likewise.
18596         (umips_load_store_pair_p): Likewise.
18597         (umips_output_load_store_pair): Likewise.
18598         (umips_movep_target_p): Likewise.
18599         (umips_12bit_offset_address_p): Likewise.
18600         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18601         (mips_base_mips16): Rename this...
18602         (mips_base_compression_flags): ...to this. Update all uses.
18603         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18604         (mips_mips16_decl_p): Delete.
18605         (mips_nomips16_decl_p): Delete.
18606         (mips_get_compress_on_flags): New function.
18607         (mips_get_compress_off_flags): New function.
18608         (mips_get_compress_mode): New function.
18609         (mips_get_compress_on_name): New function.
18610         (mips_get_compress_off_name): New function.
18611         (mips_insert_attributes): Support multiple compression types.
18612         (mips_merge_decl_attributes): Likewise.
18613         (umips_12bit_offset_address_p): New function.
18614         (mips_start_function_definition): Emit .set micromips directive.
18615         (mips_call_may_need_jalx_p): New function.
18616         (mips_function_ok_for_sibcall): Add microMIPS support.
18617         (mips_print_operand_punctuation): Support short delay slots and
18618         compact jumps.
18619         (umips_swm_mask, umips_swm_encoding): New.
18620         (umips_build_save_restore): New function.
18621         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18622         (was_mips16_p): Remove.
18623         (old_compression_mode): New.
18624         (mips_set_compression_mode): New function.
18625         (mips_set_current_function): Add microMIPS support.
18626         (mips_option_override): Likewise.
18627         (umips_save_restore_pattern_p): New function.
18628         (umips_output_save_restore): New function.
18629         (umips_load_store_pair_p_1): New function.
18630         (umips_load_store_pair_p): New function.
18631         (umips_output_load_store_pair_1): New function.
18632         (umips_output_load_store_pair): New function.
18633         (umips_movep_target_p) New function.
18634         (mips_prepare_pch_save): Add microMIPS support.
18635         * config/mips/mips.h (TARGET_COMPRESSION): New.
18636         (TARGET_CPU_CPP_BUILTINS): Update macro
18637         to use new compression flags and to support microMIPS.
18638         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18639         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18640         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18641         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18642         (ASM_SPEC): Support mmicromips and mno-micromips.
18643         (M16STORE_REG_P): New macro.
18644         (MIPS_CALL): Support TARGET_MICROMIPS.
18645         (MICROMIPS_J): New macro.
18646         (mips_base_mips16): Rename this...
18647         (mips_base_compression_flags): ...to this.
18648         (UMIPS_12BIT_OFFSET_P): New macro.
18649         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18650         (MULTILIB_DIRNAMES): Likewise.
18652 2013-02-25  Tom de Vries  <tom@codesourcery.com>
18654         PR rtl-optimization/56131
18655         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
18656         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
18657         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
18659 2013-02-25  Tobias Burnus  <burnus@net-b.de>
18661         * doc/invoke.texi (-fsanitize=): Move from optimization
18662         to debugging options.
18664 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18666         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
18668 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18669             Alexander Monakov  <amonakov@ispras.ru>
18671         PR middle-end/56077
18672         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
18673         flush pending lists also on non-jumps.  Adjust comment.
18675 2013-02-24  Catherine Moore  <clm@codesourcery.com>
18676             Maciej W. Rozycki  <macro@codesourcery.com>
18677             Tom de Vries  <tom@codesourcery.com>
18678             Nathan Sidwell  <nathan@codesourcery.com>
18679             Iain Sandoe  <iain@codesourcery.com>
18680             Nathan Froyd  <froydnj@codesourcery.com>
18681             Chao-ying Fu  <fu@mips.com>
18683         * doc/extend.texi: (micromips, nomicromips, nocompression):
18684         Document new function attributes.
18685         * doc/invoke.texi (minterlink-compressed, mmicromips,
18686         m14k, m14ke, m14kec): Document new options.
18687         (minterlink-mips16): Update documentation.
18688         * doc/md.texi (ZC, ZD): Document new constraints.
18689         * configure.ac (gcc_cv_as_micromips): Check if linker
18690         supports the .set micromips directive.
18691         * configure: Regenerate.
18692         * config.in: Regenerate.
18693         * config/mips/mips-tables.opt: Regenerate.
18694         * config/mips/micromips.md: New file.
18695         * constraints.md (ZC, AD): New constraints.
18696         * config/mips/predicates.md (movep_src_register): New predicate.
18697         (movep_src_operand): New predicate.
18698         (non_volatile_mem_operand): New predicate.
18699         * config/mips/mips.md (multimem): New type.
18700         (length): Differentiate between 17-bit and 18-bit branch offsets.
18701         (MOVEP1, MOVEP2): New mode iterator.
18702         (mov_<load>l): Use ZC constraint.
18703         (mov_<load>r): Likewise.
18704         (mov_<store>l): Likewise.
18705         (mov_<store>r): Likewise.
18706         (*branch_equality<mode>_inverted): Add microMIPS support.
18707         (*branch_equality<mode>): Likewise.
18708         (*jump_absolute): Likewise.
18709         (indirect_jump_<mode>): Likewise.
18710         (tablejump_<mode>): Likewise.
18711         (<optab>_internal): Likewise.
18712         (sibcall_internal): Likewise.
18713         (sibcall_value_internal): Likewise.
18714         (prefetch): Use constraint ZD.
18715         * config/mips/mips.opt (minterlink-compressed): New option.
18716         (minterlink-mips16): Now an alias for minterlink-compressed.
18717         (mmicromips): New option.
18718         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18719         (compare_and_swap_12): Likewise.
18720         (sync_add<mode>): Likewise.
18721         (sync_<optab>_12): Likewise.
18722         (sync_old_<optab>_12): Likewise.
18723         (sync_new_<optab>_12): Likewise.
18724         (sync_nand_12): Likewise.
18725         (sync_old_nand_12): Likewise.
18726         (sync_new_nand_12): Likewise.
18727         (sync_sub<mode>): Likewise.
18728         (sync_old_add<mode>): Likewise.
18729         (sync_old_sub<mode>): Likewise.
18730         (sync_new_add<mode>): Likewise.
18731         (sync_new_sub<mode>): Likewise.
18732         (sync_<optab><mode>): Likewise.
18733         (sync_old_<optab><mode>): Likewise.
18734         (sync_new_<optab><mode>): Likewise.
18735         (sync_nand<mode>): Likewise.
18736         (sync_old_nand<mode>): Likewise.
18737         (sync_new_nand<mode>): Likewise.
18738         (sync_lock_test_and_set<mode>): Likewise.
18739         (test_and_set_12): Likewise.
18740         (atomic_compare_and_swap<mode>): Likewise.
18741         (atomic_exchange<mode>_llsc): Likewise.
18742         (atomic_fetch_add<mode>_llsc): Likewise.
18743         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18744         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18745         (umips_save_restore_pattern_p): Likewise.
18746         (umips_load_store_pair_p): Likewise.
18747         (umips_output_load_store_pair): Likewise.
18748         (umips_movep_target_p): Likewise.
18749         (umips_12bit_offset_address_p): Likewise.
18750         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18751         (mips_base_mips16): Rename this...
18752         (mips_base_compression_flags): ...to this. Update all uses.
18753         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18754         (mips_mips16_decl_p): Delete.
18755         (mips_nomips16_decl_p): Delete.
18756         (mips_get_compress_on_flags): New function.
18757         (mips_get_compress_off_flags): New function.
18758         (mips_get_compress_mode): New function.
18759         (mips_get_compress_on_name): New function.
18760         (mips_get_compress_off_name): New function.
18761         (mips_insert_attributes): Support multiple compression types.
18762         (mips_merge_decl_attributes): Likewise.
18763         (umips_12bit_offset_address_p): New function.
18764         (mips_start_function_definition): Emit .set micromips directive.
18765         (mips_call_may_need_jalx_p): New function.
18766         (mips_function_ok_for_sibcall): Add microMIPS support.
18767         (mips_print_operand_punctuation): Support short delay slots and
18768         compact jumps.
18769         (umips_swm_mask, umips_swm_encoding): New.
18770         (umips_build_save_restore): New function.
18771         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18772         (was_mips16_p): Remove.
18773         (old_compression_mode): New.
18774         (mips_set_compression_mode): New function.
18775         (mips_set_current_function): Add microMIPS support.
18776         (mips_option_override): Likewise.
18777         (umips_save_restore_pattern_p): New function.
18778         (umips_output_save_restore): New function.
18779         (umips_load_store_pair_p_1): New function.
18780         (umips_load_store_pair_p): New function.
18781         (umips_output_load_store_pair_1): New function.
18782         (umips_output_load_store_pair): New function.
18783         (umips_movep_target_p) New function.
18784         (mips_prepare_pch_save): Add microMIPS support.
18785         * config/mips/mips.h (TARGET_COMPRESSION): New.
18786         (TARGET_CPU_CPP_BUILTINS): Update macro
18787         to use new compression flags and to support microMIPS.
18788         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18789         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18790         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18791         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18792         (ASM_SPEC): Support mmicromips and mno-micromips.
18793         (M16STORE_REG_P): New macro.
18794         (MIPS_CALL): Support TARGET_MICROMIPS.
18795         (MICROMIPS_J): New macro.
18796         (mips_base_mips16): Rename this...
18797         (mips_base_compression_flags): ...to this.
18798         (UMIPS_12BIT_OFFSET_P): New macro.
18799         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18800         (MULTILIB_DIRNAMES): Likewise.
18802 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
18804         PR target/52555
18805         * target-globals.c (save_target_globals): For init_reg_sets and
18806         target_reinit remporarily set this_fn_optabs to this_target_optabs.
18808 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
18810         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
18811         * config/aarch64/t-aarch64
18812         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
18814 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
18816         PR inline-asm/56148
18817         * lra-constraints.c (process_alt_operands): Reload operand
18818         conflicting with earlier clobber only if no more other conflicting
18819         operands.
18821 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
18823         PR sanitizer/56393
18824         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
18825         if not linking a shared library.
18827 2013-02-22  Seth LaForge  <sethml@google.com>
18829         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
18831 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
18833         * config/arm/arm.md (split for extendsidi): Update condition.
18834         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
18835         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
18836         (qhs_zextenddi_cstr): Likewise.
18838 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18840         PR middle-end/56420
18841         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
18842         avoid signed wrapping.
18843         (expand_mult): Handle properly multiplication by
18844         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
18845         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
18846         in the compiler if coeff is HOST_WIDE_INT_MIN.
18847         (expand_divmod): Don't make ext_op1 static, change it's type to
18848         uhwi.  Avoid undefined behavior in -INTVAL (op1).
18850         PR rtl-optimization/50339
18851         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
18852         field.
18853         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
18854         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
18855         into splitting_ashiftrt field.
18856         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
18857         ASHIFTRT.
18858         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
18859         choices.
18861 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
18863         PR middle-end/56108
18864         * trans-mem.c (execute_tm_mark): Do not expand transactions that
18865         are sure to go irrevocable.
18867 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
18869         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
18870         scalars are valid operands.
18872 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18874         PR tree-optimization/56310
18875         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
18876         only matching indices and non-negative final offsets.
18877         (intersect_aggregates_with_edge): Pass src_idx to
18878         agg_replacements_to_vector.  Pass src_idx insstead of index to
18879         intersect_with_agg_replacements.
18881 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18883         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
18884         instead of hard-wired defaults.
18886 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
18888         * doc/invoke.texi (MIPS Options): Update documentation of the
18889         floating-point multiply-accumulate instruction restrictions.
18891 2013-02-21  Kostya Serebryany  <kcc@google.com>
18893         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
18894         asan_shadow_offset on x86_64 linux.
