* cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
[official-gcc.git] / gcc / ChangeLog
blobde4bb3d0aa2e2f49546b34844315a828e57f71ae
1 2013-09-12  DJ Delorie  <dj@redhat.com>
3         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
4         PSImode.
6 2013-09-12  Martin Jambor  <mjambor@suse.cz>
8         PR ipa/58389
9         * ipa-prop.c (remove_described_reference): Give up if the edge in the
10         reference descriptor is NULL.
11         (ipa_edge_removal_hook): If owning a reference descriptor, set its
12         edge to NULL.
14 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
16         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
17         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
18         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
19         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
20         make_temp_ssa_name): move to tree-ssanames.h
21         * tree-ssa-alias.h: Move prototype.
22         * tree-ssa.h: Include tree-ssanames.h.
23         * tree-ssanames.c (FREE_SSANAMES): Move to here.
24         * tree-ssanames.h: New.  Move items from tree-flow*.h
25         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
27 2013-09-12  Richard Biener  <rguenther@suse.de>
29         PR tree-optimization/58404
30         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
31         propagate non-invariant addresses into dereferences wrapped
32         in component references.
34 2013-09-12  Richard Biener  <rguenther@suse.de>
36         PR tree-optimization/58402
37         * passes.def: Move pass_late_warn_uninitialized later.
39 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
41         * tree-ssa.h: New.  Move content from tree-flow.h and
42         tree-flow-inline.h.
43         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
44         Move prototypes belonging to tree-ssa.c.
45         * tree-flow-inline.h (redirect_edge_var_map_def, 
46         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
47         tree-ssa.h.
48         * gimple.h: Adjust prototypes.
49         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move to...
50         * gimple.c (useless_type_conversion_p, types_compatible_p):  Here.
51         * tree.h: Move prototype to tree-ssa.h.
52         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
53         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
54         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
55         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
56         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
57         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
58         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
59         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
60         graphite-blocking.c, graphite-clast-to-gimple.c, graphite-dependences.c,
61         graphite-interchange.c, graphite-optimize-isl.c, graphite-poly.c,
62         graphite-scop-detection.c, graphite-sese-to-poly.c, graphite.c,
63         ipa-cp.c, ipa-inline-analysis.c, ipa-inline-transform.c, ipa-inline.c,
64         ipa-prop.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa-utils.c,
65         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
66         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
67         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
68         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
69         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
70         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
71         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
72         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
73         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
74         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
75         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
76         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
77         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
78         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
79         value-prof.c, var-tracking.c,
80         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
82 2013-09-12  Richard Biener  <rguenther@suse.de>
84         PR tree-optimization/58396
85         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
86         (build_rdg): Take a loop-nest parameter, fix memleaks.
87         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
89 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
91         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
92         for SLM.
94 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
96         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
97         The FMA instruction names should have a 'v' prefix.
99 2013-09-12  Richard Biener  <rguenther@suse.de>
101         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
102         (dot_rdg): Use popen instead of system in optional code.
103         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
104         (already_processed_vertex_p): Adjust.
105         (has_anti_or_output_dependence, predecessor_has_mem_write,
106         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
107         rdg_flag_uses): Remove.
108         (rdg_flag_vertex): Simplify.
109         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
110         remove recursion.
111         (build_rdg_partition_for_component): Process the first vertex
112         of a component only.
113         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
115 2013-09-12  Alan Modra  <amodra@gmail.com>
117         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
119 2013-09-11  DJ Delorie  <dj@redhat.com>
120             Nick Clifton  <nickc@redhat.com>
122         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
123         (rl78_stack_based_mem): New.
124         * config/rl78/constraints.md (Iv08): New.
125         (Iv16): New.
126         (Iv24): New.
127         (Is09): New.
128         (Is17): New.
129         (Is25): New.
130         (ISsi): New.
131         (IShi): New.
132         (ISqi): New.
133         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
134         (movhi): Likewise.
135         (movsi): Change from expand to insn-and-split.
136         (ashrsi3): Clobber AX.
137         (lshrsi3): New.
138         (ashlsi3): New.
139         (cbranchsi4): New.
140         * config/rl78/rl78.md (CC_REG): Fix.
141         (addsi3): Allow memory and immediate operands.
142         (addsi3_internal): Split into...
143         (addsi3_internal_virt): ...new, and ...
144         (addsi3_internal_real): ...new.
145         (subsi): New.
146         (subsi3_internal_virt): New.
147         (subsi3_internal_real): New.
148         (mulsi3): Add memory operand.
149         (mulsi3_rl78): Likewise.
150         (mulsi3_g13): Likewise.
151         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
152         (cbranchqi4_real): Add more constraint options.
153         (cbranchhi4_real): Expand pattern.
154         (cbranchhi4_real_signed): New.
155         (cbranchhi4_real_inverted): New.
156         (cbranchsi4_real_lt): New.
157         (cbranchsi4_real_ge): New.
158         (cbranchsi4_real_signed): New.
159         (cbranchsi4_real): New.
160         (peephole2): New.
161         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
162         constant shifts.
163         (lshrsi3_virt): Likewise.
164         (ashlsi3_virt): Likewise.
165         (cbranchqi4_virt_signed): New.
166         (cbranchhi4_virt_signed): New.
167         (cbranchsi4_virt): New.
168         * config/rl78/rl78.c: Whitespace fixes throughout.
169         (move_elim_pass): New.
170         (pass_data_rl78_move_elim): New.
171         (pass_rl78_move_elim): New.
172         (make_pass_rl78_move_elim): New.
173         (rl78_devirt_info): Run devirt earlier.
174         (rl78_move_elim_info): New.
175         (rl78_asm_file_start): Register it.
176         (rl78_split_movsi): New.
177         (rl78_as_legitimate_address): Allow virtual base registers when
178         appropriate.
179         (rl78_addr_space_convert): Remove spurious debug stuff.
180         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
181         (rl78_print_operand): More cases for not printing '#'.
182         (rl78_expand_compare): Remove most of the logic.
183         (content_memory): New.
184         (clear_content_memory): New.
185         (get_content_index): New.
186         (get_content_name): New.
187         (display_content_memory): New.
188         (update_content): New.
189         (record_content): New.
190         (already_contains): New.
191         (insn_ok_now): Re-recog insns with virtual registers.
192         (add_postponed_content_update): New.
193         (process_postponed_content_update): New.
194         (gen_and_emit_move): New.
195         (transcode_memory_rtx): Record new location content.
196         Use gen_and_emit_move.
197         (force_into_acc): New.
198         (move_to_acc): Use gen_and_emit_move.
199         (move_from_acc): Likewise.
200         (move_acc_to_reg): Likewise.
201         (move_to_x): Likewise.
202         (move_to_hl): Likewise.
203         (move_to_de): Likewise.
204         (rl78_alloc_physical_registers_op1): Record location content.
205         (has_constraint): New.
206         (rl78_alloc_physical_registers_op2): Record location content.
207         Optimize use of HL.
208         (rl78_alloc_physical_registers_ro1): Likewise.
209         (rl78_alloc_physical_registers_cmp): Likewise.
210         (rl78_alloc_physical_registers_umul): Likewise.
211         (rl78_alloc_address_registers_macax): New.
212         (rl78_alloc_physical_registers): Initialize and set location
213         content memory as needed.
214         (rl78_reorg): Make sure split2 is called.
215         (rl78_rtx_costs): New.
217 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
219         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
220         for (not (neg ...)) and (neg (not ...)) cases.
222 2013-09-11  Richard Biener  <rguenther@suse.de>
224         PR middle-end/58377
225         * passes.def: Split critical edges before late uninit warning passes.
226         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
228 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
230         PR tree-optimization/58385
231         * fold-const.c (build_range_check): If both low and high are NULL,
232         use omit_one_operand_loc to preserve exp side-effects.
234 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
236         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
238 2013-09-11  Richard Biener  <rguenther@suse.de>
240         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
241         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
242         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
243         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
244         stmts_from_loop, known_dependences_p, build_empty_rdg,
245         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
246         * tree-loop-distribution.c: ... here.
247         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
248         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
249         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
250         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
251         * tree-loop-distribution.c: ... here.
252         * tree-loop-distribution.c: Include gimple-pretty-print.h.
253         (struct partition_s): Add loops member.
254         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
255         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
256         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
258 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
259             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
260             Sergey Lega  <sergey.s.lega@intel.com>
261             Anna Tikhonova  <anna.tikhonova@intel.com>
262             Ilya Tocar  <ilya.tocar@intel.com>
263             Andrey Turetskiy  <andrey.turetskiy@intel.com>
264             Ilya Verbin  <ilya.verbin@intel.com>
265             Kirill Yukhin  <kirill.yukhin@intel.com>
266             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
268         * config/i386/constraints.md (k): New.
269         (Yk): Ditto.
270         * config/i386/i386.c (const regclass_map): Add new mask registers.
271         (dbx_register_map): Ditto.
272         (dbx64_register_map): Ditto.
273         (svr4_dbx_register_map): Ditto.
274         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
275         disabled.
276         (ix86_preferred_reload_class): Disable constants for mask registers.
277         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
278         (ix86_hard_regno_mode_ok): Support new mask registers.
279         (x86_order_regs_for_local_alloc): Ditto.
280         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
281         (FIXED_REGISTERS): Add new mask registers.
282         (CALL_USED_REGISTERS): Ditto.
283         (REG_ALLOC_ORDER): Ditto.
284         (VALID_MASK_REG_MODE): New.
285         (FIRST_MASK_REG): Ditto.
286         (LAST_MASK_REG): Ditto.
287         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
288         (MAYBE_MASK_CLASS_P): New.
289         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
290         (REG_CLASS_CONTENTS): Ditto.
291         (MASK_REGNO_P): New.
292         (ANY_MASK_REG_P): Ditto.
293         (HI_REGISTER_NAMES): Add new mask registers.
294         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
295         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
296         mask registers.
297         (attribute "type"): Add mskmov, msklog.
298         (attribute "length_immediate"): Support them.
299         (attribute "memory"): Ditto.
300         (attribute "prefix_0f"): Ditto.
301         (*movhi_internal): Support new mask registers.
302         (*movqi_internal): Ditto.
303         (define_split): Split out clobber pattern is a logic
304         insn on mask registers.
305         (*k<logic><mode>): New.
306         (*andhi_1): Extend to support mask regs.
307         (*andqi_1): Extend to support mask regs.
308         (kandn<mode>): New.
309         (define_split): Split and-not to and and not if operands
310         are not mask regs.
311         (*<code><mode>_1): Separate HI mode to new pattern...
312         (*<code>hi_1): This.
313         (*<code>qi_1): Extend to support mask regs.
314         (kxnor<mode>): New.
315         (kortestzhi): Ditto.
316         (kortestchi): Ditto.
317         (kunpckhi): Ditto.
318         (*one_cmpl<mode>2_1): Remove HImode and handle it...
319         (*one_cmplhi2_1): ...Here, now with mask registers support.
320         (*one_cmplqi2_1): Support new mask registers.
321         (HI/QImode arithmetics splitter): Don't split if mask registers
322         are used.
323         (HI/QImode not splitter): Ditto.
324         * config/i386/predicated.md (mask_reg_operand): New.
325         (general_reg_operand): Ditto.
327 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
329         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
330         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
332 2013-09-10  Jeff Law  <law@redhat.com>
334         PR tree-optimization/58380
335         * tree-ssa-threadupdate.c (thread_block): Recognize another case
336         of threading through a buried loop header.
338         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
339         return value for single successor case.
341 2013-09-10  Jan Hubicka  <jh@suse.cz>
343         * ipa-devirt.c (ipa_devirt): Enable with LTO.
345 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
347         PR target/58361
348         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
349         support conditional execution.
350         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
352 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
354         * lra.c (lra): Clear lra_optional_reload_pseudos before every
355         constraint pass.
356         * lra-constraints.c (curr_insn_transform): Switch on optional
357         reloads.  Check destination too to check move insn.
358         (undo_optional_reloads): Add check that the original peudo did not
359         changed its allocation and the optional reload was inherited on
360         last inheritance pass.  Break loop after deciding to keep optional
361         reload.
362         (lra_undo_inheritance): Add check that inherited pseudo still in
363         memory.
365 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
367         * config/aarch64/aarch64.md (generic_sched): New.
368         * config/aarch64/aarch64-generic.md (load): Make conditional
369         on generic_sched attribute.
370         (nonload): Likewise.
372 2013-09-10  Jan Hubicka  <jh@suse.cz>
374         * lto-cgraph.c: Include ipa-utils.h.
375         (compute_ltrans_boundary): Also add possible targets into the boundary.
377 2013-09-10  Jan Hubicka  <jh@suse.cz>
379         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
380         VAR_DECL of vtable rather than full expression.
382 2013-09-10  Jan Hubicka  <jh@suse.cz>
383             Paolo Carlini  <paolo.carlini@oracle.com>
385         * cgraphunit.c (analyze_functions): Save input_location, set it
386         to UNKNOWN_LOCATION and restore it at the end.
388 2013-09-10  Martin Jambor  <mjambor@suse.cz>
390         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
391         represented by a thunk.
393 2013-09-10  Jeff Law  <law@redhat.com>
395         PR tree-optimization/58343
396         * tree-ssa-threadupdate.c (thread_block): Identify and disable
397         jump threading requests through loop headers buried in the middle
398         of a jump threading path.
400         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
401         in return value/type.
403 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
405         PR rtl-optimization/58365
406         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
407         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
408         it differs.
410 2013-09-10  Richard Biener  <rguenther@suse.de>
412         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
413         * tree-data-ref.c (create_rdg_vertices): Collect all data
414         references, signal failure to the caller, use data-ref API.
415         (build_rdg): Compute data references only once.  Maintain lifetime
416         of data references and data dependences from within RDG.
417         (free_rdg): Free dependence relations.
418         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
419         inventing extra dependences.
420         (distribute_loop): Update for RDG API changes.
422 2013-09-10  Kai Tietz  <ktietz@redhat.com>
424         * doc/invoke.texi (fms-extensions): Document changed
425         behavior for ms-abi targets.
426         * config/i386/i386.c (ix86_option_override_internal):
427         Set default value of option -fms-extension for ms-abi targets.
429 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
431         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
433 2013-09-10  Alan Modra  <amodra@gmail.com>
435         PR target/58330
436         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
438 2013-09-10  Alan Modra  <amodra@gmail.com>
440         * config/rs6000/predicates.md (add_cint_operand): New.
441         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
442         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
443         using add_cint_operand.
444         (largetoc_high_plus_aix): Likewise.
446 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
448         PR tree-optimization/58364
449         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
450         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
451         the current range can't be an unconditional true or false.
453 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
455         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
457 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
459         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
461 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
463         PR c++/43452
464         * doc/invoke.texi (-Wdelete-incomplete): Document it.
466 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
468         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
469         NO_REGS for immediate that can't be moved directly into FP_REGS.
471 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
473         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
474         comparison with negated operand.
475         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
476         RTL form.
478 2013-09-09  Richard Biener  <rguenther@suse.de>
480         PR middle-end/58326
481         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
482         of a subloop record all its block as affecting loop-closed SSA form.
484 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
486         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
487         of an rtx/bitpos pair.
488         (store_fixed_bit_field): Update accordingly.
490 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
492         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
493         GEN_INT.
494         * builtins.c (expand_errno_check): Likewise.
495         * dwarf2cfi.c (init_return_column_size): Likewise.
496         * except.c (sjlj_mark_call_sites): Likewise.
497         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
498         * lra-constraints.c (emit_inc): Likewise.
499         * ree.c (combine_set_extension): Likewise.
500         * regmove.c (fixup_match_2): Likewise.
501         * reload1.c (inc_for_reload): Likewise.
503 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
505         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
506         (force_to_mode, simplify_shift_const_1, simplify_comparison):
507         Use gen_int_mode with the mode of the associated simplify_* call.
508         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
509         * expmed.c (expand_shift_1): Likewise.
510         * function.c (instantiate_virtual_regs_in_insn): Likewise.
511         * loop-iv.c (iv_number_of_iterations): Likewise.
512         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
513         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
515 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
517         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
518         of the associated expand_* call.
519         (asan_emit_stack_protection): Likewise.
520         * builtins.c (round_trampoline_addr): Likewise.
521         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
522         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
523         (emit_store_flag): Likewise.
524         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
525         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
526         Likewise.
527         * function.c (instantiate_virtual_regs_in_insn): Likewise.
528         * ifcvt.c (noce_try_store_flag_constants): Likewise.
529         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
530         * modulo-sched.c (generate_prolog_epilog): Likewise.
531         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
532         (expand_ctz, expand_ffs, expand_unop): Likewise.
534 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
536         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
537         of the associated gen_rtx_* call.
538         * caller-save.c (init_caller_save): Likewise.
539         * combine.c (find_split_point, make_extraction): Likewise.
540         (make_compound_operation): Likewise.
541         * dwarf2out.c (mem_loc_descriptor): Likewise.
542         * explow.c (plus_constant, probe_stack_range): Likewise.
543         * expmed.c (expand_mult_const): Likewise.
544         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
545         * reload1.c (init_reload): Likewise.
546         * valtrack.c (cleanup_auto_inc_dec): Likewise.
547         * var-tracking.c (adjust_mems): Likewise.
548         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
549         rather than gen_rtx_fmt_ee.
551 2013-09-09  Jan Hubicka  <jh@suse.cz>
553         PR middle-end/58294
554         * value-prof.c (gimple_ic): Copy also abnormal edges.
556 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
558         * asan.c (asan_shadow_cst): Use gen_int_mode.
560 2013-09-08  Jan Hubicka  <jh@suse.cz>
562         * ipa-profile.c: Add toplevel comment.
563         (ipa_propagate_frequency_1): Be more conservative when profile is read.
564         (contains_hot_call_p): New function.
565         (ipa_propagate_frequency): Set frequencies based on counts when
566         profile is read.
567         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
568         profile; do not tamper with profile after inlining if it is read.
570 2013-09-08  Jan Hubicka  <jh@suse.cz>
572         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
573         speculative edges.
575 2013-09-08  Jan Hubicka  <jh@suse.cz>
577         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
578         summary generation.
580 2013-09-08  Jeff Law  <law@redhat.com>
582         PR bootstrap/58340
583         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
584         of 'found'.
586 2013-09-08  Andi Kleen  <ak@linux.intel.com>
588         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
590 2013-09-08  Jan Hubicka  <jh@suse.cz>
592         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
594 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
596         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
597         for non-debug insns.
598         * lra.c (new_insn_reg): Take the containing insn as a parameter.
599         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
600         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
601         accordingly.
603 2013-09-08  Jan Hubicka  <jh@suse.cz>
605         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
606         targets and devirtualize to BUILT_IN_UNREACHABLE.
607         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
608         * ipa.c (walk_polymorphic_call_targets): New function.
609         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
610         functions; use the new timevar.
611         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
612         was removed from the program.
613         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
614         not consider it in the walk when its vtable is dead.
615         (possible_polymorphic_call_targets_1): Pass anonymous flag to
616         record_binfo.
617         (devirt_variable_node_removal_hook): New function.
618         (possible_polymorphic_call_targets): Also register
619         devirt_variable_node_removal_hook.
620         (ipa_devirt): Do not do non-speculative devirtualization.
621         (gate_ipa_devirt): One execute if devirtualizing speculatively.
623 2013-09-08  Jan Hubicka  <jh@suse.cz>
625         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
626         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
627         varpool_remove_variable_insertion_hook): Declare.
628         * varpool.c (varpool_node_hook_list): New structure.
629         (first_varpool_node_removal_hook,
630         first_varpool_variable_insertion_hook): New variables.
631         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
632         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
633         varpool_remove_variable_insertion_hook,
634         varpool_call_variable_insertion_hooks): New functions.
635         (varpool_remove_node): Use it.
637 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
639         PR c++/54941
640         * diagnostic.c (diagnostic_build_prefix): When s.file is
641         "<built-in>" don't output line and column numbers.
643 2013-09-06  Jan Hubicka  <jh@suse.cz>
645         * cgraphunit.c (expand_thunk): Get body before touching arguments.
646         * lto-streamer-out.c: Stream thunks, too.
647         * lto-streamer-in.c (input_function): Pop cfun here
648         (lto_read_body): Instead of here.
650 2013-09-06  Caroline Tice  <cmtice@google.com>
652         * doc/install.texi: Add documentation for the --enable-vtable-verify
653         and the --disable-libvtv configure options.
655 2013-09-06  Jeff Law  <law@redhat.com>
657         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
658         edge implied equivalences into successor phis.
660 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
662         * resource.c (mark_referenced_resources): Handle COND_EXEC.
664 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
666         * resource.c (mark_target_live_regs): Compute resources taking
667         into account if a call is predicated or not.
669 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
671         * toplev.c (output_stack_usage): Be prepared for suffixes created by
672         the compiler in the function names.
674 2013-09-06  Jan Hubicka  <jh@suse.cz>
676         PR middle-end/58094
677         * ipa-inline.c (has_caller_p): New function.
678         (want_inline_function_to_all_callers_p): Use it.
679         (sum_callers, inline_to_all_callers): Break out from ...
680         (ipa_inline): ... here.
682 2013-09-06  Jan Hubicka  <jh@suse.cz>
684         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
685         only when AVX is enabled.
687 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
689         * config/aarch64/aarch64.md
690         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
691         is fpsimd_<load/store>2.
692         (load_pair<mode>): Likewise.
693         (store_pair<mode>): Likewise.
695 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
697         * config/arm/types.md (type): Add "mrs" type.
698         * config/aarch64/aarch64.md
699         (aarch64_load_tp_hard): Make type "mrs".
700         * config/arm/arm.md
701         (load_tp_hard): Make type "mrs".
702         * config/arm/cortex-a15.md: Update with new attributes.
703         * config/arm/cortex-a5.md: Update with new attributes.
704         * config/arm/cortex-a53.md: Update with new attributes.
705         * config/arm/cortex-a7.md: Update with new attributes.
706         * config/arm/cortex-a8.md: Update with new attributes.
707         * config/arm/cortex-a9.md: Update with new attributes.
708         * config/arm/cortex-m4.md: Update with new attributes.
709         * config/arm/cortex-r4.md: Update with new attributes.
710         * config/arm/fa526.md: Update with new attributes.
711         * config/arm/fa606te.md: Update with new attributes.
712         * config/arm/fa626te.md: Update with new attributes.
713         * config/arm/fa726te.md: Update with new attributes.
715 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
717         * config/aarch64/aarch64.md
718         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
719         (*movtf_aarch64): Likewise.
720         * config/arm/arm.md
721         (thumb1_movdi_insn): Use "multiple" for type where more than one
722         instruction is used for a move.
723         (*arm32_movhf): Likewise.
724         (*thumb_movdf_insn): Likewise.
726 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
728         * config/arm/types.md (type): Rename fcpys to fmov.
729         * config/arm/vfp.md
730         (*arm_movsi_vfp): Rename type fcpys as fmov.
731         (*thumb2_movsi_vfp): Likewise
732         (*movhf_vfp_neon): Likewise
733         (*movhf_vfp): Likewise
734         (*movsf_vfp): Likewise
735         (*thumb2_movsf_vfp): Likewise
736         (*movsfcc_vfp): Likewise
737         (*thumb2_movsfcc_vfp): Likewise
738         * config/aarch64/aarch64-simd.md
739         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
740         * config/aarch64/aarch64.md
741         (*movsi_aarch64): Replace type mov_reg with fmovs.
742         (*movdi_aarch64): Likewise
743         (*movsf_aarch64): Likewise
744         (*movdf_aarch64): Likewise
745         * config/arm/arm.c
746         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
747         * config/arm/iwmmxt.md
748         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
749         * config/arm/arm1020e.md: Update with new attributes.
750         * config/arm/cortex-a15-neon.md: Update with new attributes.
751         * config/arm/cortex-a5.md: Update with new attributes.
752         * config/arm/cortex-a53.md: Update with new attributes.
753         * config/arm/cortex-a7.md: Update with new attributes.
754         * config/arm/cortex-a8-neon.md: Update with new attributes.
755         * config/arm/cortex-a9.md: Update with new attributes.
756         * config/arm/cortex-m4-fpu.md: Update with new attributes.
757         * config/arm/cortex-r4f.md: Update with new attributes.
758         * config/arm/marvell-pj4.md: Update with new attributes.
759         * config/arm/vfp11.md: Update with new attributes.
761 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
763         * config/aarch64/aarch64.md
764         (*madd<mode>): Fix type attribute.
765         (*maddsi_uxtw): Likewise.
766         (*msub<mode>): Likewise.
767         (*msubsi_uxtw): Likewise.
768         (<su_optab>maddsidi4): Likewise.
769         (<su_optab>msubsidi4): Likewise.
771 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
773         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
774         * config/arm/arm.md (core_cycles): Remove fdiv.
775         * config/arm/vfp.md:
776         (*sqrtsf2_vfp): Update for attribute changes.
777         (*sqrtdf2_vfp): Likewise.
778         * config/aarch64/aarch64.md:
779         (sqrt<mode>2): Update for attribute changes.
780         * config/arm/arm1020e.md: Update with new attributes.
781         * config/arm/cortex-a15-neon.md: Update with new attributes.
782         * config/arm/cortex-a5.md: Update with new attributes.
783         * config/arm/cortex-a53.md: Update with new attributes.
784         * config/arm/cortex-a7.md: Update with new attributes.
785         * config/arm/cortex-a8-neon.md: Update with new attributes.
786         * config/arm/cortex-a9.md: Update with new attributes.
787         * config/arm/cortex-m4-fpu.md: Update with new attributes.
788         * config/arm/cortex-r4f.md: Update with new attributes.
789         * config/arm/marvell-pj4.md: Update with new attributes.
790         * config/arm/vfp11.md: Update with new attributes.
792 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
794         * config/arm/types.md
795         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
796         * config/aarch64/aarch64.md
797         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
798         new attributes.
799         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
800         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
801         (float<GPI:mode><GPF:mode>2): Likewise.
802         * config/arm/vfp.md
803         (*truncsisf2_vfp): Update with new attributes.
804         (*truncsidf2_vfp): Likewise.
805         (fixuns_truncsfsi2): Likewise.
806         (fixuns_truncdfsi2): Likewise.
807         (*floatsisf2_vfp): Likewise.
808         (*floatsidf2_vfp): Likewise.
809         (floatunssisf2): Likewise.
810         (floatunssidf2): Likewise.
811         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
812         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
813         * config/arm/arm1020e.md: Update with new attributes.
814         * config/arm/cortex-a15-neon.md: Update with new attributes.
815         * config/arm/cortex-a5.md: Update with new attributes.
816         * config/arm/cortex-a53.md: Update with new attributes.
817         * config/arm/cortex-a7.md: Update with new attributes.
818         * config/arm/cortex-a8-neon.md: Update with new attributes.
819         * config/arm/cortex-a9.md: Update with new attributes.
820         * config/arm/cortex-m4-fpu.md: Update with new attributes.
821         * config/arm/cortex-r4f.md: Update with new attributes.
822         * config/arm/marvell-pj4.md: Update with new attributes.
823         * config/arm/vfp11.md: Update with new attributes.
825 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
827         * config/aarch64/arm_neon.h
828         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
829         (vqtbx<1,2,3,4><q>_s8): Likewise.
831 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
833         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
834         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
835         (add<mode>3): Add type attribute.
836         (add<mode>3): Likewise.
837         (usadd<mode>3): Likewise.
838         (ssadd<mode>3): Likewise.
839         (sub<mode>3): Likewise.
840         (sub<mode>3): Likewise.
841         (ussub<mode>3): Likewise.
842         (sssub<mode>3): Likewise.
843         (ssmulsa3): Likewise.
844         (usmulusa3): Likewise.
845         (arm_usatsihi): Likewise.
846         * config/arm/vfp.md
847         (*movdi_vfp): Add types for all instructions.
848         (*movdi_vfp_cortexa8): Likewise.
849         (*movhf_vfp_neon): Likewise.
850         (*movhf_vfp): Likewise.
851         (*movdf_vfp): Likewise.
852         (*thumb2_movdf_vfp): Likewise.
853         (*thumb2_movdfcc_vfp): Likewise.
854         * config/arm/arm.md: Add type attribute to all insn patterns.
855         (*thumb1_adddi3): Add type attribute.
856         (*arm_adddi3): Likewise.
857         (*adddi_sesidi_di): Likewise.
858         (*adddi_zesidi_di): Likewise.
859         (*thumb1_addsi3): Likewise.
860         (addsi3_compare0): Likewise.
861         (*addsi3_compare0_scratch): Likewise.
862         (*compare_negsi_si): Likewise.
863         (cmpsi2_addneg): Likewise.
864         (*addsi3_carryin_<optab>): Likewise.
865         (*addsi3_carryin_alt2_<optab>): Likewise.
866         (*addsi3_carryin_clobercc_<optab>): Likewise.
867         (*subsi3_carryin): Likewise.
868         (*subsi3_carryin_const): Likewise.
869         (*subsi3_carryin_compare): Likewise.
870         (*subsi3_carryin_compare_const): Likewise.
871         (*arm_subdi3): Likewise.
872         (*thumb_subdi3): Likewise.
873         (*subdi_di_zesidi): Likewise.
874         (*subdi_di_sesidi): Likewise.
875         (*subdi_zesidi_di): Likewise.
876         (*subdi_sesidi_di): Likewise.
877         (*subdi_zesidi_ze): Likewise.
878         (thumb1_subsi3_insn): Likewise.
879         (*arm_subsi3_insn): Likewise.
880         (*anddi3_insn): Likewise.
881         (*anddi_zesidi_di): Likewise.
882         (*anddi_sesdi_di): Likewise.
883         (*ne_zeroextracts): Likewise.
884         (*ne_zeroextracts): Likewise.
885         (*ite_ne_zeroextr): Likewise.
886         (*ite_ne_zeroextr): Likewise.
887         (*anddi_notdi_di): Likewise.
888         (*anddi_notzesidi): Likewise.
889         (*anddi_notsesidi): Likewise.
890         (andsi_notsi_si): Likewise.
891         (thumb1_bicsi3): Likewise.
892         (*iordi3_insn): Likewise.
893         (*iordi_zesidi_di): Likewise.
894         (*iordi_sesidi_di): Likewise.
895         (*thumb1_iorsi3_insn): Likewise.
896         (*xordi3_insn): Likewise.
897         (*xordi_zesidi_di): Likewise.
898         (*xordi_sesidi_di): Likewise.
899         (*arm_xorsi3): Likewise.
900         (*andsi_iorsi3_no): Likewise.
901         (*smax_0): Likewise.
902         (*smax_m1): Likewise.
903         (*arm_smax_insn): Likewise.
904         (*smin_0): Likewise.
905         (*arm_smin_insn): Likewise.
906         (*arm_umaxsi3): Likewise.
907         (*arm_uminsi3): Likewise.
908         (*minmax_arithsi): Likewise.
909         (*minmax_arithsi_): Likewise.
910         (*satsi_<SAT:code>): Likewise.
911         (arm_ashldi3_1bit): Likewise.
912         (arm_ashrdi3_1bit): Likewise.
913         (arm_lshrdi3_1bit): Likewise.
914         (*arm_negdi2): Likewise.
915         (*thumb1_negdi2): Likewise.
916         (*arm_negsi2): Likewise.
917         (*thumb1_negsi2): Likewise.
918         (*negdi_extendsid): Likewise.
919         (*negdi_zero_extend): Likewise.
920         (*arm_abssi2): Likewise.
921         (*thumb1_abssi2): Likewise.
922         (*arm_neg_abssi2): Likewise.
923         (*thumb1_neg_abss): Likewise.
924         (one_cmpldi2): Likewise.
925         (extend<mode>di2): Likewise.
926         (*compareqi_eq0): Likewise.
927         (*arm_extendhisi2addsi): Likewise.
928         (*arm_movdi): Likewise.
929         (*thumb1_movdi_insn): Likewise.
930         (*arm_movt): Likewise.
931         (*thumb1_movsi_insn): Likewise.
932         (pic_add_dot_plus_four): Likewise.
933         (pic_add_dot_plus_eight): Likewise.
934         (tls_load_dot_plus_eight): Likewise.
935         (*thumb1_movhi_insn): Likewise.
936         (*thumb1_movsf_insn): Likewise.
937         (*movdf_soft_insn): Likewise.
938         (*thumb_movdf_insn): Likewise.
939         (cbranchsi4_insn): Likewise.
940         (cbranchsi4_scratch): Likewise.
941         (*negated_cbranchsi4): Likewise.
942         (*tbit_cbranch): Likewise.
943         (*tlobits_cbranch): Likewise.
944         (*tstsi3_cbranch): Likewise.
945         (*cbranchne_decr1): Likewise.
946         (*addsi3_cbranch): Likewise.
947         (*addsi3_cbranch_scratch): Likewise.
948         (*arm_cmpdi_insn): Likewise.
949         (*arm_cmpdi_unsig): Likewise.
950         (*arm_cmpdi_zero): Likewise.
951         (*thumb_cmpdi_zero): Likewise.
952         (*deleted_compare): Likewise.
953         (*mov_scc): Likewise.
954         (*mov_negscc): Likewise.
955         (*mov_notscc): Likewise.
956         (*cstoresi_eq0_thumb1_insn): Likewise.
957         (cstoresi_nltu_thumb1): Likewise.
958         (cstoresi_ltu_thu): Likewise.
959         (thumb1_addsi3_addgeu): Likewise.
960         (*arm_jump): Likewise.
961         (*thumb_jump): Likewise.
962         (*check_arch2): Likewise.
963         (arm_casesi_internal): Likewise.
964         (thumb1_casesi_dispatch): Likewise.
965         (*arm_indirect_jump): Likewise.
966         (*thumb1_indirect_jump): Likewise.
967         (nop): Likewise.
968         (*and_scc): Likewise.
969         (*ior_scc): Likewise.
970         (*compare_scc): Likewise.
971         (*cond_move): Likewise.
972         (*cond_arith): Likewise.
973         (*cond_sub): Likewise.
974         (*cmp_ite0): Likewise.
975         (*cmp_ite1): Likewise.
976         (*cmp_and): Likewise.
977         (*cmp_ior): Likewise.
978         (*ior_scc_scc): Likewise.
979         (*ior_scc_scc_cmp): Likewise.
980         (*and_scc_scc): Likewise.
981         (*and_scc_scc_cmp): Likewise.
982         (*and_scc_scc_nod): Likewise.
983         (*negscc): Likewise.
984         (movcond_addsi): Likewise.
985         (movcond): Likewise.
986         (*ifcompare_plus_move): Likewise.
987         (*if_plus_move): Likewise.
988         (*ifcompare_move_plus): Likewise.
989         (*if_move_plus): Likewise.
990         (*ifcompare_arith_arith): Likewise.
991         (*if_arith_arith): Likewise.
992         (*ifcompare_arith_move): Likewise.
993         (*if_arith_move): Likewise.
994         (*ifcompare_move_arith): Likewise.
995         (*if_move_arith): Likewise.
996         (*ifcompare_move_not): Likewise.
997         (*if_move_not): Likewise.
998         (*ifcompare_not_move): Likewise.
999         (*if_not_move): Likewise.
1000         (*ifcompare_shift_move): Likewise.
1001         (*if_shift_move): Likewise.
1002         (*ifcompare_move_shift): Likewise.
1003         (*if_move_shift): Likewise.
1004         (*ifcompare_shift_shift): Likewise.
1005         (*ifcompare_not_arith): Likewise.
1006         (*ifcompare_arith_not): Likewise.
1007         (*if_arith_not): Likewise.
1008         (*ifcompare_neg_move): Likewise.
1009         (*if_neg_move): Likewise.
1010         (*ifcompare_move_neg): Likewise.
1011         (*if_move_neg): Likewise.
1012         (prologue_thumb1_interwork): Likewise.
1013         (*cond_move_not): Likewise.
1014         (*sign_extract_onebit): Likewise.
1015         (*not_signextract_onebit): Likewise.
1016         (stack_tie): Likewise.
1017         (align_4): Likewise.
1018         (align_8): Likewise.
1019         (consttable_end): Likewise.
1020         (consttable_1): Likewise.
1021         (consttable_2): Likewise.
1022         (consttable_4): Likewise.
1023         (consttable_8): Likewise.
1024         (consttable_16): Likewise.
1025         (*thumb1_tablejump): Likewise.
1026         (prefetch): Likewise.
1027         (force_register_use): Likewise.
1028         (thumb_eh_return): Likewise.
1029         (load_tp_hard): Likewise.
1030         (load_tp_soft): Likewise.
1031         (tlscall): Likewise.
1032         (*arm_movtas_ze): Likewise.
1033         (*arm_rev): Likewise.
1034         (*arm_revsh): Likewise.
1035         (*arm_rev16): Likewise.
1036         * config/arm/thumb2.md
1037         (*thumb2_smaxsi3): Likewise.
1038         (*thumb2_sminsi3): Likewise.
1039         (*thumb32_umaxsi3): Likewise.
1040         (*thumb2_uminsi3): Likewise.
1041         (*thumb2_negdi2): Likewise.
1042         (*thumb2_abssi2): Likewise.
1043         (*thumb2_neg_abss): Likewise.
1044         (*thumb2_movsi_insn): Likewise.
1045         (tls_load_dot_plus_four): Likewise.
1046         (*thumb2_movhi_insn): Likewise.
1047         (*thumb2_mov_scc): Likewise.
1048         (*thumb2_mov_negs): Likewise.
1049         (*thumb2_mov_negs): Likewise.
1050         (*thumb2_mov_nots): Likewise.
1051         (*thumb2_mov_nots): Likewise.
1052         (*thumb2_movsicc_): Likewise.
1053         (*thumb2_movsfcc_soft_insn): Likewise.
1054         (*thumb2_indirect_jump): Likewise.
1055         (*thumb2_and_scc): Likewise.
1056         (*thumb2_ior_scc): Likewise.
1057         (*thumb2_ior_scc_strict_it): Likewise.
1058         (*thumb2_cond_move): Likewise.
1059         (*thumb2_cond_arith): Likewise.
1060         (*thumb2_cond_ari): Likewise.
1061         (*thumb2_cond_sub): Likewise.
1062         (*thumb2_negscc): Likewise.
1063         (*thumb2_movcond): Likewise.
1064         (thumb2_casesi_internal): Likewise.
1065         (thumb2_casesi_internal_pic): Likewise.
1066         (*thumb2_alusi3_short): Likewise.
1067         (*thumb2_mov<mode>_shortim): Likewise.
1068         (*thumb2_addsi_short): Likewise.
1069         (*thumb2_subsi_short): Likewise.
1070         (thumb2_addsi3_compare0): Likewise.
1071         (*thumb2_cbz): Likewise.
1072         (*thumb2_cbnz): Likewise.
1073         (*thumb2_one_cmplsi2_short): Likewise.
1074         (*thumb2_negsi2_short): Likewise.
1075         (*orsi_notsi_si): Likewise.
1076         * config/arm/arm1020e.md: Update with new attributes.
1077         * config/arm/arm1026ejs.md: Update with new attributes.
1078         * config/arm/arm1136jfs.md: Update with new attributes.
1079         * config/arm/arm926ejs.md: Update with new attributes.
1080         * config/arm/cortex-a15.md: Update with new attributes.
1081         * config/arm/cortex-a5.md: Update with new attributes.
1082         * config/arm/cortex-a53.md: Update with new attributes.
1083         * config/arm/cortex-a7.md: Update with new attributes.
1084         * config/arm/cortex-a8.md: Update with new attributes.
1085         * config/arm/cortex-a9.md: Update with new attributes.
1086         * config/arm/cortex-m4.md: Update with new attributes.
1087         * config/arm/cortex-r4.md: Update with new attributes.
1088         * config/arm/fa526.md: Update with new attributes.
1089         * config/arm/fa606te.md: Update with new attributes.
1090         * config/arm/fa626te.md: Update with new attributes.
1091         * config/arm/fa726te.md: Update with new attributes.
1093 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1095         * config/aarch64/aarch64-simd.md
1096         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1097         <vwx> iterator to ensure correct register choice.
1098         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1099         (aarch64_sqdmull_n<mode>): Likewise.
1100         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1101         * config/aarch64/arm_neon.h
1102         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1103         (vml<as><q>_n_<su>16): Likewise.
1104         (vml<as>l_high_lane<q>_<su>16): Likewise.
1105         (vml<as>l_high_n_<su>16): Likewise.
1106         (vml<as>l_lane<q>_<su>16): Likewise.
1107         (vml<as>l_n_<su>16): Likewise.
1108         (vmul<q>_lane<q>_<su>16): Likewise.
1109         (vmul<q>_n_<su>16): Likewise.
1110         (vmull_lane<q>_<su>16): Likewise.
1111         (vmull_n_<su>16): Likewise.
1112         (vmull_high_lane<q>_<su>16): Likewise.
1113         (vmull_high_n_<su>16): Likewise.
1114         (vqrdmulh<q>_n_s16): Likewise.
1116 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1118         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1119         have the correct lane parameter.
1121 2013-09-06  Richard Biener <rguenther@suse.de>
1123         * cfganal.c (control_dependences::~control_dependences):
1124         Properly free all of the vector.
1126 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1128         PR target/58269
1129         * config/i386/i386.c (ix86_conditional_register_usage):
1130         Proper initialize extended SSE registers.
1132 2013-09-06  Jan Hubicka  <jh@suse.cz>
1134         PR tree-optimization/58311
1135         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1137 2013-09-06  Jan Hubicka  <jh@suse.cz>
1139         * Makefile.in (tree-sra.o): Update dependencies.
1140         * tree-sra.c: Include ipa-utils.h
1141         (scan_function): Use recursive_call_p.
1142         (has_caller_p): New function.
1143         (cgraph_for_node_and_aliases): Count also callers of aliases.
1145 2013-09-06  Jan Hubicka  <jh@suse.cz>
1147         PR middle-end/58094
1148         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1149         * tree-tailcall.c: Include ipa-utils.h.
1150         (find_tail_calls): Use it.
1151         * ipa-pure-const.c (check_call): Likewise.
1152         * ipa-utils.c (recursive_call_p): New function.
1153         * ipa-utils.h (recursive_call_p): Dclare.
1154         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1155         (symtab_semantically_equivalent_p): New function.
1156         * Makefile.in (tree-tailcall.o): Update dependencies.
1158 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1160         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1161         non-inlinable part.
1163 2013-09-06  Richard Biener  <rguenther@suse.de>
1165         * lto-streamer.h (lto_global_var_decls): Remove.
1166         * Makefile.in (OBJS): Remove lto-symtab.o.
1167         (lto-symtab.o): Remove.
1168         (GTFILES): Remove lto-symtab.c
1169         * lto-symtab.c: Move to lto/
1171 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1173         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1174         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1175         (UNSPEC_FPINT_RINT): New constant definitions.
1176         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1177         definition with 2 attributes.
1178         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1179         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1180         definitions.
1182 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1184         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1185         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1186         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1187         attribute to "z196_alone".
1188         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1189         "zEC12_simple".
1191 2013-09-06  Richard Biener  <rguenther@suse.de>
1193         * basic-block.h (class control_dependences): New.
1194         * tree-ssa-dce.c (control_dependence_map): Remove.
1195         (cd): New global.
1196         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1197         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1198         find_pdom, find_control_dependence, find_all_control_dependences):
1199         Move to cfganal.c.
1200         (mark_control_dependent_edges_necessary,
1201         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1202         tree_dce_done, perform_tree_ssa_dce): Adjust.
1203         * cfganal.c (set_control_dependence_map_bit,
1204         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1205         find_all_control_dependences): Move from tree-ssa-dce.c and
1206         implement as methods of control_dependences class.
1207         (control_dependences::control_dependences): New.
1208         (control_dependences::~control_dependences): Likewise.
1209         (control_dependences::get_edges_dependent_on): Likewise.
1210         (control_dependences::get_edge): Likewise.
1212 2013-09-04  Jan Hubicka  <jh@suse.cz>
1214         * tree.c (types_same_for_odr): Drop overactive check.
1215         * ipa-devirt.c (hash_type_name): Likewise.
1217 2013-09-04  Jan Hubicka  <jh@suse.cz>
1219         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1220         (analyze_functions): ... here.
1222 2013-09-04  Jan Hubicka  <jh@suse.cz>
1224         PR middle-end/58201
1225         * cgraphunit.c (analyze_functions): Clear AUX fields
1226         after processing; initialize assembler name has.
1228 2013-09-05  Jeff Law  <law@redhat.com>
1230         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1231         from thread_around_empty_block.  Record threading path into PATH.
1232         Recurse if threading through the initial block is successful.
1233         (thread_across_edge): Corresponding changes to slightly simplify.
1235 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1237         * config/aarch64/aarch64.md
1238         (type): Remove frecpe, frecps, frecpx.
1239         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1240         fix to be a TARGET_SIMD instruction.
1241         (aarch64_frecps): Remove.
1242         * config/aarch64/aarch64-simd.md
1243         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1244         (aarch64_frecps<mode>): Handle all float/vector of float modes.
1246 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1247             Sofiane Naci  <sofiane.naci@arm.com>
1249         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1250         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1251         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1252         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1253         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
1254         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1255         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
1256         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
1257         into "clz, "rbit".  Rename "shift" to "shift_imm".
1258         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1259         changes.  Update for attribute changes all occurrences of arlo_* and
1260         shift* types.
1261         * config/arm/arm-fixed.md: Update for attribute changes
1262         all occurrences of arlo_* types.
1263         * config/arm/thumb2.md: Update for attribute changes all occurrences
1264         of arlo_* types.
1265         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
1266         (cortexa7_older_only): Likewise.
1267         (cortexa7_younger):  Likewise.
1268         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1269         (1020alu_shift_op): Likewise.
1270         (1020alu_shift_reg_op): Likewise.
1271         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1272         (alu_shift_op): Likewise.
1273         (alu_shift_reg_op): Likewise.
1274         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1275         (11_alu_shift_op): Likewise.
1276         (11_alu_shift_reg_op): Likewise.
1277         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1278         (9_alu_shift_reg_op): Likewise.
1279         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1280         attribute changes.
1281         (cortex_a15_alu_shift): Likewise.
1282         (cortex_a15_alu_shift_reg): Likewise.
1283         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1284         attribute changes.
1285         (cortex_a5_alu_shift): Likewise.
1286         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1287         attribute changes.
1288         (cortex_a53_alu_shift): Likewise.
1289         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1290         attribute changes.
1291         (cortex_a7_alu_reg): Likewise.
1292         (cortex_a7_alu_shift): Likewise.
1293         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1294         attribute changes.
1295         (cortex_a8_alu_shift): Likewise.
1296         (cortex_a8_alu_shift_reg): Likewise.
1297         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1298         (cortex_a9_dp_shift): Likewise.
1299         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1300         attribute changes.
1301         * config/arm/cortex-r4.md
1302         (cortex_r4_alu): Update for attribute changes.
1303         (cortex_r4_mov): Likewise.
1304         (cortex_r4_alu_shift_reg): Likewise.
1305         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1306         (526_alu_shift_op): Likewise.
1307         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1308         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1309         (626te_alu_shift_op): Likewise.
1310         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1311         (726te_alu_shift_op): Likewise.
1312         (726te_alu_shift_reg_op): Likewise.
1313         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1314         (mp626_alu_shift_op): Likewise.
1315         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1316         (pj4_alu_conds): Likewise.
1317         (pj4_shift): Likewise.
1318         (pj4_shift_conds): Likewise.
1319         (pj4_alu_shift): Likewise.
1320         (pj4_alu_shift_conds): Likewise.
1321         * config/aarch64/aarch64.md: Update for attribute change
1322         all occurrences of arlo_* and shift* types.
1324 2013-09-05  Mike Stump  <mikestump@comcast.net>
1326         * tree.h: Move documentation for tree_function_decl to tree-core.h
1327         with the declaration.
1329 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
1331         PR target/58139
1332         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1333         looking for widest mode.
1335 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
1337         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1339 2013-09-05  Richard Biener  <rguenther@suse.de>
1341         PR tree-optimization/58137
1342         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1343         Do not create vectors of pointers.
1344         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1345         types for the components of the vector initializer.
1346         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1347         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1349 2013-09-05  Martin Jambor  <mjambor@suse.cz>
1351         * ipa-prop.c (remove_described_reference): Accept missing references,
1352         return false if that hppens, otherwise return true.
1353         (cgraph_node_for_jfunc): New function.
1354         (try_decrement_rdesc_refcount): Likewise.
1355         (try_make_edge_direct_simple_call): Use them.
1356         (ipa_edge_removal_hook): Remove references from rdescs.
1357         (ipa_edge_duplication_hook): Clone rdescs and their references
1358         when the new edge has the same caller as the old one.
1359         * cgraph.c (cgraph_resolve_speculation): Remove speculative
1360         reference before removing any edges.
1362 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
1364         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1365         initial store.
1366         * thumb2.md (thumb2_storewb_parisi): New pattern.
1368 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
1370         * config/aarch64/aarch64-option-extensions.def: Add
1371         AARCH64_OPT_EXTENSION of 'crc'.
1372         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1373         (AARCH64_ISA_CRC): Ditto.
1374         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1375         description of the CRC extension.
1377 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1379         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1380         * config/rs6000/linux.h: Ditto.
1381         * alpha/linux.h: Ditto.
1382         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1383         no_c99_libc_has_function.
1384         * config/c6x/uclinux-elf.h: Ditto.
1385         * config/lm32/uclinux-elf.h: Ditto.
1386         * config/m68k/uclinux.h: Ditto.
1387         * config/moxie/uclinux.h: Ditto.
1388         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1389         (crisv32-*-linux*, cris-*-linux*): Ditto.
1390         * config/bfin/bfin.c: Include "tm_p.h".
1392 2013-09-05  Richard Biener  <rguenther@suse.de>
1394         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1395         check for a definition without a basic-block.
1397 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1398             Sofiane Naci <sofiane.naci@arm.com>
1400         * config/aarch64/aarch64.md
1401         (*movti_aarch64): Rename r_2_f and f_2_r.
1402         (*movsf_aarch64): Likewise.
1403         (*movdf_aarch64): Likewise.
1404         (*movtf_aarch64): Likewise.
1405         (aarch64_movdi_<mode>low): Likewise.
1406         (aarch64_movdi_<mode>high): Likewise.
1407         (aarch64_mov<mode>high_di): Likewise.
1408         (aarch64_mov<mode>low_di): Likewise.
1409         (aarch64_movtilow_tilow): Likewise.
1410         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
1411         values to config/arm/types.md
1412         (attribute "conds"): Update for attribute change.
1413         (anddi3_insn): Likewise.
1414         (iordi3_insn): Likewise.
1415         (xordi3_insn): Likewise.
1416         (one_cmpldi2): Likewise.
1417         * config/arm/types.md (type): Add Neon types.
1418         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1419         use "type" attribute.
1420         (movmisalign<mode>_neon_store): Likewise.
1421         (movmisalign<mode>_neon_load): Likewise.
1422         (vec_set<mode>_internal): Likewise.
1423         (vec_setv2di_internal): Likewise.
1424         (vec_extract<mode>): Likewise.
1425         (vec_extractv2di): Likewise.
1426         (add<mode>3_neon): Likewise.
1427         (adddi3_neon): Likewise.
1428         (sub<mode>3_neon): Likewise.
1429         (subdi3_neon): Likewise.
1430         (mul<mode>3_neon): Likewise.
1431         (mul<mode>3add<mode>_neon): Likewise.
1432         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1433         (fma<VCVTF:mode>4)): Likewise.
1434         (fma<VCVTF:mode>4_intrinsic): Likewise.
1435         (fmsub<VCVTF:mode>4)): Likewise.
1436         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1437         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1438         (ior<mode>3): Likewise.
1439         (and<mode>3): Likewise.
1440         (anddi3_neon): Likewise.
1441         (orn<mode>3_neon): Likewise.
1442         (orndi3_neon): Likewise.
1443         (bic<mode>3_neon): Likewise.
1444         (bicdi3_neon): Likewise.
1445         (xor<mode>3): Likewise.
1446         (one_cmpl<mode>2): Likewise.
1447         (abs<mode>2): Likewise.
1448         (neg<mode>2): Likewise.
1449         (umin<mode>3_neon): Likewise.
1450         (umax<mode>3_neon): Likewise.
1451         (smin<mode>3_neon): Likewise.
1452         (smax<mode>3_neon): Likewise.
1453         (vashl<mode>3): Likewise.
1454         (vashr<mode>3_imm): Likewise.
1455         (vlshr<mode>3_imm): Likewise.
1456         (ashl<mode>3_signed): Likewise.
1457         (ashl<mode>3_unsigned): Likewise.
1458         (neon_load_count): Likewise.
1459         (ashldi3_neon_noclobber): Likewise.
1460         (signed_shift_di3_neon): Likewise.
1461         (unsigned_shift_di3_neon): Likewise.
1462         (ashrdi3_neon_imm_noclobber): Likewise.
1463         (lshrdi3_neon_imm_noclobber): Likewise.
1464         (widen_ssum<mode>3): Likewise.
1465         (widen_usum<mode>3): Likewise.
1466         (quad_halves_<code>v4si): Likewise.
1467         (quad_halves_<code>v4sf): Likewise.
1468         (quad_halves_<code>v8hi): Likewise.
1469         (quad_halves_<code>v16qi): Likewise.
1470         (reduc_splus_v2di): Likewise.
1471         (neon_vpadd_internal<mode>): Likewise.
1472         (neon_vpsmin<mode>): Likewise.
1473         (neon_vpsmax<mode>): Likewise.
1474         (neon_vpumin<mode>): Likewise.
1475         (neon_vpumax<mode>): Likewise.
1476         (ss_add<mode>_neon): Likewise.
1477         (us_add<mode>_neon): Likewise.
1478         (ss_sub<mode>_neon): Likewise.
1479         (us_sub<mode>_neon): Likewise.
1480         (neon_vadd<mode>_unspec): Likewise.
1481         (neon_vaddl<mode>): Likewise.
1482         (neon_vaddw<mode>): Likewise.
1483         (neon_vhadd<mode>): Likewise.
1484         (neon_vqadd<mode>): Likewise.
1485         (neon_vaddhn<mode>): Likewise.
1486         (neon_vmul<mode>): Likewise.
1487         (neon_vmla<mode>): Likewise.
1488         (neon_vmlal<mode>): Likewise.
1489         (neon_vmls<mode>): Likewise.
1490         (neon_vmlsl<mode>): Likewise.
1491         (neon_vqdmulh<mode>): Likewise.
1492         (neon_vqdmlal<mode>): Likewise.
1493         (neon_vqdmlsl<mode>): Likewise.
1494         (neon_vmull<mode>): Likewise.
1495         (neon_vqdmull<mode>): Likewise.
1496         (neon_vsub<mode>_unspec): Likewise.
1497         (neon_vsubl<mode>): Likewise.
1498         (neon_vsubw<mode>): Likewise.
1499         (neon_vqsub<mode>): Likewise.
1500         (neon_vhsub<mode>): Likewise.
1501         (neon_vsubhn<mode>): Likewise.
1502         (neon_vceq<mode>): Likewise.
1503         (neon_vcge<mode>): Likewise.
1504         (neon_vcgeu<mode>): Likewise.
1505         (neon_vcgt<mode>): Likewise.
1506         (neon_vcgtu<mode>): Likewise.
1507         (neon_vcle<mode>): Likewise.
1508         (neon_vclt<mode>): Likewise.
1509         (neon_vcage<mode>): Likewise.
1510         (neon_vcagt<mode>): Likewise.
1511         (neon_vtst<mode>): Likewise.
1512         (neon_vabd<mode>): Likewise.
1513         (neon_vabdl<mode>): Likewise.
1514         (neon_vaba<mode>): Likewise.
1515         (neon_vabal<mode>): Likewise.
1516         (neon_vmax<mode>): Likewise.
1517         (neon_vmin<mode>): Likewise.
1518         (neon_vpaddl<mode>): Likewise.
1519         (neon_vpadal<mode>): Likewise.
1520         (neon_vpmax<mode>): Likewise.
1521         (neon_vpmin<mode>): Likewise.
1522         (neon_vrecps<mode>): Likewise.
1523         (neon_vrsqrts<mode>): Likewise.
1524         (neon_vqabs<mode>): Likewise.
1525         (neon_vqneg<mode>): Likewise.
1526         (neon_vcls<mode>): Likewise.
1527         (clz<mode>2): Likewise.
1528         (popcount<mode>2): Likewise.
1529         (neon_vrecpe): Likewise.
1530         (neon_vrsqrte): Likewise.
1531         (neon_vget_lane<mode>_sext_internal): Likewise.
1532         (neon_vget_lane<mode>_zext_internal): Likewise.
1533         (neon_vdup_n<mode>): Likewise.
1534         (neon_vdup_nv2di): Likewise.
1535         (neon_vdpu_lane<mode>_internal): Likewise.
1536         (neon_vswp<mode>): Likewise.
1537         (float<mode><V_cvtto>2): Likewise.
1538         (floatuns<mode><V_cvtto>2): Likewise.
1539         (fix_trunc<mode><V_cvtto>)2): Likewise
1540         (fixuns_trunc<mode><V_cvtto)2): Likewise.
1541         (neon_vcvt<mode>): Likewise.
1542         (neon_vcvtv4sfv4hf): Likewise.
1543         (neon_vcvtv4hfv4sf): Likewise.
1544         (neon_vcvt_n<mode>): Likewise.
1545         (neon_vmovn<mode>): Likewise.
1546         (neon_vqmovn<mode>): Likewise.
1547         (neon_vqmovun<mode>): Likewise.
1548         (neon_vmovl<mode>): Likewise.
1549         (neon_vmul_lane<mode>): Likewise.
1550         (neon_vmull_lane<mode>): Likewise.
1551         (neon_vqdmull_lane<mode>): Likewise.
1552         (neon_vqdmulh_lane<mode>): Likewise.
1553         (neon_vmla_lane<mode>): Likewise.
1554         (neon_vmlal_lane<mode>): Likewise.
1555         (neon_vqdmlal_lane<mode>): Likewise.
1556         (neon_vmls_lane<mode>): Likewise.
1557         (neon_vmlsl_lane<mode>): Likewise.
1558         (neon_vqdmlsl_lane<mode>): Likewise.
1559         (neon_vext<mode>): Likewise.
1560         (neon_vrev64<mode>): Likewise.
1561         (neon_vrev32<mode>): Likewise.
1562         (neon_vrev16<mode>): Likewise.
1563         (neon_vbsl<mode>_internal): Likewise.
1564         (neon_vshl<mode>): Likewise.
1565         (neon_vqshl<mode>): Likewise.
1566         (neon_vshr_n<mode>): Likewise.
1567         (neon_vshrn_n<mode>): Likewise.
1568         (neon_vqshrn_n<mode>): Likewise.
1569         (neon_vqshrun_n<mode>): Likewise.
1570         (neon_vshl_n<mode>): Likewise.
1571         (neon_vqshl_n<mode>): Likewise.
1572         (neon_vqshlu_n<mode>): Likewise.
1573         (neon_vshll_n<mode>): Likewise.
1574         (neon_vsra_n<mode>): Likewise.
1575         (neon_vsri_n<mode>): Likewise.
1576         (neon_vsli_n<mode>): Likewise.
1577         (neon_vtbl1v8qi): Likewise.
1578         (neon_vtbl2v8qi): Likewise.
1579         (neon_vtbl3v8qi): Likewise.
1580         (neon_vtbl4v8qi): Likewise.
1581         (neon_vtbx1v8qi): Likewise.
1582         (neon_vtbx2v8qi): Likewise.
1583         (neon_vtbx3v8qi): Likewise.
1584         (neon_vtbx4v8qi): Likewise.
1585         (neon_vtrn<mode>_internal): Likewise.
1586         (neon_vzip<mode>_internal): Likewise.
1587         (neon_vuzp<mode>_internal): Likewise.
1588         (neon_vld1<mode>): Likewise.
1589         (neon_vld1_lane<mode>): Likewise.
1590         (neon_vld1_dup<mode>): Likewise.
1591         (neon_vld1_dupv2di): Likewise.
1592         (neon_vst1<mode>): Likewise.
1593         (neon_vst1_lane<mode>): Likewise.
1594         (neon_vld2<mode>): Likewise.
1595         (neon_vld2_lane<mode>): Likewise.
1596         (neon_vld2_dup<mode>): Likewise.
1597         (neon_vst2<mode>): Likewise.
1598         (neon_vst2_lane<mode>): Likewise.
1599         (neon_vld3<mode>): Likewise.
1600         (neon_vld3qa<mode>): Likewise.
1601         (neon_vld3qb<mode>): Likewise.
1602         (neon_vld3_lane<mode>): Likewise.
1603         (neon_vld3_dup<mode>): Likewise.
1604         (neon_vst3<mode>): Likewise.
1605         (neon_vst3qa<mode>): Likewise.
1606         (neon_vst3qb<mode>): Likewise.
1607         (neon_vst3_lane<mode>): Likewise.
1608         (neon_vld4<mode>): Likewise.
1609         (neon_vld4qa<mode>): Likewise.
1610         (neon_vld4qb<mode>): Likewise.
1611         (neon_vld4_lane<mode>): Likewise.
1612         (neon_vld4_dup<mode>): Likewise.
1613         (neon_vst4<mode>): Likewise.
1614         (neon_vst4qa<mode>): Likewise.
1615         (neon_vst4qb<mode>): Likewise.
1616         (neon_vst4_lane<mode>): Likewise.
1617         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1618         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1619         (neon_vec_<US>mult_lo_<mode>): Likewise.
1620         (neon_vec_<US>mult_hi_<mode>): Likewise.
1621         (neon_vec_<US>shiftl_<mode>): Likewise.
1622         (neon_unpack<US>_<mode>): Likewise.
1623         (neon_vec_<US>mult_<mode>): Likewise.
1624         (vec_pack_trunc_<mode>): Likewise.
1625         (neon_vec_pack_trunk_<mode>): Likewise.
1626         (neon_vabd<mode>_2): Likewise.
1627         (neon_vabd<mode>_3): Likewise.
1628         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
1629         (thumb2_movsi_vfp): Likewise.
1630         (movdi_vfp): Likewise.
1631         (movdi_vfp_cortexa8): Likewise.
1632         (movhf_vfp_neon): Likewise.
1633         (movhf_vfp): Likewiwse.
1634         (movsf_vfp): Likewiwse.
1635         (thumb2_movsf_vfp): Likewiwse.
1636         (movdf_vfp): Likewise.
1637         (thumb2_movdf_vfp): Likewise.
1638         (movsfcc_vfp): Likewise.
1639         (thumb2_movsfcc_vfp): Likewise.
1640         (movdfcc_vfp): Likewise.
1641         (thumb2_movdfcc_vfp): Likewise.
1642         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
1643         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
1644         (v10_v2c): Likewise.
1645         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
1646         attribute change.
1647         (cortex_a15_neon_int_2): Likewise.
1648         (cortex_a15_neon_int_3): Likewise.
1649         (cortex_a15_neon_int_4): Likewise.
1650         (cortex_a15_neon_int_5): Likewise.
1651         (cortex_a15_neon_vqneg_vqabs): Likewise.
1652         (cortex_a15_neon_vmov): Likewise.
1653         (cortex_a15_neon_vaba): Likewise.
1654         (cortex_a15_neon_vaba_qqq): Likewise.
1655         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1656         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1657         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
1658         scalar_64_32_long_scalar): Likewise.
1659         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1660         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1661         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1662         lotype_qdd_64_32_long): Likewise.
1663         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1664         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1665         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1666         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1667         (cortex_a15_neon_shift_1): Likewise.
1668         (cortex_a15_neon_shift_2): Likewise.
1669         (cortex_a15_neon_shift_3): Likewise.
1670         (cortex_a15_neon_vshl_ddd): Likewise.
1671         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1672         (cortex_a15_neon_vsra_vrsra): Likewise.
1673         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1674         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1675         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1676         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1677         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1678         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1679         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1680         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1681         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1682         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1683         (cortex_a15_neon_bp_simple): Likewise.
1684         (cortex_a15_neon_bp_2cycle): Likewise.
1685         (cortex_a15_neon_bp_3cycle): Likewise.
1686         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1687         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1688         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1689         (cortex_a15_neon_vld2_4_regs): Likewise.
1690         (cortex_a15_neon_vld3_vld4): Likewise.
1691         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1692         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1693         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
1694         (cortex_a15_neon_vst3_vst4): Likewise.
1695         (cortex_a15_neon_vld1_vld2_lane): Likewise.
1696         (cortex_a15_neon_vld3_vld4_lane" 10
1697         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1698         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1699         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1700         (cortex_a15_neon_ldm_2): Likewise.0
1701         (cortex_a15_neon_stm_2): Likewise.
1702         (cortex_a15_neon_mcr): Likewise.
1703         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1704         (cortex_a15_neon_mrc): Likewise.
1705         (cortex_a15_neon_mrrc): Likewise.
1706         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
1707         change.
1708         (cortex_a15_alu_shift): Likewise.
1709         (cortex_a15_alu_shift_reg): Likewise.
1710         (cortex_a15_mult32): Likewise.
1711         (cortex_a15_mult64): Likewise.
1712         (cortex_a15_block): Likewise.
1713         (cortex_a15_branch): Likewise.
1714         (cortex_a15_load1): Likewise.
1715         (cortex_a15_load3): Likewise.
1716         (cortex_a15_store1): Likewise.
1717         (cortex_a15_store3): Likewise.
1718         (cortex_a15_call): Likewise.
1719         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
1720         (cortex_a5_f2r): Likewise.
1721         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
1722         change.
1723         (cortex_a53_f2r): Likewise.
1724         * config/arm/cortex-a7.md
1725         (cortex_a7_branch): Update for attribute change.
1726         (cortex_a7_call): Likewise.
1727         (cortex_a7_alu_imm): Likewise.
1728         (cortex_a7_alu_reg): Likewise.
1729         (cortex_a7_alu_shift): Likewise.
1730         (cortex_a7_mul): Likewise.
1731         (cortex_a7_load1): Likewise.
1732         (cortex_a7_store1): Likewise.
1733         (cortex_a7_load2): Likewise.
1734         (cortex_a7_store2): Likewise.
1735         (cortex_a7_load3): Likewise.
1736         (cortex_a7_store3): Likewise.
1737         (cortex_a7_load4): Likewise.
1738         (cortex_a7_store4): Likewise.
1739         (cortex_a7_fpalu): Likewise.
1740         (cortex_a7_fconst): Likewise.
1741         (cortex_a7_fpmuls): Likewise.
1742         (cortex_a7_neon_mul): Likewise.
1743         (cortex_a7_fpmacs): Likewise.
1744         (cortex_a7_neon_mla: Likewise.
1745         (cortex_a7_fpmuld: Likewise.
1746         (cortex_a7_fpmacd: Likewise.
1747         (cortex_a7_fpfmad: Likewise.
1748         (cortex_a7_fdivs: Likewise.
1749         (cortex_a7_fdivd: Likewise.
1750         (cortex_a7_r2f: Likewise.
1751         (cortex_a7_f2r: Likewise.
1752         (cortex_a7_f_flags: Likewise.
1753         (cortex_a7_f_loads: Likewise.
1754         (cortex_a7_f_loadd: Likewise.
1755         (cortex_a7_f_stores: Likewise.
1756         (cortex_a7_f_stored: Likewise.
1757         (cortex_a7_neon): Likewise.
1758         * config/arm/cortex-a8-neon.md
1759         (cortex_a8_neon_mrc): Update for attribute change.
1760         (cortex_a8_neon_mrrc): Likewise.
1761         (cortex_a8_neon_int_1): Likewise.
1762         (cortex_a8_neon_int_2): Likewise.
1763         (cortex_a8_neon_int_3): Likewise.
1764         (cortex_a8_neon_int_4): Likewise.
1765         (cortex_a8_neon_int_5): Likewise.
1766         (cortex_a8_neon_vqneg_vqabs): Likewise.
1767         (cortex_a8_neon_vmov): Likewise.
1768         (cortex_a8_neon_vaba): Likewise.
1769         (cortex_a8_neon_vaba_qqq): Likewise.
1770         (cortex_a8_neon_vsma): Likewise.
1771         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1772         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1773         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1774         long_scalar): Likewise.
1775         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1776         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1777         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1778         long_scalar_qdd_64_32_long): Likewise.
1779         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1780         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1781         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1782         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1783         (cortex_a8_neon_shift_1): Likewise.
1784         (cortex_a8_neon_shift_2): Likewise.
1785         (cortex_a8_neon_shift_3): Likewise.
1786         (cortex_a8_neon_vshl_ddd): Likewise.
1787         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1788         (cortex_a8_neon_vsra_vrsra): Likewise.
1789         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1790         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1791         (cortex_a8_neon_fp_vsum): Likewise.
1792         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1793         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1794         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1795         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1796         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1797         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1798         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1799         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1800         (cortex_a8_neon_bp_simple): Likewise.
1801         (cortex_a8_neon_bp_2cycle): Likewise.
1802         (cortex_a8_neon_bp_3cycle): Likewise.
1803         (cortex_a8_neon_ldr): Likewise.
1804         (cortex_a8_neon_str): Likewise.
1805         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1806         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1807         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1808         (cortex_a8_neon_vld2_4_regs): Likewise.
1809         (cortex_a8_neon_vld3_vld4): Likewise.