18896 2013-02-21  Richard Biener  <rguenther@suse.de>
18898         PR tree-optimization/56415
18899         Revert
18900         2013-02-11  Richard Biener  <rguenther@suse.de>
18902         PR tree-optimization/56273
18903         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
18904         first VRP run.
18906 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18908         PR bootstrap/56258
18909         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
18910         instead of @itemx.
18912         PR inline-asm/56405
18913         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
18914         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
18916 2013-02-20  Jan Hubicka  <jh@suse.cz>
18918         PR tree-optimization/56265
18919         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
18920         when target is referenced for first time.
18922 2013-02-20  Richard Biener  <rguenther@suse.de>
18924         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
18925         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
18926         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
18927         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
18928         not return anything.
18929         (rename_ssa_copies): Do not remove unused locals.
18930         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
18931         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
18932         * passes.c (execute_function_todo): Do not schedule unused locals
18933         removal if cleanup_tree_cfg did something.
18934         * tree-ssa-live.c (remove_unused_locals): Dump statistics
18935         about the number of removed locals.
18937 2013-02-20  Richard Biener  <rguenther@suse.de>
18939         PR tree-optimization/56398
18940         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
18942 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18944         PR tree-optimization/55334
18945         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
18946         restricted pointers to arrays.
18948 2013-02-20  Richard Biener  <rguenther@suse.de>
18949             Jakub Jelinek  <jakub@redhat.com>
18951         PR tree-optimization/56396
18952         * tree-ssa-ccp.c (n_const_val): New static variable.
18953         (get_value): Return NULL for SSA names we don't have a lattice
18954         entry for.
18955         (ccp_initialize): Initialize n_const_val.
18956         * tree-ssa-copy.c (n_copy_of): New static variable.
18957         (init_copy_prop): Initialize n_copy_of.
18958         (get_value): Return NULL_TREE for SSA names we don't have a
18959         lattice entry for.
18961 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18963         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
18965 2013-02-20  Richard Biener  <rguenther@suse.de>
18967         * genpreds.c (write_lookup_constraint): Do not compare first
18968         letter of the constraint again.
18970 2013-02-20  Richard Biener  <rguenther@suse.de>
18972         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
18973         and ceil_log2.
18974         (get_use_iv_cost): Terminate hashtable walk when coming across
18975         an empty entry.
18977 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
18979         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
18980         reassociation for avx2 targets.
18982 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18984         * config/microblaze/microblaze.c: microblaze_has_clz = 0
18985         Add version check for v8.10.a to enable microblaze_has_clz
18986         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
18987         version and TARGET_PATTERN_COMPARE check
18988         * config/microblaze/microblaze.md: New clzsi2 instruction
18990 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18992         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
18993         function before branching.
18995 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
18997         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
18998         DUMP_INSN_RTX_UID.
18999         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19001 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19003         PR middle-end/55889
19004         * sel-sched.c: Include ira.h.
19005         (implicit_clobber_conflict_p): New function.
19006         (moveup_expr): Use it.
19007         * Makefile.in (sel-sched.o): Depend on ira.h.
19009 2013-02-19  Richard Biener  <rguenther@suse.de>
19011         PR tree-optimization/56384
19012         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19013         (vn_hash_type): Split out from ...
19014         (vn_hash_constant_with_type): ... here.
19015         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19016         (vn_phi_eq): Compare types from vn_phi_s structure.
19017         (vn_phi_lookup): Populate vn_phi_s type.
19018         (vn_phi_insert): Likewise.
19020 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19022         PR tree-optimization/56350
19023         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19024         if haven't found reduction or nested cycle operand, rather than
19025         asserting we must find it.
19027         PR tree-optimization/56381
19028         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19029         to fold_build3.
19031 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19032             Jakub Jelinek  <jakub@redhat.com>
19034         PR target/52555
19035         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19036         (swap_optab_enable): Same.
19037         (init_all_optabs): Use argument instead of global.
19038         * tree.h (struct tree_optimization_option): New field target_optabs.
19039         * expr.h (init_all_optabs): Add argument to prototype.
19040         (TREE_OPTIMIZATION_OPTABS): New.
19041         (save_optabs_if_changed): Protoize.
19042         * optabs.h: Declare this_fn_optabs.
19043         * optabs.c (save_optabs_if_changed): New.
19044         Declare this_fn_optabs.
19045         (init_optabs): Add argument to init_all_optabs() call.
19046         * function.c (invoke_set_current_function_hook): Handle per
19047         function optabs.
19048         * function.h (struct function): New field optabs.
19049         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19050         optimization_current_node has changed.
19051         * target-globals.h (save_target_globals_default_opts): Protoize.
19052         * target-globals.c (save_target_globals_default_opts): New.
19054 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19056         PR target/56347
19057         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19058         registers %fr12 and %fr12R as call used.
19060         PR target/56214
19061         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19062         and HImode, require all displacements to be an integer multiple of
19063         their mode size.
19064         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19065         only allow QImode and HImode when reload is in progress and strict is
19066         true.  Likewise for symbolic addresses.  Use base14_operand to check
19067         displacements in REG+BASE addresses.
19069 2013-02-18  Richard Biener  <rguenther@suse.de>
19071         PR tree-optimization/56366
19072         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19073         handle sign-conversion of outer-loop initial induction value.
19075 2013-02-18  Richard Biener  <rguenther@suse.de>
19077         PR middle-end/56349
19078         * cfghooks.c (merge_blocks): If we merge a latch into another
19079         block adjust references to it.
19080         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19081         (verify_loop_structure): Verify that a recorded latch is in fact
19082         a latch.
19084 2013-02-18  Richard Biener  <rguenther@suse.de>
19086         PR tree-optimization/56321
19087         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19088         order SSA name release and virtual operand unlinking.
19090 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19092         * config/microblaze/microblaze.md (save_stack_block): Define.
19093         (restore_stack_block): Likewise.
19095 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19097         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19098         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19099         * config/microblaze/microblaze.c (microblaze_option_override):
19100         Bail out early for PIC modes when target does not support PIC.
19102 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19104         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19105         Replace with a microblaze version.
19106         (microblaze_trampoline_init): Adapt for microblaze.
19107         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19108         microblaze.
19110 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19111             Dodji Seketeli  <dodji@redhat.com>
19113         PR asan/56330
19114         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19115         (instrument_mem_region_access): Do not forget to always put
19116         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19117         0) statement, even for cases where either 'base' or 'base + len'
19118         are not instrumented -- because they have been previously
19119         instrumented.  Simplify the logic by putting all the statements
19120         instrument 'base + len' inside a sequence, and then insert that
19121         sequence right before the current insertion point.  Then, to
19122         instrument 'base + len', just get an iterator on that statement.
19123         And do not forget to update the pointer to iterator the function
19124         received as argument.
19126 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19128         PR rtl-optimization/56348
19129         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19131 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19133         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19134         (clean_graph_dump_file): Pass base to start_graph_dump.
19136 2013-02-14  Richard Henderson  <rth@redhat.com>
19138         PR target/55941
19139         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19141 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19143         * collect2-aix.h: Define F_LOADONLY.
19145 2013-02-14  Richard Biener  <rguenther@suse.de>
19147         PR lto/50494
19148         * varasm.c (output_constant_def_1): Get the decl representing
19149         the constant as argument.
19150         (output_constant_def): Wrap output_constant_def_1.
19151         (make_decl_rtl): Use output_constant_def_1 with the decl
19152         representing the constant.
19153         (build_constant_desc): Optionally re-use a decl already
19154         representing the constant.
19155         (tree_output_constant_def): Adjust.
19157 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19159         Fix an asan crash
19160         * asan.c (instrument_builtin_call):  Really put the length of the
19161         second source argument into src1_len.
19163 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19165         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19166         argument.  If it is false, don't create edge from then_bb to
19167         fallthru_bb.
19168         (insert_if_then_before_iter): Pass true to it.
19169         (build_check_stmt): Pass false to it.
19170         (transform_statements): Flush hash table only on extended basic
19171         block boundaries, rather than at the beginning of every bb.
19172         Don't flush hash table on nonfreeing_call_p calls.
19173         * tree-flow.h (nonfreeing_call_p): New prototype.
19174         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19176 2013-02-13  David S. Miller  <davem@davemloft.net>
19178         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19180 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19182         PR target/56184
19183         * ira.c (max_regno_before_ira): Move from ...
19184         (ira): ... here.
19185         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19186         vec_safe_length.
19188 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19190         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19192 2013-02-13  Richard Biener  <rguenther@suse.de>
19194         PR lto/56295
19195         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19196         globals in MEM_REFs.
19198 2013-02-13  Richard Biener  <rguenther@suse.de>
19200         * loop-init.c (loop_optimizer_init): Clear loop state when
19201         re-initializing preserved loops.
19202         * loop-unswitch.c (unswitch_single_loop): Return whether
19203         we unswitched the loop.  Do not verify loop state here.
19204         (unswitch_loops): When we unswitched a loop discover new loops.
19206 2013-02-13  Kostya Serebryany  <kcc@google.com>
19208         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19209         on x86_64 linux.
19210         * sanitizer.def: Rename __asan_init to __asan_init_v1.
19212 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
19214         Avoid instrumenting duplicated memory access in the same basic block
19215         * Makefile.in (asan.o): Add new dependency on hash-table.h
19216         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19217         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19218         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19219         (free_mem_ref_resources, has_mem_ref_been_instrumented)
19220         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19221         (get_mem_ref_of_assignment): New functions.
19222         (get_mem_refs_of_builtin_call): Extract from
19223         instrument_builtin_call and tweak a little bit to make it fit with
19224         the new signature.
19225         (instrument_builtin_call): Use the new
19226         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
19227         of is_gimple_builtin_call.
19228         (instrument_derefs, instrument_mem_region_access): Insert the
19229         instrumented memory reference into the hash table.
19230         (maybe_instrument_assignment): Renamed instrument_assignment into
19231         this, and change it to advance the iterator when instrumentation
19232         actually happened and return true in that case.  This makes it
19233         homogeneous with maybe_instrument_assignment, and thus give a
19234         chance to callers to be more 'regular'.
19235         (transform_statements): Clear the memory reference hash table
19236         whenever we enter a new BB, when we cross a function call, or when
19237         we are done transforming statements.  Use
19238         maybe_instrument_assignment instead of instrumentation.  No more
19239         need to special case maybe_instrument_assignment and advance the
19240         iterator after calling it; it's now handled just like
19241         maybe_instrument_call.  Update comment.
19243 2013-02-13  Richard Biener  <rguenther@suse.de>
19245         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19246         Fix loop discovery code.
19248 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
19250         PR inline-asm/56148
19251         * lra-constraints.c (process_alt_operands): Match early clobber
19252         operand with itself.  Check conflicts with earlyclobber only if
19253         the operand is not reloaded.  Prefer to reload conflicting operand
19254         if earlyclobber and matching operands are the same.
19256 2013-02-12  Richard Biener  <rguenther@suse.de>
19258         PR lto/56297
19259         * lto-streamer-out.c (write_symbol): Do not output symbols
19260         for hard register variables.
19262 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
19264         PR target/54222
19265         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19266         (umulsidi3_insn, mulsidi3_insn): New insns.
19268 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
19270         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19271         (struct tune_params): Add vec_costs field.
19272         * config/arm/arm.c (arm_builtin_vectorization_cost)
19273         (arm_add_stmt_cost): New functions.
19274         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19275         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19276         (arm_default_vec_cost): New struct of type cpu_vec_costs.
19277         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19278         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19279         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19280         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19282 2013-02-12  Richard Biener  <rguenther@suse.de>
19284         PR lto/56295
19285         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19286         decls again if possible.