1810         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1811         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1812         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1813         (cortex_a8_neon_vst3_vst4): Likewise.
1814         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1815         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1816         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1817         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1818         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1819         (cortex_a8_neon_mcr): Likewise.
1820         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1821         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
1822         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
1823         change.
1824         (ca9_neon_mrrc): Likewise.
1825         (cortex_a9_neon_int_1): Likewise.
1826         (cortex_a9_neon_int_2): Likewise.
1827         (cortex_a9_neon_int_3): Likewise.
1828         (cortex_a9_neon_int_4): Likewise.
1829         (cortex_a9_neon_int_5): Likewise.
1830         (cortex_a9_neon_vqneg_vqabs): Likewise.
1831         (cortex_a9_neon_vmov): Likewise.
1832         (cortex_a9_neon_vaba): Likewise.
1833         (cortex_a9_neon_vaba_qqq): Likewise.
1834         (cortex_a9_neon_vsma): Likewise.
1835         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1836         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1837         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1838         long_scalar): Likewise.
1839         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1840         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1841         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1842         long_scalar_qdd_64_32_long): Likewise.
1843         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1844         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1845         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1846         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1847         (cortex_a9_neon_shift_1): Likewise.
1848         (cortex_a9_neon_shift_2): Likewise.
1849         (cortex_a9_neon_shift_3): Likewise.
1850         (cortex_a9_neon_vshl_ddd): Likewise.
1851         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1852         (cortex_a9_neon_vsra_vrsra): Likewise.
1853         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1854         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1855         (cortex_a9_neon_fp_vsum): Likewise.
1856         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1857         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1858         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1859         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1860         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1861         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1862         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1863         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1864         (cortex_a9_neon_bp_simple): Likewise.
1865         (cortex_a9_neon_bp_2cycle): Likewise.
1866         (cortex_a9_neon_bp_3cycle): Likewise.
1867         (cortex_a9_neon_ldr): Likewise.
1868         (cortex_a9_neon_str): Likewise.
1869         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1870         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1871         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1872         (cortex_a9_neon_vld2_4_regs): Likewise.
1873         (cortex_a9_neon_vld3_vld4): Likewise.
1874         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1875         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1876         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1877         (cortex_a9_neon_vst3_vst4): Likewise.
1878         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1879         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1880         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1881         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1882         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1883         (cortex_a9_neon_mcr): Likewise.
1884         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1885         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
1886         (cortex_a9_fps): Likewise.
1887         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
1888         change.
1889         (cortex_m4_fmuls): Likewise.
1890         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
1891         change.
1892         (cortex_r4_mrc): Likewise.
1893         * config/arm/iterators.md: Update comment referring to neon_type.
1894         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
1895         (iwmmxt_movsi_insn): Likewise.
1896         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
1897         attribute change.
1898         (pj4_core_to_vfp): Likewise.
1899         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
1900         attribute change.
1901         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
1902         (vfp_fstore): Likewise.
1903         * doc/md.texi: Change references to neon_type to refer to type.
1905 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1907         * tree.h (DECL_BUILT_IN): Fix typo in comment.
1909 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
1911         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
1912         lglobl if not weak.
1914 2013-09-04  Easwaran Raman  <eraman@google.com>
1916         PR middle-end/57370
1917         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
1918         (build_and_add_sum): Use it.
1919         (appears_later_in_bb): Simplify code.
1921 2013-09-04  Teresa Johnson  <tejohnson@google.com>
1923         * dumpfile.c (dump_finish): Don't close stderr/stdout.
1925 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
1927         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
1929 2013-09-04  Jan Hubicka  <jh@suse.cz>
1931         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
1932         * ipa-devirt.c: Include diganostic.h
1933         (odr_type_d): Add types and types_set.
1934         (hash_type_name): Work for types with vtables during LTO.
1935         (odr_hasher::remove): Fix comment; destroy types_set.
1936         (add_type_duplicate): New function,
1937         (get_odr_type): Use it.
1938         (dump_type_inheritance_graph): Dump type duplicates.
1939         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
1940         graph.
1941         * tree.c (types_same_for_odr): Give exact answers on types with
1942         virtual tables.
1944 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1946         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
1947         explaining their differences.
1949 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
1951         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
1953 2013-09-03  Jeff Law  <law@redhat.com>
1955         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
1956         when not threading through a joiner block.  Pass joiner/no joiner
1957         state to register_jump_thread.
1958         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
1959         state from argument rather than implying on path length.
1960         Dump the entire jump thread path into debugging dump.
1961         * tree-flow.h (register_jump_thread): Update prototype.
1963 2013-08-29  Xinliang David Li  <davidxl@google.com>
1965         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1966         Remove a trivial gcc_assert.
1968 2013-08-29  Xinliang David Li  <davidxl@google.com>
1970         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
1971         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
1972         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1973         Delay base decl alignment adjustment.
1974         * tree-vectorizer.c (vect_destroy_datarefs): New function.
1975         * tree-vectorizer.h: New data structure.
1976         (set_dr_misalignment): New function.
1977         (dr_misalignment): Ditto.
1978         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
1979         (vectorizable_load): Ditto.
1980         (ensure_base_align): New function.
1981         (vectorize_loops): Add dbg_cnt support.
1982         (execute_vect_slp): Ditto.
1983         * dbgcnt.def: New debug counter.
1984         * Makefile: New dependency.
1986 2013-09-03  Meador Inge  <meadori@codesourcery.com>
1988         Revert:
1990         2013-08-30  Meador Inge  <meadori@codesourcery.com>
1992         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
1994 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
1996         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
1997         function descriptor.
1999 2013-09-03  Richard Biener  <rguenther@suse.de>
2001         * tree-affine.c (add_elt_to_tree): Fix association issue,
2002         avoid useless converts and make sure to always return a
2003         properly typed result.
2005 2013-09-03  Richard Biener  <rguenther@suse.de>
2007         PR middle-end/57656
2008         * fold-const.c (negate_expr_p): Fix division case.
2009         (negate_expr): Likewise.
2011 2013-09-03  Richard Biener  <rguenther@suse.de>
2013         PR lto/58285
2014         * tree-streamer-out.c: Include tm.h.
2015         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2017 2013-09-03  Jan Hubicka  <jh@suse.cz>
2019         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2021 2013-09-03  Alan Modra  <amodra@gmail.com>
2023         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2024         multilibs to big-endian target and vice versa.
2025         * config/rs6000/t-linux64: Use := assignment on all vars.
2026         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2027         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2028         * config/rs6000/t-linux64le: New file.
2029         * config/rs6000/t-linux64bele: New file.
2030         * config/rs6000/t-linux64lebe: New file.
2032 2013-09-02  Jan Hubicka  <jh@suse.cz>
2034         * ipa-inline-transform.c (inline_transform): Do not
2035         optimize_inline_calls when not optimizing.
2037 2013-09-02  Jan Hubicka  <jh@suse.cz>
2039         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2040         duplicated nodes for assembler names.
2041         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2042         hard registers from assembler name hash.
2044 2013-09-02  Jan Hubicka  <jh@suse.cz>
2046         * ipa-split.c (execute_split_functions): Split externally visible
2047         functions called once.
2049 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2051         PR ipa/58106
2052         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2053         linked list.  When finding the correct duplicate, also consider also
2054         the caller in additon to its inlined_to node.
2056 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2058         * config/aarch64/aarch64-simd-builtins.def
2059         (dup_lane_scalar): Remove.
2060         * config/aarch64/aarch64-simd.md
2061         (aarch64_simd_dup): Add 'w->w' alternative.
2062         (aarch64_dup_lane<mode>): Allow for VALL.
2063         (aarch64_dup_lane_scalar<mode>): Remove.
2064         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2065         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2066         (aarch64_get_lane_unsigned<mode>): Likewise.
2067         (aarch64_get_lane<mode>): Likewise.
2068         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2069         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2070         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2071         (VCON): Change container of V2SF.
2072         (vswap_width_name): Likewise.
2073         * config/aarch64/arm_neon.h
2074         (__aarch64_vdup_lane_any): New.
2075         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2076         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2077         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2079 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2081         PR middle-end/56382
2082         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2083         the source or the destination is a single hard register.
2085 2013-09-02  Richard Biener  <rguenther@suse.de>
2087         PR middle-end/57511
2088         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2089         non-linear SCEVs.
2091 2013-09-02  Richard Biener  <rguenther@suse.de>
2093         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2094         arithmetic to sizetype.
2096 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2098         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2099         Find auto-increment use both before and after candidate.
2101 2013-09-02  Marek Polacek  <polacek@redhat.com>
2103         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2105 2013-09-01  Jan Hubicka  <jh@suse.cz>
2107         * Makefile.in: Add ipa-profile.o
2108         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2109         * cgraph.c (struct cgraph_propagate_frequency_data,
2110         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2111         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2112         * cgraph.h (cgraph_propagate_frequency): Remove.
2113         * ipa-inline-analysis.c: Include ipa-utils.h;
2114         drop duplicated cfgloop.h.
2115         (inline_update_callee_summaries): Update.
2116         * ipa-profile.c: New file.
2117         * ipa-utils.h (ipa_propagate_frequency): Declare.
2118         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2119         data-streamer.h, value-prof.h.
2120         (symtab_remove_unreachable_nodes): Update profile.
2121         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2122         account_time_size, cmp_counts, dump_histogram,
2123         ipa_profile_generate_summary, ipa_profile_write_summary,
2124         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2125         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2126         Move to ipa-profile.c.
2128 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2130         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2132 2013-09-01  Jan Hubicka  <jh@suse.cz>
2134         * common.opt (fdevirtualize-speculatively): New function.
2135         * invoke.texi (fdevirtualize-speculatively): Document.
2136         * ipa-devirt.c: Include ipa-inline.h
2137         (likely_target_p): New function.
2138         (ipa_devirt): New function.
2139         (gate_ipa_devirt): New function.
2140         (pass_data_ipa_devirt): New static var.
2141         (pass_ipa_devirt): Likewise.
2142         (make_pass_ipa_devirt): New function.
2143         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2144         (common_handle_option): Disable devirtualization when
2145         value range profiling is available.
2146         * passes.def (pass_ipa_devirt): Add.
2147         * timever.def (TV_IPA_DEVIRT): New timevar.
2148         * tree-pass.h (make_pass_ipa_devirt):
2150 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2152         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2153         include sanitize(undefined).
2155 2013-08-31  Diego Novillo  <dnovillo@google.com>
2157         * Makefile.in (TREE_CORE_H): Define.
2158         (TREE_H): Use.
2159         (GTFILES): Add tree-core.h.
2160         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2161         size the array.
2162         * tree-core.h: New file.
2163         Move all data structures, enum, typedefs, global
2164         declarations and constants from ...
2165         * tree.h: ... here.
2167 2013-08-31  Jan Hubicka  <jh@suse.cz>
2169         * bulitins.c (expand_builtin): Do not early exit for gcov
2170         instrumented functions.
2172 2013-08-31  Marek Polacek  <polacek@redhat.com>
2174         * ubsan.c: Include tm_p.h.
2176 2013-08-31  Jan Hubicka  <jh@suse.cz>
2178         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2179         warning.
2181         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2182         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2183         * tree-cfg.c (verify_gimple_label): ... here.
2184         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2185         (ipa_merge_profiles): New function.
2186         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2187         (lto_input_function_body): Likewise.
2188         * ipa-utils.h (ipa_merge_profiles): Declare.
2189         * lto-streamer.h (lto_input_function_body): Update prototype.
2190         (emit_label_in_global_context_p): Remove.
2191         * lto-symtab.c: Include ipa-utils.h
2192         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2194 2013-08-31  Jan Hubicka  <jh@suse.cz>
2196         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2198 2013-08-31  Jan Hubicka  <jh@suse.cz>
2200         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2202 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2204         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2205         "cmp" RTX before signed_comparison_operator check to account
2206         for "code" changes.
2208 2013-08-30  Jan Hubicka  <jh@suse.cz>
2210         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2211         (detect_type_change_1): Rename to ...
2212         (detect_type_change): ... this one; early return on non-polymorphic
2213         types.
2214         (detect_type_change_ssa): Add comp_type parameter; update
2215         use of detect_type_change.
2216         (compute_complex_assign_jump_func): Add param_type parameter;
2217         update use of detect_type_change_ssa.
2218         (compute_complex_ancestor_jump_func): Likewise.
2219         (ipa_get_callee_param_type): New function.
2220         (ipa_compute_jump_functions_for_edge): Compute parameter type;
2221         update calls to the jump function computation functions.
2223 2013-08-30  Teresa Johnson  <tejohnson@google.com>
2224             Steven Bosscher  <steven@gcc.gnu.org>
2226         * cfgrtl.c (fixup_new_cold_bb): New routine.
2227         (commit_edge_insertions): Invoke fixup_partitions.
2228         (find_partition_fixes): New routine.
2229         (fixup_partitions): Ditto.
2230         (verify_hot_cold_block_grouping): Update comments.
2231         (rtl_verify_edges): Invoke find_partition_fixes.
2232         (rtl_verify_bb_pointers): Update comments.
2233         (rtl_verify_bb_layout): Ditto.
2234         * basic-block.h (probably_never_executed_edge_p): Declare.
2235         (fixup_partitions): Ditto.
2236         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2237         * bb-reorder.c (sanitize_hot_paths): New function.
2238         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2239         sanitize_hot_paths.
2240         * predict.c (probably_never_executed_edge_p): New routine.
2241         * cfg.c (check_bb_profile): Add partition insanity warnings.
2243 2013-08-30  Meador Inge  <meadori@codesourcery.com>
2245         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2247 2013-08-30  Marek Polacek  <polacek@redhat.com>
2249         * Makefile.in (ubsan.o): Add.
2250         (c-family/c-ubsan.o): Add.
2251         (builtins.o): Add ubsan.h dependency.
2252         * ubsan.h: New file.
2253         * ubsan.c: New file.
2254         * common.opt: Add -fsanitize=undefined option.
2255         (flag_sanitize): Add variable.
2256         (fsanitize=): Add option.  Add Driver.
2257         (fsanitize=thread): Remove option.
2258         (fsanitize=address): Likewise.
2259         (static-libubsan): New option.
2260         * doc/invoke.texi: Document the new flag and -static-libubsan.
2261         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2262         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2263         * builtin-attrs.def (ATTR_COLD): Define.
2264         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2265         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2266         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2267         * flag-types.h (sanitize_code): New enum.
2268         * opts.c (common_handle_option): Parse command line arguments
2269         of -fsanitize=.  Add -fsanitize=unreachable option.
2270         * varasm.c (get_variable_section): Adjust.
2271         (assemble_noswitch_variable): Likewise.
2272         (assemble_variable): Likewise.
2273         (output_constant_def_contents): Likewise.
2274         (categorize_decl_for_section): Likewise.
2275         (place_block_symbol): Likewise.
2276         (output_object_block): Likewise.
2277         * builtins.def: Likewise.
2278         * toplev.c (compile_file): Likewise.
2279         (process_options): Likewise.
2280         * cppbuiltin.c: Likewise.
2281         * tsan.c (tsan_pass): Likewise.
2282         (tsan_gate): Likewise.
2283         (tsan_gate_O0): Likewise.
2284         * cfgexpand.c (partition_stack_vars): Likewise.
2285         (expand_stack_vars): Likewise.
2286         (defer_stack_allocation): Likewise.
2287         (expand_used_vars): Likewise.
2288         * cfgcleanup.c (old_insns_match_p): Likewise.
2289         * asan.c (asan_finish_file): Likewise.
2290         (asan_instrument): Likewise.
2291         (gate_asan): Likewise.
2292         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2293         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2294         (asan_global_struct): Use pointer_sized_int_node instead
2295         calling build_nonstandard_integer_type.
2296         (initialize_sanitizer_builtins): Likewise.
2297         (asan_finish_file): Likewise.
2298         * gcc.c: Document %{%:function(args):X}.
2299         (static_spec_functions): Add sanitize.
2300         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2301         store funcval != NULL there.
2302         (do_spec_1): Adjust handle_spec_function caller.
2303         (handle_braces): Allow %:function(args) as condition.
2304         (sanitize_spec_function): New function.
2305         (ADD_STATIC_LIBUBSAN_LIBS): Define.
2306         (LIBUBSAN_SPEC): Likewise.
2307         (LIBUBSAN_EARLY_SPEC): Likewise.
2308         (SANITIZER_SPEC): Handle libubsan.
2309         (SANITIZER_EARLY_SPEC): Likewise.
2310         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2311         instead of fsanitize=address.
2312         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2313         instead of fsanitize=address*.
2314         * builtins.c: Include ubsan.h.
2315         (fold_builtin_0): Instrument __builtin_unreachable.
2316         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2317         instead of flag_asan.
2318         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2319         (pointer_sized_int_node): Define.
2320         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2322 2013-08-30  Mike Stump  <mikestump@comcast.net>
2324         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2325         with RE patterns.
2327 2013-08-29  Jan Hubicka  <jh@suse.cz>
2329         * cgraph.c (cgraph_function_body_availability): Handle weakref
2330         correctly.
2331         * passes.def: Remove pass_fixup_cfg.
2332         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2333         track when we need to remove functions.
2334         (gate_ipa_inline): Execute inlining always; add comment why.
2335         (pass_data_ipa_inline): Remove TODO_remove_functions.
2336         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2337         do not produce summaries.
2338         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2339         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2340         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2341         constant pool and vtable.
2343 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
2345         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2346         New arm_neon.h's internal macros to specify 64-bit constants.
2347         Avoid using stdint.h's macros.
2349 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2351         * recog.c (verify_changes): Verify that changes[i].old is non-zero
2352         before applying REG_P.
2354 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
2356         PR tree-optimization/58277
2357         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2358         after seeing too many stmts with vdef in between dombb and current
2359         bb, invalidate everything.
2361 2013-08-30  Richard Biener  <rguenther@suse.de>
2363         * fold-const.c (fold_single_bit_test): Fix overflow test.
2365 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
2367         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2368         and which can live in a register, always retrieve the value on entry.
2369         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2370         passed by invisible reference specially.
2371         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2372         (vt_add_function_parameter): Correctly deal with a parameter passed by
2373         invisible reference.
2375 2013-08-30  Jan Hubicka  <jh@suse.cz>
2377         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2379 2013-08-30  Richard Biener  <rguenther@suse.de>
2381         PR tree-optimization/58228
2382         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2383         allow invariant loads in nested loop vectorization.
2385 2013-08-30  Richard Biener  <rguenther@suse.de>
2387         PR tree-optimization/58223
2388         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2389         (has_anti_or_output_dependence): ... this and adjust to also
2390         look for output dependences.
2391         (mark_nodes_having_upstream_mem_writes): Adjust.
2392         (rdg_flag_uses): Likewise.
2394 2013-08-30  Richard Biener  <rguenther@suse.de>
2396         PR tree-optimization/58010
2397         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2398         assert that we have a loop-closed PHI.
2400 2013-08-29  Jan Hubicka  <jh@suse.cz>
2402         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2403         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2404         * lto-section-in.c (lto_free_function_in_decl_state): New function.
2405         (lto_free_function_in_decl_state_for_node): New function.
2407 2013-08-29  Xinliang David Li  <davidxl@google.com>
2409         * loop-unroll.c (report_unroll_peel): Minor message change.
2410         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2411         Emit alignment peeling message with default -fopt-info.
2412         (vect_loop_versioning): Emit loop version info message.
2413         * tree-vectorizer.c (vectorize_loops): Minor message change.
2414         (execute_vect_slp): Ditto.
2416 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
2418         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2419         of the clone from the DECL_NAME of the original function.
2421 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
2423         * passes.c (register_pass): Add overload.
2424         * tree-pass.h (register_pass): Forward declare it.  Add comment.
2426 2013-08-29  Jan Hubicka  <jh@suse.cz>
2428         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2429         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2430         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2431         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2432         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2433         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2434         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2435         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2437 2013-08-29  Teresa Johnson  <tejohnson@google.com>
2439         * dumpfile.c (dump_loc): Output column number.
2440         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2441         * doc/invoke.texi: Document optall -fopt-info flag.
2442         * profile.c (read_profile_edge_counts): Use new dump framework.
2443         (compute_branch_probabilities): Ditto.
2444         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2445         when pass not in any opt group.
2446         * pass_manager.h (pass_manager::get_pass_profile): New method.
2447         * value-prof.c (check_counter): Use new dump framework.
2448         (check_ic_target): Ditto.
2449         * coverage.c (get_coverage_counts): Ditto.
2450         (coverage_init): Setup new dump framework.
2452 2013-08-29  Richard Biener  <rguenther@suse.de>
2454         PR tree-optimization/58246
2455         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2456         handle the dominance check inside a basic-block.
2458 2013-08-29  Richard Biener  <rguenther@suse.de>
2460         PR middle-end/57287
2461         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2462         of default defs that appear in abnormal PHI nodes.
2464 2013-08-29  Richard Biener  <rguenther@suse.de>
2466         PR tree-optimization/57685
2467         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2468         single-use operands to avoid exponential complexity.
2470 2013-08-28  Dehao Chen  <dehao@google.com>
2472         * ipa-inline.c (edge_badness): Fix integer underflow.
2474 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
2476         * gtm-builtins.def (_ITM_free): Declare leaf.
2478 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2480         PR target/58067
2481         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2482         (*tls_local_dynamic_base_64_largepic): Likewise.
2483         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2484         Remove predicate from call operand.
2485         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2486         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2488 2013-08-28  Jeff Law  <law@redhat.com>
2490         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2491         checks for the number of predecessors and successors allowed.
2492         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2493         which require copying a joiner block if there is a request which
2494         is a subpath that requires no joiner block copying.
2496 2013-08-28  Jan Hubicka  <jh@suse.cz>
2498         * lto-streamer-out.c (DFS_write_tree_body): Drop
2499         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2500         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2501         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2502         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2503         Do not read DECL_ERROR_ISSUED.
2504         (unpack_ts_decl_with_vis_value_fields): Do not read
2505         DECL_DEFER_OUTPUT.
2506         (lto_input_ts_binfo_tree_pointers): Do not read
2507         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2508         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2509         write DECL_ERROR_ISSUED..
2510         (pack_ts_decl_with_vis_value_fields): Do not write
2511         DECL_DEFER_OUTPUT.
2512         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2513         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2514         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2515         * tree.h (tree_decl_common): Update comment.
2516         (DECL_ERROR_ISSUED): Remove.
2518 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2520         PR middle-end/58257
2521         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2523 2013-08-28  Jan Hubicka  <jh@suse.cz>
2525         * builtins.def (free): Declare leaf.
2527 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
2529         * gdbhooks.py: New.
2530         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2531         * configure: Regenerate.
2533 2013-08-27  Martin Jambor  <mjambor@suse.cz>
2535         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2536         (ipa_ancestor_jf_data): Likewise.
2537         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2538         (ipa_get_jf_pass_through_type_preserved): New function.
2539         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2540         (ipa_get_jf_ancestor_type_preserved): New function.
2541         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2542         (ipa_get_jf_ancestor_result): Likewise.
2543         (propagate_vals_accross_pass_through): Use
2544         ipa_get_jf_pass_through_result to do all the value mappings.
2545         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2546         type_preserved flag.
2547         (ipa_set_jf_cst_copy): New function.
2548         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2549         (ipa_set_jf_arith_pass_through): Likewise.
2550         (ipa_set_ancestor_jf): Likewise.
2551         (compute_complex_assign_jump_func): Set type_preserved instead of
2552         punting.
2553         (ipa_compute_jump_functions_for_edge): Likewise.
2554         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2555         (update_jump_functions_after_inlining): Update type_preserved.
2556         Explicitely create jump functions when combining one with pass_through.
2557         (ipa_write_jump_function): Stream the type_preserved flags.
2558         (ipa_read_jump_function): Likewise.
2560 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
2561             Aldy Hernandez  <aldyh@redhat.com>
2563         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
2564         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
2565         * function.h (struct function): Add has_force_vect_loops and
2566         has_simduid_loops.
2567         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
2568         * gimple.c (gimple_build_omp_critical): Add KIND argument and
2569         handle it.
2570         * gimple.def: Update CLAUSES comments.
2571         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
2572         (gimple_build_omp_for): Add argument to prototype.
2573         (gimple_omp_for_kind): New.
2574         (gimple_omp_for_set_kind): New.
2575         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
2576         GOVD_DATA_SHARE_CLASS.
2577         (enum omp_region_type): Add ORT_SIMD.
2578         (gimple_add_tmp_var): Handle ORT_SIMD.
2579         (gimplify_var_or_parm_decl): Same.
2580         (is_gimple_stmt): Same.
2581         (omp_firstprivatize_variable): Same.
2582         (omp_add_variable): Only use splay_tree_insert if lookup failed.
2583         (omp_notice_variable): Handle ORT_SIMD.
2584         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
2585         (omp_check_private): Handle ORT_SIMD.
2586         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2587         OMP_CLAUSE_SAFELEN.
2588         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
2589         Handle OMP_CLAUSE_LASTPRIVATE.
2590         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2591         OMP_CLAUSE_SAFELEN.
2592         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
2593         (gimplify_expr): Handle OMP_SIMD.
2594         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
2595         (expand_GOMP_SIMD_VF): New.
2596         (expand_GOMP_SIMD_LAST_LANE): New.
2597         * internal-fn.def (GOMP_SIMD_LANE): New.
2598         (GOMP_SIMD_VF): New.
2599         (GOMP_SIMD_LAST_LANE): New.
2600         * omp-low.c: Include target.h.
2601         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
2602         OMP_CLAUSE_SAFELEN.
2603         (check_omp_nesting_restrictions): Same.
2604         (omp_max_vf): New.
2605         (lower_rec_simd_input_clauses): New.
2606         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
2607         OMP_CLAUSE_LINEAR.
2608         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
2609         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
2610         (expand_omp_build_assign): New.
2611         (expand_omp_for_init_counts): New.
2612         (expand_omp_for_init_vars): New.
2613         (extract_omp_for_update_vars): New.
2614         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
2615         and rewrite accordingly.
2616         (expand_omp_simd): New.
2617         (expand_omp_for): Use expand_omp_simd.
2618         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
2619         (lower_omp_for): Do not lower the body.
2620         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
2621         in their own loops.
2622         * tree-flow.h (find_omp_clause): Remove prototype.
2623         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
2624         forcing vectorization of the loop, or if flag_tree_vectorize.
2625         (gate_tree_if_conversion): Similarly.
2626         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
2627         gimple_build_omp_for.
2628         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
2629         * tree-parloops (create_parallel_loop): Pass kind argument to
2630         gimple_build_omp_for.
2631         * tree-pretty-print.c (dump_omp_clause): Add cases for
2632         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
2633         OMP_CLAUSE__SIMDUID_.
2634         (dump_generic_node): Handle OMP_SIMD.
2635         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
2636         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
2637         unroll OMP_SIMD loops here.
2638         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
2639         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
2640         loop->safelen.
2641         (vect_analyze_data_refs): Handle simd loops.
2642         * tree-vect-loop.c (vectorizable_live_operation): Handle
2643         IFN_GOMP_SIMD*.
2644         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
2645         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
2646         (vectorizable_load): Same.
2647         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
2648         (struct simduid_to_vf): New.
2649         (simduid_to_vf::hash): New.
2650         (simduid_to-vf::equal): New.
2651         (struct simd_array_to_simduid): New.
2652         (simd_array_to_simduid::hash): New.
2653         (simd_array_to_simduid::equal): New.
2654         (adjust_simduid_builtins): New.
2655         (struct note_simd_array_uses_struct): New.
2656         (note_simd_array_uses_cb): New.
2657         (note_simd_array_uses): New.
2658         (vectorize_loops): Handle simd hints and adjust simd builtins
2659         accordingly.
2660         * tree-vectorizer.h (struct _stmt_vec_info): Add
2661         simd_lane_access_p field.
2662         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
2663         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
2664         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
2665         (omp_clause_code_name): Same.
2666         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
2667         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
2668         * tree.def (OMP_SIMD): New entry.
2669         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
2670         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
2671         (OMP_CLAUSE_DECL): Adjust range for new clauses.
2672         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
2673         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
2674         (OMP_CLAUSE_LINEAR_STEP): New.
2675         (OMP_CLAUSE_SAFELEN_EXPR): New.
2676         (OMP_CLAUSE__SIMDUID__DECL): New.
2677         (find_omp_clause): New prototype.
2679 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
2681         * config/i386/driver-i386.c (host_detect_local_cpu): Update
2682         Haswell processor detection.
2684 2013-08-27  Christian Widmer  <shadow@umbrox.de>
2686         PR target/57927
2687         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
2688         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
2689         AVX2 capable processors.
2691 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
2693         * config/aarch64/arm_neon.h: Replace all inline asm implementations
2694         of vget_low_* with implementations in terms of other intrinsics.
2696 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
2698         PR middle-end/57219
2699         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
2700         values to -1, 0 and 1.
2702 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
2704         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
2705         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
2706         (<optab><mode>3_insn): Remove.
2707         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
2708         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
2709         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
2710         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
2711         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
2712         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
2713         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
2714         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
2715         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
2716         (ror<mode>3_insn): Likewise.
2717         * config/aarch64/predicates.md (aarch64_simd_register): New.
2719 2013-08-27  Richard Biener  <rguenther@suse.de>
2721         PR tree-optimization/57521
2722         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
2723         one edge is non-critical.
2724         (find_phi_replacement_condition): Make sure to use a non-critical
2725         edge.  Cleanup and remove old bug workarounds.
2726         (bb_postdominates_preds): Remove.
2727         (if_convertible_loop_p_1): Do not compute post-dominators.
2728         (combine_blocks): Do not free post-dominators.
2729         (main_tree_if_conversion): Likewise.
2730         (pass_data_if_conversion): Add TODO_verify_ssa.
2732 2013-08-27  DJ Delorie  <dj@redhat.com>
2734         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
2736 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
2738         * function.c (assign_parm_find_data_types): Set passed_mode and
2739         nominal_mode to the TYPE_MODE of nominal_type for the built
2740         pointer type in case of the struct-pass-by-reference.
2742 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2744         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
2745         (UINT16_TYPE): Change default to "unsigned int".
2747         * config/avr/avr.opt (mfract-convert-truncate): New option.
2748         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
2749         is set, round negative fractional integers according to n1169
2750         when converting to integer types.
2752 2013-08-26  Jan Hubicka  <jh@suse.cz>
2754         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
2755         methods can not be called indirectly when their address is not taken.
2757 2013-08-26  Jan Hubicka  <jh@suse.cz>
2759         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
2760         ctor_for_folding.
2762 2013-08-26  Jan Hubicka  <jh@suse.cz>
2764         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
2765         can be unshared.
2767 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2769         * reload.c (find_valid_class): Allow classes that do not include
2770         FIRST_PSEUDO_REGISTER - 1.
2772 2013-08-26  Jan Hubicka  <jh@suse.cz>
2774         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
2775         fix edge count/frequency when speculation failed; fix type check
2776         for the direct call.
2778 2013-08-26  Jan Hubicka  <jh@suse.cz>
2780         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
2782 2013-08-26  Jan Hubicka  <jh@suse.cz>
2784         * ipa-inline-transform.c (inline_transform): Be ready for basic block
2785         to be changed by edge redirection.
2787 2013-08-26  Jan Hubicka  <jh@suse.cz>
2789         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
2790         formating; add sanity check.
2791         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
2792         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
2793         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
2794         (resolve_noninline_speculation): Update callee keys, too.
2796 2013-08-26  Jan Hubicka  <jh@suse.cz>
2798         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
2799         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
2801 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2803         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
2804         into proper place.
2806 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
2808         * config/i386/i386.c (ix86_debug_options): Remove prototype.
2809         (x86_64_elf_select_section): Ditto.
2810         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
2811         arguments.
2812         (ix86_pass_by_reference): Ditto.
2813         (output_set_got): Ditto.
2814         (ix86_unary_operator_ok): Ditto.
2815         (ix86_expand_builtin): Ditto.
2817 2013-08-23  Jan Hubicka  <jh@suse.cz>
2819         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
2821 2013-08-23  Jan Hubicka  <jh@suse.cz>
2823         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
2824         (tree_decl_with_vis): Add FINAL field.
2826 2013-08-23  Jeff Law  <law@redhat.com>
2828         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
2829         the debugging dump when the expression is fully redundant.
2831 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2833         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
2834         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
2835         * pretty-print.c (pp_formatted_text_data): Likewise.
2836         (pp_write_text_to_stream): Likewise.
2837         (pp_write_text_as_dot_label_to_stream): Likewise.
2838         (pp_append_r): Likewise.
2839         (pp_format): Likewise.
2840         (pp_flush): Likewise.
2841         (pp_clear_output_area): Likewise.
2842         (pp_append_text): Likewise.
2843         (pp_formatted_text): Likewise.
2844         (pp_remaining_character_count_for_line): Likewise.
2845         (pp_newline): Likewise.
2846         (pp_character): Likewise.
2847         (output_buffer::~output_buffer): Define.
2848         (pretty_printer::~pretty_printer): Destruct output buffer.
2849         * pretty-print.h (output_buffer::~output_buffer): Declare.
2850         (pretty_printer::~pretty_printer): Declare virtual.
2852 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
2854         PR other/57324
2855         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
2856         HOST_WIDE_INT_M1U): New macros.
2857         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
2858         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
2859         unsigned -1 for lshift.
2860         * cse.c (cse_insn): Likewise.
2861         * double-int.c (rshift_double, lshift_double): Likewise.
2862         * builtins.c (fold_builtin_bitop): Likewise.
2863         * combine.c (force_to_mode): Likewise.
2864         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
2865         * simplify-rtx.c (simplify_const_unary_operation,
2866         simplify_const_binary_operation): Likewise.
2867         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
2868         check_va_list_escapes): Likewise.
2869         * rtlanal.c (nonzero_bits1): Likewise.
2870         * expmed.c (expand_smod_pow2): Likewise.
2871         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
2873 2013-08-23  Jan Hubicka  <jh@suse.cz>
2875         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
2876         as having address taken.
2878 2013-08-23  Jan Hubicka  <jh@suse.cz>
2880         * ipa-utils.h (method_class_type): Declare.
2881         * ipa-devirt.c (method_class_type): Export.
2883         * cgraphunit.c (analyze_functions): Do basic devirtualization;
2884         do not walk base classes of anonymous types.
2886 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
2888         PR rtl-optimization/58220
2889         PR regression/58221
2890         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
2891         handle SEQUENCE insns properly.
2893 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2895         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
2896         definition.
2897         (pp_newline_and_indent): Likewise.
2898         (pp_separate_with): Likewise.
2899         * pretty-print.c (pp_newline_and_flush): Define.
2900         (pp_newline_and_indent): Likewise.
2901         (pp_separate_with): Likewise.