19288 2013-02-12  Richard Biener  <rguenther@suse.de>
19290         PR middle-end/56288
19291         * tree-ssa.c (verify_ssa_name): Fix check, move
19292         SSA_NAME_IN_FREE_LIST check up.
19294 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
19295             Steven Bosscher   <steven@gcc.gnu.org>
19297         PR rtl-optimization/56151
19298         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19299         equal to op0 or op1, and last_insn pattern is CODE operation
19300         with MEM dest and one of the operands matches that MEM.
19302 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
19304         * doc/extend.texi: Document Function Multiversioning and "default"
19305         parameter string to target attribute.
19306         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19307         target attribute parameter is "default".
19308         (ix86_compare_version_priority): Remove checks for target attribute.
19309         (ix86_mangle_function_version_assembler_name): Change error to sorry.
19310         Remove check for target attribute equal to NULL. Add assert.
19311         (ix86_generate_version_dispatcher_body): Change error to sorry.
19313 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
19314             Jack Howarth  <howarth@bromo.med.uc.edu>
19315             Patrick Marlier  <patrick.marlier@gmail.com>
19317         PR libitm/55693
19318         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19319         define ENDFILE_SPEC as TM_DESTRUCTOR.
19320         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19322 2013-02-11  Alexander Potapenko  <glider@google.com>
19323             Jack Howarth  <howarth@bromo.med.uc.edu>
19324             Jakub Jelinek  <jakub@redhat.com>
19326         PR sanitizer/55617
19327         * config/darwin.c (cdtor_record): Rename ctor_record.
19328         (sort_cdtor_records): Rename sort_ctor_records.
19329         (finalize_dtors): New routine to sort destructors by
19330         priority before use in assemble_integer.
19331         (machopic_asm_out_destructor): Use finalize_dtors if needed.
19333 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
19335         PR rtl-optimization/56275
19336         * simplify-rtx.c (avoid_constant_pool_reference): Check that
19337         offset is non-negative and less than cmode size before
19338         calling simplify_subreg.
19340 2013-02-11  Richard Biener  <rguenther@suse.de>
19342         PR tree-optimization/56264
19343         * cfgloop.h (fix_loop_structure): Adjust prototype.
19344         * loop-init.c (fix_loop_structure): Return the number of
19345         newly discovered loops.
19346         * tree-cfgcleanup.c (repair_loop_structures): When new loops
19347         are discovered, do a full loop-closed SSA rewrite.
19349 2013-02-11  Richard Biener  <rguenther@suse.de>
19351         PR tree-optimization/56273
19352         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19353         first VRP run.
19354         (check_array_ref): Fix missing newline in dumps.
19355         (search_for_addr_array): Likewise.
19357 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
19359         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19361 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
19363         PR target/56256
19364         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19366 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
19368         PR rtl-optimization/56246
19369         * lra-constraints.c (simplify_operand_subreg): Try to reuse
19370         reload pseudo.
19371         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19372         constraints are satisfied.
19374 2013-02-08  Jeff Law  <law@redhat.com>
19376         PR debug/53948
19377         * emit-rtl.c (reg_is_parm_p): New function.
19378         * regs.h (reg_is_parm_p): New prototype.
19379         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19380         callee-clobbered registers.
19382 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
19384         PR target/56043
19385         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19386         If there is no implicit builtin declaration, just return NULL.
19388 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19390         * config/i386/sse.md (FMAMODEM): New mode iterator.
19391         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19392         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19394 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19396         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19397         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19398         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19400 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19402         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19403         (microblaze*-*-elf): Likewise.
19404         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19405         LINK_SPEC.
19406         * config/microblaze/microblaze-c.c: Add builtin defines for
19407         _LITTLE_ENDIAN and _BIG_ENDIAN.
19408         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19409         add to TARGET_DEFAULT flags.
19410         Expand ASM_SPEC and LINK_SPEC.
19411         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19412         * config/microblaze/microblaze.md: Update extendsidi2 and
19413         movdi_internal instructions to use low-order / high-order reg
19414         print_operands.
19415         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19416         options and inversemask / mask of LITTLE_ENDIAN.
19417         * config/microblaze/t-microblaze: Expand multilib options to
19418         include mlittle-endian (le) and update exceptions patterns.
19420 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
19422         PR rtl-optimization/56195
19423         * lra-constraints.c (get_reload_reg): Don't reuse regs
19424         if they have smaller mode than requested, if they have
19425         wider mode than requested, try to return a SUBREG.
19427         PR tree-optimization/56250
19428         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19429         if type is unsigned and code isn't MULT_EXPR.
19431 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19433         PR tree-optimization/56064
19434         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19435         bits according to mode.
19436         * fixed-value.h (fixed_from_double_int)
19437         (const_fixed_from_double_int): Adjust comments.
19439 2013-02-08  Richard Biener  <rguenther@suse.de>
19441         PR lto/56231
19442         * lto-streamer.h (struct data_in): Remove current_file, current_line
19443         and current_col members.
19444         * lto-streamer-out.c (lto_output_location): Stream changed bits
19445         en-block for efficiency.
19446         * lto-streamer-in.c (clear_line_info): Remove.
19447         (lto_input_location): Cache current file, line and column
19448         globally via local statics.  Read changed bits en-block.
19449         (input_function): Do not call clear_line_info.
19450         (lto_read_body): Likewise.
19451         (lto_input_toplevel_asms): Likewise.
19453 2013-02-08  Michael Matz  <matz@suse.de>
19455         PR tree-optimization/52448
19456         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19457         (nt_call_phase): New static.
19458         (add_or_mark_expr): Only mark accesses with newer phase than any
19459         call seen.
19460         (nonfreeing_call_p): New.
19461         (nt_init_block): Update nt_call_phase, mark blocks as visited.
19462         (nt_fini_block): Keep blocks marked as visited.
19463         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19465 2013-02-08  Richard Biener  <rguenther@suse.de>
19467         * ira.c (ira): Free broken dominator information.
19469 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19471         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19473 2013-02-08  Marek Polacek  <polacek@redhat.com>
19475         * cfgloop.c (verify_loop_structure): Add more checking of headers.
19477 2013-02-08  Richard Biener  <rguenther@suse.de>
19479         PR middle-end/56181
19480         * cfgloop.h (flow_loops_find): Adjust.
19481         (bb_loop_header_p): Declare.
19482         * cfgloop.c (bb_loop_header_p): New function split out from ...
19483         (flow_loops_find): ... here.  Adjust function signature,
19484         support incremental loop structure update.
19485         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
19486         * cfgloopmanip.c (fix_loop_structure): Move ...
19487         * loop-init.c (fix_loop_structure): ... here.
19488         (apply_loop_flags): Split out from ...
19489         (loop_optimizer_init): ... here.
19490         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
19491         in incremental mode, only remove dead loops here.
19493 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19495         PR target/54222
19496         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19497         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19498         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19499         (*round<mode>3.libgcc): New insns for fixed-modes.
19500         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19501         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19502         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19503         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19504         implementations.  Define to __builtin_avr_absFX,
19505         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19506         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19507         __builtin_avr_countlsFX, respectively.
19508         * config/avr/avr-c.c (target.h): Include it.
19509         (enum avr_builtin_id): New enum.
19510         (avr_resolve_overloaded_builtin): New static function.
19511         (avr_register_target_pragmas): Use it to set
19512         targetm.resolve_overloaded_builtin.
19513         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19514         tree nodes used by DEF_BUILTIN.
19515         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19516         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19517         <AVR_BUILTIN_xxBITS>: Same.
19519 2013-02-08  Richard Biener  <rguenther@suse.de>
19521         * cfgloop.c (verify_loop_structure): Properly handle
19522         a loop exiting to another loop header.
19523         * ira-int.h (ira_loops): Remove.
19524         * ira.c (ira_loops): Remove.
19525         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19526         (do_reload): Use loop_optimizer_finalize.
19527         * ira-build.c (create_loop_tree_nodes): Use get_loops and
19528         number_of_loops to access the loop tree.
19529         (more_one_region_p): Likewise.
19530         (finish_loop_tree_nodes): Likewise.
19531         (rebuild_regno_allocno_maps): Likewise.
19532         (mark_loops_for_removal): Likewise.
19533         (mark_all_loops_for_removal): Likewise.
19534         (remove_unnecessary_regions): Likewise.
19535         (ira_build): Likewise.
19536         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19538 2013-02-08  Richard Biener  <rguenther@suse.de>
19540         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
19541         * ipa-pure-const.c (analyze_function): Avoid calling
19542         mark_irreducible_loops twice.
19543         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
19545 2013-02-07  David S. Miller  <davem@davemloft.net>
19547         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
19548         on 'reg'.
19549         * var-tracking.c (vt_add_function_parameter): Test the presence of
19550         HAVE_window_save properly and do not remap argument registers when
19551         we have a leaf function.
19553 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
19555         PR bootstrap/56227
19556         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
19557         instead of "ll".
19558         * config/i386/i386.c (ix86_print_operand): Ditto.
19560 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19562         * lra-constraints.c (process_alt_operands): Fix recently added comment.
19564 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19566         PR rtl-optimization/56225
19567         * lra-constraints.c (process_alt_operands): Check that reload hard
19568         reg can hold value for strict_low_part.
19570 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19572         PR debug/56154
19573         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
19574         dwarf2out_end_function.
19575         (in_first_function_p, maybe_at_text_label_p,
19576         first_loclabel_num_not_at_text_label): New variables.
19577         (dwarf2out_var_location): In the first function find out
19578         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
19579         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
19580         functions.
19582 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
19584         PR rtl-optimization/56178
19585         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
19586         SUBREG of a register.  Tidy up related block of code.
19587         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
19588         note if the source is a register or a SUBREG of a register.
19590 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19592         PR target/56228
19593         * config/rs6000/rs6000.md (ptrm): New mode attr.
19594         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
19595         call_value_indirect_aix<pttrsize>,
19596         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
19597         m in constraints.
19599 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
19601         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
19602         if -bnortl. Convert to strcmp and strncmp.
19604 2013-02-07  Alan Modra  <amodra@gmail.com>
19606         PR target/54009
19607         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
19608         addresses won't wrap when offsetting.
19609         (rs6000_secondary_reload): Provide secondary reloads needed for
19610         wrapping LO_SUM addresses.
19612 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
19614         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
19615         MACH, just __MACH__.
19617 2013-02-06  Richard Biener  <rguenther@suse.de>
19619         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
19620         instead of calling fix_loop_structure.
19622 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
19624         PR middle-end/56217
19625         * omp-low.c (use_pointer_for_field): Return false if
19626         lower_send_shared_vars doesn't generate any copy-out code.
19628 2013-02-06  Tom de Vries  <tom@codesourcery.com>
19630         PR rtl-optimization/56131
19631         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
19632         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
19633         of the label is NULL.  Add comment.
19635 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
19637         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
19639         PR sanitizer/55374
19640         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
19641         (STATIC_LIBTSAN_LIBS): Likewise.
19642         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
19643         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
19644         is defined, don't add anything else beyond that.
19645         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
19646         (LINK_COMMAND_SPEC): Use them.
19648         PR tree-optimization/56205
19649         * tree-stdarg.c (check_all_va_list_escapes): Return true if
19650         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
19651         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
19653 2013-02-05  Richard Biener  <rguenther@suse.de>
19655         PR tree-optimization/53342
19656         PR tree-optimization/53185
19657         * tree-vectorizer.h (vect_check_strided_load): Remove.
19658         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
19659         not disallow peeling for vectorized strided loads.
19660         (vect_check_strided_load): Make static and simplify.
19661         (vect_analyze_data_refs): Adjust.
19662         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
19663         correctly when vectorizing strided loads.
19665 2013-02-05  Richard Biener  <rguenther@suse.de>
19667         * doc/install.texi: Refer to ISL, not PPL.
19669 2013-02-05  Jan Hubicka  <jh@suse.cz>
19671         PR tree-optimization/55789
19672         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
19674 2013-02-05  Jan Hubicka  <jh@suse.cz>
19676         PR tree-optimization/55789
19677         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
19678         the dead call anyway.
19680 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
19682         PR sanitizer/55374
19683         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
19685 2013-02-04  Alexander Potapenko  <glider@google.com>
19686             Jack Howarth  <howarth@bromo.med.uc.edu>
19687             Jakub Jelinek  <jakub@redhat.com>
19689         PR sanitizer/55617
19690         * config/darwin.c (sort_ctor_records): Stabilized qsort
19691         on constructor priority by using original position.
19692         (finalize_ctors): New routine to sort constructors by
19693         priority before use in assemble_integer.
19694         (machopic_asm_out_constructor): Use finalize_ctors if needed.
19696 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
19698         PR libstdc++/54314
19699         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
19700         about visibility on artificial decls.
19701         * config/sol2.c (solaris_assemble_visibility): Likewise.
19703 2013-02-04  Kai Tietz  <ktietz@redhat.com>
19705         PR target/56186
19706         * config/i386/i386.c (function_value_ms_64): Add additional valtype
19707         argument and improve checking of return-argument types for 16-byte
19708         modes.
19709         (ix86_function_value_1): Add additional valtype argument on call
19710         of function_value_64.
19711         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
19712         handling infunction_value_64 function.
19714 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
19716         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
19718 2013-02-04  Richard Biener  <rguenther@suse.de>
19720         PR tree-optimization/56188
19721         * tree-ssa-structalias.c (label_visit): Consider case with
19722         initially non-empty points-to set.
19723         (perform_var_substitution): Dump node mapping and clean up.
19725 2013-02-04  Richard Guenther  <rguenther@suse.de>
19727         PR lto/56168
19728         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
19729         node prevail as last resort.
19730         (lto_symtab_merge_decls): Remove guard on LTRANS here.
19731         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
19733 2013-02-04  Richard Biener  <rguenther@suse.de>
19735         PR tree-optimization/56113
19736         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
19737         Merge into ...
19738         (equiv_class_lookup_or_add): ... this.
19739         (label_visit): Adjust and fix error in previous patch.
19740         (perform_var_substitution): Adjust.
19742 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
19744         * config/sh/divtab.c: Fix formatting and comments throughout the file.
19745         * config/sh/sh4-300.md: Likewise.
19746         * config/sh/sh4a.md: Likewise.
19747         * config/sh/constraints.md: Likewise.
19748         * config/sh/sh.md: Likewise.
19749         * config/sh/netbsd-elf.h: Likewise.
19750         * config/sh/predicates.md: Likewise.
19751         * config/sh/sh-protos.h: Likewise.
19752         * config/sh/ushmedia.h: Likewise.
19753         * config/sh/linux.h: Likewise.
19754         * config/sh/sh.c: Likewise.
19755         * config/sh/superh.h: Likewise.
19756         * config/sh/elf.h: Likewise.
19757         * config/sh/sh4.md: Likewise.
19758         * config/sh/sh.h: Likewise.
19760 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19762         * config/pa/constraints.md: Adjust unused letters.  Change "T"
19763         constraint to match_test floating_point_store_memory_operand().
19764         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
19765         (base14_operand): New.
19766         (floating_point_store_memory_operand): New.
19767         (integer_store_memory_operand): Revise to use base14_operand and
19768         reg_plus_base_memory_operand.
19769         (move_dest_operand): Allow symbolic_memory_operands.
19770         (symbolic_memory_operand): Check for LO_SOM.
19771         (symbolic_operand): Change default case to break.
19772         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
19773         CONST_DOUBLE values to be reloaded by putting them into memory when
19774         the destination is a floating point register.
19775         (movdf): Remove code to handle CONST_DOUBLE.
19776         (movsf): Likewise.
19777         (reload_indf_r1): New.
19778         (reload_insf_r1): New.
19779         Consistently use "Q" and "T" constraints with integer and floating
19780         point move instructions, respectively.
19781         (movdi): Remove FAIL.
19782         Change predicate for source operand unamed DImode move from
19783         general_operand to move_src_operand.
19784         (umulsidi3): Change predicate for destination operand to
19785         register_operand.
19786         Likewise for similar unamed patterns.
19787         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
19788         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
19789         (hppa_legitimize_address): Simplify mask calculation.
19790         (pa_emit_move_sequence): Revised handling of secondary reloads from
19791         REG+D addresses for floating point loads and stores.  Directly handle
19792         loading CONST0_RTX (mode) to a floating point register.
19793         (pa_secondary_reload): Handle reloading DF and SFmode constant values
19794         to floating point registers.  Don't restrict secondary reloads to
19795         floating point registers to integer modes.  Revise some comments and
19796         cleanup some code.
19797         (TARGET_LEGITIMATE_ADDRESS_P): Define.
19798         (pa_legitimate_address_p): New.
19799         (pa_legitimize_reload_address): New.
19800         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
19801         (STRICT_REG_OK_FOR_BASE_P): New.
19802         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
19803         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
19805 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
19806             Andrew Dixie  <andrewd@gentrack.com>
19808         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
19809         flag set.
19811 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
19813         * expmed.c (extract_bit_field_1): Pass the full width of the
19814         structure to get_best_reg_extraction_insn.
19816 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19818         PR target/54601
19819         * configure.ac (use_cxa_atexit): Add AIX.
19820         * configure: Regenerate.
19822         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
19824 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
19826         PR debug/54793
19827         * final.c (need_profile_function): New variable.
19828         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
19829         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
19830         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
19831         notes, targetm.asm_out.function_prologue doesn't emit anything,
19832         HAVE_prologue and profiler should be emitted before prologue,
19833         set need_profile_function instead of emitting it.
19834         (final_scan_insn): If need_profile_function, emit
19835         profile_function on the first NOTE_INSN_BASIC_BLOCK or
19836         NOTE_INSN_FUNCTION_BEG note.
19838 2013-02-01  Richard Henderson  <rth@redhat.com>
19840         * config/rs6000/rs6000.md (smulditi3): New.
19841         (umulditi3): New.
19843         * config/alpha/alpha.md (umulditi3): New.
19845 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19847         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
19848         (ASM_OUTPUT_ALIGNED_LOCAL): New.
19850 2013-02-01  Richard Biener  <rguenther@suse.de>
19852         PR tree-optimization/56113
19853         * tree-ssa-structalias.c (label_visit): Reduce work for
19854         single-predecessor nodes.
19856 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
19858         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
19859         range isn't testing for zero.
19861 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
19863         PR middle-end/56113
19864         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
19866 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
19867             Nick Clifton  <nickc@redhat.com>
19869         * config/v850/constraints.md (Q): Define as a memory constraint.
19870         * config/v850/predicates.md (label_ref_operand): New predicate.
19871         (e3v5_shift_operand): New predicate.
19872         (ior_operator): New predicate.
19873         * config/v850/t-v850: Add e3v5 multilib.
19874         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
19875         (v850_gen_movdi): Prototype.
19876         * config/v850/v850.c: Add support for e3v5 architecture.
19877         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
19878         TARGET_V850E_UP.
19879         (construct_save_jarl): Add e3v5 long JARL support.
19880         (v850_adjust_insn_length): New function.  Adjust length of call
19881         insns when using e3v5 instructions.
19882         (v850_gen_movdi): New function: Generate instructions to move a
19883         DImode value.
19884         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
19885         (CPP_SPEC): Define __v850e3v5__ as appropriate.
19886         (TARGET_USE_FPU): Enable for e3v5.
19887         (CONST_OK_FOR_W): New macro.
19888         (ADJUST_INSN_LENGTH): Define.
19889         * config/v850/v850.md (UNSPEC_LOOP): Define.
19890         (attr cpu): Add v850e3v5.
19891         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
19892         (movdi): New pattern.
19893         (movdi_internal): New pattern.
19894         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
19895         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
19896         (cstoresf4): Likewise.
19897         (cstoredf4): Likewise.
19898         (insv): New pattern.
19899         (rotlso3_a): New pattern.
19900         (rotlsi3_b): New pattern
19901         (rotlsi3_v850e3v5): New pattern.
19902         (doloop_begin): New pattern.
19903         (fix_loop_counter): New pattern.
19904         (doloop_end): New pattern.
19905         (branch_normal): Add e3v5 long branch support.
19906         (branch_invert): Likewise.
19907         (branch_z_normal): Likewise.
19908         (branch_z_invert): Likewise.
19909         (branch_nz_normal): Likewise.
19910         (branch_nz_invert): Likewise.
19911         (call_internal_short): Add e3v5 register-indirect JARL support.
19912         (call_internal_long): Likewise.
19913         (call_value_internal_short): Likewise.
19914         (call_value_internal_long): Likewise.
19915         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
19916         (mloop): New option.
19917         * config.gcc: Add support for configuring v840e3v5 target.
19918         * doc/invoke.texi: Document new v850 specific command line options.
19920 2013-01-31  Paul Koning  <ni1d@arrl.net>
19922         PR debug/55059
19923         PR debug/54508
19924         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
19925         children if parent is a class.
19926         (prune_unused_types_prune): Don't add DW_AT_declaration.
19928 2013-01-31  Richard Biener  <rguenther@suse.de>
19930         PR tree-optimization/56157
19931         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
19932         match up operand with SLP child.
19934 2013-01-31  Jason Merrill  <jason@redhat.com>
19936         PR debug/54410
19937         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
19938         parameters the first time.
19939         (gen_scheduled_generic_parms_dies): Check completeness here.
19941 2013-01-31  Richard Biener  <rguenther@suse.de>
19943         PR middle-end/53073
19944         * common.opt (faggressive-loop-optimizations): New flag,
19945         enabled by default.
19946         * doc/invoke.texi (faggressive-loop-optimizations): Document.
19947         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
19948         infer_loop_bounds_from_undefined by it.
19950 2013-01-31  Richard Biener  <rguenther@suse.de>
19952         PR tree-optimization/56150
19953         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
19954         visit virtual operands.
19955         (find_uses_to_rename_bb): Likewise.
19957 2013-01-31  Richard Biener  <rguenther@suse.de>
19959         PR tree-optimization/56150
19960         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
19961         mixed store non-store stmts.
19963 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
19965         PR sanitizer/55374
19966         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
19967         LIBASAN_EARLY_SPEC is defined.
19968         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
19969         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
19970         before %o.
19971         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
19973         PR c++/55742
19974         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
19975         invalid args instead of ICEing on it.
19976         (ix86_valid_target_attribute_tree): Return error_mark_node if
19977         ix86_valid_target_attribute_inner_p failed.
19978         (ix86_valid_target_attribute_p): Return false only if
19979         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
19980         target("default") attribute.
19981         (sorted_attr_string): Change argument from const char * to tree,
19982         merge in all target attribute arguments rather than just one.
19983         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
19984         instead of free.  Avoid using strcat.
19985         (ix86_mangle_function_version_assembler_name): Mangle
19986         target("default") as if no target attribute is present.  Adjust
19987         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
19988         instead of xmalloc and XDELETEVEC instead of free.
19989         (ix86_function_versions): Don't return true if one of the decls
19990         doesn't have target attribute.  If they don't and one of the decls
19991         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
19992         sorted_attr_string caller.  Use XDELETEVEC instead of free.
19993         (ix86_supports_function_versions): Remove.
19994         (make_name): Fix up formatting.
19995         (make_dispatcher_decl): Remove resolver_name and its initialization.
19996         Avoid leaking memory.
19997         (is_function_default_version): Return true if there is
19998         target("default") attribute rather than no target attribute at all.
19999         (make_resolver_func): Avoid leaking memory.
20000         (ix86_generate_version_dispatcher_body): Likewise.
20001         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20002         * target.def (supports_function_versions): Remove.
20003         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20004         * doc/tm.texi: Regenerated.
20006 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20008         PR rtl-optimization/56144
20009         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20010         for values with side effects.