2903 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2905         PR target/58218
2906         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
2907         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
2909 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2911         * config/i386/predicates.md (ext_sse_reg_operand): New.
2912         * config/i386/i386.md (*movti_internal): Use
2913         predicate to determine if EVEX is needed.
2914         (*movsi_internal): Ditto.
2915         (*movdf_internal): Ditto.
2916         (*movsf_internal): Ditto.
2917         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
2919 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2921         PR tree-optimization/58209
2922         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
2923         (find_tail_calls): Give up for pointer result types if m is non-NULL.
2924         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
2925         emit POINTER_PLUS_EXPR.
2926         (create_tailcall_accumulator): For pointer result type accumulate in
2927         sizetype type.
2929 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
2931         * configure.ac: Add backslashes missing from the last change.
2932         * configure: Regenerate.
2934 2013-08-22  Jan Hubicka  <jh@susue.cz>
2936         * ipa.c (function_and_variable_visibility): First remember function
2937         was global and then make it local.
2939 2013-08-22  Julian Brown  <julian@codesourcery.com>
2941         * configure.ac: Add aarch64 to list of arches which use "nop" in
2942         debug_line test.
2943         * configure: Regenerate.
2945 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2947         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
2948         gnu_libc_has_function.
2949         * config/s390/tpf.h: Likewise.
2951 2013-08-22  Jan Hubicka  <jh@susue.cz>
2953         * timevar.c (validate_phases): Add cast.
2955 2013-08-22  Jan Hubicka  <jh@susue.cz>
2957         * timevar.c (validate_phases): Use size_t for memory.
2958         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
2960 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2962         * pretty-print.h (output_buffer::output_buffer): Declare.
2963         (pretty_printer::pretty_printer): Likewise.
2964         (pp_construct): Remove.
2965         * pretty-print.c (output_buffer::output_buffer): Define.
2966         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
2967         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
2968         (print_gimple_expr): Likewise.
2969         (print_gimple_seq): Likewise.
2970         (gimple_dump_bb): Likewise.
2971         * sched-vis.c (dump_value_slim): Likewise.
2972         (dump_insn_slim): Likewise.
2973         (dump_rtl_slim): Likewise.
2974         (str_pattern_slim): Likewise.
2975         * tree-mudflap.c (mf_varname_tree): Likewise.
2976         * graph.c (print_graph_cfg): Likewise.
2977         (start_graph_dump): Likewise.
2978         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
2979         placement-new.
2980         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
2981         pretty printer initialization.
2982         * coretypes.h (diagnostic_context): Remove superflous type alias
2983         declaration.
2984         (pretty_printer): Likewise.  Declare directly as a class.
2985         (pretty_print_info): Remove declaration as class.
2986         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
2987         and pp_clear_output_area.
2988         (asan_add_global): Likewise.
2990 2013-08-22  Jan Hubicka  <jh@suse.cz>
2992         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
2993         * ipa-utils.h (update_type_inheritance_graph): Declare.
2994         (possible_polymorphic_call_target_p): Declare.
2995         (possible_polymorphic_call_target_p): New.
2996         * ipa-devirt.c: Update toplevel comments.
2997         (cached_polymorphic_call_targets): Move up.
2998         (odr_type_d): Move ID down.
2999         (polymorphic_type_binfo_p): Update comment.
3000         (odr_hasher::remove): Likewise;
3001         (get_odr_type): Set anonymous_namespace.
3002         (dump_odr_type): Dump it.
3003         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3004         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3005         (record_binfo): Add comment.
3006         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3007         built.
3008         (devirt_node_removal_hook): Do not iCE when cache is freed.
3009         (possible_polymorphic_call_target_p): New predicate.
3010         (update_type_inheritance_graph): New function.
3012 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3013             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3014             Sergey Lega  <sergey.s.lega@intel.com>
3015             Anna Tikhonova  <anna.tikhonova@intel.com>
3016             Ilya Tocar  <ilya.tocar@intel.com>
3017             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3018             Ilya Verbin  <ilya.verbin@intel.com>
3019             Kirill Yukhin  <kirill.yukhin@intel.com>
3020             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3022         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3023         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3024         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3025         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3026         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3027         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3028         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3029         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3030         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3031         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3032         OPT_mavx512pf, OPT_mavx512er cases.
3033         * config/i386/constraints.md (v): New constraint.
3034         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3035         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3036         (bit_AVX512CD): New.
3037         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3038         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3039         * config/i386/i386-c.c (ix86_target_macros_internal):
3040         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3041         __AVX512PF__.
3042         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3043         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3044         * config/i386/i386.c (regclass_map, dbx_register_map)
3045         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3046         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3047         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3048         -mavx512pf options.
3049         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3050         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3051         -mavx512pf options.  Fix formatting.
3052         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3053         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3054         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3055         -mavx512cd, -mavx512pf options.
3056         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3057         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3058         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3059         ALL_SSE_REGS.
3060         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3061         (ix86_set_reg_reg_cost): Ditto.
3062         (x86_order_regs_for_local_alloc): Ditto.
3063         (MAX_VECT_LEN): Extend to 64-byte.
3064         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3065         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3066         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3067         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3068         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3069         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3070         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3071         (SSE_REG_MODE_P): Support new modes.
3072         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3073         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3074         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3075         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3076         (REG_CLASS_CONTENTS): Add new registers.
3077         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3078         (EXT_REX_SSE_REGNO_P): New.
3079         (HI_REGISTER_NAMES): Add new registers.
3080         * config/i386/i386.md: Define constants for new registers.
3081         (mode): Add new 512-bit modes.
3082         (prefix): Support evex prefix.
3083         (isa): Support avx512f, noavx512f, fma_avx512f.
3084         (ssemodesuffix): Add new 512-bit modes.
3085         (movxi): New.
3086         (*movxi_internal_avx512f): Ditto.
3087         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3088         Support MODE_XI.
3089         (*movsi_internal): Likewise.
3090         (*movdf_internal): Likewise.
3091         (*movsf_internal): Likewise.
3092         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3093         (<code><mode>3): Likewise.
3094         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3095         New.
3096         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3097         with the new constraint "v".
3098         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3099         modes.
3100         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3101         with the new constraint "v".
3102         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3103         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3104         (*<plusminus_insn><mode>3): Likewise.
3105         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3106         (*mul<mode>3): Likewise.
3107         (<sse>_vmmul<mode>3): Likewise.
3108         (<sse>_div<mode>3): Likewise.
3109         (<sse>_vmdiv<mode>3): Likewise.
3110         (<sse>_sqrt<mode>2): Likewise.
3111         (<sse>_vmsqrt<mode>2): Likewise.
3112         (*<code><mode>3_finite): Likewise.
3113         (*<code><mode>3) <smaxmin>: Likewise.
3114         (<sse>_vm<code><mode>3): Likewise.
3115         (*<code><mode>3) <any_logic>: Likewise.
3116         (*fma_fmadd_<mode>): Likewise.
3117         (*fma_fmsub_<mode>): Likewise.
3118         (*fma_fnmadd_<mode>): Likewise.
3119         (*fma_fnmsub_<mode>): Likewise.
3120         (*fma_fmaddsub_<mode>): Likewise.
3121         (*fma_fmsubadd_<mode>): Likewise.
3122         (*fmai_fmadd_<mode>): Likewise.
3123         (*fmai_fmsub_<mode>): Likewise.
3124         (*fmai_fnmadd_<mode>): Likewise.
3125         (*fmai_fnmsub_<mode>): Likewise.
3126         (sse_cvtsi2ss): Likewise.
3127         (sse_cvtsi2ssq): Likewise.
3128         (sse_cvtss2si): Likewise.
3129         (sse_cvtss2si_2): Likewise.
3130         (sse_cvtss2siq): Likewise.
3131         (sse_cvtss2siq_2): Likewise.
3132         (sse_cvttss2si): Likewise.
3133         (sse_cvtss2siq_2): Likewise.
3134         (float<sseintvecmodelower><mode>2): Likewise.
3135         (sse2_cvtsd2si_2): Likewise.
3136         (sse2_cvtsd2siq_2): Likewise.
3137         (*<plusminus_insn><mode>3): Likewise.
3138         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3139         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3140         (ashr<mode>3): Likewise.
3141         (<shift_insn><mode>3): Likewise.
3142         (avx2_<code><mode>3): Likewise.
3143         (*avx2_<code><mode>3): Likewise.
3144         (*andnot<mode>3): Likewise.
3145         (*<code><mode>3) <any_logic>: Likewise.
3146         (abs<mode>2): Likewise.
3147         (avx2_permvar<mode>): Likewise.
3148         (avx2_perm<mode>_1): Likewise.
3149         (*avx_vpermilp<mode>): Likewise.
3150         (avx_vpermilvar<mode>3): Likewise.
3151         (avx2_ashrv<mode>): Likewise.
3152         (avx2_<shift_insn>v<mode>): Likewise.
3153         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3154         -mavx512cd.
3155         * doc/rtl.texi: Document XImode.
3157 2013-08-21  Jeff Law  <law@redhat.com>
3159         * tree-flow.h (register_jump_thread): Pass vector of edges
3160         instead of each important edge.
3161         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3162         thread path into a vector and pass that to register_jump_thread.
3163         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3164         passed in edge vector to the current 3-edge form.
3166         Revert:
3167         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3169         * dce.c (fini_dce): Call df_analyze again just in case
3170         delete_unmarked_insns removed anything.
3172 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3174         * reload.h (struct reg_equivs): Rename to ..
3175         (struct reg_equivs_s): .. this.
3177 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3179         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3181 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3183         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3185 2013-08-21  Jeff Law  <law@redhat.com>
3187         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3188         simplify assignments too.  If the RHS collapses to a singleton
3189         range, then return the value for the range.
3191 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3193         * config/i386/sse.md (V16): Rename to...
3194         (VMOVE): this.
3195         (mov<mode>): Update iterator name.
3196         (*mov<mode>_internal): Ditto.
3197         (push<mode>1): Ditto.
3198         (movmisalign<mode>): Ditto.
3200 2013-08-20  Jan Hubicka  <jh@suse.cz>
3202         PR bootstrap/58186
3203         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3204         entry for direct edges.
3205         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3207 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3209         Revert my last two changes, r201865 and r201864:
3211         Revert r201865:
3212         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3214         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3215         instances can own GC refs.
3217         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3218         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3219         (gcc::context::gt_pch_nx): Likewise.
3220         (gcc::context::gt_pch_nx):  Likewise.
3221         * ggc.h (gt_ggc_mx <T>): New.
3222         (gt_pch_nx_with_op <T>): New.
3223         (gt_pch_nx <T>): New.
3224         * passes.c (opt_pass::gt_ggc_mx): New.
3225         (opt_pass::gt_pch_nx): New.
3226         (opt_pass::gt_pch_nx_with_op): New.
3227         (pass_manager::gt_ggc_mx): New.
3228         (pass_manager::gt_pch_nx): New.
3229         (pass_manager::gt_pch_nx_with_op): New.
3230         (pass_manager::operator new): Use
3231         ggc_internal_cleared_alloc_stat rather than xcalloc.
3232         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3233         (pass_manager::gt_ggc_mx): New.
3234         (pass_manager::gt_pch_nx): New.
3235         (pass_manager::gt_pch_nx_with_op): New.
3236         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3237         (opt_pass::operator new): New.
3238         (opt_pass::gt_ggc_mx): New.
3239         (opt_pass::gt_pch_nx): New.
3240         (opt_pass::gt_pch_nx_with_op): New.
3242         Revert r201864:
3243         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3245         * Makefile.in (GTFILES): Add context.h.
3246         * context.c (gcc::context::operator new): New.
3247         (gcc::context::gt_ggc_mx): New.
3248         (gcc::context::gt_pch_nx): New.
3249         (gcc::context::gt_pch_nx): New.
3250         * context.h (gcc::context): Add GTY((user)) marking.
3251         (gcc::context::operator new): New.
3252         (gcc::context::gt_ggc_mx): New.
3253         (gcc::context::gt_pch_nx): New.
3254         (gcc::context::gt_pch_nx): New.
3255         (g): Add GTY marking.
3256         (gt_ggc_mx (gcc::context *)): New.
3257         (gt_pch_nx (gcc::context *)): New.
3258         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3259         void *cookie)): New.
3260         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3262 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3264         * dce.c (fini_dce): Call df_analyze again just in case
3265         delete_unmarked_insns removed anything.
3267 2013-08-20  Teresa Johnson  <tejohnson@google.com>
3269         PR rtl-optimizations/57451
3270         * final.c (reemit_insn_block_notes): Prevent lexical blocks
3271         from crossing split section boundaries.
3273 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3275         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3276         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3277         with MULTLIB_DEFAULTS.
3279 2013-08-20  Nick Clifton  <nickc@redhat.com>
3281         * target.def (narrow_volatile_bitfield): Note that the default
3282         value is false, not !TARGET_STRICT_ALIGN.
3283         * doc/tm.texi: Regenerate.
3285 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
3287         Fix LIB_SPEC for systems without libpthread.
3289         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3290         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3291         for Android.
3292         * config/i386/linux-common.h: Likewise.
3293         * config/mips/linux-common.h: Likewise.
3295 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
3297         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3298         checks.
3300 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3302         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3303         instances can own GC refs.
3305         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3306         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3307         (gcc::context::gt_pch_nx): Likewise.
3308         (gcc::context::gt_pch_nx):  Likewise.
3309         * ggc.h (gt_ggc_mx <T>): New.
3310         (gt_pch_nx_with_op <T>): New.
3311         (gt_pch_nx <T>): New.
3312         * passes.c (opt_pass::gt_ggc_mx): New.
3313         (opt_pass::gt_pch_nx): New.
3314         (opt_pass::gt_pch_nx_with_op): New.
3315         (pass_manager::gt_ggc_mx): New.
3316         (pass_manager::gt_pch_nx): New.
3317         (pass_manager::gt_pch_nx_with_op): New.
3318         (pass_manager::operator new): Use
3319         ggc_internal_cleared_alloc_stat rather than xcalloc.
3320         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3321         (pass_manager::gt_ggc_mx): New.
3322         (pass_manager::gt_pch_nx): New.
3323         (pass_manager::gt_pch_nx_with_op): New.
3324         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3325         (opt_pass::operator new): New.
3326         (opt_pass::gt_ggc_mx): New.
3327         (opt_pass::gt_pch_nx): New.
3328         (opt_pass::gt_pch_nx_with_op): New.
3330 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3332         * Makefile.in (GTFILES): Add context.h.
3333         * context.c (gcc::context::operator new): New.
3334         (gcc::context::gt_ggc_mx): New.
3335         (gcc::context::gt_pch_nx): New.
3336         (gcc::context::gt_pch_nx): New.
3337         * context.h (gcc::context): Add GTY((user)) marking.
3338         (gcc::context::operator new): New.
3339         (gcc::context::gt_ggc_mx): New.
3340         (gcc::context::gt_pch_nx): New.
3341         (gcc::context::gt_pch_nx): New.
3342         (g): Add GTY marking.
3343         (gt_ggc_mx (gcc::context *)): New.
3344         (gt_pch_nx (gcc::context *)): New.
3345         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3346         void *cookie)): New.
3347         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3349 2013-08-20  Alan Modra  <amodra@gmail.com>
3351         PR target/57865
3352         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3353         (rs6000_emit_epilogue): Likewise.
3355 2013-08-19  Dehao Chen  <dehao@google.com>
3357         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3359 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
3360             Jakub Jelinek  <jakub@redhat.com>
3362         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3363         (BUILT_IN_FABSD64): Likewise.
3364         (BUILT_IN_FABSD128): Likewise.
3365         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3366         (fold_builtin_1): Likewise.
3367         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3368         destination and source operands.
3369         (*abstd2_fpr): Likewise.
3370         (*nabstd2_fpr): Likewise.
3372 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
3374         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3375         JUMP_P and INSN_P.
3377 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
3379         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3380         incomplete.
3382 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3384         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3385         * builtins.c (default_libc_has_function): New.
3386         (gnu_libc_has_function): Ditto.
3387         (no_c99_libc_has_function): Ditto.
3388         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3389         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3390         (fold_builtin_sincos): Likewise.
3391         (fold_builtin_cexp): Likewise.
3392         * builtins.def (DEF_C94_BUILTIN): Likewise.
3393         (DEF_C99_BUILTIN): Likewise.
3394         (DEF_C99_C90RES_BUILTIN): Likewise.
3395         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3396         definitions to using this define.
3397         * config/darwin-protos.h (darwin_libc_has_function): New.
3398         * config/darwin.c (darwin_libc_has_function): Ditto.
3399         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3400         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3401         * config/darwin.h: Ditto.
3402         * config/elfos.h: Ditto.
3403         * config/freebsd.h: Ditto.
3404         * config/i386/cygming.h: Ditto.
3405         * config/i386/djgpp.h: Ditto.
3406         * config/i386/i386-interix.h: Ditto.
3407         * config/microblaze/microblaze.h: Ditto.
3408         * config/mmix/mmix.h: Ditto.
3409         * config/gnu-user.h: Ditto.
3410         * config/ia64/hpux.h: Ditto.
3411         * config/pa/pa-hpux.h: Ditto.
3412         * config/pdp11/pdp11.h: Ditto.
3413         * config/picochip/picochip.h: Ditto.
3414         * config/linux.h: Ditto.
3415         * config/netbsd.h: Ditto.
3416         * config/openbsd.h: Ditto.
3417         * config/rs6000/aix43.h: Ditto.
3418         * config/rs6000/aix51.h: Ditto.
3419         * config/rs6000/aix52.h: Ditto.
3420         * config/rs6000/aix53.h: Ditto.
3421         * config/rs6000/aix61.h: Ditto.
3422         * config/rs6000/darwin.h: Ditto.
3423         * config/rs6000/linux.h: Ditto.
3424         * config/rs6000/linux64.h: Ditto.
3425         * config/s390/tpf.h: Ditto.
3426         * config/sol2-10.h: Ditto.
3427         * config/sol2.h: Ditto.
3428         * config/vms/vms.h: Ditto.
3429         * config/vxworks.h: Ditto.
3430         * config/linux-android.c (linux_android_libc_has_function):
3431         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3432         * config/linux-protos.h (linux_android_libc_has_function):
3433         New declaration.
3434         * config/i386/i386.c (ix86_libc_has_function): New.
3435         * config/i386/i386-protos.h
3436         (ix86_libc_has_function): New declaration.
3437         * config/i386/i386.md
3438         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3439         ("isinf<mode>2): Likewise.
3440         * convert.c (convert_to_integer): Using new target hook
3441         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3442         TARGET_C99_FUNCTIONS.
3443         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3444         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3445         * coretypes.h (function_class): New enum for different
3446         classes of functions.
3447         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3448         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3449         (TARGET_HAS_SINCOS): Likewise.
3450         (TARGET_LIBC_HAS_FUNCTION): New.
3451         * doc/tm.texi: Regenerated.
3452         * targhooks.h (default_libc_has_function): New declaration.
3453         (no_c99_libc_has_function): Ditto.
3454         (gnu_libc_has_function): Ditto.
3455         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3456         and TARGET_HAS_SINCOS.
3458 2013-08-18  Jan Hubicka  <jh@suse.cz>
3460         * Makeifle-in (ipa-devirt.o): New.
3461         (GTFILES): Add ipa-utils.h and ipa-devirt.c
3462         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3463         (analyze_functions): Look into possible targets of polymorphic call.
3464         * dumpfile.c (dump_files): Add type-inheritance dump.
3465         * dumpfile.h (TDI_inheritance): New.
3466         * ipa-devirt.c: New file.
3467         * ipa-utils.h (odr_type_d): Forward declare.
3468         (odr_type): New type.
3469         (build_type_inheritance_graph): Declare.
3470         (possible_polymorphic_call_targets): Declare and introduce inline
3471         variant when only edge is pased.
3472         (dump_possible_polymorphic_call_targets): Likewise.
3473         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3474         * tree.c (type_in_anonymous_namespace_p): Break out from ...
3475         (types_same_for_odr): ... here.
3476         * tree.h (type_in_anonymous_namespace_p): Declare.
3478 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
3480         PR tree-optimization/58006
3481         * tree-parloops.c (take_address_of): Don't ICE if get_name
3482         returns NULL.
3483         (eliminate_local_variables_stmt): Remove clobber stmts.
3485 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
3487         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3488         error message is issued for an alias to undefined symbol.
3490 2013-08-18  Jan Hubicka  <jh@suse.cz>
3492         * cgraph.c (cgraph_create_indirect_edge): Discover
3493         polymorphic calls and record basic info into indirect_info.
3494         * gimple-fold.c (gimple_fold_call): When doing BINFO based
3495         devirtualization, ignore objc function calls.
3496         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3497         call with no parm index info.
3498         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3499         * tree.c (virtual_method_call_p): New function.
3500         * tree.h (virtual_method_call_p): Declare.
3502 2013-08-16  Jan Hubicka  <jh@suse.cz>
3504         PR middle-end/58179
3505         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3507 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
3509         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3510         attribute.
3512 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3514         * gengtype.c (type_for_name): Add special-case support for
3515         locating types within the "gcc::" namespace.
3516         (open_base_files): Emit a "using namespace gcc" directive.
3518 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3520         PR target/58160
3521         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3522         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3524         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3525         array instead of each individual operand as a separate argument.
3526         (emit_fusion_gpr_load): Likewise.
3527         (expand_fusion_gpr_load): Add new function declaration.
3529         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3530         signature to have the operands passed as an array, instead of as
3531         separate arguments.  Allow ZERO_EXTEND to be in the memory
3532         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
3533         depend on the register live/dead flags when peepholes are run.
3534         (expand_fusion_gpr_load): New function to be called from the
3535         peephole2 pass, to change the register that addis sets to be the
3536         target register.
3537         (emit_fusion_gpr_load): Change the calling signature to have the
3538         operands passed as an array, instead of as separate arguments.
3539         Allow ZERO_EXTEND to be in the memory address, and also
3540         SIGN_EXTEND if -mpower8-fusion-sign.
3542         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3543         unspec enumeration.
3544         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3545         adjust the register addis loads up in the peephole2 pass.  Do not
3546         depend on the register live/dead state when the peephole pass is done.
3548 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3550         * gengtype.c (create_user_defined_type): Ensure that the kind
3551         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3552         declaration is seen before the GTY((user)) marking.
3554 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3556         PR target/58105
3557         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
3559 2013-08-16  Jan Hubicka  <jh@suse.cz>
3561         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
3562         arugment expected_type.
3563         (gimple_fold_call): Use it.
3564         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
3565         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
3566         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
3567         (try_make_edge_direct_virtual_call): Likewise.
3568         * tree.c (obj_type_ref_class): New.
3569         * tree.h (obj_type_ref_class): Use it.
3571 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3573         * sched-vis.c (rtl_slim_pp_initialized): Remove.
3574         (rtl_slim_pp): Likewise.
3575         (init_rtl_slim_pretty_print): Likewise.
3576         (dump_value_slim):  Don't call it.  Use local pretty printer.
3577         (dump_insn_slim): Likewise.
3578         (dump_rtl_slim): Likewise.
3579         (str_pattern_slim): Likewise.
3580         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
3581         Simplify.
3583 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
3585         PR tree-optimization/58164
3586         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
3587         walk gimple_goto_dest of GIMPLE_GOTO.
3589         PR tree-optimization/58165
3590         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
3591         bi_call must be the last stmt in a bb, don't split_block, instead
3592         use fallthru edge from it and give up if there is none.
3593         Release conds vector when returning early.
3595 2013-08-14  Xinliang David Li  <davidxl@google.com>
3597         * config/i386/i386.c (ix86_option_override_internal):
3598         Remove unused variable and field.
3600 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3602         PR target/57949
3603         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
3604         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
3605         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
3606         and Linux, correct BLKmode alignment when 128-bit alignment is
3607         required and compatibility flag is not set.
3608         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
3609         for zero-size arguments when compatibility flag is not set.
3611 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3613         PR tree-optimization/58145
3614         * tree-sra.c (build_ref_for_offset): If prev_base has
3615         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
3617 2013-08-14  Xinliang David Li  <davidxl@google.com>
3619         * config/i386/i386.c (ix86_option_override_internal):
3620         Fix uninitialized variable error.
3622 2013-08-14  Xinliang David Li  <davidxl@google.com>
3624         * config/i386/i386.opt: Define two new options.
3625         * config/i386/x86-tune.def: Add arch selector field in macros.
3626         * config/i386/i386.h: Adjust macro definition.
3627         * config/i386/i386.c (ix86_option_override_internal):
3628         Refactor the code.
3629         (parse_mtune_ctrl_str): New function.
3630         (set_ix86_tune_features): New function.
3631         (ix86_function_specific_restore): Call the new helper function.
3633 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
3635         PR rtl-optimization/57662
3636         * sel-sched.c (code_motion_process_successors): When the current insn
3637         is removed after the recursive traversal, break from the loop.
3638         Add comments and debug printouts.
3640 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3641             Alexandre Oliva  <aoliva@redhat.com>
3643         PR target/58067
3644         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
3645         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
3646         there also UNSPEC_PLTOFF.
3648 2013-08-14  Marek Polacek  <polacek@redhat.com>
3650         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
3651         NUM_CONDITIONS bit positions.
3653 2013-08-13  Cary Coutant  <ccoutant@google.com>
3655         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
3656         (attr_checksum): Hash vector contents instead of pointer.
3657         (attr_checksum_ordered): Likewise.
3659 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
3661         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
3662         when Pmode != word_mode.  Add length_address attribute.
3663         (sse3_monitor_<mode>): Merge from sse3_monitor and
3664         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
3665         Pmode != word_mode.  Update insn length attribute.
3666         * config/i386/i386.c (ix86_option_override_internal): Update
3667         ix86_gen_monitor selection for merged sse3_monitor insn.
3669 2013-08-13  Julian Brown  <julian@codesourcery.com>
3671         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
3672         perform invalid legitimization on greater-than-word-size modes for
3673         TARGET_E500_DOUBLE.
3675 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
3677         * ira.c (setup_class_translate_array): Use aclass instead of cl
3678         for classes not fully covered by allocno classes.
3680 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
3682         PR tree-optimization/57661
3683         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
3684         * tree-inline.c (tree_function_versioning): Initialize it.
3685         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
3686         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
3687         that is not being copied.
3689         PR sanitizer/56417
3690         * asan.c (instrument_strlen_call): Fix typo in comment.
3691         Use char * type even for the lhs of POINTER_PLUS_EXPR.
3693 2013-08-13  Steve Ellcey  <sellcey@mips.com>
3695         * config/mips/mips.md (prefetch): Use lw instead of ld on
3696         loongson in 32bit mode.
3698 2013-08-13  Nick Clifton  <nickc@redhat.com>
3700         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
3702 2013-08-13  Jan Hubicka  <jh@suse.cz>
3704         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
3705         introduced edge; fix typo in sanity check.
3706         (cgraph_resolve_speculation): Export; improve diagnostic.
3707         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
3708         speculation at type mismatch.
3709         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
3710         (cgraph_resolve_speculation): Declare.
3711         (symtab_can_be_discarded): New function.
3712         * value-prof.c (gimple_ic_transform): Remove actual transform code.
3713         * ipa-inline-transform.c (speculation_removed): New global var.
3714         (clone_inlined_nodes): See if speculation can be removed.
3715         (inline_call): If speculations was removed, we growths may not match.
3716         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
3717         (speculation_useful_p): New function.
3718         (resolve_noninline_speculation): New function.
3719         (inline_small_functions): Resolve useless speculations.
3720         * ipa-inline.h (speculation_useful_p): Declare
3721         * ipa.c (can_replace_by_local_alias): Simplify.
3722         (ipa_profile): Produce speculative calls in non-lto, too;
3723         add simple cost model; produce local aliases.
3725 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
3727         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
3728         PASS_MANAGER_H.
3730 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
3732         * config/i386/i386.c (ix86_function_versions): Use error + inform.
3734 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
3736         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
3737         iterator instead of X87MODEF.
3739 2013-08-12  Perez Read  <netfirewall@gmail.com>
3741         PR target/58132
3742         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
3743         operand 0 for intel asm alternative.
3744         (*movabs<mode>_2): Ditto for operand 1.
3746 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
3748         * config/aarch64/arm_none.h
3749         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
3751 2013-08-12  Nick Clifton  <nickc@redhat.com>
3753         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
3755 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
3757         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
3758         expand for QI/HImode operand to produce more effictive code for
3759         unsigned char(short) --> float(double) conversion.
3761 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
3763         * doc/invoke.texi: Mention that -ftls-model does not force the final
3764         model.
3766 2013-08-12  Marek Polacek  <polacek@redhat.com>
3767             Marc Glisse  <marc.glisse@inria.fr>
3769         PR tree-optimization/57980
3770         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
3771         when creating -1 constant.
3773 2013-08-10  Jan Hubicka  <jh@suse.cz>
3775         Workaround binutils PR14342.
3776         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
3777         (gimple_init_edge_profiler): Likewise.
3778         (gimple_gen_ic_func_profiler): Likewise.
3780 2013-08-09  Jan Hubicka  <jh@suse.cz>
3782         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
3784 2013-08-09  Xinliang David Li  <davidxl@google.com>
3786         * config/i386/stringop.def: New file.
3787         * config/i386/stringop.opt: New file.
3788         * config/i386/i386-opts.h: Include stringopt.def.
3789         * config/i386/i386.opt: Include stringopt.opt.
3790         * config/i386/i386.c (ix86_option_override_internal):
3791         Override default size based stringop inline strategies with options.
3792         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
3793         New function.
3795 2013-08-09  Jan Hubicka  <jh@suse.cz>
3797         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
3799 2013-08-09  Jan Hubicka  <jh@suse.cz>
3801         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
3802         CGRAPH_FREQ_MAX.
3803         (dump_cgraph_node): Dump profile-id.
3804         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
3805         and common_target_probability.
3806         * lto-cgraph.c (lto_output_edge): Stream common targets.
3807         (lto_output_node): Stream profile ids.
3808         (input_node): Stream profile ids.
3809         (input_edge): Stream common targets.
3810         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
3811         * ipa.c: Include value-prof.h
3812         (ipa_profile_generate_summary): Turn indirect call statement histograms
3813         into common targets.
3814         (ipa_profile): Turn common targets into speculative edges.
3816 2013-08-09  Jan Hubicka  <jh@suse.cz>
3818         * cgraph.h (cgraph_node): Add profile_id.
3819         * value-prof.c (cgraph_node_map): Turn into pointer_map.
3820         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
3821         (del_node_map): Update.
3822         (find_func_by_funcdef_no): Replace by ...
3823         (find_func_by_profile_id): ... this one.
3824         (gimple_ic_transform): Do not remove useful histograms when
3825         speculation is not done; dump info when indirect call removal
3826         can happen at LTO.
3827         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
3828         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
3829         (__gcov_indirect_call_profiler_v2): .. this one.
3830         * profile.h (init_node_map): Update.
3831         * coverage.c (coverage_compute_profile_id): New function.
3832         * coverage.h (coverage_compute_profile_id): Declare.
3833         * tree-profile.c (init_ic_make_global_vars): Make
3834         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
3835         (gimple_init_edge_profiler): Update prototype of
3836         __gcov_indirect_call_profiler.
3837         (gimple_gen_ic_func_profiler): Simplify.
3838         (tree_profiling): Use init_node_map
3840 2013-08-09  Jan Hubicka  <jh@suse.cz>
3842         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
3843         non-speculative refs.
3844         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
3845         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
3846         (cgraph_set_call_stmt): Likewise.
3847         (cgraph_create_edge_1): Fix release checking compilatoin;
3848         clear lto_stmt_uid.
3849         (cgraph_free_edge): Free indirect info.
3850         (cgraph_turn_edge_to_speculative): New function.
3851         (cgraph_speculative_call_info): New function.
3852         (cgraph_make_edge_direct): Return direct edge; handle speculation.
3853         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
3854         (dump_cgraph_node): Dump speculation.
3855         (verify_edge_count_and_frequency): Accept speculative edges.
3856         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
3857         (verify_cgraph_node): Handle speculation.
3858         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
3859         (cgraph_set_call_stmt): Update prototype.
3860         (cgraph_make_edge_direct): Update prototype.
3861         (cgraph_speculative_call_info): Declare.
3862         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
3863         to change; update call of ipa_find_references.
3864         * ipa-ref.c (ipa_record_reference): Fix return value; clear
3865         lto_stmt_uid and speculative flags.
3866         (ipa_dump_references): Dump speculation.
3867         (ipa_clone_references): Clone speculative flag.
3868         (ipa_clone_referring): Likewise.
3869         (ipa_clone_ref): New function.
3870         (ipa_find_reference): Look into lto_stmt_uids
3871         (ipa_clear_stmts_in_references): Do not clear speculative calls.
3872         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
3873         (ipa_find_reference): Update declaration.
3874         (ipa_clone_ref): Declare.
3875         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
3876         stream speculative flag.
3877         (lto_output_ref): Stream statements uids and speculation.
3878         (input_ref): Likewise.
3879         (input_edge): Stream speuclation.
3880         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
3881         (cgraph_set_call_stmt_including_clones): Handle speculation.
3882         * ipa-inline.c (heap_edge_removal_hook): New function.
3883         (inline_small_functions): Register it.
3884         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
3885         also initialize refs.
3886         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
3887         edge to change.
3888         (try_make_edge_direct_simple_call): Likewise.
3889         (try_make_edge_direct_simple_call): Likewise.
3890         (update_indirect_edges_after_inlining): Likewise.
3891         (remove_described_reference): Look proper lto_stmt_uid.
3892         (propagate_controlled_uses): Likewise.
3893         (propagate_controlled_uses): Liekwise.
3894         * tree-inline.c (copy_bb): Copy speculative edges.
3895         (redirect_all_calls): New function.
3896         (copy_cfg_body): Do redirection after loop info is updated.
3897         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
3899 2013-08-09  Jan Hubicka  <jh@suse.cz>
3901         * lto-streamer-out.c (output_function): Renumber PHIs.
3902         * lto-streamer-in.c (input_function): Likewise.
3904 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
3906         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
3907         (get_lane_unsigned): Likewise.
3908         (dup_lane_scalar): Likewise.
3909         (get_lane): enable for VALL.
3910         * config/aarch64/aarch64-simd.md
3911         (aarch64_dup_lane_scalar<mode>): Remove.
3912         (aarch64_get_lane_signed<mode>): Likewise.
3913         (aarch64_get_lane_unsigned<mode>): Likewise.
3914         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
3915         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
3916         (aarch64_get_lane<mode>): Enable for all vector modes.
3917         (aarch64_get_lanedi): Remove misleading constraints.
3918         * config/aarch64/arm_neon.h
3919         (__aarch64_vget_lane_any): Define.