20012 2013-01-30  Richard Biener  <rguenther@suse.de>
20014         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20015         (sparseset_pop): Likewise.
20016         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20017         to be able to use quick_push in the worker loop.
20019 2013-01-30  Marek Polacek  <polacek@redhat.com>
20021         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20023 2013-01-30  Richard Biener  <rguenther@suse.de>
20025         PR lto/56147
20026         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20028 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20030         PR tree-optimization/56064
20031         * fixed-value.c (fixed_from_double_int): New function.
20032         * fixed-value.h (fixed_from_double_int): New prototype.
20033         (const_fixed_from_double_int): New static inline function.
20034         * fold-const.c (native_interpret_fixed): New static function.
20035         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20036         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20037         (native_encode_fixed): New static function.
20038         (native_encode_expr) <FIXED_CST>: Use it.
20039         (native_interpret_int): Move double_int worker code to...
20040         * double-int.c (double_int::from_buffer): ...this new static method.
20041         * double-int.h (double_int::from_buffer): Prototype it.
20043 2013-01-30  Richard Biener  <rguenther@suse.de>
20045         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20046         New pointer-map and obstack.
20047         (init_alias_vars): Allocate pointer-map and obstack.
20048         (delete_points_to_sets): Free them.
20049         (find_what_var_points_to): Cache result.
20050         (find_what_p_points_to): Adjust for changed interface of
20051         find_what_var_points_to.
20052         (compute_points_to_sets): Likewise.
20053         (ipa_pta_execute): Likewise.
20055 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20057         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20058         * configure: Regenerate.
20059         * config.in: Regenerate.
20060         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20061         #nobits/#progbits if supported.
20063 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20065         PR target/56121
20066         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20067         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20068         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20070 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20072         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20073         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20075 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20077         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20078         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20080 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20082         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20083         declaration.
20084         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20085         * config/arm/cortex-a7.md: New bypasses using
20086         arm_mac_accumulator_is_result.
20088 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20090         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20091         (cortex_a7_neon_mla): Likewise.
20092         (cortex_a7_fpfmad): New reservation.
20093         (cortex_a7_fpmacs): Use ffmas and update required units.
20094         (cortex_a7_fpmuld): Update required units and latency.
20095         (cortex_a7_fpmacd): Likewise.
20096         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20097         (cortex_a7_neon). Likewise.
20098         (bypass) Update participating units.
20100 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20102         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20103         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20104         from fmac to ffma.
20105         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20106         (vfp_fmul): Use ffmad.
20107         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20108         (cortex_r4_fmacd): Use ffmad.
20109         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20110         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20111         (cortex_a9_fmacd): Use ffmad.
20112         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20113         (cortex_a8_vfp_macd): Use ffmad.
20114         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20115         (cortex_a5_fpmacd): Use ffmad.
20116         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20117         (cortex_a15_vfp_macd): Use ffmad.
20118         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20120 2013-01-29  Jason Merrill  <jason@redhat.com>
20122         PR libstdc++/54314
20123         * varasm.c (default_assemble_visibility): Don't warn about
20124         visibility on artificial decls.
20126 2013-01-29  Richard Biener  <rguenther@suse.de>
20128         PR tree-optimization/56113
20129         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20130         the bitmap leader.
20131         (label_visit): Free duplicate bitmaps and record the leader instead.
20132         (perform_var_substitution): Adjust.
20134 2013-01-29  Richard Biener  <rguenther@suse.de>
20136         PR tree-optimization/55270
20137         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20138         the CFG, schedule loops for fixup.
20140 2013-01-29  Nick Clifton  <nickc@redhat.com>
20142         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20143         SP_REG.
20145 2013-01-28  Leif Ekblad  <leif@rdos.net>
20147         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20148         * config/i386/i386.h (TARGET_RDOS): New macro.
20149         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20150         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20151         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20152         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20153         DEFAULT_LARGE_SECTION_THRESHOLD.
20154         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20155         * config/i386/rdos.h: New file.
20156         * config/i386/rdos64.h: New file.
20158 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20160         PR other/54814
20161         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20162         TEST_HARD_REG_BIT.
20164 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20166         PR rtl-optimization/56117
20167         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20168         call cselib_lookup_from_insn on the MEM before calling
20169         add_insn_mem_dependence.
20171 2013-01-28  Richard Biener  <rguenther@suse.de>
20173         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20174         to a stmt that didn't have one.
20175         (copy_phis_for_bb): Likewise for PHI arguments.
20176         (copy_debug_stmt): Likewise for debug stmts.
20178 2013-01-28  Richard Biener  <rguenther@suse.de>
20180         PR tree-optimization/56034
20181         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20182         (partition_builtin_p): Adjust.
20183         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20184         it is the last partition.
20185         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20186         up the vertex for the definition.
20187         (classify_partition): Classify whether a partition is a
20188         PKIND_REDUCTION, thus has uses outside of the loop.
20189         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20190         Merge all PKIND_REDUCTION partitions into the last partition.
20191         (tree_loop_distribution): Seed partitions from reductions as well.
20193 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20195         PR tree-optimization/56125
20196         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20197         pow(x,c) into sqrt(x) * powi(x, n/2) or
20198         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20199         optimizing for size.
20200         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20201         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20202         integer.
20204         PR tree-optimization/56094
20205         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20206         to UNKNOWN_LOCATION while gimplifying expr.
20208 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
20210         PR target/56114
20211         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20212         operand 0 in movabs insn template for -masm=intel asm alternative.
20213         (*movabs<mode>_2): Ditto for operand 1.
20215 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
20217         PR target/54663
20218         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20219         of microblaze-c.o
20221 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20223         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20224         tm_file.
20226 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20228         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20229         Undef to avoid warning.
20231 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20233         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20234         * configure: Regenerate.
20236 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
20238         PR tree-optimization/56098
20239         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20240         for stmts with volatile ops.
20241         (cond_store_replacement): Don't optimize if assign has volatile ops.
20242         (cond_if_else_store_replacement_1): Don't optimize if either
20243         then_assign or else_assign have volatile ops.
20244         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20245         volatile ops.
20247 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20249         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20251 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20253         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20254         missing ':' in asm example.
20256 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
20258         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20259         entries into lane and laneq entries.
20260         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20261         Remove AdvSIMD scalar modes.
20262         (aarch64_sq<r>dmulh_laneq<mode>): New.
20263         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20264         modes.
20265         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20266         builtin implementations to relfect changes in RTL in aarch64-simd.md.
20267         * config/aarch64/iterators.md (VCOND): New.
20268         (VCONQ): New.
20270 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20272         PR target/54222
20273         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20274         Add NULL LIBNAME argument to existing definitions.
20275         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20276         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20277         * config/avr/avr.c (DEF_BUILTIN): Same.
20278         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20279         (avr_expand_builtin): Expand to a vanilla call if a libgcc
20280         implementation is available (DECL_ASSEMBLER_NAME is set).
20281         (avr_fold_absfx): New static function.
20282         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20283         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20284         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20285         AVR_BUILTIN_ABSLLK.
20286         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20287         (abshk, absk, abslk, absllk): Provide as static inline functions.
20289 2013-01-25  Marek Polacek  <polacek@redhat.com>
20291         PR tree-optimization/56035
20292         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20294 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
20296         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20297         (*movtf_internal_rex64): Add (!o,C) alternative
20298         (*movxf_internal_rex64): Ditto.
20299         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20301 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
20303         * doc/invoke.texi: fix typo.
20304         * doc/objc.texi: fix typo.
20306 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
20308         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20309         for the first two alternatives.
20311 2013-01-24  Diego Novillo  <dnovillo@google.com>
20313         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
20314         (ggc-zone.o): Remove.
20315         * configure.ac: Remove option --with-gc.
20316         * configure: Re-generate.
20317         * doc/install.texi: Remove documentation for --with-gc.
20318         * gengtype.c (write_enum_defn): Remove.  Update all users.
20319         (write_Types_process_field): Remove generation of gt_e_* argument.
20320         (output_type_enum): Remove.  Update all users.
20321         (write_enum_defn): Remove.  Update all users.
20322         (enum alloc_zone): Remove.  Update all users.
20323         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20324         * ggc-common.c (ggc_splay_alloc): Remove first argument.
20325         Update all callers.
20326         (struct ptr_data): Remove field TYPE.  Update all users.
20327         (gt_pch_note_object): Remove argument TYPE.  Update all users.
20328         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20329         Update all users.
20330         * ggc-none.c (ggc_alloc_typed_stat): Remove.
20331         (struct alloc_zone): Remove.
20332         (ggc_internal_alloc_zone_stat): Remove.
20333         (ggc_internal_cleared_alloc_zone_stat): Remove.
20334         * ggc-page.c (ggc_alloc_typed_stat): Remove.
20335         (ggc_pch_count_object): Remove last argument.  Update all users.
20336         (ggc_pch_alloc_object): Remove last argument.  Update all users.
20337         (struct alloc_zone): Remove.
20338         * ggc-zone.c: Remove.
20339         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
20340         (struct alloc_zone): Remove.
20341         (ggc_alloc_typed_stat): Remove.
20342         (ggc_alloc_typed): Remove.
20343         (ggc_splay_alloc): Remove first argument.
20344         (rtl_zone): Remove.  Update all users.
20345         (tree_zone): Remove.  Update all users.
20346         (tree_id_zone): Remove.  Update all users.
20347         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
20348         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
20349         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
20350         * tree-ssanames.c: Remove references to zone allocator in comments.
20352 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
20354         * config/avr/avr.c (avr_out_fract): Make register numbers that
20355         might be outside of source operand signed.
20357 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
20359         * config/i386/constraints.md (Yf): New constraint.
20360         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20361         of f constraint to conditionaly disable x87 register preferences.
20362         (*movdf_internal): Ditto.
20363         (*movsf_internal): Ditto.
20365 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
20367         PR inline-asm/55934
20368         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20369         that have operands with impossible constraints.
20370         Add a FIXME for a speed-up opportunity.
20371         * lra-constraints.c (process_alt_operands): Verify that a class
20372         selected from constraints on asms is valid for the operand mode.
20373         (curr_insn_transform): Remove incorrect comment.
20375 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
20377         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20378         TOC operand is a valid symbol ref in the constant pool.
20380 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20382         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20384 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
20386         PR target/54222
20387         * config/avr/stdfix.h: New file.
20388         * t-avr (stdfix-gcc.h): New rule to build it.
20389         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20391 2013-01-23  Kostya Serebryany  <kcc@google.com>
20393         * config/darwin.h: remove dependency on
20394         CoreFoundation (asan on Mac OS).
20396 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
20398         PR target/49069
20399         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20400         instead of cmpdi_operand for first comparison operand.
20401         Don't assert that comparison operands aren't both constants.
20403 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20405         * doc/install.texi (Downloading the Source): Update references to
20406         downloading separate components.
20408 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20410         * doc/extend.texi (__int128): Improve grammar.
20412 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
20414         PR target/56028
20415         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20416         alternative to (o,r).
20417         (*movdi_internal_rex64): Remove (!o,n) alternative.
20418         (DImode immediate->memory splitter): Remove.
20419         (DImode immediate->memory peephole2): Remove.
20420         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20421         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20422         alternative to (!o,*r).
20423         (*movtf_internal_sse): New pattern.
20424         (*movxf_internal_rex64): New pattern.
20425         (*movxf_internal): Disable for TARGET_64BIT.
20426         (*movdf_internal_rex64): Remove (!o,F) alternative.
20428 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
20430         PR middle-end/56074
20431         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20432         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20433         * tree-vect-loop-manip.c (find_loop_location): Also ignore
20434         stmt locations where LOCATION_LOCUS of the stmt location is
20435         UNKNOWN_LOCATION or BUILTINS_LOCATION.
20437         PR target/55686
20438         * config/i386/i386.md (UNSPEC_STOS): New.
20439         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20440         *strsetqi_1): Add UNSPEC_STOS.
20442 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
20444         PR c++/56067
20445         * doc/invoke.texi: Remove left over -Wsynth example.
20447 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20449         PR tree-optimization/56051
20450         * fold-const.c (fold_binary_loc): Don't fold
20451         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20452         a narrowing conversion, or widening conversion from signed
20453         to unsigned.
20455 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20457         PR rtl-optimization/56023
20458         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20459         dependent on debug instruction.
20461 2013-01-21  Martin Jambor  <mjambor@suse.cz>
20463         PR middle-end/56022
20464         * function.c (allocate_struct_function): Call
20465         invoke_set_current_function_hook earlier.
20467 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20469         * reload1.c (init_reload): Only initialize reload_obstack
20470         during the first call.
20472 2013-01-21  Marek Polacek  <polacek@redhat.com>
20474         * cfgloop.c (verify_loop_structure): Fix up grammar.
20476 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
20478         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20479         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20481 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20483         PR target/56058
20484         * config/arm/marvell-pj4.md: Update copyright year.
20485         Fix up use of alu to alu_reg and simple_alu_imm.
20487 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20489         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20491 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
20493         PR target/55433
20494         * lra-constraints.c (curr_insn_transform): Don't reuse original
20495         insn for secondary memory move when memory mode should be different.
20497 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20499         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20500         atomic_storedi_1): New patterns.
20502 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
20504         btver2 pipeline descriptions.
20505         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20506         descriptions.
20507         * config/i386/i386.md (btver2_decode): New type attributes.
20508         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20509         type attributes.
20510         * config/i386/btver2.md: New file describing btver2 pipelines.
20512 2013-01-19  Andrew Pinski  <apinski@cavium.com>
20514         PR tree-optimization/52631
20515         * tree-ssa-sccvn (visit_use): Before looking up the original
20516         statement, try looking up the simplified expression.
20518 2013-01-19  Anthony Green  <green@moxielogic.com>
20520         * config/moxie/moxie.c (moxie_expand_prologue): Set
20521         current_function_static_stack_size.
20523 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20525         PR tree-optimization/56029
20526         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20527         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20529 2013-01-18  Sharad Singhai  <singhai@google.com>
20531         PR tree-optimization/55995
20532         * dumpfile.c (dump_loc): Print location only if available.
20533         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
20535 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
20537         PR target/55433
20538         * lra-constraints.c (curr_insn_transform): Reuse original insn for
20539         secondary memory move.
20540         (inherit_reload_reg): Use rclass instead of cl for
20541         check_secondary_memory_needed_p.
20543 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20545         PR middle-end/56015
20546         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
20547         the case where writing real complex part of target modifies op1.
20549 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
20551         * config/aarch64/aarch64-simd.md
20552         (aarch64_vcond_internal<mode>): Handle unordered cases.
20553         * config/aarch64/iterators.md (v_cmp_result): New.
20555 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
20556             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20558         * config/arm/marvell-pj4.md: New file.
20559         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
20560         * config/arm/arm.md (generic_sched): Add marvell_pj4.
20561         (generic_vfp): Likewise.
20562         * config/arm/arm-cores.def: Add marvell-pj4.
20563         * config/arm/arm-tune.md: Regenerate.
20564         * config/arm/arm-tables.opt: Regenerate.
20565         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
20566         * doc/invoke.texi: Document marvell-pj4.
20568 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
20570         * config/aarch64/arm_neon.h: Map scalar types to standard types.
20572 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20574         PR debug/54114
20575         PR debug/54402
20576         PR debug/49888
20577         * var-tracking.c (negative_power_of_two_p): New.
20578         (global_get_addr_cache, local_get_addr_cache): New.
20579         (get_addr_from_global_cache, get_addr_from_local_cache): New.
20580         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
20581         heading comment.
20582         (vt_stack_offset_p): Remove.
20583         (vt_canon_true_dep): Always canonicalize loc's address.
20584         (clobber_overlapping_mems): Make sure we have a MEM.
20585         (local_get_addr_clear_given_value): New.
20586         (val_reset): Clear local cached entries.
20587         (compute_bb_dataflow): Create and release the local cache.
20588         Disable duplicate MEMs clobbering.
20589         (emit_notes_in_bb): Clobber MEMs likewise.
20590         (vt_emit_notes): Create and release the local cache.
20591         (vt_initialize, vt_finalize): Create and release the global
20592         cache, respectively.
20593         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
20595 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20597         PR libmudflap/53359
20598         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
20599         not found in the symtab.
20601 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20603         PR debug/56006
20604         PR rtl-optimization/55547
20605         PR rtl-optimization/53827
20606         PR debug/53671
20607         PR debug/49888
20608         * alias.c (offset_overlap_p): New, factored out of...
20609         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
20610         the conservative special case for symbolic constants.  Don't
20611         adjust zero sizes on alignment.
20613 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
20615         PR rtl-optimization/52573
20616         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
20617         REG_UNUSED for the same register.
20619 2013-01-17  Richard Biener  <rguenther@suse.de>
20620             Marek Polacek  <polacek@redhat.com>
20622         PR rtl-optimization/55833
20623         * loop-unswitch.c (unswitch_loops): Move loop verification...
20624         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
20625         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
20626         Set it to true when we're removing a loop from hierarchy tree in
20627         an irreducible region.
20628         (fix_bb_placements): Adjust caller.
20629         (fix_loop_placements): Likewise.
20631 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
20633         * config/avr/builtins.def (DEF_BUILTIN): Factor out
20634         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
20635         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
20636         Remove ID.  Adjust comments.
20637         * config/avr/avr-c.c (avr_builtin_name): Remove.
20638         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
20639         * config/avr/avr.c (avr_tolower): New static function.
20640         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
20641         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
20642         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
20643         default expansion.
20645 2013-01-17  Jan Hubicka  <jh@suse.cz>
20647         PR tree-optimization/55273
20648         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
20650 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
20652         PR target/55981
20653         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
20654         store through atomic_store<mode>_1.
20655         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
20657 2013-01-17  Martin Jambor  <mjambor@suse.cz>
20659         PR tree-optimizations/55264
20660         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
20661         for virtual methods.
20662         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
20663         virtual methods before inlining is over.
20664         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
20665         virtual functions.
20666         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
20667         non-virtual.
20669 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
20671         PR rtl-optimization/56005
20672         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
20673         pending reads for prefetch.
20675 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
20677         * config/aarch64/aarch64.md
20678         (*cstoresi_neg_uxtw): New pattern.
20679         (*cmovsi_insn_uxtw): New pattern.
20680         (*<optab>si3_uxtw): New pattern.
20681         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
20682         (*<optab>si3_insn_uxtw): New pattern.
20683         (*bswapsi2_uxtw): New pattern.
20685 2013-01-16  Richard Biener  <rguenther@suse.de>
20687         * tree-inline.c (tree_function_versioning): Remove set but
20688         never used variable.
20690 2013-01-16  Richard Biener  <rguenther@suse.de>
20692         PR tree-optimization/55964
20693         * tree-flow.h (rename_variables_in_loop): Remove.
20694         (rename_variables_in_bb): Likewise.
20695         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
20696         (copy_loop_before): Adjust and delete update-ssa status.
20697         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
20698         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
20699         (rename_variables_in_loop): Remove.
20700         (slpeel_update_phis_for_duplicate_loop): Likewise.
20701         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
20702         use available cfg machinery instead of duplicating it.
20703         Update PHI nodes and perform poor-mans SSA update here.
20704         (slpeel_tree_peel_loop_to_edge): Adjust.
20706 2013-01-16  Richard Biener  <rguenther@suse.de>
20708         PR tree-optimization/54767
20709         PR tree-optimization/53465
20710         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
20711         (vrp_visit_phi_node): For PHI arguments coming via backedges
20712         drop all symbolical range information.
20713         (execute_vrp): Compute backedges.
20715 2013-01-16  Richard Biener  <rguenther@suse.de>
20717         * doc/install.texi: Update CLooG and ISL requirements to
20718         0.18.0 and 0.11.1.
20720 2013-01-16  Christian Bruel  <christian.bruel@st.com>
20722         PR target/55301
20723         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
20724         (broken_move): Handle UNSPECV_SP_SWITCH_B.
20725         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
20727 2013-01-16  DJ Delorie  <dj@redhat.com>
20729         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
20730         (UNSPECV_SP_SWITCH_E): New.
20731         (sp_switch_1): Change to an unspec.
20732         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
20733         replace $r15.
20735 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
20737         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
20738         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
20739         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
20740         (expand_mem_thread_fence): Ditto.
20741         (expand_mem_signal_fence): Ditto.
20742         (expand_atomic_load): Ditto.
20743         (expand_atomic_store): Ditto.
20745 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
20747         PR rtl-optimization/55547
20748         PR rtl-optimization/53827
20749         PR debug/53671
20750         PR debug/49888
20751         * alias.c (memrefs_conflict_p): Set sizes to negative after
20752         AND adjustments.
20754 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20756         PR target/55940
20757         * function.c (thread_prologue_and_epilogue_insns): Always
20758         add crtl->drap_reg to set_up_by_prologue.set, even if
20759         stack_realign_drap is false.
20761 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
20763         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
20764         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
20765         *call): Fix indention.
20767 2013-01-15  Tom de Vries  <tom@codesourcery.com>
20769         PR target/55876
20770         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
20771         Update comment.
20773 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
20775         PR rtl-optimization/55153
20776         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
20778 2013-01-15  Martin Jambor  <mjambor@suse.cz>
20780         PR tree-optimization/55920
20781         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
20782         accesses as grp_to_be_debug_replaced.
20784 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20786         PR tree-optimization/55920
20787         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
20788         there is non-useless type conversion needed from debug rhs to lhs,
20789         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
20791 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
20792             Mikael Pettersson  <mikpe@it.uu.se>
20794         PR target/43961
20795         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
20796         Thumb.
20797         (ASM_OUTPUT_CASE_LABEL): Remove.
20798         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
20799         * final.c (shorten_branches): Update alignment of labels before
20800         jump tables if CASE_VECTOR_SHORTEN_MODE.
20802 2013-01-15  Richard Biener  <rguenther@suse.de>
20804         PR bootstrap/55961
20805         * system.h: Do not include gmp.h for building host tools.
20807 2013-01-15  Richard Biener  <rguenther@suse.de>
20809         PR middle-end/55882
20810         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
20811         account for bitpos when computing alignment.
20813 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
20815         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
20816         (ix86_target_macros_internal): Likewise.
20818         * config/i386/i386.c (m_CORE2I7): Removed.
20819         (m_CORE_HASWELL): New macro.
20820         (m_CORE_ALL): Likewise.
20821         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
20822         (initial_ix86_arch_features): Likewise.
20823         (processor_target_table): Initializations for Core avx2.
20824         (cpu_names): New names "core-avx2".
20825         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
20826         PROCESSOR_CORE_HASWELL.
20827         (ix86_issue_rate): New case.
20828         (ia32_multipass_dfa_lookahead): Likewise.
20829         (ix86_sched_init_global): Likewise.
20831         * config/i386/i386.h (TARGET_HASWELL): New macro.
20832         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
20833         (processor_type): New PROCESSOR_HASWELL.
20835 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20837         PR tree-optimization/55955
20838         * tree-vect-loop.c (vectorizable_reduction): Give up early on
20839         *SHIFT_EXPR and *ROTATE_EXPR codes.
20841         PR tree-optimization/48766
20842         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
20843         -ftrapv disable -fwrapv.
20845 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20847         PR target/55974
20848         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
20849         etc. to 1 and not to __flash.
20850         Use LL suffix for __INT24_MAX__ with -mint8.
20851         Use ULL suffix for __UINT24_MAX__ with -mint8.
20853 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20855         * config/avr/avr-arch.h
20856         (struct base_arch_s): Use typedef avr_arch_t instead.