3920         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
3921         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
3922         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
3923         * config/aarch64/iterators.md (VDQQH): New.
3924         (VDQQHS): Likewise.
3925         (vwcore): Likewise.
3927 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
3929         * configure.ac: Add GAS check for LEON instructions on SPARC.
3930         * configure: Regenerate.
3931         * config.in: Likewise.
3932         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
3933         sparc*-*-* block.
3934         * config/sparc/sparc.opt (LEON, LEON3): New masks.
3935         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
3936         for LEON or LEON3.
3937         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
3938         (AS_LEON_FLAG): New macro.
3939         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
3940         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
3941         Deal with LEON and LEON3 for the memory model.
3942         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
3943         (atomic_compare_and_swap<mode>_1): Likewise.
3944         (*atomic_compare_and_swap<mode>_1): Likewise.
3946 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3948         * config/arm/neon.md (vcond): Fix floating-point vector
3949         comparisons against 0.
3951 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
3953         * lra-constraints.c (emit_spill_move): Remove assert.
3954         (process_alt_operands): Add more debugging
3955         output.  Increase reject for spilling into memory.  Decrease
3956         reject for reloading scratch.
3957         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
3959 2013-08-08  Steve Ellcey  <sellcey@mips.com>
3961         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
3962         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
3963         micromips incompatible.  Add nan2008.
3964         (MULTILIB_DIRNAMES): Add nan2008.
3965         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3966         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
3967         and micromips incompatible.  Add nan2008.
3968         (MULTILIB_DIRNAMES): Add nan2008.
3969         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3971 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
3973         PR rtl-optimization/58079
3974         * combine.c (combine_simplify_rtx): Avoid using SUBST if
3975         simplify_comparison has widened a comparison with an integer.
3977 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3979         * config/arm/neon.md (movmisalign<mode>): Disable when we
3980         don't allow unaligned accesses.
3981         (*movmisalign<mode>_neon_store): Likewise.
3982         (*movmisalign<mode>_neon_load): Likewise.
3983         (*movmisalign<mode>_neon_store): Likewise.
3984         (*movmisalign<mode>_neon_load): Likewise.
3986 2013-08-08  Jan Hubicka  <jh@suse.cz>
3988         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
3989         (make_pass_rebuild_cgraph_edges): Also clear references.
3990         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
3991         * ipa-inline-transform.c (inline_transform): Remove all references
3992         after inlining.
3993         * cgraphunit.c (expand_function): Remove all references after
3994         expansion.
3995         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
3996         (ipa_find_reference): Rewrite to iterator.
3997         (remove_stmt_references): Likewise.
3998         (ipa_clear_stmts_in_references): New function.
3999         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4000         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4001         clear references.
4002         * ipa-split.c (split_function): Remove references in split function.
4004 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4006         PR target/57431
4007         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4008         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4010 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4012         PR target/56979
4013         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4014         suggested mode for the assignment isn't compatible with the
4015         registers required.
4017 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4019         PR target/58065
4020         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4022 2013-08-07  Xinliang David Li  <davidxl@google.com>
4024         * config/i386/i386.opt: New option -mtune-ctrl=.
4025         * config/i386/x86-tune.def: New file.
4026         * config/i386/i386.h: include x86-tune.def.
4027         * config/i386/i386.c (ix86_option_override_internal):
4028         Parsing -mtune-ctrl= option and set tune features.
4030 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4032         PR other/12081
4033         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4034         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4035         to gen_2arg_fn_t.
4037 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4039         * rtl.h (update_alignments): Declare.
4040         * final.c (grow_label_align): New function extracted from...
4041         (shorten_branches): ...here.  Call it.
4042         (update_alignments): New function.
4043         * reorg.c (sibling_labels): New variable.
4044         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4045         the new label along with it onto the sibling_labels vector.
4046         (fill_simple_delay_slots): Adjust call to get_label_before.
4047         (fill_slots_from_thread): Likewise.
4048         (relax_delay_slots): Likewise.
4049         (make_return_insns): Likewise.
4050         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4052 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4054         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4055         document its semantics.
4056         (diagnostic_report_diagnostic): Adjust accordingly.
4058 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4060         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4061         (sparc_option_override): ...and port to new C++ pass API.
4062         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4064 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4066         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4068 2013-08-06  Caroline Tice  <cmtice@google.com>
4070         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4071         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4072         * tree-pass.h: Add pass_vtable_verify.
4073         * varasm.c (assemble_variable): Add code to properly set the comdat
4074         section and name for the .vtable_map_vars section.
4075         (assemble_vtyv_preinit_initializer): New function.
4076         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4077         LINK_ONCE flag.
4078         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4079         * vtable-verify.c: New file.
4080         * vtable-verify.h: New file.
4081         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4082         initialiation levels.
4083         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4084         * passes.def: Insert pass_vtable_verify.
4085         * aclocal.m4: Reorder includes.
4086         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4087         -fvtv-counts options.
4088         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4089         as appropriate, if -fvtable-verify=... is used.
4090         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4091         -fvtable-verify=... is used.
4092         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4093         (vtable-verify.o): Add new build rule.
4094         (GTFILES): Add vtable-verify.c to list.
4095         * common.opt (fvtable-verify=): New flag.
4096         (vtv_priority): Values for fvtable-verify= flag.
4097         (fvtv-counts): New flag.
4098         (fvtv-debug): New flag.
4099         * tree.h (save_vtable_map_decl): New extern function decl.
4101 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4103         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4104         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4105         (pass_data_rl78_devirt): ...new pass_data instance and...
4106         (make_pass_rl78_devirt): ...new function.
4107         (rl78_asm_file_start): Port pass registration to new C++ API.
4109 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4111         * coretypes.h (rtl_opt_pass): Add.
4112         (gcc::context): Add.
4113         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4114         (epiphany_init): Port to new C++ pass API.
4115         (epiphany_optimize_mode_switching): Likewise.
4116         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4117         (pass_manager::get_pass_mode_switching): New.
4118         (pass_manager::get_pass_peephole2): New.
4119         * mode-switching.c (pass_mode_switching): Add clone method.
4120         * recog.c (pass_peephole2): Add clone method.
4121         (pass_split_all_insns): Add clone method.
4123 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4125         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4126         (mips_option_override): ...here, porting to new C++ API for passes.
4128 2013-08-06  Jan Hubicka  <jh@suse.cz>
4130         * cgraph.c (cgraph_get_body): New function based on lto.c
4131         implementation.
4132         * cgraph.h (cgraph_get_body): Declare.
4133         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4134         LTO paths.
4135         * cgraphunit.c (expand_function): Get body prior expanding.
4136         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4137         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4138         really need.
4139         * passes.c (do_per_function_toporder): Get body.
4140         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4141         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4143 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4145         PR fortran/57987
4146         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4147         is not re-finalized.  Rename second parameter to no_collect.
4149 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4151         PR middle-end/58041
4152         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4153         MEM_REF has proper alignment information.
4155 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4157         PR other/12081
4158         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4159         class insn_gen_fn.
4160         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4161         rtx (*) (rtx, ...) with insn_gen_fn.
4162         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4163         insn_gen_fn::stored_funcptr.  Add initializer braces.
4165 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4167         Rewrite how instances of passes are cloned to remove assumptions
4168         about their sizes (thus allowing pass subclasses to have
4169         additional data fields, albeit non-GC-managed ones at this point).
4171         * passes.c (make_pass_instance): Now that passes have clone
4172         methods, rewrite this function to eliminate XNEW and memcpy
4173         calls that used hardcoded sizes.  Since this function no longer
4174         creates pass instances, rename it to...
4175         (add_pass_instance): ...this.  Document the old way that passes were
4176         numbered and flagged, and rework this function to continue using it.
4177         (next_pass_1): Add an initial_pass argument for use by
4178         add_pass_instance.
4179         (position_pass): When adding multiple instances of a pass, use
4180         the pass's clone method, rather than relying on the XNEW/memcpy
4181         within the former make_pass_instance (now add_pass_instance).
4182         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4183         the initial instance of the current pass within the pass manager.
4185 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4187         This is the automated part of the conversion of passes from C
4188         structs to C++ classes.
4190         Patch autogenerated by refactor_passes.py from
4191         https://github.com/davidmalcolm/gcc-refactoring-scripts
4192         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4194         * asan.c (pass_asan): Convert from a global struct to a subclass of
4195         gimple_opt_pass along with...
4196         (pass_data_asan): ...new pass_data instance and...
4197         (make_pass_asan): ...new function.
4198         (pass_asan_O0): Convert from a global struct to a subclass of
4199         gimple_opt_pass along with...
4200         (pass_data_asan_O0): ...new pass_data instance and...
4201         (make_pass_asan_O0): ...new function.
4202         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4203         subclass of rtl_opt_pass along with...
4204         (pass_data_inc_dec): ...new pass_data instance and...
4205         (make_pass_inc_dec): ...new function.
4206         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4207         a subclass of rtl_opt_pass along with...
4208         (pass_data_reorder_blocks): ...new pass_data instance and...
4209         (make_pass_reorder_blocks): ...new function.
4210         (pass_duplicate_computed_gotos): Convert from a global struct to a
4211         subclass of rtl_opt_pass along with...
4212         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4213         (make_pass_duplicate_computed_gotos): ...new function.
4214         (pass_partition_blocks): Convert from a global struct to a subclass of
4215         rtl_opt_pass along with...
4216         (pass_data_partition_blocks): ...new pass_data instance and...
4217         (make_pass_partition_blocks): ...new function.
4218         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4219         struct to a subclass of rtl_opt_pass along with...
4220         (pass_data_branch_target_load_optimize1): ...new pass_data instance
4221         and...
4222         (make_pass_branch_target_load_optimize1): ...new function.
4223         (pass_branch_target_load_optimize2): Convert from a global struct to a
4224         subclass of rtl_opt_pass along with...
4225         (pass_data_branch_target_load_optimize2): ...new pass_data instance
4226         and...
4227         (make_pass_branch_target_load_optimize2): ...new function.
4228         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4229         of rtl_opt_pass along with...
4230         (pass_data_jump): ...new pass_data instance and...
4231         (make_pass_jump): ...new function.
4232         (pass_jump2): Convert from a global struct to a subclass of
4233         rtl_opt_pass along with...
4234         (pass_data_jump2): ...new pass_data instance and...
4235         (make_pass_jump2): ...new function.
4236         * cfgexpand.c (pass_expand): Convert from a global struct to a
4237         subclass of rtl_opt_pass along with...
4238         (pass_data_expand): ...new pass_data instance and...
4239         (make_pass_expand): ...new function.
4240         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4241         of rtl_opt_pass along with...
4242         (pass_data_free_cfg): ...new pass_data instance and...
4243         (make_pass_free_cfg): ...new function.
4244         (pass_into_cfg_layout_mode): Convert from a global struct to a
4245         subclass of rtl_opt_pass along with...
4246         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4247         (make_pass_into_cfg_layout_mode): ...new function.
4248         (pass_outof_cfg_layout_mode): Convert from a global struct to a
4249         subclass of rtl_opt_pass along with...
4250         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4251         (make_pass_outof_cfg_layout_mode): ...new function.
4252         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4253         struct to a subclass of gimple_opt_pass along with...
4254         (pass_data_build_cgraph_edges): ...new pass_data instance and...
4255         (make_pass_build_cgraph_edges): ...new function.
4256         (pass_rebuild_cgraph_edges): Convert from a global struct to a
4257         subclass of gimple_opt_pass along with...
4258         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4259         (make_pass_rebuild_cgraph_edges): ...new function.
4260         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4261         subclass of gimple_opt_pass along with...
4262         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4263         and...
4264         (make_pass_remove_cgraph_callee_edges): ...new function.
4265         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4266         struct to a subclass of rtl_opt_pass along with...
4267         (pass_data_stack_adjustments): ...new pass_data instance and...
4268         (make_pass_stack_adjustments): ...new function.
4269         * combine.c (pass_combine): Convert from a global struct to a subclass
4270         of rtl_opt_pass along with...
4271         (pass_data_combine): ...new pass_data instance and...
4272         (make_pass_combine): ...new function.
4273         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4274         global struct to a subclass of rtl_opt_pass along with...
4275         (pass_data_compare_elim_after_reload): ...new pass_data instance
4276         and...
4277         (make_pass_compare_elim_after_reload): ...new function.
4278         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4279         of rtl_opt_pass along with...
4280         (pass_data_rtl_cprop): ...new pass_data instance and...
4281         (make_pass_rtl_cprop): ...new function.
4282         * cse.c (pass_cse): Convert from a global struct to a subclass of
4283         rtl_opt_pass along with...
4284         (pass_data_cse): ...new pass_data instance and...
4285         (make_pass_cse): ...new function.
4286         (pass_cse2): Convert from a global struct to a subclass of
4287         rtl_opt_pass along with...
4288         (pass_data_cse2): ...new pass_data instance and...
4289         (make_pass_cse2): ...new function.
4290         (pass_cse_after_global_opts): Convert from a global struct to a
4291         subclass of rtl_opt_pass along with...
4292         (pass_data_cse_after_global_opts): ...new pass_data instance and...
4293         (make_pass_cse_after_global_opts): ...new function.
4294         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4295         of rtl_opt_pass along with...
4296         (pass_data_ud_rtl_dce): ...new pass_data instance and...
4297         (make_pass_ud_rtl_dce): ...new function.
4298         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4299         rtl_opt_pass along with...
4300         (pass_data_fast_rtl_dce): ...new pass_data instance and...
4301         (make_pass_fast_rtl_dce): ...new function.
4302         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4303         a subclass of rtl_opt_pass along with...
4304         (pass_data_df_initialize_opt): ...new pass_data instance and...
4305         (make_pass_df_initialize_opt): ...new function.
4306         (pass_df_initialize_no_opt): Convert from a global struct to a
4307         subclass of rtl_opt_pass along with...
4308         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4309         (make_pass_df_initialize_no_opt): ...new function.
4310         (pass_df_finish): Convert from a global struct to a subclass of
4311         rtl_opt_pass along with...
4312         (pass_data_df_finish): ...new pass_data instance and...
4313         (make_pass_df_finish): ...new function.
4314         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4315         rtl_opt_pass along with...
4316         (pass_data_rtl_dse1): ...new pass_data instance and...
4317         (make_pass_rtl_dse1): ...new function.
4318         (pass_rtl_dse2): Convert from a global struct to a subclass of
4319         rtl_opt_pass along with...
4320         (pass_data_rtl_dse2): ...new pass_data instance and...
4321         (make_pass_rtl_dse2): ...new function.
4322         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4323         subclass of rtl_opt_pass along with...
4324         (pass_data_dwarf2_frame): ...new pass_data instance and...
4325         (make_pass_dwarf2_frame): ...new function.
4326         * except.c (pass_set_nothrow_function_flags): Convert from a global
4327         struct to a subclass of rtl_opt_pass along with...
4328         (pass_data_set_nothrow_function_flags): ...new pass_data instance
4329         and...
4330         (make_pass_set_nothrow_function_flags): ...new function.
4331         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4332         subclass of rtl_opt_pass along with...
4333         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4334         and...
4335         (make_pass_convert_to_eh_region_ranges): ...new function.
4336         * final.c (pass_compute_alignments): Convert from a global struct to a
4337         subclass of rtl_opt_pass along with...
4338         (pass_data_compute_alignments): ...new pass_data instance and...
4339         (make_pass_compute_alignments): ...new function.
4340         (pass_final): Convert from a global struct to a subclass of
4341         rtl_opt_pass along with...
4342         (pass_data_final): ...new pass_data instance and...
4343         (make_pass_final): ...new function.
4344         (pass_shorten_branches): Convert from a global struct to a subclass of
4345         rtl_opt_pass along with...
4346         (pass_data_shorten_branches): ...new pass_data instance and...
4347         (make_pass_shorten_branches): ...new function.
4348         (pass_clean_state): Convert from a global struct to a subclass of
4349         rtl_opt_pass along with...
4350         (pass_data_clean_state): ...new pass_data instance and...
4351         (make_pass_clean_state): ...new function.
4352         * function.c (pass_instantiate_virtual_regs): Convert from a global
4353         struct to a subclass of rtl_opt_pass along with...
4354         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4355         (make_pass_instantiate_virtual_regs): ...new function.
4356         (pass_leaf_regs): Convert from a global struct to a subclass of
4357         rtl_opt_pass along with...
4358         (pass_data_leaf_regs): ...new pass_data instance and...
4359         (make_pass_leaf_regs): ...new function.
4360         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4361         subclass of rtl_opt_pass along with...
4362         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4363         and...
4364         (make_pass_thread_prologue_and_epilogue): ...new function.
4365         (pass_match_asm_constraints): Convert from a global struct to a
4366         subclass of rtl_opt_pass along with...
4367         (pass_data_match_asm_constraints): ...new pass_data instance and...
4368         (make_pass_match_asm_constraints): ...new function.
4369         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4370         subclass of rtl_opt_pass along with...
4371         (pass_data_rtl_fwprop): ...new pass_data instance and...
4372         (make_pass_rtl_fwprop): ...new function.
4373         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4374         rtl_opt_pass along with...
4375         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4376         (make_pass_rtl_fwprop_addr): ...new function.
4377         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4378         rtl_opt_pass along with...
4379         (pass_data_rtl_pre): ...new pass_data instance and...
4380         (make_pass_rtl_pre): ...new function.
4381         (pass_rtl_hoist): Convert from a global struct to a subclass of
4382         rtl_opt_pass along with...
4383         (pass_data_rtl_hoist): ...new pass_data instance and...
4384         (make_pass_rtl_hoist): ...new function.
4385         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4386         subclass of gimple_opt_pass along with...
4387         (pass_data_lower_cf): ...new pass_data instance and...
4388         (make_pass_lower_cf): ...new function.
4389         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4390         from a global struct to a subclass of gimple_opt_pass along with...
4391         (pass_data_strength_reduction): ...new pass_data instance and...
4392         (make_pass_strength_reduction): ...new function.
4393         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4394         of rtl_opt_pass along with...
4395         (pass_data_rtl_ifcvt): ...new pass_data instance and...
4396         (make_pass_rtl_ifcvt): ...new function.
4397         (pass_if_after_combine): Convert from a global struct to a subclass of
4398         rtl_opt_pass along with...
4399         (pass_data_if_after_combine): ...new pass_data instance and...
4400         (make_pass_if_after_combine): ...new function.
4401         (pass_if_after_reload): Convert from a global struct to a subclass of
4402         rtl_opt_pass along with...
4403         (pass_data_if_after_reload): ...new pass_data instance and...
4404         (make_pass_if_after_reload): ...new function.
4405         * init-regs.c (pass_initialize_regs): Convert from a global struct to
4406         a subclass of rtl_opt_pass along with...
4407         (pass_data_initialize_regs): ...new pass_data instance and...
4408         (make_pass_initialize_regs): ...new function.
4409         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4410         of ipa_opt_pass_d along with...
4411         (pass_data_ipa_cp): ...new pass_data instance and...
4412         (make_pass_ipa_cp): ...new function.
4413         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4414         global struct to a subclass of gimple_opt_pass along with...
4415         (pass_data_inline_parameters): ...new pass_data instance and...
4416         (make_pass_inline_parameters): ...new function.
4417         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4418         subclass of gimple_opt_pass along with...
4419         (pass_data_early_inline): ...new pass_data instance and...
4420         (make_pass_early_inline): ...new function.
4421         (pass_ipa_inline): Convert from a global struct to a subclass of
4422         ipa_opt_pass_d along with...
4423         (pass_data_ipa_inline): ...new pass_data instance and...
4424         (make_pass_ipa_inline): ...new function.
4425         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4426         struct to a subclass of gimple_opt_pass along with...
4427         (pass_data_local_pure_const): ...new pass_data instance and...
4428         (make_pass_local_pure_const): ...new function.
4429         (pass_ipa_pure_const): Convert from a global struct to a subclass of
4430         ipa_opt_pass_d along with...
4431         (pass_data_ipa_pure_const): ...new pass_data instance and...
4432         (make_pass_ipa_pure_const): ...new function.
4433         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4434         to a subclass of ipa_opt_pass_d along with...
4435         (pass_data_ipa_reference): ...new pass_data instance and...
4436         (make_pass_ipa_reference): ...new function.
4437         * ipa-split.c (pass_split_functions): Convert from a global struct to
4438         a subclass of gimple_opt_pass along with...
4439         (pass_data_split_functions): ...new pass_data instance and...
4440         (make_pass_split_functions): ...new function.
4441         (pass_feedback_split_functions): Convert from a global struct to a
4442         subclass of gimple_opt_pass along with...
4443         (pass_data_feedback_split_functions): ...new pass_data instance and...
4444         (make_pass_feedback_split_functions): ...new function.
4445         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4446         global struct to a subclass of simple_ipa_opt_pass along with...
4447         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4448         instance and...
4449         (make_pass_ipa_function_and_variable_visibility): ...new function.
4450         (pass_ipa_free_inline_summary): Convert from a global struct to a
4451         subclass of simple_ipa_opt_pass along with...
4452         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4453         (make_pass_ipa_free_inline_summary): ...new function.
4454         (pass_ipa_whole_program_visibility): Convert from a global struct to a
4455         subclass of ipa_opt_pass_d along with...
4456         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4457         and...
4458         (make_pass_ipa_whole_program_visibility): ...new function.
4459         (pass_ipa_profile): Convert from a global struct to a subclass of
4460         ipa_opt_pass_d along with...
4461         (pass_data_ipa_profile): ...new pass_data instance and...
4462         (make_pass_ipa_profile): ...new function.
4463         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4464         ipa_opt_pass_d along with...
4465         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4466         (make_pass_ipa_cdtor_merge): ...new function.
4467         * ira.c (pass_ira): Convert from a global struct to a subclass of
4468         rtl_opt_pass along with...
4469         (pass_data_ira): ...new pass_data instance and...
4470         (make_pass_ira): ...new function.
4471         (pass_reload): Convert from a global struct to a subclass of
4472         rtl_opt_pass along with...
4473         (pass_data_reload): ...new pass_data instance and...
4474         (make_pass_reload): ...new function.
4475         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4476         subclass of rtl_opt_pass along with...
4477         (pass_data_cleanup_barriers): ...new pass_data instance and...
4478         (make_pass_cleanup_barriers): ...new function.
4479         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4480         of rtl_opt_pass along with...
4481         (pass_data_loop2): ...new pass_data instance and...
4482         (make_pass_loop2): ...new function.
4483         (pass_rtl_loop_init): Convert from a global struct to a subclass of
4484         rtl_opt_pass along with...
4485         (pass_data_rtl_loop_init): ...new pass_data instance and...
4486         (make_pass_rtl_loop_init): ...new function.
4487         (pass_rtl_loop_done): Convert from a global struct to a subclass of
4488         rtl_opt_pass along with...
4489         (pass_data_rtl_loop_done): ...new pass_data instance and...
4490         (make_pass_rtl_loop_done): ...new function.
4491         (pass_rtl_move_loop_invariants): Convert from a global struct to a
4492         subclass of rtl_opt_pass along with...
4493         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4494         (make_pass_rtl_move_loop_invariants): ...new function.
4495         (pass_rtl_unswitch): Convert from a global struct to a subclass of
4496         rtl_opt_pass along with...
4497         (pass_data_rtl_unswitch): ...new pass_data instance and...
4498         (make_pass_rtl_unswitch): ...new function.
4499         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4500         subclass of rtl_opt_pass along with...
4501         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4502         and...
4503         (make_pass_rtl_unroll_and_peel_loops): ...new function.
4504         (pass_rtl_doloop): Convert from a global struct to a subclass of
4505         rtl_opt_pass along with...
4506         (pass_data_rtl_doloop): ...new pass_data instance and...
4507         (make_pass_rtl_doloop): ...new function.
4508         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4509         a subclass of rtl_opt_pass along with...
4510         (pass_data_lower_subreg): ...new pass_data instance and...
4511         (make_pass_lower_subreg): ...new function.
4512         (pass_lower_subreg2): Convert from a global struct to a subclass of
4513         rtl_opt_pass along with...
4514         (pass_data_lower_subreg2): ...new pass_data instance and...
4515         (make_pass_lower_subreg2): ...new function.
4516         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4517         struct to a subclass of ipa_opt_pass_d along with...
4518         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4519         (make_pass_ipa_lto_gimple_out): ...new function.
4520         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4521         of ipa_opt_pass_d along with...
4522         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4523         (make_pass_ipa_lto_finish_out): ...new function.
4524         * mode-switching.c (pass_mode_switching): Convert from a global struct
4525         to a subclass of rtl_opt_pass along with...
4526         (pass_data_mode_switching): ...new pass_data instance and...
4527         (make_pass_mode_switching): ...new function.
4528         * modulo-sched.c (pass_sms): Convert from a global struct to a
4529         subclass of rtl_opt_pass along with...
4530         (pass_data_sms): ...new pass_data instance and...
4531         (make_pass_sms): ...new function.
4532         * omp-low.c (pass_expand_omp): Convert from a global struct to a
4533         subclass of gimple_opt_pass along with...
4534         (pass_data_expand_omp): ...new pass_data instance and...
4535         (make_pass_expand_omp): ...new function.
4536         (pass_lower_omp): Convert from a global struct to a subclass of
4537         gimple_opt_pass along with...
4538         (pass_data_lower_omp): ...new pass_data instance and...
4539         (make_pass_lower_omp): ...new function.
4540         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4541         of gimple_opt_pass along with...
4542         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4543         (make_pass_diagnose_omp_blocks): ...new function.
4544         * passes.c (pass_early_local_passes): Convert from a global struct to
4545         a subclass of simple_ipa_opt_pass along with...
4546         (pass_data_early_local_passes): ...new pass_data instance and...
4547         (make_pass_early_local_passes): ...new function.
4548         (pass_all_early_optimizations): Convert from a global struct to a
4549         subclass of gimple_opt_pass along with...
4550         (pass_data_all_early_optimizations): ...new pass_data instance and...
4551         (make_pass_all_early_optimizations): ...new function.
4552         (pass_all_optimizations): Convert from a global struct to a subclass
4553         of gimple_opt_pass along with...
4554         (pass_data_all_optimizations): ...new pass_data instance and...
4555         (make_pass_all_optimizations): ...new function.
4556         (pass_all_optimizations_g): Convert from a global struct to a subclass
4557         of gimple_opt_pass along with...
4558         (pass_data_all_optimizations_g): ...new pass_data instance and...
4559         (make_pass_all_optimizations_g): ...new function.
4560         (pass_rest_of_compilation): Convert from a global struct to a subclass
4561         of rtl_opt_pass along with...
4562         (pass_data_rest_of_compilation): ...new pass_data instance and...
4563         (make_pass_rest_of_compilation): ...new function.
4564         (pass_postreload): Convert from a global struct to a subclass of
4565         rtl_opt_pass along with...
4566         (pass_data_postreload): ...new pass_data instance and...
4567         (make_pass_postreload): ...new function.
4568         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
4569         subclass of rtl_opt_pass along with...
4570         (pass_data_gcse2): ...new pass_data instance and...
4571         (make_pass_gcse2): ...new function.
4572         * postreload.c (pass_postreload_cse): Convert from a global struct to
4573         a subclass of rtl_opt_pass along with...
4574         (pass_data_postreload_cse): ...new pass_data instance and...
4575         (make_pass_postreload_cse): ...new function.
4576         * predict.c (pass_profile): Convert from a global struct to a subclass
4577         of gimple_opt_pass along with...
4578         (pass_data_profile): ...new pass_data instance and...
4579         (make_pass_profile): ...new function.
4580         (pass_strip_predict_hints): Convert from a global struct to a subclass
4581         of gimple_opt_pass along with...
4582         (pass_data_strip_predict_hints): ...new pass_data instance and...
4583         (make_pass_strip_predict_hints): ...new function.
4584         * recog.c (pass_peephole2): Convert from a global struct to a subclass
4585         of rtl_opt_pass along with...
4586         (pass_data_peephole2): ...new pass_data instance and...
4587         (make_pass_peephole2): ...new function.
4588         (pass_split_all_insns): Convert from a global struct to a subclass of
4589         rtl_opt_pass along with...
4590         (pass_data_split_all_insns): ...new pass_data instance and...
4591         (make_pass_split_all_insns): ...new function.
4592         (pass_split_after_reload): Convert from a global struct to a subclass
4593         of rtl_opt_pass along with...
4594         (pass_data_split_after_reload): ...new pass_data instance and...
4595         (make_pass_split_after_reload): ...new function.
4596         (pass_split_before_regstack): Convert from a global struct to a
4597         subclass of rtl_opt_pass along with...
4598         (pass_data_split_before_regstack): ...new pass_data instance and...
4599         (make_pass_split_before_regstack): ...new function.
4600         (pass_split_before_sched2): Convert from a global struct to a subclass
4601         of rtl_opt_pass along with...
4602         (pass_data_split_before_sched2): ...new pass_data instance and...
4603         (make_pass_split_before_sched2): ...new function.
4604         (pass_split_for_shorten_branches): Convert from a global struct to a
4605         subclass of rtl_opt_pass along with...
4606         (pass_data_split_for_shorten_branches): ...new pass_data instance
4607         and...
4608         (make_pass_split_for_shorten_branches): ...new function.
4609         * ree.c (pass_ree): Convert from a global struct to a subclass of
4610         rtl_opt_pass along with...
4611         (pass_data_ree): ...new pass_data instance and...
4612         (make_pass_ree): ...new function.
4613         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
4614         subclass of rtl_opt_pass along with...
4615         (pass_data_stack_regs): ...new pass_data instance and...
4616         (make_pass_stack_regs): ...new function.
4617         (pass_stack_regs_run): Convert from a global struct to a subclass of
4618         rtl_opt_pass along with...
4619         (pass_data_stack_regs_run): ...new pass_data instance and...
4620         (make_pass_stack_regs_run): ...new function.
4621         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
4622         subclass of rtl_opt_pass along with...
4623         (pass_data_cprop_hardreg): ...new pass_data instance and...
4624         (make_pass_cprop_hardreg): ...new function.
4625         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
4626         subclass of rtl_opt_pass along with...
4627         (pass_data_reginfo_init): ...new pass_data instance and...
4628         (make_pass_reginfo_init): ...new function.
4629         * regmove.c (pass_regmove): Convert from a global struct to a subclass
4630         of rtl_opt_pass along with...
4631         (pass_data_regmove): ...new pass_data instance and...
4632         (make_pass_regmove): ...new function.
4633         * regrename.c (pass_regrename): Convert from a global struct to a
4634         subclass of rtl_opt_pass along with...
4635         (pass_data_regrename): ...new pass_data instance and...
4636         (make_pass_regrename): ...new function.
4637         * reorg.c (pass_delay_slots): Convert from a global struct to a
4638         subclass of rtl_opt_pass along with...
4639         (pass_data_delay_slots): ...new pass_data instance and...
4640         (make_pass_delay_slots): ...new function.
4641         (pass_machine_reorg): Convert from a global struct to a subclass of
4642         rtl_opt_pass along with...
4643         (pass_data_machine_reorg): ...new pass_data instance and...
4644         (make_pass_machine_reorg): ...new function.
4645         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
4646         of rtl_opt_pass along with...
4647         (pass_data_sched): ...new pass_data instance and...
4648         (make_pass_sched): ...new function.
4649         (pass_sched2): Convert from a global struct to a subclass of
4650         rtl_opt_pass along with...
4651         (pass_data_sched2): ...new pass_data instance and...
4652         (make_pass_sched2): ...new function.
4653         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
4654         to a subclass of rtl_opt_pass along with...
4655         (pass_data_stack_ptr_mod): ...new pass_data instance and...
4656         (make_pass_stack_ptr_mod): ...new function.
4657         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
4658         to a subclass of rtl_opt_pass along with...
4659         (pass_data_rtl_store_motion): ...new pass_data instance and...
4660         (make_pass_rtl_store_motion): ...new function.
4661         * tracer.c (pass_tracer): Convert from a global struct to a subclass
4662         of gimple_opt_pass along with...
4663         (pass_data_tracer): ...new pass_data instance and...
4664         (make_pass_tracer): ...new function.
4665         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
4666         to a subclass of gimple_opt_pass along with...
4667         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
4668         (make_pass_diagnose_tm_blocks): ...new function.
4669         (pass_lower_tm): Convert from a global struct to a subclass of
4670         gimple_opt_pass along with...
4671         (pass_data_lower_tm): ...new pass_data instance and...
4672         (make_pass_lower_tm): ...new function.
4673         (pass_tm_init): Convert from a global struct to a subclass of
4674         gimple_opt_pass along with...
4675         (pass_data_tm_init): ...new pass_data instance and...
4676         (make_pass_tm_init): ...new function.
4677         (pass_tm_mark): Convert from a global struct to a subclass of
4678         gimple_opt_pass along with...
4679         (pass_data_tm_mark): ...new pass_data instance and...
4680         (make_pass_tm_mark): ...new function.
4681         (pass_tm_edges): Convert from a global struct to a subclass of
4682         gimple_opt_pass along with...
4683         (pass_data_tm_edges): ...new pass_data instance and...
4684         (make_pass_tm_edges): ...new function.
4685         (pass_tm_memopt): Convert from a global struct to a subclass of
4686         gimple_opt_pass along with...
4687         (pass_data_tm_memopt): ...new pass_data instance and...
4688         (make_pass_tm_memopt): ...new function.
4689         (pass_ipa_tm): Convert from a global struct to a subclass of
4690         simple_ipa_opt_pass along with...
4691         (pass_data_ipa_tm): ...new pass_data instance and...
4692         (make_pass_ipa_tm): ...new function.
4693         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
4694         subclass of gimple_opt_pass along with...
4695         (pass_data_call_cdce): ...new pass_data instance and...
4696         (make_pass_call_cdce): ...new function.
4697         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
4698         subclass of gimple_opt_pass along with...
4699         (pass_data_build_cfg): ...new pass_data instance and...
4700         (make_pass_build_cfg): ...new function.
4701         (pass_split_crit_edges): Convert from a global struct to a subclass of
4702         gimple_opt_pass along with...
4703         (pass_data_split_crit_edges): ...new pass_data instance and...
4704         (make_pass_split_crit_edges): ...new function.
4705         (pass_warn_function_return): Convert from a global struct to a
4706         subclass of gimple_opt_pass along with...
4707         (pass_data_warn_function_return): ...new pass_data instance and...
4708         (make_pass_warn_function_return): ...new function.
4709         (pass_warn_function_noreturn): Convert from a global struct to a
4710         subclass of gimple_opt_pass along with...
4711         (pass_data_warn_function_noreturn): ...new pass_data instance and...
4712         (make_pass_warn_function_noreturn): ...new function.
4713         (pass_warn_unused_result): Convert from a global struct to a subclass
4714         of gimple_opt_pass along with...
4715         (pass_data_warn_unused_result): ...new pass_data instance and...