20857         (struct arch_info_s): Use typedef avr_arch_info_t instead.
20858         (struct mcu_type_s): Use typedef avr_mcu_t instead.
20859         * config/avr/avr.c: Same.
20860         * config/avr/avr-devices.c: Same.
20861         * config/avr/driver-avr.c: Same.
20862         * config/avr/gen-avr-mmcu-texi.c: Same.
20863         * config/avr/avr-mcus.def: Adjust comment.
20865 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
20867         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
20868         * config/aarch64/iterators.md (VALLDI): New.
20870 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
20871             Andi Kleen  <ak@linux.intel.com>
20873         PR target/55948
20874         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
20875         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
20876         memmodel flag.
20878 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20880         * config/avr/avr-stdint.h: Remove trailing blanks.
20881         * config/avr/avr-log.h: Same.
20882         * config/avr/avr-arch.h: Same.
20883         * config/avr/avr-devices.c: Same.
20884         * config/avr/avr-dimode.md: Same.
20885         * config/avr/predicates.md: Same.
20886         * config/avr/avr-c.c: Same.  And fix typo.
20888         * config/avr/avr-protos.h: Same.  And:
20889         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
20890         (init_cumulative_args): Rename to avr_init_cumulative_args.
20891         (expand_prologue): Rename to avr_expand_prologue.
20892         (expand_epilogue): Rename to avr_expand_epilogue.
20893         (adjust_insn_length): Rename to avr_adjust_insn_length.
20894         (notice_update_cc): Rename to avr_notice_update_cc.
20895         (final_prescan_insn): Rename to avr_final_prescan_insn.
20896         * config/avr/avr.c: Same.
20897         * config/avr/avr.h: Same.
20898         * config/avr/avr.md: Remove trailing blanks.
20899         (prologue): Use avr_expand_prologue.
20900         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
20902 2013-01-14  Richard Biener  <rguenther@suse.de>
20904         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
20905         verify_location, collect_subblocks): New functions.
20906         (verify_gimple_in_cfg): Verify that locations only reference
20907         BLOCKs in the functions BLOCK tree.
20909 2013-01-14  Richard Biener  <rguenther@suse.de>
20911         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
20912         PHI argument.
20913         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
20914         unshare reference.
20915         (insert_out_of_ssa_copy_on_edge): Likewise.
20916         (rewrite_close_phi_out_of_ssa): Likewise.
20917         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
20918         debug expressions.
20919         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
20920         propagated constants.
20921         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
20922         can not be shared.
20924 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20926         * config/avr/avr-modes.def: Add GPL copyright notice.
20928 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
20930         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
20931         MEMMODEL_MASK to determine memory model.
20932         (atomic_store<mode>): Ditto from operands[2].
20933         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
20935 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
20937         PR fortran/55935
20938         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
20939         (fold_gimple_assign): Don't call unshare_expr here.
20940         (fold_ctor_reference): Call unshare_expr.
20942 2013-01-13  Terry Guo  <terry.guo@arm.com>
20944         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
20945         * doc/fragments.texi: Document MULTILIB_REUSE.
20946         * gcc.c (multilib_reuse): New internal spec.
20947         (set_multilib_dir): Also search multilib from multilib_reuse.
20948         * genmultilib (tmpmultilib3): Refactor code.
20949         (tmpmultilib4): Ditto.
20950         (multilib_reuse): New multilib argument.
20952 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
20954         * Makefile.in: Update copyright.
20956 2013-01-12  Tom de Vries  <tom@codesourcery.com>
20958         PR middle-end/55890
20959         * calls.c (expand_call): Check if arg_nr is valid.
20961 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
20963         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
20964         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
20965         documentation.  Add missing '__' in front of
20966         __builtin_ia32_packssdw256.
20968 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20970         PR target/55719
20971         * config/s390/s390.c (s390_preferred_reload_class): Do not return
20972         NO_REGS for larl operands.
20973         (s390_reload_larl_operand): Use s390_load_address instead of
20974         emit_move_insn.
20976 2013-01-11  Richard Biener  <rguenther@suse.de>
20978         * tree-cfg.c (verify_node_sharing_1): Split out from ...
20979         (verify_node_sharing): ... here.
20980         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
20982 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
20984         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
20985         Substitute TREECHECKING.
20986         * configure: Regenerate.
20987         * Makefile.in (TREECHECKING): New.
20989 2013-01-11  Richard Guenther  <rguenther@suse.de>
20991         PR tree-optimization/44061
20992         * tree-vrp.c (extract_range_basic): Compute zero as
20993         value-range for __builtin_constant_p of function parameters.
20995 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
20997         Update copyright years.
20999 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21001         PR rtl-optimization/55672
21002         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21003         const to be eliminable.
21005 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21007         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21008         * configure: Regenerate.
21010 2013-01-10  Richard Biener  <rguenther@suse.de>
21012         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21014 2013-01-10  Richard Biener  <rguenther@suse.de>
21016         PR bootstrap/55792
21017         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21018         locations for virtual PHI arguments.
21019         (rewrite_update_phi_arguments): Likewise.
21021 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21023         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21024         on to assembler.
21026 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21028         PR tree-optimization/55921
21029         * tree-complex.c (expand_complex_asm): New function.
21030         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21032 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21034         PR target/55718
21035         * config/s390/s390.c (s390_symref_operand_p)
21036         (s390_loadrelative_operand_p): Merge the two functions.
21037         (s390_check_qrst_address, print_operand_address): Add parameters
21038         to s390_loadrelative_operand_p invokation.
21039         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21040         (s390_reload_larl_operand, s390_secondary_reload): Use
21041         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21042         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21044 2013-01-09  Mike Stump  <mikestump@comcast.net>
21046         * dse.c (record_store): Remove unnecessary assert.
21048 2013-01-09  Jan Hubicka  <jh@suse.cz>
21050         PR tree-optimization/55569
21051         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21052         * cfgloop.h (scale_loop_profile): Likewise.
21054 2013-01-09  Jan Hubicka  <jh@suse.cz>
21056         PR lto/45375
21057         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21058         functions.
21059         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21061 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21063         PR middle-end/55114
21064         * expr.h (maybe_emit_group_store): Declare.
21065         * expr.c (maybe_emit_group_store): New function.
21066         * builtins.c (expand_builtin_int_roundingfn): Call it.
21067         (expand_builtin_int_roundingfn_2): Likewise.
21069 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21071         PR rtl-optimization/55829
21072         * lra-constraints.c (match_reload): Add code for absent output.
21073         (curr_insn_transform): Add code for reloads of matched inputs
21074         without output.
21076 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21078         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21079         attribute of movddup insn to DF.
21080         (*vec_interleave_lowv2df): Ditto.
21081         (vec_dupv2df): Ditto.
21083 2013-01-09  Jan Hubicka  <jh@suse.cz>
21085         PR tree-optimiation/55875
21086         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21087         EVERY_ITERATION parameter.
21088         (number_of_iterations_exit): Check if exit is executed every iteration.
21089         (idx_infer_loop_bounds): Similarly here.
21090         (n_of_executions_at_most): Simplify
21091         to only test for cases where statement is dominated by the
21092         particular bound; handle correctly the "postdominance" test.
21093         (scev_probably_wraps_p): Use max loop iterations info
21094         as a global bound first.
21096 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21097             Nick Clifton  <nickc@redhat.com>
21099         * config/v850/v850.md (cbranchsf4): New pattern.
21100         (cstoresf4): New pattern.
21101         (cbranchdf4): New pattern.
21102         (cstoredf4): New pattern.
21103         (movsicc): Disallow floating point comparisons.
21104         (cmpsf_le_insn): Fix order of operators.
21105         (cmpsf_lt_insn): Likewise.
21106         (cmpsf_eq_insn): Likewise.
21107         (cmpdf_le_insn): Likewise.
21108         (cmpdf_lt_insn): Likewise.
21109         (cmpdf_eq_insn): Likewise.
21110         (cmpsf_ge_insn): Use LE comparison.
21111         (cmpdf_ge_insn): Likewise.
21112         (cmpsf_gt_insn): Use LT comparison.
21113         (cmpdf_gt_insn): Likewise.
21114         (cmpsf_ne_insn): Delete pattern.
21115         (cmpdf_ne_insn): Delete pattern.
21116         * config/v850/v850.c (v850_gen_float_compare): Use
21117         gen_cmpdf_eq_insn for NE comparison.
21118         (v850_float_z_comparison_operator)
21119         (v850_float_nz_comparison_operator): Move from here ...
21120         * config/v850/predicates.md: ... to here.  Move GT and GE
21121         comparisons into v850_float_z_comparison_operator.
21122         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21123         Delete prototype.
21124         (v850_float_nz_comparison_operator): Likewise.
21126 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21128         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21129         with calls to gen_insvsi/gen_insvdi.
21131 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21133         * config/i386/i386.c (initial_ix86_tune_features): Set up
21134         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21136 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21137             Jakub Jelinek  <jakub@redhat.com>
21139         PR tree-optimization/48189
21140         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21141         If nitercst is 0, don't predict the exit edge.
21143 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21145         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21146         in asm_fprintf with reg_names.
21147         (aarch64_print_operand_address): Likewise.
21148         (aarch64_return_addr): Likewise.
21149         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21151 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21153         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21154         (INT_U6_BITS): Likewise.
21155         * config/pa/predicates.md (uint6_operand): New predicate.
21156         (shift5_operand, shift6_operand): Likewise.
21157         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21158         arith32_operand.
21159         (lshrdi3): Use shift6_operand.
21160         (shrpsi4, shrpdi4): New insn patterns.
21161         (extzv): Delete expander.
21162         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21163         predicates in unamed zero extract patterns.  Tighten common constraint.
21164         (extv): Delete expander.
21165         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21166         predicates in unamed sign extract patterns.  Tighten common constraint.
21167         (insv): Delete expander.
21168         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21169         predicates in unamed insert patterns.  Tighten common constraint.
21170         Change uint32_operand predicate to uint6_operand predicate in unamed
21171         DImode pattern to insert constant values of type 1...1xxxx.
21173 2013-01-04  Jan Hubicka  <jh@suse.cz>
21175         PR tree-optimization/55823
21176         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21177         issue.
21179 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21180             Uros Bizjak  <ubizjak@gmail.com>
21182         PR rtl-optimization/55845
21183         * df-problems.c (can_move_insns_across): Stop scanning at
21184         volatile_insn_p source instruction or give up if
21185         across_from .. across_to range contains any volatile_insn_p
21186         instructions.
21188 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21190         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21191         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21192         Declare.
21193         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21194         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21196 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21198         PR fortran/55341
21199         * asan.c (asan_clear_shadow): New function.
21200         (asan_emit_stack_protection): Use it.
21202 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21204         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21205         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21206         with tab instead of space.
21208 2013-01-08  Nick Clifton  <nickc@redhat.com>
21210         * config/rl78/rl78.c (rl78_expand_prologue): Always select
21211         register bank 0 at the start of an interrupt handler.
21212         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21213         MDBH registers.
21215 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21217         * config/aarch64/aarch64-simd.md
21218         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21219         (aarch64_simd_bsl): Likewise.
21220         (aarch64_vcond_internal<mode>): Likewise.
21221         (vcond<mode><mode>): Likewise.
21222         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21223         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21225 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21227         * config/aarch64/aarch64-builtins.c
21228         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21230 2013-01-08  Martin Jambor  <mjambor@suse.cz>
21232         PR debug/55579
21233         * tree-sra.c (analyze_access_subtree): Return true also after
21234         potentially creating a debug-only replacement.
21236 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21238         PR middle-end/55890
21239         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21241         PR tree-optimization/54120
21242         * tree-vrp.c (range_fits_type_p): Don't allow
21243         src_precision < precision from signed vr to unsigned_p
21244         if vr->min or vr->max is negative.