4716         (make_pass_warn_unused_result): ...new function.
4717         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
4718         a subclass of gimple_opt_pass along with...
4719         (pass_data_merge_phi): ...new pass_data instance and...
4720         (make_pass_merge_phi): ...new function.
4721         * tree-complex.c (pass_lower_complex): Convert from a global struct to
4722         a subclass of gimple_opt_pass along with...
4723         (pass_data_lower_complex): ...new pass_data instance and...
4724         (make_pass_lower_complex): ...new function.
4725         (pass_lower_complex_O0): Convert from a global struct to a subclass of
4726         gimple_opt_pass along with...
4727         (pass_data_lower_complex_O0): ...new pass_data instance and...
4728         (make_pass_lower_complex_O0): ...new function.
4729         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
4730         subclass of gimple_opt_pass along with...
4731         (pass_data_lower_eh): ...new pass_data instance and...
4732         (make_pass_lower_eh): ...new function.
4733         (pass_refactor_eh): Convert from a global struct to a subclass of
4734         gimple_opt_pass along with...
4735         (pass_data_refactor_eh): ...new pass_data instance and...
4736         (make_pass_refactor_eh): ...new function.
4737         (pass_lower_resx): Convert from a global struct to a subclass of
4738         gimple_opt_pass along with...
4739         (pass_data_lower_resx): ...new pass_data instance and...
4740         (make_pass_lower_resx): ...new function.
4741         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
4742         of gimple_opt_pass along with...
4743         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
4744         (make_pass_lower_eh_dispatch): ...new function.
4745         (pass_cleanup_eh): Convert from a global struct to a subclass of
4746         gimple_opt_pass along with...
4747         (pass_data_cleanup_eh): ...new pass_data instance and...
4748         (make_pass_cleanup_eh): ...new function.
4749         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
4750         to a subclass of simple_ipa_opt_pass along with...
4751         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
4752         (make_pass_ipa_lower_emutls): ...new function.
4753         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
4754         a subclass of gimple_opt_pass along with...
4755         (pass_data_if_conversion): ...new pass_data instance and...
4756         (make_pass_if_conversion): ...new function.
4757         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
4758         subclass of gimple_opt_pass along with...
4759         (pass_data_build_ssa): ...new pass_data instance and...
4760         (make_pass_build_ssa): ...new function.
4761         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
4762         global struct to a subclass of gimple_opt_pass along with...
4763         (pass_data_loop_distribution): ...new pass_data instance and...
4764         (make_pass_loop_distribution): ...new function.
4765         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
4766         subclass of gimple_opt_pass along with...
4767         (pass_data_mudflap_1): ...new pass_data instance and...
4768         (make_pass_mudflap_1): ...new function.
4769         (pass_mudflap_2): Convert from a global struct to a subclass of
4770         gimple_opt_pass along with...
4771         (pass_data_mudflap_2): ...new pass_data instance and...
4772         (make_pass_mudflap_2): ...new function.
4773         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
4774         subclass of gimple_opt_pass along with...
4775         (pass_data_mudflap_1): ...new pass_data instance and...
4776         (make_pass_mudflap_1): ...new function.
4777         (pass_mudflap_2): Convert from a global struct to a subclass of
4778         gimple_opt_pass along with...
4779         (pass_data_mudflap_2): ...new pass_data instance and...
4780         (make_pass_mudflap_2): ...new function.
4781         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
4782         gimple_opt_pass along with...
4783         (pass_data_nrv): ...new pass_data instance and...
4784         (make_pass_nrv): ...new function.
4785         (pass_return_slot): Convert from a global struct to a subclass of
4786         gimple_opt_pass along with...
4787         (pass_data_return_slot): ...new pass_data instance and...
4788         (make_pass_return_slot): ...new function.
4789         * tree-object-size.c (pass_object_sizes): Convert from a global struct
4790         to a subclass of gimple_opt_pass along with...
4791         (pass_data_object_sizes): ...new pass_data instance and...
4792         (make_pass_object_sizes): ...new function.
4793         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
4794         global struct to a subclass of gimple_opt_pass along with...
4795         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
4796         and...
4797         (make_pass_cleanup_cfg_post_optimizing): ...new function.
4798         (pass_fixup_cfg): Convert from a global struct to a subclass of
4799         gimple_opt_pass along with...
4800         (pass_data_fixup_cfg): ...new pass_data instance and...
4801         (make_pass_fixup_cfg): ...new function.
4802         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
4803         (make_pass_mudflap_1): ...new function.
4804         (pass_mudflap_2): Replace declaration with that of...
4805         (make_pass_mudflap_2): ...new function.
4806         (pass_asan): Replace declaration with that of...
4807         (make_pass_asan): ...new function.
4808         (pass_asan_O0): Replace declaration with that of...
4809         (make_pass_asan_O0): ...new function.
4810         (pass_tsan): Replace declaration with that of...
4811         (make_pass_tsan): ...new function.
4812         (pass_tsan_O0): Replace declaration with that of...
4813         (make_pass_tsan_O0): ...new function.
4814         (pass_lower_cf): Replace declaration with that of...
4815         (make_pass_lower_cf): ...new function.
4816         (pass_refactor_eh): Replace declaration with that of...
4817         (make_pass_refactor_eh): ...new function.
4818         (pass_lower_eh): Replace declaration with that of...
4819         (make_pass_lower_eh): ...new function.
4820         (pass_lower_eh_dispatch): Replace declaration with that of...
4821         (make_pass_lower_eh_dispatch): ...new function.
4822         (pass_lower_resx): Replace declaration with that of...
4823         (make_pass_lower_resx): ...new function.
4824         (pass_build_cfg): Replace declaration with that of...
4825         (make_pass_build_cfg): ...new function.
4826         (pass_early_tree_profile): Replace declaration with that of...
4827         (make_pass_early_tree_profile): ...new function.
4828         (pass_cleanup_eh): Replace declaration with that of...
4829         (make_pass_cleanup_eh): ...new function.
4830         (pass_sra): Replace declaration with that of...
4831         (make_pass_sra): ...new function.
4832         (pass_sra_early): Replace declaration with that of...
4833         (make_pass_sra_early): ...new function.
4834         (pass_early_ipa_sra): Replace declaration with that of...
4835         (make_pass_early_ipa_sra): ...new function.
4836         (pass_tail_recursion): Replace declaration with that of...
4837         (make_pass_tail_recursion): ...new function.
4838         (pass_tail_calls): Replace declaration with that of...
4839         (make_pass_tail_calls): ...new function.
4840         (pass_tree_loop): Replace declaration with that of...
4841         (make_pass_tree_loop): ...new function.
4842         (pass_tree_loop_init): Replace declaration with that of...
4843         (make_pass_tree_loop_init): ...new function.
4844         (pass_lim): Replace declaration with that of...
4845         (make_pass_lim): ...new function.
4846         (pass_tree_unswitch): Replace declaration with that of...
4847         (make_pass_tree_unswitch): ...new function.
4848         (pass_predcom): Replace declaration with that of...
4849         (make_pass_predcom): ...new function.
4850         (pass_iv_canon): Replace declaration with that of...
4851         (make_pass_iv_canon): ...new function.
4852         (pass_scev_cprop): Replace declaration with that of...
4853         (make_pass_scev_cprop): ...new function.
4854         (pass_empty_loop): Replace declaration with that of...
4855         (make_pass_empty_loop): ...new function.
4856         (pass_record_bounds): Replace declaration with that of...
4857         (make_pass_record_bounds): ...new function.
4858         (pass_graphite): Replace declaration with that of...
4859         (make_pass_graphite): ...new function.
4860         (pass_graphite_transforms): Replace declaration with that of...
4861         (make_pass_graphite_transforms): ...new function.
4862         (pass_if_conversion): Replace declaration with that of...
4863         (make_pass_if_conversion): ...new function.
4864         (pass_loop_distribution): Replace declaration with that of...
4865         (make_pass_loop_distribution): ...new function.
4866         (pass_vectorize): Replace declaration with that of...
4867         (make_pass_vectorize): ...new function.
4868         (pass_slp_vectorize): Replace declaration with that of...
4869         (make_pass_slp_vectorize): ...new function.
4870         (pass_complete_unroll): Replace declaration with that of...
4871         (make_pass_complete_unroll): ...new function.
4872         (pass_complete_unrolli): Replace declaration with that of...
4873         (make_pass_complete_unrolli): ...new function.
4874         (pass_parallelize_loops): Replace declaration with that of...
4875         (make_pass_parallelize_loops): ...new function.
4876         (pass_loop_prefetch): Replace declaration with that of...
4877         (make_pass_loop_prefetch): ...new function.
4878         (pass_iv_optimize): Replace declaration with that of...
4879         (make_pass_iv_optimize): ...new function.
4880         (pass_tree_loop_done): Replace declaration with that of...
4881         (make_pass_tree_loop_done): ...new function.
4882         (pass_ch): Replace declaration with that of...
4883         (make_pass_ch): ...new function.
4884         (pass_ccp): Replace declaration with that of...
4885         (make_pass_ccp): ...new function.
4886         (pass_phi_only_cprop): Replace declaration with that of...
4887         (make_pass_phi_only_cprop): ...new function.
4888         (pass_build_ssa): Replace declaration with that of...
4889         (make_pass_build_ssa): ...new function.
4890         (pass_build_alias): Replace declaration with that of...
4891         (make_pass_build_alias): ...new function.
4892         (pass_build_ealias): Replace declaration with that of...
4893         (make_pass_build_ealias): ...new function.
4894         (pass_dominator): Replace declaration with that of...
4895         (make_pass_dominator): ...new function.
4896         (pass_dce): Replace declaration with that of...
4897         (make_pass_dce): ...new function.
4898         (pass_dce_loop): Replace declaration with that of...
4899         (make_pass_dce_loop): ...new function.
4900         (pass_cd_dce): Replace declaration with that of...
4901         (make_pass_cd_dce): ...new function.
4902         (pass_call_cdce): Replace declaration with that of...
4903         (make_pass_call_cdce): ...new function.
4904         (pass_merge_phi): Replace declaration with that of...
4905         (make_pass_merge_phi): ...new function.
4906         (pass_split_crit_edges): Replace declaration with that of...
4907         (make_pass_split_crit_edges): ...new function.
4908         (pass_pre): Replace declaration with that of...
4909         (make_pass_pre): ...new function.
4910         (pass_profile): Replace declaration with that of...
4911         (make_pass_profile): ...new function.
4912         (pass_strip_predict_hints): Replace declaration with that of...
4913         (make_pass_strip_predict_hints): ...new function.
4914         (pass_lower_complex_O0): Replace declaration with that of...
4915         (make_pass_lower_complex_O0): ...new function.
4916         (pass_lower_complex): Replace declaration with that of...
4917         (make_pass_lower_complex): ...new function.
4918         (pass_lower_vector): Replace declaration with that of...
4919         (make_pass_lower_vector): ...new function.
4920         (pass_lower_vector_ssa): Replace declaration with that of...
4921         (make_pass_lower_vector_ssa): ...new function.
4922         (pass_lower_omp): Replace declaration with that of...
4923         (make_pass_lower_omp): ...new function.
4924         (pass_diagnose_omp_blocks): Replace declaration with that of...
4925         (make_pass_diagnose_omp_blocks): ...new function.
4926         (pass_expand_omp): Replace declaration with that of...
4927         (make_pass_expand_omp): ...new function.
4928         (pass_expand_omp_ssa): Replace declaration with that of...
4929         (make_pass_expand_omp_ssa): ...new function.
4930         (pass_object_sizes): Replace declaration with that of...
4931         (make_pass_object_sizes): ...new function.
4932         (pass_strlen): Replace declaration with that of...
4933         (make_pass_strlen): ...new function.
4934         (pass_fold_builtins): Replace declaration with that of...
4935         (make_pass_fold_builtins): ...new function.
4936         (pass_stdarg): Replace declaration with that of...
4937         (make_pass_stdarg): ...new function.
4938         (pass_early_warn_uninitialized): Replace declaration with that of...
4939         (make_pass_early_warn_uninitialized): ...new function.
4940         (pass_late_warn_uninitialized): Replace declaration with that of...
4941         (make_pass_late_warn_uninitialized): ...new function.
4942         (pass_cse_reciprocals): Replace declaration with that of...
4943         (make_pass_cse_reciprocals): ...new function.
4944         (pass_cse_sincos): Replace declaration with that of...
4945         (make_pass_cse_sincos): ...new function.
4946         (pass_optimize_bswap): Replace declaration with that of...
4947         (make_pass_optimize_bswap): ...new function.
4948         (pass_optimize_widening_mul): Replace declaration with that of...
4949         (make_pass_optimize_widening_mul): ...new function.
4950         (pass_warn_function_return): Replace declaration with that of...
4951         (make_pass_warn_function_return): ...new function.
4952         (pass_warn_function_noreturn): Replace declaration with that of...
4953         (make_pass_warn_function_noreturn): ...new function.
4954         (pass_cselim): Replace declaration with that of...
4955         (make_pass_cselim): ...new function.
4956         (pass_phiopt): Replace declaration with that of...
4957         (make_pass_phiopt): ...new function.
4958         (pass_forwprop): Replace declaration with that of...
4959         (make_pass_forwprop): ...new function.
4960         (pass_phiprop): Replace declaration with that of...
4961         (make_pass_phiprop): ...new function.
4962         (pass_tree_ifcombine): Replace declaration with that of...
4963         (make_pass_tree_ifcombine): ...new function.
4964         (pass_dse): Replace declaration with that of...
4965         (make_pass_dse): ...new function.
4966         (pass_nrv): Replace declaration with that of...
4967         (make_pass_nrv): ...new function.
4968         (pass_rename_ssa_copies): Replace declaration with that of...
4969         (make_pass_rename_ssa_copies): ...new function.
4970         (pass_sink_code): Replace declaration with that of...
4971         (make_pass_sink_code): ...new function.
4972         (pass_fre): Replace declaration with that of...
4973         (make_pass_fre): ...new function.
4974         (pass_check_data_deps): Replace declaration with that of...
4975         (make_pass_check_data_deps): ...new function.
4976         (pass_copy_prop): Replace declaration with that of...
4977         (make_pass_copy_prop): ...new function.
4978         (pass_vrp): Replace declaration with that of...
4979         (make_pass_vrp): ...new function.
4980         (pass_uncprop): Replace declaration with that of...
4981         (make_pass_uncprop): ...new function.
4982         (pass_return_slot): Replace declaration with that of...
4983         (make_pass_return_slot): ...new function.
4984         (pass_reassoc): Replace declaration with that of...
4985         (make_pass_reassoc): ...new function.
4986         (pass_rebuild_cgraph_edges): Replace declaration with that of...
4987         (make_pass_rebuild_cgraph_edges): ...new function.
4988         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
4989         (make_pass_remove_cgraph_callee_edges): ...new function.
4990         (pass_build_cgraph_edges): Replace declaration with that of...
4991         (make_pass_build_cgraph_edges): ...new function.
4992         (pass_local_pure_const): Replace declaration with that of...
4993         (make_pass_local_pure_const): ...new function.
4994         (pass_tracer): Replace declaration with that of...
4995         (make_pass_tracer): ...new function.
4996         (pass_warn_unused_result): Replace declaration with that of...
4997         (make_pass_warn_unused_result): ...new function.
4998         (pass_diagnose_tm_blocks): Replace declaration with that of...
4999         (make_pass_diagnose_tm_blocks): ...new function.
5000         (pass_lower_tm): Replace declaration with that of...
5001         (make_pass_lower_tm): ...new function.
5002         (pass_tm_init): Replace declaration with that of...
5003         (make_pass_tm_init): ...new function.
5004         (pass_tm_mark): Replace declaration with that of...
5005         (make_pass_tm_mark): ...new function.
5006         (pass_tm_memopt): Replace declaration with that of...
5007         (make_pass_tm_memopt): ...new function.
5008         (pass_tm_edges): Replace declaration with that of...
5009         (make_pass_tm_edges): ...new function.
5010         (pass_split_functions): Replace declaration with that of...
5011         (make_pass_split_functions): ...new function.
5012         (pass_feedback_split_functions): Replace declaration with that of...
5013         (make_pass_feedback_split_functions): ...new function.
5014         (pass_strength_reduction): Replace declaration with that of...
5015         (make_pass_strength_reduction): ...new function.
5016         (pass_ipa_lower_emutls): Replace declaration with that of...
5017         (make_pass_ipa_lower_emutls): ...new function.
5018         (pass_ipa_function_and_variable_visibility): Replace declaration with
5019         that of...
5020         (make_pass_ipa_function_and_variable_visibility): ...new function.
5021         (pass_ipa_tree_profile): Replace declaration with that of...
5022         (make_pass_ipa_tree_profile): ...new function.
5023         (pass_early_local_passes): Replace declaration with that of...
5024         (make_pass_early_local_passes): ...new function.
5025         (pass_ipa_whole_program_visibility): Replace declaration with that
5026         of...
5027         (make_pass_ipa_whole_program_visibility): ...new function.
5028         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5029         (make_pass_ipa_lto_gimple_out): ...new function.
5030         (pass_ipa_increase_alignment): Replace declaration with that of...
5031         (make_pass_ipa_increase_alignment): ...new function.
5032         (pass_ipa_inline): Replace declaration with that of...
5033         (make_pass_ipa_inline): ...new function.
5034         (pass_ipa_free_lang_data): Replace declaration with that of...
5035         (make_pass_ipa_free_lang_data): ...new function.
5036         (pass_ipa_free_inline_summary): Replace declaration with that of...
5037         (make_pass_ipa_free_inline_summary): ...new function.
5038         (pass_ipa_cp): Replace declaration with that of...
5039         (make_pass_ipa_cp): ...new function.
5040         (pass_ipa_reference): Replace declaration with that of...
5041         (make_pass_ipa_reference): ...new function.
5042         (pass_ipa_pure_const): Replace declaration with that of...
5043         (make_pass_ipa_pure_const): ...new function.
5044         (pass_ipa_pta): Replace declaration with that of...
5045         (make_pass_ipa_pta): ...new function.
5046         (pass_ipa_lto_finish_out): Replace declaration with that of...
5047         (make_pass_ipa_lto_finish_out): ...new function.
5048         (pass_ipa_tm): Replace declaration with that of...
5049         (make_pass_ipa_tm): ...new function.
5050         (pass_ipa_profile): Replace declaration with that of...
5051         (make_pass_ipa_profile): ...new function.
5052         (pass_ipa_cdtor_merge): Replace declaration with that of...
5053         (make_pass_ipa_cdtor_merge): ...new function.
5054         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5055         of...
5056         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5057         (pass_init_datastructures): Replace declaration with that of...
5058         (make_pass_init_datastructures): ...new function.
5059         (pass_fixup_cfg): Replace declaration with that of...
5060         (make_pass_fixup_cfg): ...new function.
5061         (pass_expand): Replace declaration with that of...
5062         (make_pass_expand): ...new function.
5063         (pass_instantiate_virtual_regs): Replace declaration with that of...
5064         (make_pass_instantiate_virtual_regs): ...new function.
5065         (pass_rtl_fwprop): Replace declaration with that of...
5066         (make_pass_rtl_fwprop): ...new function.
5067         (pass_rtl_fwprop_addr): Replace declaration with that of...
5068         (make_pass_rtl_fwprop_addr): ...new function.
5069         (pass_jump): Replace declaration with that of...
5070         (make_pass_jump): ...new function.
5071         (pass_jump2): Replace declaration with that of...
5072         (make_pass_jump2): ...new function.
5073         (pass_lower_subreg): Replace declaration with that of...
5074         (make_pass_lower_subreg): ...new function.
5075         (pass_cse): Replace declaration with that of...
5076         (make_pass_cse): ...new function.
5077         (pass_fast_rtl_dce): Replace declaration with that of...
5078         (make_pass_fast_rtl_dce): ...new function.
5079         (pass_ud_rtl_dce): Replace declaration with that of...
5080         (make_pass_ud_rtl_dce): ...new function.
5081         (pass_rtl_dce): Replace declaration with that of...
5082         (make_pass_rtl_dce): ...new function.
5083         (pass_rtl_dse1): Replace declaration with that of...
5084         (make_pass_rtl_dse1): ...new function.
5085         (pass_rtl_dse2): Replace declaration with that of...
5086         (make_pass_rtl_dse2): ...new function.
5087         (pass_rtl_dse3): Replace declaration with that of...
5088         (make_pass_rtl_dse3): ...new function.
5089         (pass_rtl_cprop): Replace declaration with that of...
5090         (make_pass_rtl_cprop): ...new function.
5091         (pass_rtl_pre): Replace declaration with that of...
5092         (make_pass_rtl_pre): ...new function.
5093         (pass_rtl_hoist): Replace declaration with that of...
5094         (make_pass_rtl_hoist): ...new function.
5095         (pass_rtl_store_motion): Replace declaration with that of...
5096         (make_pass_rtl_store_motion): ...new function.
5097         (pass_cse_after_global_opts): Replace declaration with that of...
5098         (make_pass_cse_after_global_opts): ...new function.
5099         (pass_rtl_ifcvt): Replace declaration with that of...
5100         (make_pass_rtl_ifcvt): ...new function.
5101         (pass_into_cfg_layout_mode): Replace declaration with that of...
5102         (make_pass_into_cfg_layout_mode): ...new function.
5103         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5104         (make_pass_outof_cfg_layout_mode): ...new function.
5105         (pass_loop2): Replace declaration with that of...
5106         (make_pass_loop2): ...new function.
5107         (pass_rtl_loop_init): Replace declaration with that of...
5108         (make_pass_rtl_loop_init): ...new function.
5109         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5110         (make_pass_rtl_move_loop_invariants): ...new function.
5111         (pass_rtl_unswitch): Replace declaration with that of...
5112         (make_pass_rtl_unswitch): ...new function.
5113         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5114         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5115         (pass_rtl_doloop): Replace declaration with that of...
5116         (make_pass_rtl_doloop): ...new function.
5117         (pass_rtl_loop_done): Replace declaration with that of...
5118         (make_pass_rtl_loop_done): ...new function.
5119         (pass_web): Replace declaration with that of...
5120         (make_pass_web): ...new function.
5121         (pass_cse2): Replace declaration with that of...
5122         (make_pass_cse2): ...new function.
5123         (pass_df_initialize_opt): Replace declaration with that of...
5124         (make_pass_df_initialize_opt): ...new function.
5125         (pass_df_initialize_no_opt): Replace declaration with that of...
5126         (make_pass_df_initialize_no_opt): ...new function.
5127         (pass_reginfo_init): Replace declaration with that of...
5128         (make_pass_reginfo_init): ...new function.
5129         (pass_inc_dec): Replace declaration with that of...
5130         (make_pass_inc_dec): ...new function.
5131         (pass_stack_ptr_mod): Replace declaration with that of...
5132         (make_pass_stack_ptr_mod): ...new function.
5133         (pass_initialize_regs): Replace declaration with that of...
5134         (make_pass_initialize_regs): ...new function.
5135         (pass_combine): Replace declaration with that of...
5136         (make_pass_combine): ...new function.
5137         (pass_if_after_combine): Replace declaration with that of...
5138         (make_pass_if_after_combine): ...new function.
5139         (pass_ree): Replace declaration with that of...
5140         (make_pass_ree): ...new function.
5141         (pass_partition_blocks): Replace declaration with that of...
5142         (make_pass_partition_blocks): ...new function.
5143         (pass_match_asm_constraints): Replace declaration with that of...
5144         (make_pass_match_asm_constraints): ...new function.
5145         (pass_regmove): Replace declaration with that of...
5146         (make_pass_regmove): ...new function.
5147         (pass_split_all_insns): Replace declaration with that of...
5148         (make_pass_split_all_insns): ...new function.
5149         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5150         (make_pass_fast_rtl_byte_dce): ...new function.
5151         (pass_lower_subreg2): Replace declaration with that of...
5152         (make_pass_lower_subreg2): ...new function.
5153         (pass_mode_switching): Replace declaration with that of...
5154         (make_pass_mode_switching): ...new function.
5155         (pass_sms): Replace declaration with that of...
5156         (make_pass_sms): ...new function.
5157         (pass_sched): Replace declaration with that of...
5158         (make_pass_sched): ...new function.
5159         (pass_ira): Replace declaration with that of...
5160         (make_pass_ira): ...new function.
5161         (pass_reload): Replace declaration with that of...
5162         (make_pass_reload): ...new function.
5163         (pass_clean_state): Replace declaration with that of...
5164         (make_pass_clean_state): ...new function.
5165         (pass_branch_prob): Replace declaration with that of...
5166         (make_pass_branch_prob): ...new function.
5167         (pass_value_profile_transformations): Replace declaration with that
5168         of...
5169         (make_pass_value_profile_transformations): ...new function.
5170         (pass_postreload_cse): Replace declaration with that of...
5171         (make_pass_postreload_cse): ...new function.
5172         (pass_gcse2): Replace declaration with that of...
5173         (make_pass_gcse2): ...new function.
5174         (pass_split_after_reload): Replace declaration with that of...
5175         (make_pass_split_after_reload): ...new function.
5176         (pass_branch_target_load_optimize1): Replace declaration with that
5177         of...
5178         (make_pass_branch_target_load_optimize1): ...new function.
5179         (pass_thread_prologue_and_epilogue): Replace declaration with that
5180         of...
5181         (make_pass_thread_prologue_and_epilogue): ...new function.
5182         (pass_stack_adjustments): Replace declaration with that of...
5183         (make_pass_stack_adjustments): ...new function.
5184         (pass_peephole2): Replace declaration with that of...
5185         (make_pass_peephole2): ...new function.
5186         (pass_if_after_reload): Replace declaration with that of...
5187         (make_pass_if_after_reload): ...new function.
5188         (pass_regrename): Replace declaration with that of...
5189         (make_pass_regrename): ...new function.
5190         (pass_cprop_hardreg): Replace declaration with that of...
5191         (make_pass_cprop_hardreg): ...new function.
5192         (pass_reorder_blocks): Replace declaration with that of...
5193         (make_pass_reorder_blocks): ...new function.
5194         (pass_branch_target_load_optimize2): Replace declaration with that
5195         of...
5196         (make_pass_branch_target_load_optimize2): ...new function.
5197         (pass_leaf_regs): Replace declaration with that of...
5198         (make_pass_leaf_regs): ...new function.
5199         (pass_split_before_sched2): Replace declaration with that of...
5200         (make_pass_split_before_sched2): ...new function.
5201         (pass_compare_elim_after_reload): Replace declaration with that of...
5202         (make_pass_compare_elim_after_reload): ...new function.
5203         (pass_sched2): Replace declaration with that of...
5204         (make_pass_sched2): ...new function.
5205         (pass_stack_regs): Replace declaration with that of...
5206         (make_pass_stack_regs): ...new function.
5207         (pass_stack_regs_run): Replace declaration with that of...
5208         (make_pass_stack_regs_run): ...new function.
5209         (pass_df_finish): Replace declaration with that of...
5210         (make_pass_df_finish): ...new function.
5211         (pass_compute_alignments): Replace declaration with that of...
5212         (make_pass_compute_alignments): ...new function.
5213         (pass_duplicate_computed_gotos): Replace declaration with that of...
5214         (make_pass_duplicate_computed_gotos): ...new function.
5215         (pass_variable_tracking): Replace declaration with that of...
5216         (make_pass_variable_tracking): ...new function.
5217         (pass_free_cfg): Replace declaration with that of...
5218         (make_pass_free_cfg): ...new function.
5219         (pass_machine_reorg): Replace declaration with that of...
5220         (make_pass_machine_reorg): ...new function.
5221         (pass_cleanup_barriers): Replace declaration with that of...
5222         (make_pass_cleanup_barriers): ...new function.
5223         (pass_delay_slots): Replace declaration with that of...
5224         (make_pass_delay_slots): ...new function.
5225         (pass_split_for_shorten_branches): Replace declaration with that of...
5226         (make_pass_split_for_shorten_branches): ...new function.
5227         (pass_split_before_regstack): Replace declaration with that of...
5228         (make_pass_split_before_regstack): ...new function.
5229         (pass_convert_to_eh_region_ranges): Replace declaration with that
5230         of...
5231         (make_pass_convert_to_eh_region_ranges): ...new function.
5232         (pass_shorten_branches): Replace declaration with that of...
5233         (make_pass_shorten_branches): ...new function.
5234         (pass_set_nothrow_function_flags): Replace declaration with that of...
5235         (make_pass_set_nothrow_function_flags): ...new function.
5236         (pass_dwarf2_frame): Replace declaration with that of...
5237         (make_pass_dwarf2_frame): ...new function.
5238         (pass_final): Replace declaration with that of...
5239         (make_pass_final): ...new function.
5240         (pass_rtl_seqabstr): Replace declaration with that of...
5241         (make_pass_rtl_seqabstr): ...new function.
5242         (pass_release_ssa_names): Replace declaration with that of...
5243         (make_pass_release_ssa_names): ...new function.
5244         (pass_early_inline): Replace declaration with that of...
5245         (make_pass_early_inline): ...new function.
5246         (pass_inline_parameters): Replace declaration with that of...
5247         (make_pass_inline_parameters): ...new function.
5248         (pass_update_address_taken): Replace declaration with that of...
5249         (make_pass_update_address_taken): ...new function.
5250         (pass_convert_switch): Replace declaration with that of...
5251         (make_pass_convert_switch): ...new function.
5252         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5253         to a subclass of simple_ipa_opt_pass along with...
5254         (pass_data_ipa_tree_profile): ...new pass_data instance and...
5255         (make_pass_ipa_tree_profile): ...new function.
5256         * tree-sra.c (pass_sra_early): Convert from a global struct to a
5257         subclass of gimple_opt_pass along with...
5258         (pass_data_sra_early): ...new pass_data instance and...
5259         (make_pass_sra_early): ...new function.
5260         (pass_sra): Convert from a global struct to a subclass of
5261         gimple_opt_pass along with...
5262         (pass_data_sra): ...new pass_data instance and...
5263         (make_pass_sra): ...new function.
5264         (pass_early_ipa_sra): Convert from a global struct to a subclass of
5265         gimple_opt_pass along with...
5266         (pass_data_early_ipa_sra): ...new pass_data instance and...
5267         (make_pass_early_ipa_sra): ...new function.
5268         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5269         subclass of gimple_opt_pass along with...
5270         (pass_data_ccp): ...new pass_data instance and...
5271         (make_pass_ccp): ...new function.
5272         (pass_fold_builtins): Convert from a global struct to a subclass of
5273         gimple_opt_pass along with...
5274         (pass_data_fold_builtins): ...new pass_data instance and...
5275         (make_pass_fold_builtins): ...new function.
5276         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5277         subclass of gimple_opt_pass along with...
5278         (pass_data_copy_prop): ...new pass_data instance and...
5279         (make_pass_copy_prop): ...new function.
5280         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5281         global struct to a subclass of gimple_opt_pass along with...
5282         (pass_data_rename_ssa_copies): ...new pass_data instance and...
5283         (make_pass_rename_ssa_copies): ...new function.
5284         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5285         subclass of gimple_opt_pass along with...
5286         (pass_data_dce): ...new pass_data instance and...
5287         (make_pass_dce): ...new function.
5288         (pass_dce_loop): Convert from a global struct to a subclass of
5289         gimple_opt_pass along with...
5290         (pass_data_dce_loop): ...new pass_data instance and...
5291         (make_pass_dce_loop): ...new function.
5292         (pass_cd_dce): Convert from a global struct to a subclass of
5293         gimple_opt_pass along with...
5294         (pass_data_cd_dce): ...new pass_data instance and...
5295         (make_pass_cd_dce): ...new function.
5296         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5297         subclass of gimple_opt_pass along with...
5298         (pass_data_dominator): ...new pass_data instance and...
5299         (make_pass_dominator): ...new function.
5300         (pass_phi_only_cprop): Convert from a global struct to a subclass of
5301         gimple_opt_pass along with...
5302         (pass_data_phi_only_cprop): ...new pass_data instance and...
5303         (make_pass_phi_only_cprop): ...new function.
5304         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5305         subclass of gimple_opt_pass along with...
5306         (pass_data_dse): ...new pass_data instance and...
5307         (make_pass_dse): ...new function.
5308         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5309         a subclass of gimple_opt_pass along with...
5310         (pass_data_forwprop): ...new pass_data instance and...
5311         (make_pass_forwprop): ...new function.
5312         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5313         struct to a subclass of gimple_opt_pass along with...
5314         (pass_data_tree_ifcombine): ...new pass_data instance and...
5315         (make_pass_tree_ifcombine): ...new function.
5316         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5317         subclass of gimple_opt_pass along with...
5318         (pass_data_ch): ...new pass_data instance and...
5319         (make_pass_ch): ...new function.
5320         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5321         subclass of gimple_opt_pass along with...
5322         (pass_data_tree_loop): ...new pass_data instance and...
5323         (make_pass_tree_loop): ...new function.
5324         (pass_tree_loop_init): Convert from a global struct to a subclass of
5325         gimple_opt_pass along with...
5326         (pass_data_tree_loop_init): ...new pass_data instance and...
5327         (make_pass_tree_loop_init): ...new function.
5328         (pass_lim): Convert from a global struct to a subclass of
5329         gimple_opt_pass along with...
5330         (pass_data_lim): ...new pass_data instance and...
5331         (make_pass_lim): ...new function.
5332         (pass_tree_unswitch): Convert from a global struct to a subclass of
5333         gimple_opt_pass along with...
5334         (pass_data_tree_unswitch): ...new pass_data instance and...
5335         (make_pass_tree_unswitch): ...new function.
5336         (pass_predcom): Convert from a global struct to a subclass of
5337         gimple_opt_pass along with...
5338         (pass_data_predcom): ...new pass_data instance and...
5339         (make_pass_predcom): ...new function.
5340         (pass_vectorize): Convert from a global struct to a subclass of
5341         gimple_opt_pass along with...
5342         (pass_data_vectorize): ...new pass_data instance and...
5343         (make_pass_vectorize): ...new function.
5344         (pass_graphite): Convert from a global struct to a subclass of
5345         gimple_opt_pass along with...
5346         (pass_data_graphite): ...new pass_data instance and...
5347         (make_pass_graphite): ...new function.
5348         (pass_graphite_transforms): Convert from a global struct to a subclass
5349         of gimple_opt_pass along with...
5350         (pass_data_graphite_transforms): ...new pass_data instance and...
5351         (make_pass_graphite_transforms): ...new function.
5352         (pass_check_data_deps): Convert from a global struct to a subclass of
5353         gimple_opt_pass along with...
5354         (pass_data_check_data_deps): ...new pass_data instance and...
5355         (make_pass_check_data_deps): ...new function.
5356         (pass_iv_canon): Convert from a global struct to a subclass of
5357         gimple_opt_pass along with...
5358         (pass_data_iv_canon): ...new pass_data instance and...
5359         (make_pass_iv_canon): ...new function.
5360         (pass_scev_cprop): Convert from a global struct to a subclass of
5361         gimple_opt_pass along with...
5362         (pass_data_scev_cprop): ...new pass_data instance and...
5363         (make_pass_scev_cprop): ...new function.
5364         (pass_record_bounds): Convert from a global struct to a subclass of
5365         gimple_opt_pass along with...
5366         (pass_data_record_bounds): ...new pass_data instance and...
5367         (make_pass_record_bounds): ...new function.
5368         (pass_complete_unroll): Convert from a global struct to a subclass of
5369         gimple_opt_pass along with...
5370         (pass_data_complete_unroll): ...new pass_data instance and...
5371         (make_pass_complete_unroll): ...new function.
5372         (pass_complete_unrolli): Convert from a global struct to a subclass of
5373         gimple_opt_pass along with...
5374         (pass_data_complete_unrolli): ...new pass_data instance and...
5375         (make_pass_complete_unrolli): ...new function.
5376         (pass_parallelize_loops): Convert from a global struct to a subclass
5377         of gimple_opt_pass along with...
5378         (pass_data_parallelize_loops): ...new pass_data instance and...
5379         (make_pass_parallelize_loops): ...new function.
5380         (pass_loop_prefetch): Convert from a global struct to a subclass of
5381         gimple_opt_pass along with...
5382         (pass_data_loop_prefetch): ...new pass_data instance and...
5383         (make_pass_loop_prefetch): ...new function.
5384         (pass_iv_optimize): Convert from a global struct to a subclass of
5385         gimple_opt_pass along with...
5386         (pass_data_iv_optimize): ...new pass_data instance and...
5387         (make_pass_iv_optimize): ...new function.
5388         (pass_tree_loop_done): Convert from a global struct to a subclass of
5389         gimple_opt_pass along with...
5390         (pass_data_tree_loop_done): ...new pass_data instance and...
5391         (make_pass_tree_loop_done): ...new function.
5392         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5393         struct to a subclass of gimple_opt_pass along with...
5394         (pass_data_cse_reciprocals): ...new pass_data instance and...
5395         (make_pass_cse_reciprocals): ...new function.
5396         (pass_cse_sincos): Convert from a global struct to a subclass of
5397         gimple_opt_pass along with...
5398         (pass_data_cse_sincos): ...new pass_data instance and...
5399         (make_pass_cse_sincos): ...new function.
5400         (pass_optimize_bswap): Convert from a global struct to a subclass of
5401         gimple_opt_pass along with...
5402         (pass_data_optimize_bswap): ...new pass_data instance and...
5403         (make_pass_optimize_bswap): ...new function.
5404         (pass_optimize_widening_mul): Convert from a global struct to a
5405         subclass of gimple_opt_pass along with...
5406         (pass_data_optimize_widening_mul): ...new pass_data instance and...
5407         (make_pass_optimize_widening_mul): ...new function.
5408         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5409         subclass of gimple_opt_pass along with...
5410         (pass_data_phiopt): ...new pass_data instance and...
5411         (make_pass_phiopt): ...new function.
5412         (pass_cselim): Convert from a global struct to a subclass of
5413         gimple_opt_pass along with...
5414         (pass_data_cselim): ...new pass_data instance and...
5415         (make_pass_cselim): ...new function.
5416         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5417         subclass of gimple_opt_pass along with...
5418         (pass_data_phiprop): ...new pass_data instance and...
5419         (make_pass_phiprop): ...new function.
5420         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5421         subclass of gimple_opt_pass along with...
5422         (pass_data_pre): ...new pass_data instance and...
5423         (make_pass_pre): ...new function.
5424         (pass_fre): Convert from a global struct to a subclass of
5425         gimple_opt_pass along with...
5426         (pass_data_fre): ...new pass_data instance and...
5427         (make_pass_fre): ...new function.
5428         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5429         subclass of gimple_opt_pass along with...
5430         (pass_data_reassoc): ...new pass_data instance and...
5431         (make_pass_reassoc): ...new function.
5432         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5433         subclass of gimple_opt_pass along with...
5434         (pass_data_sink_code): ...new pass_data instance and...
5435         (make_pass_sink_code): ...new function.
5436         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5437         subclass of gimple_opt_pass along with...
5438         (pass_data_strlen): ...new pass_data instance and...
5439         (make_pass_strlen): ...new function.
5440         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5441         struct to a subclass of gimple_opt_pass along with...
5442         (pass_data_build_alias): ...new pass_data instance and...
5443         (make_pass_build_alias): ...new function.
5444         (pass_build_ealias): Convert from a global struct to a subclass of
5445         gimple_opt_pass along with...
5446         (pass_data_build_ealias): ...new pass_data instance and...
5447         (make_pass_build_ealias): ...new function.
5448         (pass_ipa_pta): Convert from a global struct to a subclass of
5449         simple_ipa_opt_pass along with...
5450         (pass_data_ipa_pta): ...new pass_data instance and...
5451         (make_pass_ipa_pta): ...new function.
5452         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5453         subclass of gimple_opt_pass along with...
5454         (pass_data_uncprop): ...new pass_data instance and...
5455         (make_pass_uncprop): ...new function.
5456         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5457         global struct to a subclass of gimple_opt_pass along with...
5458         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5459         (make_pass_late_warn_uninitialized): ...new function.
5460         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5461         to a subclass of gimple_opt_pass along with...
5462         (pass_data_init_datastructures): ...new pass_data instance and...
5463         (make_pass_init_datastructures): ...new function.
5464         (pass_early_warn_uninitialized): Convert from a global struct to a
5465         subclass of gimple_opt_pass along with...
5466         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5467         (make_pass_early_warn_uninitialized): ...new function.
5468         (pass_update_address_taken): Convert from a global struct to a
5469         subclass of gimple_opt_pass along with...
5470         (pass_data_update_address_taken): ...new pass_data instance and...
5471         (make_pass_update_address_taken): ...new function.
5472         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5473         struct to a subclass of gimple_opt_pass along with...
5474         (pass_data_release_ssa_names): ...new pass_data instance and...
5475         (make_pass_release_ssa_names): ...new function.
5476         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5477         subclass of gimple_opt_pass along with...
5478         (pass_data_stdarg): ...new pass_data instance and...
5479         (make_pass_stdarg): ...new function.
5480         * tree-switch-conversion.c (pass_convert_switch): Convert from a
5481         global struct to a subclass of gimple_opt_pass along with...
5482         (pass_data_convert_switch): ...new pass_data instance and...
5483         (make_pass_convert_switch): ...new function.
5484         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5485         to a subclass of gimple_opt_pass along with...
5486         (pass_data_tail_recursion): ...new pass_data instance and...
5487         (make_pass_tail_recursion): ...new function.
5488         (pass_tail_calls): Convert from a global struct to a subclass of
5489         gimple_opt_pass along with...
5490         (pass_data_tail_calls): ...new pass_data instance and...
5491         (make_pass_tail_calls): ...new function.
5492         * tree-vect-generic.c (pass_lower_vector): Convert from a global
5493         struct to a subclass of gimple_opt_pass along with...
5494         (pass_data_lower_vector): ...new pass_data instance and...
5495         (make_pass_lower_vector): ...new function.
5496         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5497         gimple_opt_pass along with...
5498         (pass_data_lower_vector_ssa): ...new pass_data instance and...
5499         (make_pass_lower_vector_ssa): ...new function.
5500         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5501         to a subclass of gimple_opt_pass along with...
5502         (pass_data_slp_vectorize): ...new pass_data instance and...
5503         (make_pass_slp_vectorize): ...new function.
5504         (pass_ipa_increase_alignment): Convert from a global struct to a
5505         subclass of simple_ipa_opt_pass along with...
5506         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5507         (make_pass_ipa_increase_alignment): ...new function.
5508         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5509         gimple_opt_pass along with...
5510         (pass_data_vrp): ...new pass_data instance and...
5511         (make_pass_vrp): ...new function.
5512         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5513         subclass of simple_ipa_opt_pass along with...
5514         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5515         (make_pass_ipa_free_lang_data): ...new function.
5516         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5517         gimple_opt_pass along with...
5518         (pass_data_tsan): ...new pass_data instance and...
5519         (make_pass_tsan): ...new function.
5520         (pass_tsan_O0): Convert from a global struct to a subclass of
5521         gimple_opt_pass along with...
5522         (pass_data_tsan_O0): ...new pass_data instance and...
5523         (make_pass_tsan_O0): ...new function.
5524         * var-tracking.c (pass_variable_tracking): Convert from a global
5525         struct to a subclass of rtl_opt_pass along with...
5526         (pass_data_variable_tracking): ...new pass_data instance and...
5527         (make_pass_variable_tracking): ...new function.
5528         * web.c (pass_web): Convert from a global struct to a subclass of
5529         rtl_opt_pass along with...
5530         (pass_data_web): ...new pass_data instance and...
5531         (make_pass_web): ...new function.
5532         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5533         declaration with that of...
5534         (make_pass_mode_switch_use): ...new function.
5535         (pass_resolve_sw_modes): Replace declaration with that of...
5536         (make_pass_resolve_sw_modes): ...new function.
5537         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5538         from a global struct to a subclass of rtl_opt_pass along with...
5539         (pass_data_mode_switch_use): ...new pass_data instance and...
5540         (make_pass_mode_switch_use): ...new function.
5541         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5542         from a global struct to a subclass of rtl_opt_pass along with...
5543         (pass_data_resolve_sw_modes): ...new pass_data instance and...
5544         (make_pass_resolve_sw_modes): ...new function.
5545         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5546         struct to a subclass of rtl_opt_pass along with...
5547         (pass_data_insert_vzeroupper): ...new pass_data instance and...
5548         (make_pass_insert_vzeroupper): ...new function.
5549         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5550         global struct to a subclass of rtl_opt_pass along with...
5551         (pass_data_work_around_errata): ...new pass_data instance and...
5552         (make_pass_work_around_errata): ...new function.
5553         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5554         struct to a subclass of rtl_opt_pass along with...
5555         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
5556         (make_pass_mips_machine_reorg2): ...new function.
5558 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5560         * passes.c (pass_manager::operator new): New.
5562 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5564         Handwritten part of conversion of passes to C++ classes.
5566         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
5567         (toplev.o): Add dep on PASS_MANAGER_H.
5568         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
5569         of early local pases to reflect this moving from a global to a
5570         member of gcc::pass_manager.
5571         (cgraph_add_new_function): Likewise.
5572         * lto-cgraph.c (lto_output_node): Update for conversion of
5573         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
5574         * passes.c (opt_pass::clone): New.
5575         (opt_pass::gate): New.
5576         (opt_pass::execute): New.
5577         (opt_pass::opt_pass): New.
5578         (pass_manager::execute_early_local_passes): New.
5579         (pass_manager::execute_pass_mode_switching): new.
5580         (finish_optimization_passes): Convert to...
5581         (pass_manager::finish_optimization_passes): ...this.
5582         (finish_optimization_passes): Update for conversion of passes to
5583         C++ classes.
5584         (register_dump_files_1): Use has_gate since we cannot portably
5585         check a vtable entry against NULL.
5586         (dump_one_pass): Likewise.
5587         (ipa_write_summaries_2): Likewise.
5588         (ipa_write_optimization_summaries_1): Likewise.
5589         (ipa_read_summaries_1): Likewise.
5590         (ipa_read_optimization_summaries_1): Likewise.
5591         (execute_ipa_stmt_fixups): Likewise.
5592         (pass_manager::pass_manager): Rewrite pass-creation, invoking
5593         pass-creation functions rather than wiring up globals, and
5594         storing the results in fields of pass_manager generated using
5595         pass-instances.def.
5596         (pass_manager::dump_profile_report): Update for conversion of
5597         passes to C++ classes.
5598         (pass_manager::execute_ipa_summary_passes): Likewise.
5599         (execute_one_ipa_transform_pass): Likewise.
5600         (execute_one_pass): Use has_gate and has_execute since we cannot
5601         portably check a vtable entry against NULL.
5602         * pass_manager.h (pass_manager::finish_optimization_passes): New.
5603         (pass_manager): Use pass-instances.def to add fields for the
5604         various pass instances.
5605         * toplev.c (finalize): Update for move of
5606         finish_optimization_passes to a method of gcc::pass_manager.
5607         * toplev.h (finish_optimization_passes): Move to method of class
5608         pass_manager.
5609         * tree-pass.h (struct pass_data): New.
5610         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
5611         (opt_pass::gate): Convert to virtual function.
5612         (opt_pass::~opt_pass): New.
5613         (opt_pass::clone): New.
5614         (opt_pass::execute): Convert to virtual function.
5615         (opt_pass::opt_pass): New.
5616         (opt_pass::ctxt_): new.
5617         (gimple_opt_pass): Convert to subclass of opt_pass.
5618         (gimple_opt_pass::gimple_opt_pass): New.
5619         (rtl_opt_pass): Convert to subclass of opt_pass.
5620         (rtl_opt_pass::rtl_opt_pass): New.
5621         (ipa_opt_pass_d): Convert to subclass of opt_pass.
5622         (ipa_opt_pass_d::ipa_opt_pass_d): New.
5623         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
5624         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
5625         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
5626         invocation of pass_mode_switching to reflect this moving from a
5627         global to a member of gcc::pass_manager.
5628         (ix86_option_override): Rework how pass_insert_vzeroupper is
5629         added to the pass_manager to reflect autogenerated changes.
5630         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
5632 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
5634         PR rtl-optimization/57708
5635         * recog.c (peep2_find_free_register): Validate all regs in a
5636         multi-reg mode.
5638 2013-08-05  Jan Hubicka  <jh@suse.cz>
5640         PR lto/57602
5641         * cgraph.c (verify_cgraph_node): Accept local flags from other
5642         partitions.
5643         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
5644         (function_and_variable_visibility): Likewise.
5645         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
5647 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5649         * graph.c (init_graph_slim_pretty_print): Remove.
5650         (print_graph_cfg): Do not call it.  Use local pretty printer.
5651         (start_graph_dump): Likewise.
5653 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5655         * gimple-pretty-print.c (buffer): Remove.
5656         (initialized): Likewise.
5657         (maybe_init_pretty_print): Likewise.
5658         (print_gimple_stmt): Do not call it.  Use non-static local
5659         pretty_printer variable.
5660         (print_gimple_expr): Likewise.
5661         (print_gimple_seq): Likewise.
5662         (gimple_dump_bb): Likewise.
5664 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5666         * asan.c (asan_pp): Remove.
5667         (asan_pp_initialized): Likewise.
5668         (asan_pp_initialize): Likewise.
5669         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
5670         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
5671         (asan_add_global): Likewise.
5673 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5675         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
5676         * diagnostic.h (diagnostic_flush_buffer): Adjust.
5677         * pretty-print.c (pp_formatted_text_data): Likewise.
5678         (pp_indent): Rename from pp_base_indent.
5679         (pp_format): Rename from pp_base_format.
5680         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
5681         (pp_format_verbatim): Rename from pp_base_format_verbatim.
5682         (pp_flush): Rename from pp_base_flush.
5683         (pp_set_line_maximum_length): Rename from
5684         pp_base_set_line_maximum_length.
5685         (pp_clear_output_area): Rename from pp_base_clear_output_area.
5686         (pp_set_prefix): Rename from pp_base_set_prefix.
5687         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
5688         (pp_emit_prefix): Rename from pp_base_emit_prefix.
5689         (pp_append_text): Rename from pp_base_append_text.
5690         (pp_formatted_text): Rename from pp_base_formatted_text.
5691         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
5692         (pp_remaining_character_count_for_line): Rename from
5693         pp_base_remaining_character_count_for_line.
5694         (pp_newline): Rename from pp_base_newline.
5695         (pp_character): Rename from pp_base_character.
5696         (pp_string): Rename from pp_base_string.
5697         (pp_maybe_space): Rename from pp_base_maybe_space.
5698         * asan.c (asan_pp_string): Adjust.
5699         (asan_emit_stack_protection): Likewise.
5700         (asan_add_global): Likewise.
5701         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
5702         * tree-mudflap.c (mf_varname_tree): Likewise.
5703         * tree-pretty-print.c (pp_tree_identifier): Rename from
5704         pp_base_tree_identifier.
5705         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
5706         Declare as function.
5708 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5710         * pretty-print.h (pp_bar_bar): New.
5711         (pp_ampersand_ampersand): Likewise.
5712         (pp_less_equal): Likewise.
5713         (pp_greater_equal): Likewise.
5714         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
5715         printer functions instead of pp_string or operators and punctuators.
5716         (dump_gimple_call): Likewise.
5717         (dump_gimple_omp_for): Likewise.
5718         (dump_gimple_transaction): Likewise.
5719         (dump_gimple_phi): Likewise.
5720         (pp_gimple_stmt_1): Likewise.
5721         * sched-vis.c (print_insn): Likewise.
5722         * tree-mudflap.c (mf_varname_tree): Likewise.
5723         * tree-pretty-print.c (dump_block_node): Likewise.
5724         (dump_generic_node): Likewise.
5726 2013-08-02  Jan Hubicka  <jh@suse.cz>
5728         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
5729         boundaries.
5730         * lto-streamer-out.c (tree_is_indexable): Results decls and
5731         parm decls are not indexable.
5732         (DFS_write_tree_body): Do not follow args and results.
5733         (hash_tree): Likewise.
5734         (output_functions): Rearrange so struct function is needed
5735         only when real body is output; be able to also ouptut abstract
5736         functions; output DECL_ARGUMENTS and DECL_RESULT.
5737         (lto_output): When not in WPA, ale store abstract functions.
5738         (write_symbol): Do not care about RESULT_DECL.
5739         (output_symbol_p): Handle correctly sbtract decls.
5740         * lto-streamer-in.c (input_function): Rearrange so struct
5741         function can be NULL at entry; allow streaming of
5742         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
5743         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
5744         sanity check during LTO.
5745         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
5746         RESULT_DECl and DECL_ARGUMENTS.
5747         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
5748         Likewise.
5750 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5752         * pretty-print.h (pp_underscore): New.
5753         (pp_comma): Tidy.
5754         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
5755         printer functions instead of pp_character.
5756         (dump_binary_rhs): Likewise.
5757         (dump_ternary_rhs): Likewise.
5758         (dump_gimple_call_args): Likewise.
5759         (pp_points_to_solution): Likewise.
5760         (dump_gimple_call): Likewise.
5761         (dump_gimple_switch): Likewise.
5762         (dump_gimple_cond): Likewise.
5763         (dump_gimple_bind): Likewise.
5764         (dump_gimple_try): Likewise.
5765         (dump_gimple_omp_for): Likewise.
5766         (dump_gimple_omp_continue): Likewise.
5767         (dump_gimple_omp_single): Likewise.
5768         (dump_gimple_omp_sections): Likewise.
5769         (dump_gimple_omp_block): Likewise.
5770         (dump_gimple_omp_critical): Likewise.
5771         (dump_gimple_transaction): Likewise.
5772         (dump_gimple_asm): Likewise.
5773         (dump_gimple_phi): Likewise.
5774         (dump_gimple_omp_parallel): Likewise.
5775         (dump_gimple_omp_task): Likewise.
5776         (dump_gimple_omp_atomic_load): Likewise.
5777         (dump_gimple_omp_atomic_store): Likewise.
5778         (dump_gimple_mem_ops): Likewise.
5779         (pp_gimple_stmt_1): Likewise.
5780         (pp_cfg_jump): Likewise.
5781         (dump_implicit_edges): Likewise.
5782         (gimple_dump_bb_for_graph): Likewise.
5783         * graph.c (draw_cfg_node): Likewise.
5784         * langhooks.c (lhd_print_error_function): Likewise.
5785         * sched-vis.c (print_exp): Likewise.
5786         (print_value): Likewise.
5787         (print_pattern): Likewise.
5788         (print_insn): Likewise.
5789         (rtl_dump_bb_for_graph): Likewise.
5790         * tree-pretty-print.c (dump_function_declaration): Likewise.
5791         (dump_array_domain): Likewise.
5792         (dump_omp_clause): Likewise.
5793         (dump_location): Likewise.
5794         (dump_generic_node): Likewise.
5795         (print_struct_decl): Likewise.
5796         * diagnostic.c (diagnostic_show_locus): Use pp_space.
5798 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5800         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
5801         candidate table when replacing a candidate statement.
5802         (replace_rhs_if_not_dup): Likewise.
5803         (replace_one_candidate): Likewise.
5805 2013-08-02  Jan Hubicka  <jh@suse.cz>
5806             Martin Liska  <marxin.liska@gmail.com>
5808         * cgraphunit.c (add_new_function): Fix logic when adding from
5809         late IPA pass.
5810         (assemble_thunk): Rename to ...
5811         (expand_thunk); .. this one; export; get it working with
5812         general functions; make produced gimple valid.
5813         * cgraph.h (expand_thunk): Declare.
5815 2013-08-02  Jan Hubicka  <jh@suse.cz>
5817         * ipa-cp.c (gather_context_independent_values): Use
5818         ipa_get_param_move_cost.
5819         (get_replacement_map): Remove PARAM; move parameter folding
5820         into tree-inline.c
5821         (create_specialized_node): Update.
5822         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
5823         assert that we have gimple body; update move_cost.
5824         (count_formal_params): Assert that we have gimple body.
5825         (ipa_dump_param): New function.
5826         (ipa_alloc_node_params): Break out from ...
5827         (ipa_initialize_node_params): ... here.
5828         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
5829         (ipa_write_node_info): Stream move costs.
5830         (ipa_read_node_info): Read move costs.
5831         (ipa_update_after_lto_read): Do not recompute node params.
5832         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
5833         (ipa_get_param): Check we are not in WPA.
5834         (ipa_get_param_move_cost): New.
5835         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
5836         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
5837         parm numbers to be present.
5839 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5841         PR rtl-optimization/58048
5842         * lra-constraints.c (process_alt_operands): Don't check asm
5843         operand on register.
5845 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
5847         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
5848         the implied StoreLoad barrier for atomic operations if before.
5850 2013-08-02  Jan Hubicka  <jh@suse.cz>
5851             Martin Liska  <marxin.liska@gmail.com>
5853         * cgraph.c (cgraph_function_body_availability): Do not check
5854         cgraph flags.
5855         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
5856         symtab_node_availability): Declare.
5857         * ipa.c (can_replace_by_local_alias): New.
5858         (function_and_variable_visibility): Use it.
5859         * symtab.c (symtab_for_node_and_aliases,
5860         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
5862 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5864         PR rtl-optimization/57963
5865         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
5866         (lra_constraints): Use them.
5868 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
5870         * config/arm/types.md (define_attr "type"): Add "load_acq"
5871         and "store_rel".
5872         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
5873         changes.
5874         (cortex_a53_store1): Likewise.
5876 2013-08-01  Jan Hubicka  <jh@suse.cz>
5878         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
5879         partitions are not needed.
5881 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
5883         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
5884         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
5885         MAYBE_NON_Q_CLASS_P where appropriate.
5887 2013-08-01  Jan Hubicka  <jh@suse.cz>
5889         * cgraph.h (release_function_body): Declare.
5890         * tree.c (free_lang_data_in_decl): Free, parameters and return values
5891         of unused delcarations.
5893 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5895         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
5896         RTL form when subtracting a constant.
5898 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5900         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
5901         Generate canonical plus rtx with negated immediate instead of minus
5902         where appropriate.
5903         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
5905 2013-08-01  Jan Hubicka  <jh@suse.cz>
5907         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
5908         (cgraph_release_function_body): Likewise.
5909         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
5910         * cgraph.h (cgrpah_node): Rename abstract_and_needed
5911         to used_as_abstract_origin.
5912         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
5913         symbols used as abstract origins.
5914         * cgraphunit.c (analyze_functions): Update.
5915         * ipa.c (symtab_remove_unreachable_nodes): Recompute
5916         used_as_abstract_origin.
5917         * tree-inline.c (tree_function_versioning): Update
5918         used_as_abstract_origin; be ready for DECL_RESULT and
5919         DECL_ARGUMENTS to be NULL.
5921         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
5922         for abstract functions.
5923         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
5924         real symbols.
5926 2013-08-01  Jan Hubicka  <jh@suse.cz>
5928         * profile.c (compute_value_histograms): Fix thinko.
5930 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5932         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
5933         aarch-common-protos.h to extra_headers.
5934         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5935         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
5936         * config/aarch64/t-aarch64 (aarch-common.o): Define.
5938 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5940         * config/aarch64/aarch64.md (define_attr "type"): Delete.
5941         Include "../arm/types.md".  Define "type" attribute for all patterns.
5942         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
5943         attribute changes.
5945 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
5947         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
5948         to support power8 load fusion.
5949         (fusion_gpr_mem_load): Likewise.
5951         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
5953         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
5954         declarations for power8 load fusion.
5955         (emit_fusion_gpr_load): Likewise.
5957         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
5958         tuning for power8, turn on fusion mode by default.  Turn on sign
5959         extending fusion mode if normal fusion mode is on, and we are at
5960         -O2 or -O3.
5961         (fusion_gpr_load_p): New function, return true if we can fuse an
5962         addis instruction with a dependent load to a GPR.
5963         (emit_fusion_gpr_load): Emit the instructions for power8 load
5964         fusion to GPRs.
5966         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
5967         (VSX load fusion peepholes): New peepholes to fuse together an
5968         addi instruction with a VSX load instruction.
5970         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
5971         peepholes to fuse an addis instruction with a load to a GPR base
5972         register.  If we are supporting sign extending fusions, convert
5973         sign extending loads to zero extending loads and add an explicit
5974         sign extension.
5976 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
5978         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
5979         aarch-common-protos.h to extra_headers.
5980         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5981         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
5982         (arm_early_store_addr_dep): Likewise.
5983         (arm_no_early_alu_shift_dep): Likewise.
5984         (arm_no_early_alu_shift_value_dep): Likewise.
5985         (arm_no_early_mul_dep): Likewise.
5986         (arm_no_early_store_addr_dep): Likewise.
5987         (arm_mac_accumulator_is_mul_result): Likewise.
5988         (arm_mac_accumulator_is_result): Likewise.
5989         * config/arm/aarch-common.c: ... here.  New file.
5990         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
5991         here to ...
5992         (arm_early_store_addr_dep): Likewise.
5993         (arm_no_early_alu_shift_dep): Likewise.
5994         (arm_no_early_alu_shift_value_dep): Likewise.
5995         (arm_no_early_mul_dep): Likewise.
5996         (arm_no_early_store_addr_dep): Likewise.
5997         (arm_mac_accumulator_is_mul_result): Likewise.
5998         (arm_mac_accumulator_is_result): Likewise.
5999         * config/arm/aarch-common-protos.h: ... here.  New file.
6000         * config/arm/t-arm (aarch-common.o): Define.
6002 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6004         * config/arm/arm.md: Include new file "types.md".
6005         (define_attr "type"): Move from here to ...
6006         (define_attr "mul32"): Likewise.
6007         (define_attr "mul64"): Likewise.
6008         * config/arm/types.md: ... here.  New file.
6010 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6012         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6013         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6015 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6017         * gen-pass-instances.awk: Fix offset of substr().
6019 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6021         * Makefile.in (pass-instances.def): New.
6022         (passes.o): Replace dependency on passes.def with one on
6023         pass-instances.def
6025         * gen-pass-instances.awk: New.
6027         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6028         rather than passes.def, updating local definition of NEXT_PASS
6029         macro to add an extra NUM parameter (currently unused).
6031 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6033         * Makefile.in (PASS_MANAGER_H): New.
6034         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6035         (passes.o): Likewise.
6036         (statistics.o): Likewise.
6037         (cgraphunit.o): Likewise.
6038         (context.o): Depend on PASS_MANAGER_H.
6040         * pass_manager.h: New.
6042         * cgraphunit.c (cgraph_add_new_function): Update for moves
6043         of globals to fields of pass_manager.
6044         (analyze_function): Likewise.
6045         (expand_function): Likewise.
6046         (ipa_passes): Likewise.
6047         (compile): Likewise.
6049         * context.c (context::context): New.
6050         * context.h  (context::context): New.
6051         (context::get_passes): New.
6052         (context::passes_): New.
6054         * lto-cgraph.c (input_node): Update for moves of globals to
6055         fields of pass_manager.
6057         * passes.c (all_passes): Remove, in favor of a field of the
6058         same name within the new class pass_manager.
6059         (all_small_ipa_passes): Likewise.
6060         (all_lowering_passes): Likewise.
6061         (all_regular_ipa_passes): Likewise.
6062         (all_late_ipa_passes): Likewise.
6063         (all_lto_gen_passes): Likewise.
6064         (passes_by_id): Likewise.
6065         (passes_by_id_size): Likewise.
6066         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6067         the new class pass_manager.
6068         (set_pass_for_id): Convert to...
6069         (pass_manager::set_pass_for_id): ...method.
6070         (get_pass_for_id): Convert to...
6071         (pass_manager::get_pass_for_id): ...method.
6072         (register_one_dump_file): Move body of implementation into...
6073         (pass_manager::register_one_dump_file): ...here.
6074         (register_dump_files_1): Convert to...
6075         (pass_manager::register_dump_files_1): ...method.
6076         (register_dump_files): Convert to...
6077         (pass_manager::register_dump_files): ...method.
6078         (create_pass_tab): Update for moves of globals to fields of
6079         pass_manager.
6080         (dump_passes): Move body of implementation into...
6081         (pass_manager::dump_passes): ...here.
6082         (register_pass): Move body of implementation into...
6083         (pass_manager::register_pass): ...here.
6084         (init_optimization_passes): Convert into...
6085         (pass_manager::pass_manager): ...constructor for new
6086         pass_manager class, and initialize the pass_lists array.
6087         (check_profile_consistency): Update for moves of globals to
6088         fields of pass_manager.
6089         (dump_profile_report): Move body of implementation into...
6090         (pass_manager::dump_profile_report): ...here.
6091         (ipa_write_summaries_1): Update for moves of pass lists from
6092         being globals to fields of pass_manager.
6093         (ipa_write_optimization_summaries): Likewise.
6094         (ipa_read_summaries):  Likewise.
6095         (ipa_read_optimization_summaries): Likewise.
6096         (execute_all_ipa_stmt_fixups): Likewise.
6098         * statistics.c (statistics_fini): Update for moves of globals to
6099         fields of pass_manager.
6101         * toplev.c (general_init): Replace call to
6102         init_optimization_passes with construction of the pass_manager
6103         instance.
6105         * tree-pass.h (all_passes): Remove, in favor of a field of the
6106         same name within the new class pass_manager.
6107         (all_small_ipa_passes): Likewise.
6108         (all_lowering_passes): Likewise.
6109         (all_regular_ipa_passes): Likewise.
6110         (all_lto_gen_passes): Likewise.
6111         (all_late_ipa_passes): Likewise.
6112         (passes_by_id): Likewise.
6113         (passes_by_id_size): Likewise.
6114         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6115         the new class pass_manager.
6116         (get_pass_for_id): Remove.
6118 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6120         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6121         configs.
6123 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6125         * arm.md (mulhi3): New expand pattern.
6127 2013-07-30  Jan Hubicka  <jh@suse.cz>
6128             Martin Liska  <marxin.liska@gmail.com>
6130         * profile.c (compute_value_histograms): Do not ICE when
6131         there is mismatch only on some counters.
6133 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6135         PR rtl-optimization/57637
6136         * function.c (move_insn_for_shrink_wrap): Also check the
6137         GEN set of the LIVE problem for the liveness analysis
6138         if it exists, otherwise give up.
6140 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6142         PR tree-optimization/57993
6143         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6144         replaced statement in the candidate table.
6145         (phi_add_costs): Return infinite cost when the hidden basis does
6146         not dominate all phis on which the candidate is dependent.
6147         (replace_one_candidate): Record replaced statement in the
6148         candidate table.
6150 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6152         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6153         (ashlv2si3): New expander.
6154         (*ashlv2si3_i): New define_insn_and_split.
6155         * predicates.md (float_operation): Allow patterns with three
6156         basic sub-patterns.
6158         PR rtl-optimization/58021
6159         * mode-switching.c (create_pre_exit): Always split off preceding
6160         insns if we are not at the basic block head.
6162 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6164         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6165         (UCLIBC_DYNAMIC_LINKER): New macro.
6166         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6167         `-mnan=2008'.
6168         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6169         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6170         `-mnan=2008'.
6171         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6172         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6173         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6174         for SF and DF modes.  Use ieee_quad_format for TF mode.
6175         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6176         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6177         (mips_option_override): Handle `-mnan=legacy'.
6178         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6179         `-mabs=2008' and `-mnan=2008'.
6180         (OPTION_DEFAULT_SPECS): Add "nan" default.
6181         (ASM_SPEC): Handle `-mnan='.
6182         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6183         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6184         comment accordingly.
6185         (neg<mode>2): Likewise.
6186         * config/mips/mips.opt (mabs, mnan): New options.
6187         * doc/install.texi (Configuration): Document `--with-nan=' option.
6188         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6189         `-mnan=' options.
6190         (MIPS Options): Document them.
6191         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6192         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6193         * configure: Regenerate.
6194         * config.in: Regenerate.
6196 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6198         * config/i386/i386.md (float post-reload splitters): Do not check
6199         for subregs of SSE registers.
6201 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6202             H.J. Lu  <hongjiu.lu@intel.com>
6204         PR target/57954
6205         PR target/57988
6206         * config/i386/i386.md (post-reload splitter
6207         to avoid partial SSE reg dependency stalls): New pattern.
6209 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6211         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6213 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6215         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6216         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6218 2013-07-26  Cary Coutant  <ccoutant@google.com>
6220         * dwarf2out.c (die_checksum_ordered): Don't include template
6221         instantiations in signature.
6222         (is_template_parameter): New function.
6223         (is_template_instantiation): New function.
6224         (generate_skeleton_bottom_up): Don't include template instantiations
6225         in type unit DIE.
6226         (generate_skeleton): Likewise.
6227         (break_out_comdat_types): Move recursive call to break out nested
6228         types earlier.
6229         (prune_unused_types_mark_generic_parms_dies): Call
6230         is_template_parameter.
6232 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
6234         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6235         uses vector registers.
6236         * config/aarch64/iterators.md: Add attributes rtn and vas.
6238 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6239             Richard Earnshaw  <richard.earnshaw@arm.com>
6241         * combine.c (simplify_comparison): Re-canonicalize operands
6242         where appropriate.
6243         * config/arm/arm.md (movcond_addsi): New splitter.
6245 2013-07-25  Sterling Augustine  <saugustine@google.com>
6247         * dwarf2out.c (size_of_pubnames): Move code to...
6248         (include_pubname_in_output): ...here.  New.
6249         (want_pubnames): Rearrange.