21245         (simplify_float_conversion_using_ranges): Test can_float_p
21246         against CODE_FOR_nothing.
21248 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21249             Richard Biener  <rguenther@suse.de>
21251         PR middle-end/55851
21252         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21253         types instead of just INTEGER_TYPE types.
21255 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
21257         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21258         TF_SIZE): Define.
21260 2013-01-07  Steve Ellcey  <sellcey@mips.com>
21262         PR target/42661
21263         * config/mips/mips.opt: Change mad to mmad to match documentation.
21265 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21267         PR target/55897
21268         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21269         .progmemx.data now.
21271 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21273         PR target/55897
21274         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21275         (avr_addrspace_t): Add .section_name field.
21276         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21277         array size.
21278         (avr_addrspace): Same.  Initialize .section_name.  Remove last
21279         NULL entry.  Put __memx into .progmemx.data.
21280         (progmem_section_prefix): Remove.
21281         (avr_asm_init_sections): No need to initialize progmem_section.
21282         (avr_asm_named_section): Use avr_addrspace[].section_name to get
21283         section name prefix.
21284         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
21285         retrieve the progmem section.
21286         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21287         boundary to run over avr_addrspace[].
21288         (avr_register_target_pragmas): Ditto.
21290 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
21292         * varasm.c (output_constant_def_contents): For asan_protect_global
21293         protected strings, adjust DECL_ALIGN if needed, before testing for
21294         anchored symbols.
21295         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21296         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
21297         normal decls.
21298         (output_object_block): For asan protected decls, emit asan padding
21299         after their contents.
21300         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21301         (asan_finish_file): Test it here instead.
21303 2013-01-07  Nick Clifton  <nickc@redhat.com>
21304             Matthias Klose  <doko@debian.org>
21305             Doug Kwan  <dougkwan@google.com>
21306             H.J. Lu  <hongjiu.lu@intel.com>
21308         PR driver/55470
21309         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21311         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21313         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21315         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21317         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21319 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21321         PR target/54461
21322         * doc/install.texi (Cross-Compiler-Specific Options): Document
21323         --with-avrlibc.
21325 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
21327         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21328         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21329         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21330         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21331         vqmovun_high_s64): Fix source operand number and update copyright.
21333 2013-01-07  Richard Biener  <rguenther@suse.de>
21335         PR middle-end/55890
21336         * gimple.h (gimple_call_builtin_p): New overload.
21337         * gimple.c (validate_call): New function.
21338         (gimple_call_builtin_p): Likewise.
21339         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21340         Use gimple_call_builtin_p.
21341         (find_func_clobbers): Likewise.
21342         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21343         (strlen_optimize_stmt): Likewise.
21345 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21347         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21348         (vld1q_dup_*): Likewise.
21349         (vld1_*): Likewise.
21350         (vld1q_*): Likewise.
21351         (vld1_lane_*): Likewise.
21352         (vld1q_lane_*): Likewise.
21354 2013-01-07  Richard Biener  <rguenther@suse.de>
21356         * lto-streamer.h (LTO_minor_version): Bump to 2.
21358 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21360         * config/aarch64/aarch64-protos.h
21361         (aarch64_const_double_zero_rtx_p): Rename to...
21362         (aarch64_float_const_zero_rtx_p): ...this.
21363         (aarch64_float_const_representable_p): New.
21364         (aarch64_output_simd_mov_immediate): Likewise.
21365         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21366         move immediate case.
21367         * config/aarch64/aarch64.c
21368         (aarch64_const_double_zero_rtx_p): Rename to...
21369         (aarch64_float_const_zero_rtx_p): ...this.
21370         (aarch64_print_operand): Allow printing of new constants.
21371         (aarch64_valid_floating_const): New.
21372         (aarch64_legitimate_constant_p): Check for valid floating-point
21373         constants.
21374         (aarch64_simd_valid_immediate): Likewise.
21375         (aarch64_vect_float_const_representable_p): New.
21376         (aarch64_float_const_representable_p): Likewise.
21377         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21378         (aarch64_output_simd_mov_immediate): New.
21379         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21380         (*movdf_aarch64): Likewise.
21381         * config/aarch64/constraints.md (Ufc): New.
21382         (Y): call aarch64_float_const_zero_rtx.
21383         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21385 2013-01-07  Richard Biener  <rguenther@suse.de>
21387         PR tree-optimization/55888
21388         PR tree-optimization/55862
21389         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21390         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21391         not if it is contained therein.
21393 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21395         * config/avr/t-avr: Typo.
21397 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21399         PR55243
21400         * config/avr/t-avr: Don't automatically rebuild
21401         $(srcdir)/config/avr/t-multilib
21402         $(srcdir)/config/avr/avr-tables.opt
21403         $(srcdir)/doc/avr-mmcu.texi
21404         (avr-mcus): New phony target to build them on request.
21405         (s-avr-mlib, s-avr-mmcu-texi): Remove.
21406         * avr/avr-mcus.def: Adjust comments.
21408 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
21410         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21412 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21414         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21416 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21418         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21420 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
21422         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21423         to generate profiling.
21424         * config/rs6000/aix64.h (LIB_SPEC): Same.
21426 2013-01-04  Andrew Pinski  <apinski@cavium.com>
21428         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21429         New function.
21430         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21432 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
21434         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21435         unconditionally.
21436         (ix86_expand_move): Ditto.
21437         (ix86_zero_extend_to_Pmode): Ditto.
21438         (ix86_expand_call): Ditto.
21439         (ix86_expand_special_args_builtin): Ditto.
21440         (ix86_expand_builtin): Ditto.
21442 2013-01-04  Richard Biener  <rguenther@suse.de>
21444         PR tree-optimization/55862
21445         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21446         translating them through PHI nodes.
21448 2013-01-04  Martin Jambor  <mjambor@suse.cz>
21450         PR tree-optimization/55755
21451         * tree-sra.c (sra_modify_assign): Do not check that an access has no
21452         children when trying to avoid producing a VIEW_CONVERT_EXPR.
21454 2013-01-04  Marek Polacek  <polacek@redhat.com>
21456         PR middle-end/55859
21457         * opts.c (default_options_optimization): Clarify error message.
21459 2013-01-04  Richard Biener  <rguenther@suse.de>
21461         PR middle-end/55863
21462         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21463         reassociation.
21465 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21467         PR target/53789
21468         * config/pa/pa.md (movsi): Revert previous change.
21469         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21470         references.
21472 2013-01-03  Richard Henderson  <rth@redhat.com>
21474         * config/i386/i386.c (ix86_expand_move): Always assign to op1
21475         after eliminating TLS symbols.
21477 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
21479         PR bootstrap/50167
21480         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21481         * graphite-poly.c (debug_gmp_value): Likewise.
21483 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
21485         PR target/55712
21486         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21487         selected code model, define __code_mode_small__, __code_model_medium__,
21488         __code_model_large__, __code_model_32__ or __code_model_kernel__.
21489         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21490         xchg temporary register with %k.  Declare temporary register as
21491         early clobbered.
21492         [__x86_64__]: For medium and large code models, preserve %rbx register.
21494 2013-01-03  Richard Biener  <rguenther@suse.de>
21496         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21497         (dump_subscript): Adjust.
21498         (finalize_ddr_dependent): Do not dump redundant info.
21499         (analyze_siv_subscript): Adjust.
21500         (subscript_dependence_tester): Likewise.
21501         (compute_affine_dependence): Likewise.
21503 2013-01-03  Richard Biener  <rguenther@suse.de>
21505         Revert
21506         2013-01-03  Richard Biener  <rguenther@suse.de>
21508         PR tree-optimization/55857
21509         * tree-vect-stmts.c (vectorizable_load): Do not setup
21510         re-alignment for invariant loads.
21512         2013-01-02  Richard Biener  <rguenther@suse.de>
21514         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21515         invariant load do not generate a vector load from the scalar location.
21517 2013-01-03  Richard Biener  <rguenther@suse.de>
21519         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21520         for not vectorizing.
21521         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21522         not build INDIRECT_REFs, call get_name once only.
21523         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
21524         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21526 2013-01-03  Richard Biener  <rguenther@suse.de>
21528         PR tree-optimization/55857
21529         * tree-vect-stmts.c (vectorizable_load): Do not setup
21530         re-alignment for invariant loads.
21532 2013-01-03  Richard Biener  <rguenther@suse.de>
21534         PR lto/55848
21535         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
21536         prefer a built-in decl.
21538 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21540         * gcc.c (process_command): Update copyright notice dates.
21541         * gcov.c (print_version): Likewise.
21542         * gcov-dump.c (print_version): Likewise.
21544         PR rtl-optimization/55838
21545         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
21546         iv0.step, iv1.step and step.
21548 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21549             Marc Glisse  <marc.glisse@inria.fr>
21551         PR tree-optimization/55832
21552         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
21553         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
21554         integer_{one,zero}_node.
21556 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21558         PR debug/54402
21559         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
21560         * var-tracking.c (reverse_op): Don't add reverse ops to
21561         VALUEs that have already
21562         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
21564 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
21566         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
21568 2013-01-02  Teresa Johnson  <tejohnson@google.com>
21570         * dumpfile.c (dump_loc): Print filename with location.
21571         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
21572         new location_t parameter to emit complete unroll message with
21573         new dump framework.
21574         (canonicalize_loop_induction_variables): Compute loops location
21575         and pass to try_unroll_loop_completely.
21576         * loop-unroll.c (report_unroll_peel): New function.
21577         (peel_loops_completely): Use new dump format with location
21578         for main dumpfile message, and invoke report_unroll_peel on success.
21579         (decide_unrolling_and_peeling): Ditto.
21580         (decide_peel_once_rolling): Remove old dumpfile message subsumed
21581         by report_unroll_peel.
21582         (decide_peel_completely): Ditto.
21583         (decide_unroll_constant_iterations): Ditto.
21584         (decide_unroll_runtime_iterations): Ditto.
21585         (decide_peel_simple): Ditto.
21586         (decide_unroll_stupid): Ditto.
21587         * cfgloop.c (get_loop_location): New function.
21588         * cfgloop.h (get_loop_location): Declare.
21590 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21592         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
21593         NULL.
21595 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21597         PR middle-end/55198
21598         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
21599         BLKmode objects when EXPAND_MEMORY is specified.
21601 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21603         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
21604         in loop predicate.
21605         (fold_builtin_cpu): Do not share cpu model decls across statements.
21607 2013-01-02  Jason Merrill  <jason@redhat.com>
21609         PR c++/55804
21610         * tree.c (build_array_type_1): Revert earlier change.
21612 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
21614         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
21615         "cortex-a57".
21616         * config/aarch64/aarch64-tune.md: Re-generate.
21618 2013-01-02  Richard Biener  <rguenther@suse.de>
21620         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21621         invariant load do not generate a vector load from the scalar location.
21623 2013-01-02  Richard Biener  <rguenther@suse.de>
21625         PR bootstrap/55784
21626         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
21627         * configure: Regenerate.
21629 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21631         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
21632         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
21633         (expand_builtin_int_roundingfn_2): Keep the original target around
21634         for the fallback case.
21636 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21638         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
21639         to be clear for sign changes.
21641 2013-01-01  Jan Hubicka  <jh@suse.cz>
21643         * ipa-inline-analysis.c: Fix formatting.
21645 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
21647         PR tree-optimization/55831
21648         * tree-vect-loop.c (get_initial_def_for_induction): Use
21649         gsi_after_labels instead of gsi_start_bb.
21651 Copyright (C) 2013 Free Software Foundation, Inc.
21653 Copying and distribution of this file, with or without modification,
21654 are permitted in any medium without royalty provided the copyright
21655 notice and this notice are preserved.