6250         (output_pubnames): Call include_pubname_in_output.  Move assertion.
6252 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6254         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6256 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6258         PR target/38836
6259         * doc/extend.texi: Remove obsolete builtins. Fix
6260         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6262 2013-07-25  Jan Hubicka  <jh@suse.cz>
6264         * cgraph.c (release_function_body): Break out from ...
6265         (cgraph_release_function_body): ... this one; also release DECL_RESULT
6266         and DECL_ARGUMENTS.
6267         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6268         old_tree in the map.
6269         (create_specialized_node): Update.
6270         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6271         into index.
6272         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6273         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6274         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6275         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6276         DECL_RESULT.
6278 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6280         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6281         addsi3_carryin_alt2_<optab>): Correct output template.
6283 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6285         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6286         Adjust for arm_restrict_it.
6287         Remove trailing whitespace.
6289 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
6291         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6292         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6294         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6296 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
6298         PR rtl-optimization/57960
6299         * lra-constraints.c (process_alt_operands): Use the right mode
6300         when checking strict_low.
6302 2013-07-25  Jan Hubicka  <jh@suse.cz>
6304         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6305         * cgraph.c (cgraph_remove_node): Do not release function body
6306         when in cgraph streaming.
6307         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6308         in other partitions are not considered reachable; fix handling of
6309         clones.
6311 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6313         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6315 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6317         PR target/19599
6318         PR target/57731
6319         PR target/57837
6320         * config/arm/arm.md ("*sibcall_insn): Replace use of
6321         Ss with US. Adjust output for v5 and v4t.
6322         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6324         * config/arm/constraints.md ("Ss"): Rename to US.
6326 2013-07-25  Terry Guo  <terry.guo@arm.com>
6328         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6329         shift_add/shift_sub0/shift_sub1 RTXs.
6331 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
6332             Anton Blanchard  <anton@au1.ibm.com>
6334         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6335         (altivec_vpks<VI_char>ss): Likewise.
6336         (altivec_vpks<VI_char>us): Likewise.
6337         (altivec_vpku<VI_char>us): Likewise.
6338         (altivec_vpku<VI_char>um): Likewise.
6340 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
6342         Introduce context class.
6344         * Makefile.in (CONTEXT_H): New.
6345         (OBJS): Add context.o.
6346         (toplev.o): Add CONTEXT_H to dependencies.
6347         (context.o): New.
6349         * toplev.c (general_init): Create the singleton gcc::context instance.
6351         * context.c: New.
6353         * context.h: New.
6355 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
6357         PR rtl-optimization/57968
6358         * mode-switching.c (create_pre_exit): Allow instructions that
6359         don't set a return register to need a non-exit mode.
6361 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6362             Anton Blanchard  <anton@au1.ibm.com>
6364         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6365         operands to vperm for little endian.
6366         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6367         of lvsl to create the control mask for a vperm for little endian.
6369 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6370             Anton Blanchard  <anton@au1.ibm.com>
6372         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6373         two operands for little-endian.
6375 2013-07-23  Steve Ellcey  <sellcey@mips.com>
6377         * config/mips/mips.c (mips_case_values_threshold): New.
6378         (TARGET_CASE_VALUES_THRESHOLD): Define.
6380 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6381             Anton Blanchard  <anton@au1.ibm.com>
6383         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6384         selection of field for vector splat in little endian mode.
6386 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6388         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6389         expanders to rs6000.md.
6390         (ior<mode>3): Likewise.
6391         (and<mode>3): Likewise.
6392         (one_cmpl<mode>2): Likewise.
6393         (nor<mode>3): Likewise.
6394         (andc<mode>3): Likewise.
6395         (eqv<mode>3): Likewise.
6396         (nand<mode>3): Likewise.
6397         (orc<mode>3): Likewise.
6399         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6400         declaration.
6402         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6403         to split multi-word logical operations.
6404         (rs6000_split_logical_di): Likewise.
6405         (rs6000_split_logical): Likewise.
6407         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6408         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6409         and allow TImode operations in 32-bit.
6410         (vsx_and<mode>3_64bit): Likewise.
6411         (vsx_ior<mode>3_32bit): Likewise.
6412         (vsx_ior<mode>3_64bit): Likewise.
6413         (vsx_xor<mode>3_32bit): Likewise.
6414         (vsx_xor<mode>3_64bit): Likewise.
6415         (vsx_one_cmpl<mode>2_32bit): Likewise.
6416         (vsx_one_cmpl<mode>2_64bit): Likewise.
6417         (vsx_nor<mode>3_32bit): Likewise.
6418         (vsx_nor<mode>3_64bit): Likewise.
6419         (vsx_andc<mode>3_32bit): Likewise.
6420         (vsx_andc<mode>3_64bit): Likewise.
6421         (vsx_eqv<mode>3_32bit): Likewise.
6422         (vsx_eqv<mode>3_64bit): Likewise.
6423         (vsx_nand<mode>3_32bit): Likewise.
6424         (vsx_nand<mode>3_64bit): Likewise.
6425         (vsx_orc<mode>3_32bit): Likewise.
6426         (vsx_orc<mode>3_64bit): Likewise.
6428         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6429         logical types in GPRs.
6431         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6432         logical insns to rs6000.md, and allow TImode operations in
6433         32-bit.
6434         (altivec_ior<mode>3): Likewise.
6435         (altivec_xor<mode>3): Likewise.
6436         (altivec_one_cmpl<mode>2): Likewise.
6437         (altivec_nor<mode>3): Likewise.
6438         (altivec_andc<mode>3): Likewise.
6440         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6441         attributes for moving the 128-bit logical operations into
6442         rs6000.md.
6443         (BOOL_REGS_OUTPUT): Likewise.
6444         (BOOL_REGS_OP1): Likewise.
6445         (BOOL_REGS_OP2): Likewise.
6446         (BOOL_REGS_UNARY): Likewise.
6447         (BOOL_REGS_AND_CR0): Likewise.
6448         (one_cmpl<mode>2): Add support for DI logical operations on
6449         32-bit, splitting the operations to 32-bit.
6450         (anddi3): Likewise.
6451         (iordi3): Likewise.
6452         (xordi3): Likewise.
6453         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6454         changes to combine the 32/64-bit code, allow logical operations on
6455         TI mode in 32-bit, and to use similar match_operator patterns like
6456         scalar mode uses.  Combine the Altivec and VSX code for logical
6457         operations, and move it here.
6458         (ior<mode>3, 128-bit types): Likewise.
6459         (xor<mode>3, 128-bit types): Likewise.
6460         (one_cmpl<mode>3, 128-bit types): Likewise.
6461         (nor<mode>3, 128-bit types): Likewise.
6462         (andc<mode>3, 128-bit types): Likewise.
6463         (eqv<mode>3, 128-bit types): Likewise.
6464         (nand<mode>3, 128-bit types): Likewise.
6465         (orc<mode>3, 128-bit types): Likewise.
6466         (and<mode>3_internal): Likewise.
6467         (bool<mode>3_internal): Likewise.
6468         (boolc<mode>3_internal1): Likewise.
6469         (boolc<mode>3_internal2): Likewise.
6470         (boolcc<mode>3_internal1): Likewise.
6471         (boolcc<mode>3_internal2): Likewise.
6472         (eqv<mode>3_internal1): Likewise.
6473         (eqv<mode>3_internal2): Likewise.
6474         (one_cmpl1<mode>3_internal): Likewise.
6476 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6478         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6479         Rename flag_stack_usage to flag_stack_usage_info.
6481 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6483         * config/microblaze/sync.md: New file.
6484         * config/microblaze/microblaze.md: Include sync.md
6485         * config/microblaze/microblaze.c: Add print_operand 'y'.
6486         * config/microblaze/constraints.md: Add memory_contraint
6487         'Q' which is a single register.
6489 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
6491         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6493 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6495         * reload.c (find_reloads): Exit loop once we find this operand
6496         cannot be reloaded somehow for this alternative.
6498         * reload.c (find_reloads): Exit loop once we find a hard register.
6500         * rtlanal.c (computed_jump_p): Exit loop once we find label
6501         reference is used.
6503         * i386.c (ix86_pad_returns): Exit loop after setting replace.
6505         * cfgloopmanip.c (remove_path): Exit loop after setting
6506         irred_invalidated.
6508         * gensupport.c (subst_dup): Avoid loop if code is not
6509         MATCH_DUP nor MATCH_OP_DUP.
6511 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
6513         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6515 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6517         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6518         true for SP_REGNUM if mode == ptr_mode.
6519         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6520         with value R0_REGNUM + 31.
6522 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6524         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6525         pad pointer-typed argument downward.
6527 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6529         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6530         and __ILP32__ when the ILP32 model is in use.
6532 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6534         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6535         (aarch64_load_symref_appropriately): In the case of
6536         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6537         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6538         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6539         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6540         if 'mode' doesn't equal to 'ptr_mode'.
6541         (aarch64_output_mi_thunk): Add an assertion on the alignment of
6542         'vcall_offset'; change to call aarch64_emit_move differently depending
6543         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6544         to calculate the upper bound of 'vcall_offset'.
6545         (aarch64_cannot_force_const_mem): Change to also return true if
6546         mode != ptr_mode.
6547         (aarch64_legitimize_reload_address): In the case of large
6548         displacements, add new local variable 'xmode' and an assertion
6549         based on it; change to use 'xmode' to generate the new rtx and
6550         reload.
6551         (aarch64_asm_trampoline_template): Change to generate the template
6552         differently depending on TARGET_ILP32 or not; change to use
6553         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6554         (aarch64_trampoline_size): Removed.
6555         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
6556         and replace immediate literals with it.  Change to use 'ptr_mode'
6557         instead of 'DImode' and call convert_memory_address if the mode
6558         of 'fnaddr' doesn't equal to 'ptr_mode'.
6559         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
6560         to output symbol.
6561         (aarch64_elf_asm_destructor): Likewise.
6562         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
6563         on TARGET_ILP32 instead of aarch64_trampoline_size.
6564         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
6565         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
6566         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
6567         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
6568         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
6569         (storewb_pair<GPI:mode>_<P:mode>): ... this.
6570         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
6571         depending on the value of 'mode'.
6572         (add_losym_<mode>): New.
6573         (ldr_got_small_<mode>): New, based on ldr_got_small.
6574         (ldr_got_small): Remove.
6575         (ldr_got_small_sidi): New.
6576         * config/aarch64/iterators.md (P): New.
6577         (PTR): Change to 'ptr_mode' in the condition.
6579 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6581         * config.gcc (aarch64*-*-*): Support --with-abi.
6582         (aarch64*-*-elf): Support --with-multilib-list.
6583         (aarch64*-*-linux*): Likewise.
6584         (supported_defaults): Add abi to aarch64*-*-*.
6585         * configure.ac: Mention AArch64 for --with-multilib-list.
6586         * configure: Re-generated.
6587         * config/aarch64/biarchilp32.h: New file.
6588         * config/aarch64/biarchlp64.h: New file.
6589         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
6590         (ABI_SPEC): Ditto.
6591         (MULTILIB_DEFAULTS): Ditto.
6592         (DRIVER_SELF_SPECS): Ditto.
6593         (ASM_SPEC): Update to also substitute -mabi.
6594         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
6595         file whose name depends on -mabi= and -mbig-endian.
6596         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
6597         TARGET_ILP32.
6598         (POINTER_SIZE): New define.
6599         (POINTERS_EXTEND_UNSIGNED): Ditto.
6600         (enum aarch64_abi_type): New enumeration tag.
6601         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
6602         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
6603         (TARGET_ILP32): New define.
6604         * config/aarch64/aarch64.opt (mabi): New.
6605         (aarch64_abi): New.
6606         (ilp32, lp64): New values for -mabi.
6607         * config/aarch64/t-aarch64 (comma): New define.
6608         (MULTILIB_OPTIONS): Ditto.
6609         (MULTILIB_DIRNAMES): Ditto.
6610         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
6611         * doc/invoke.texi: Document -mabi for AArch64.
6613 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
6615         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
6617 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6618             Anton Blanchard  <anton@au1.ibm.com>
6620         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
6621         endianness when selecting field to splat.
6623 2013-07-22  Eric Christopher  <echristo@gmail.com>
6625         * dwarf2out.c (die_odr_checksum): New function to use
6626         CHECKSUM_ macros and ULEB128 for DIE tag.
6627         (generate_type_signature): Use.
6629 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
6631         * config.gcc (sparc*-*-*): Accept leon3 processor.
6632         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
6633         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
6634         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
6635         * config/sparc/sparc.opt (enum processor_type): Add leon3.
6636         (mfix-ut699): Adjust comment.
6637         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
6638         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
6639         (CPP_CPU_SPEC): Likewise.
6640         (ASM_CPU_SPEC): Likewise.
6641         * config/sparc/sparc.c (leon3_cost): New constant.
6642         (sparc_option_override): Add leon3 support.
6643         (mem_ref): New function.
6644         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
6645         (sparc_do_work_around_errata): Look into the instruction in the delay
6646         slot and adjust accordingly.  Add fix for the data cache nullify issues
6647         of the UT699.  Change insertion position for the NOP.
6648         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
6649         (leon3_load): New reservation.
6650         (leon_store): Bump latency to 2.
6651         (grfpu): New automaton.
6652         (grfpu_alu): New unit.
6653         (grfpu_ds): Likewise.
6654         (leon_fp_alu): Adjust.
6655         (leon_fp_mult): Delete.
6656         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
6657         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
6658         * config/sparc/sparc.md (cpu): Add leon3.
6659         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
6660         (swapsi): Likewise.
6661         (atomic_test_and_set): Likewise.
6662         (ldstub): Likewise.
6664 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
6666         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
6667         default for R5900 targets.
6668         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
6669         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
6670         * config/mips/mips.c (mips_option_override): Report an error for
6671         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
6672         for -march=r5900 -mhard-float.
6674 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6676         * df-problems.c (can_move_insns_across): Exit loop once we
6677         find a non-fixed, non-global register.
6679         * ipa-pure-const.c (propagate_nothrow): Exit loop after
6680         setting can_throw.
6682         * omega.c (omega_eliminate_red): Break after setting red_found.
6683         (omega_problem_has_red_equations): Similarly after setting found.
6684         (omega_query_variable): Similarly after setting coupled.
6686 2013-07-22  Marek Polacek  <polacek@redhat.com>
6688         * gimplify.c: Don't include gimple.h twice.
6690 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6692         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
6693         instead of TARGET_THUMB1.
6694         (Pz): New constraint.
6695         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
6696         encodings.
6697         (compare_negsi_si): Likewise.
6698         (compare_addsi2_op0): Likewise.
6699         (compare_addsi2_op1): Likewise.
6700         (addsi3_carryin_<optab>): Likewise.
6701         (addsi3_carryin_alt2_<optab>): Likewise.
6702         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
6703         for arm_restrict_it.
6704         (subsi3_carryin): Likewise.
6705         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
6706         (minmax_arithsi): Disable for arm_restrict_it.
6707         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
6708         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
6709         (satsi_<SAT:code>_shift): Likewise.
6710         (arm_shiftsi3): Add alternative for 16-bit encoding.
6711         (arm32_movhf): Disable for arm_restrict_it.
6712         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
6713         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
6715 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
6717         * config/arm/arm.md (attribute "insn"): Delete.
6718         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
6719         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
6720         (not_shiftsi): Update for attribute change.
6721         (not_shiftsi_compare0): Likewise.
6722         (not_shiftsi_compare0_scratch): Likewise.
6723         (arm_one_cmplsi2): Likewise.
6724         (thumb1_one_cmplsi2): Likewise.
6725         (notsi_compare0): Likewise.
6726         (notsi_compare0_scratch): Likewise.
6727         (thumb1_movdi_insn): Likewise.
6728         (arm_movsi_insn): Likewise.
6729         (movhi_insn_arch4): Likewise.
6730         (movhi_bytes): Likewise.
6731         (arm_movqi_insn): Likewise.
6732         (thumb1_movqi_insn): Likewise.
6733         (arm32_movhf): Likewise.
6734         (thumb1_movhf): Likewise.
6735         (arm_movsf_soft_insn): Likewise.
6736         (thumb1_movsf_insn): Likewise.
6737         (thumb_movdf_insn): Likewise.
6738         (movsicc_insn): Likewise.
6739         (movsfcc_soft_insn): Likewise.
6740         (and_scc): Likewise.
6741         (cond_move): Likewise.
6742         (if_move_not): Likewise.
6743         (if_not_move): Likewise.
6744         (if_shift_move): Likewise.
6745         (if_move_shift): Likewise.
6746         (if_shift_shift): Likewise.
6747         (if_not_arith): Likewise.
6748         (if_arith_not): Likewise.
6749         (cond_move_not): Likewise.
6750         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
6751         (neon_mov<mode>): Likewise.
6752         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
6753         (thumb2_movsi_vfp): Likewise.
6754         (movsf_vfp): Likewise.
6755         (thumb2_movsf_vfp): Likewise.
6756         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
6757         change.
6758         (cortexa7_older_only): Likewise.
6759         (cortexa7_younger): Likewise.
6760         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
6761         (1020alu_shift_op): Likewise.
6762         (1020alu_shift_reg_op): Likewise.
6763         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
6764         (alu_shift_op): Likewise.
6765         (alu_shift_reg_op): Likewise.
6766         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
6767         (11_alu_shift_op): Likewise.
6768         (11_alu_shift_reg_op): Likewise.
6769         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
6770         (9_alu_shift_reg_op): Likewise.
6771         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6772         change.
6773         (cortex_a15_alu_shift): Likewise.
6774         (cortex_a15_alu_shift_reg): Likewise.
6775         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
6776         (cortex_a5_alu_shift): Likewise.
6777         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
6778         change.
6779         (cortex_a53_alu_shift): Likewise.
6780         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
6781         change.
6782         (cortex_a7_alu_reg): Likewise.
6783         (cortex_a7_alu_shift): Likewise.
6784         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6785         (cortex_a8_alu_shift): Likewise.
6786         (cortex_a8_alu_shift_reg): Likewise.
6787         (cortex_a8_mov): Likewise.
6788         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
6789         (cortex_a9_dp_shift): Likewise.
6790         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
6791         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
6792         (cortex_r4_mov): Likewise.
6793         (cortex_r4_alu_shift): Likewise.
6794         (cortex_r4_alu_shift_reg): Likewise.
6795         * config/arm/fa526.md (526_alu_op): Update for attribute change.
6796         (526_alu_shift_op): Likewise.
6797         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
6798         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
6799         (626te_alu_shift_op): Likewise.
6800         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
6801         (726te_alu_op): Likewise.
6802         (726te_alu_shift_op): Likewise.
6803         (726te_alu_shift_reg_op): Likewise.
6804         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
6805         (mp626_alu_shift_op): Likewise.
6806         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
6807         (pj4_alu_e1_conds): Likewise.
6808         (pj4_alu): Likewise.
6809         (pj4_alu_conds): Likewise.
6810         (pj4_shift): Likewise.
6811         (pj4_shift_conds): Likewise.
6812         (pj4_alu_shift): Likewise.
6813         (pj4_alu_shift_conds): Likewise.
6815 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6817         * config/arm/predicates.md (shiftable_operator_strict_it):
6818         New predicate.
6819         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
6820         Disable cond_exec version for arm_restrict_it.
6821         (thumb2_smaxsi3): Convert to generate cond_exec.
6822         (thumb2_sminsi3): Likewise.
6823         (thumb32_umaxsi3): Likewise.
6824         (thumb2_uminsi3): Likewise.
6825         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
6826         (thumb2_neg_abssi2): Likewise.
6827         (thumb2_mov_scc): Add alternative for 16-bit encoding.
6828         (thumb2_movsicc_insn): Adjust alternatives.
6829         (thumb2_mov_negscc): Disable for arm_restrict_it.
6830         (thumb2_mov_negscc_strict_it): New pattern.
6831         (thumb2_mov_notscc_strict_it): New pattern.
6832         (thumb2_mov_notscc): Disable for arm_restrict_it.
6833         (thumb2_ior_scc): Likewise.
6834         (thumb2_ior_scc_strict_it): New pattern.
6835         (thumb2_cond_move): Adjust for arm_restrict_it.
6836         (thumb2_cond_arith): Disable for arm_restrict_it.
6837         (thumb2_cond_arith_strict_it): New pattern.
6838         (thumb2_cond_sub): Adjust for arm_restrict_it.
6839         (thumb2_movcond): Likewise.
6840         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
6841         (thumb2_zero_extendhisi2_v6): Likewise.
6842         (thumb2_zero_extendqisi2_v6): Likewise.
6843         (orsi_notsi_si): Likewise.
6844         (orsi_not_shiftsi_si): Likewise.
6846 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
6848         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
6849         instruction sequence is 1 byte shorter.
6851 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
6853         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
6854         it is not needed after split.
6856 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
6858         PR target/51784
6859         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
6860         second label for nonlocal goto receivers. Don't output pic base labels
6861         unless we're producing PIC; mark that action unreachable().
6862         (ix86_save_reg): If the function contains a nonlocal label, save the
6863         PIC base reg.
6864         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
6865         * config/darwin.c (emitted_pic_label_num): New GTY.
6866         (update_pic_label_number_if_needed): New.
6867         (machopic_output_function_base_name): Adjust for nonlocal receiver
6868         case.
6869         (machopic_should_output_picbase_label): New.
6870         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
6871         (nonlocal_goto_receiver): New insn and split.
6873 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
6875         * config/aarch64/aarch64-builtins.c
6876         (aarch64_fold_builtin): Fold abs in all modes.
6877         * config/aarch64/aarch64-simd-builtins.def
6878         (abs): Enable for all modes.
6879         * config/aarch64/arm_neon.h
6880         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
6881         (vabs_f64): Add missing intrinsic.
6883 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
6885         * config/aarch64/arm_neon.h (vabs_s64): New function
6887 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
6889         PR target/57516
6890         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
6891         * config/avr/avr.md (adjust_len): Add `round'.
6892         * config/avr/avr-protos.h (avr_out_round): New prototype.
6893         (avr_out_plus): Add `out_label' argument.
6894         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
6895         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
6896         Handle the case where `insn' is just a pattern.
6897         (avr_out_bitop): Handle the case where `insn' is just a pattern.
6898         (avr_out_round): New function.
6899         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
6901 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
6903         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6904         Add check for flag_stack_usage to handle -fstack-usage support
6906 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
6908         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
6909         interaction for new Power8 flags and VSX.
6911 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
6913         PR middle-end/57698
6914         * tree-inline.c (expand_call_inline): Emit errors during
6915         early_inlining only if optimization is not turned on.
6917 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6919         * passes.def: New.
6921         * passes.c (init_optimization_passes): Move the construction of
6922         the pass hierarchy into a new passes.def file.
6924         * Makefile.in (passes.o): Add dependency on passes.def.
6926 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6928         * passes.c (init_optimization_passes): Introduce macros for
6929         constructing the tree of passes (INSERT_PASSES_AFTER,
6930         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
6931         TERMINATE_PASS_LIST).
6933 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
6934             Wei Mi  <wmi@google.com>
6936         PR rtl-optimization/57878
6937         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
6938         top.
6939         (reload_pseudo_compare_func): Check nregs first for reload
6940         pseudos.
6942 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6944         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
6946 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
6948         * read-rtl.c (validate_const_int): Once an invalid character is
6949         seen, quit the loop.
6951         * gengtype.c (write_roots): Similarly once we find the "deletable"
6952         or "if_marked" option.
6954 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6956         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
6957         "xtab" and "sat".  Move value "clz" from here to ...
6958         (attriubte "type"): ... here.
6959         (satsi_<SAT:code>): Delete "insn" attribute.
6960         (satsi_<SAT:code>_shift): Likewise.
6961         (arm_zero_extendqisi2addsi): Likewise.
6962         (arm_extendqisi2addsi): Likewise.
6963         (clzsi2): Update for attribute changes.
6964         (rbitsi2): Likewise.
6965         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
6966         attribute.
6967         (arm_usatsihi): Likewise.
6968         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6970 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6972         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
6973         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
6974         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
6975         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
6976         in alphabetical order.
6977         (attribute "core_cycles"): Update for attribute changes.
6978         (arm_addsi3): Likewise.
6979         (addsi3_compare0): Likewise.
6980         (addsi3_compare0_scratch): Likewise.
6981         (addsi3_compare_op1): Likewise.
6982         (addsi3_compare_op2): Likewise.
6983         (compare_addsi2_op0): Likewise.
6984         (compare_addsi2_op1): Likewise.
6985         (addsi3_carryin_shift_<optab>): Likewise.
6986         (subsi3_carryin_shift): Likewise.
6987         (rsbsi3_carryin_shift): Likewise.
6988         (arm_subsi3_insn): Likewise.
6989         (subsi3_compare0): Likewise.
6990         (subsi3_compare): Likewise.
6991         (arm_andsi3_insn): Likewise.
6992         (thumb1_andsi3_insn): Likewise.
6993         (andsi3_compare0): Likewise.
6994         (andsi3_compare0_scratch): Likewise.
6995         (zeroextractsi_compare0_scratch
6996         (andsi_not_shiftsi_si): Likewise.
6997         (iorsi3_insn): Likewise.
6998         (iorsi3_compare0): Likewise.
6999         (iorsi3_compare0_scratch): Likewise.
7000         (arm_xorsi3): Likewise.
7001         (thumb1_xorsi3_insn): Likewise.
7002         (xorsi3_compare0): Likewise.
7003         (xorsi3_compare0_scratch): Likewise.
7004         (satsi_<SAT:code>_shift): Likewise.
7005         (rrx): Likewise.
7006         (arm_shiftsi3): Likewise.
7007         (shiftsi3_compare0): Likewise.
7008         (not_shiftsi): Likewise.
7009         (not_shiftsi_compare0): Likewise.
7010         (not_shiftsi_compare0_scratch): Likewise.
7011         (arm_one_cmplsi2): Likewise.
7012         (thumb_one_complsi2): Likewise.
7013         (notsi_compare0): Likewise.
7014         (notsi_compare0_scratch): Likewise.
7015         (thumb1_zero_extendhisi2): Likewise.
7016         (arm_zero_extendhisi2): Likewise.
7017         (arm_zero_extendhisi2_v6): Likewise.
7018         (arm_zero_extendhisi2addsi): Likewise.
7019         (thumb1_zero_extendqisi2): Likewise.
7020         (thumb1_zero_extendqisi2_v6): Likewise.
7021         (arm_zero_extendqisi2): Likewise.
7022         (arm_zero_extendqisi2_v6): Likewise.
7023         (arm_zero_extendqisi2addsi): Likewise.
7024         (thumb1_extendhisi2): Likewise.
7025         (arm_extendhisi2): Likewise.
7026         (arm_extendhisi2_v6): Likewise.
7027         (arm_extendqisi): Likewise.
7028         (arm_extendqisi_v6): Likewise.
7029         (arm_extendqisi2addsi): Likewise.
7030         (thumb1_extendqisi2): Likewise.
7031         (thumb1_movdi_insn): Likewise.
7032         (arm_movsi_insn): Likewise.
7033         (movsi_compare0): Likewise.
7034         (movhi_insn_arch4): Likewise.
7035         (movhi_bytes): Likewise.
7036         (arm_movqi_insn): Likewise.
7037         (thumb1_movqi_insn): Likewise.
7038         (arm32_movhf): Likewise.
7039         (thumb1_movhf): Likewise.
7040         (arm_movsf_soft_insn): Likewise.
7041         (thumb1_movsf_insn): Likewise.
7042         (movdf_soft_insn): Likewise.
7043         (thumb_movdf_insn): Likewise.
7044         (arm_cmpsi_insn): Likewise.
7045         (cmpsi_shiftsi): Likewise.
7046         (cmpsi_shiftsi_swp): Likewise.
7047         (arm_cmpsi_negshiftsi_si): Likewise.
7048         (movsicc_insn): Likewise.
7049         (movsfcc_soft_insn): Likewise.
7050         (arith_shiftsi): Likewise.
7051         (arith_shiftsi_compare0
7052         (arith_shiftsi_compare0_scratch
7053         (sub_shiftsi): Likewise.
7054         (sub_shiftsi_compare0
7055         (sub_shiftsi_compare0_scratch
7056         (and_scc): Likewise.
7057         (cond_move): Likewise.
7058         (if_plus_move): Likewise.
7059         (if_move_plus): Likewise.
7060         (if_move_not): Likewise.
7061         (if_not_move): Likewise.
7062         (if_shift_move): Likewise.
7063         (if_move_shift): Likewise.
7064         (if_shift_shift): Likewise.
7065         (if_not_arith): Likewise.
7066         (if_arith_not): Likewise.
7067         (cond_move_not): Likewise.
7068         (thumb1_ashlsi3): Set type attribute.
7069         (thumb1_ashrsi3): Likewise.
7070         (thumb1_lshrsi3): Likewise.
7071         (thumb1_rotrsi3): Likewise.
7072         (shiftsi3_compare0_scratch): Likewise.
7073         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7074         (neon_mov<mode>): Likewise.
7075         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7076         attribute changes.
7077         (thumb2_movsi_insn): Likewise.
7078         (thumb2_cmpsi_neg_shiftsi): Likewise.
7079         (thumb2_extendqisi_v6): Likewise.
7080         (thumb2_zero_extendhisi2_v6): Likewise.
7081         (thumb2_zero_extendqisi2_v6): Likewise.
7082         (thumb2_shiftsi3_short): Likewise.
7083         (thumb2_addsi3_compare0_scratch): Likewise.
7084         (orsi_not_shiftsi_si): Likewise.
7085         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7086         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7087         changes.
7088         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7089         (1020alu_shift_op): Likewise.
7090         (1020alu_shift_reg_op): Likewise.
7091         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7092         (alu_shift_op): Likewise.
7093         (alu_shift_reg_op): Likewise.
7094         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7095         (11_alu_shift_op): Likewise.
7096         (11_alu_shift_reg_op): Likewise.
7097         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7098         (9_alu_shift_reg_op): Likewise.
7099         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7100         changes.
7101         (cortex_a15_alu_shift): Likewise.
7102         (cortex_a15_alu_shift_reg): Likewise.
7103         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7104         changes.
7105         (cortex_a5_alu_shift): Likewise.
7106         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7107         changes.
7108         (cortex_a53_alu_shift): Likewise.
7109         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7110         changes.
7111         (cortex_a7_alu_reg): Likewise.
7112         (cortex_a7_alu_shift): Likewise.
7113         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7114         changes.
7115         (cortex_a8_alu_shift): Likewise.
7116         (cortex_a8_alu_shift_reg): Likewise.
7117         (cortex_a8_mov): Likewise.
7118         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7119         (cortex_a9_dp_shift): Likewise.
7120         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7121         changes.
7122         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7123         changes.
7124         (cortex_r4_mov): Likewise.
7125         (cortex_r4_alu_shift): Likewise.
7126         (cortex_r4_alu_shift_reg): Likewise.
7127         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7128         (526_alu_shift_op): Likewise.
7129         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7130         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7131         (626te_alu_shift_op): Likewise.
7132         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7133         (726te_alu_op): Likewise.
7134         (726te_alu_shift_op): Likewise.
7135         (726te_alu_shift_reg_op): Likewise.
7136         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7137         (mp626_alu_shift_op): Likewise.
7138         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7139         (pj4_alu_e1_conds): Likewise.
7140         (pj4_alu): Likewise.
7141         (pj4_alu_conds): Likewise.
7142         (pj4_shift): Likewise.
7143         (pj4_shift_conds): Likewise.
7144         (pj4_alu_shift): Likewise.
7145         (pj4_alu_shift_conds): Likewise.
7146         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7147         changes.
7148         (cortexa7_older_only): Likewise.
7149         (cortexa7_younger): Likewise.
7151 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7153         * ipa-pure-const.c (generate_summary): Rename to...
7154         (pure_const_generate_summary): ... this.
7156 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7158         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7160 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7162         PR target/57909
7163         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7164         usage in HI mode.
7166 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7168         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7169         enabled without -march=zEC12.
7170         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7171         flags to be set.
7173 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7175         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7176         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7177         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7178         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7179         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7180         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7181         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7182         ISA_HAS_FP4.
7183         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7184         and ISA_HAS_NMADD3_NMSUB3.
7185         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7186         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7187         (nmsub4<mode>, nmsub3<mode>): Likewise.
7188         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7190 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7192         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7193         TARGET_MIPS5400 checking.
7195 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7196             Peter Bergner  <bergner@vnet.ibm.com>
7198         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7199         registers in the comment.
7200         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7201         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7202         rather than FIRST_PSEUDO_REGISTERS.
7204 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7206         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7207         enable extra ISA flags with TARGET_HTM.
7209 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7211         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7212         Fix comment typos.
7214 2013-07-15  Cong Hou  <congh@google.com>
7216         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7217         in compare function for sorting.
7219 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
7221         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7222         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7223         * config/rs6000/rs6000.opt: Add -mhtm option.
7224         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7225         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7226         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7227         __HTM__ if the HTM instructions are available.
7228         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7229         htm_spr_reg_operand): New define_predicates.
7230         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7231         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7232         Include htm.md.
7233         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7234         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7235         HTM builtin functions.
7236         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7237         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7238         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7239         (rs6000_builtin_mask_calculate): Likewise.
7240         (rs6000_option_override_internal): Likewise.
7241         (bdesc_htm): Add new HTM builtin support.
7242         (htm_spr_num): New function.
7243         (htm_spr_regno): Likewise.
7244         (rs6000_htm_spr_icode): Likewise.
7245         (htm_expand_builtin): Likewise.
7246         (htm_init_builtins): Likewise.
7247         (rs6000_expand_builtin): Add support for HTM builtin functions.
7248         (rs6000_init_builtins): Likewise.
7249         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7250         option.
7251         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7252         (TARGET_HTM, MASK_HTM): Define macros.
7253         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7254         (FIXED_REGISTERS): Likewise.
7255         (CALL_USED_REGISTERS): Likewise.
7256         (CALL_REALLY_USED_REGISTERS): Likewise.
7257         (REG_ALLOC_ORDER): Likewise.
7258         (enum reg_class): Likewise.
7259         (REG_CLASS_NAMES): Likewise.
7260         (REG_CLASS_CONTENTS): Likewise.
7261         (REGISTER_NAMES): Likewise.
7262         (ADDITIONAL_REGISTER_NAMES): Likewise.
7263         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7264         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7265         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7266         * config/rs6000/htm.md: New file.
7267         * config/rs6000/htmintrin.h: New file.
7268         * config/rs6000/htmxlintrin.h: New file.
7270 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7272         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7273         Define SYMBOL_TINY_GOT, update comment.
7274         * config/aarch64/aarch64.c
7275         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7276         (aarch64_expand_mov_immediate): Likewise.
7277         (aarch64_print_operand): Likewise.
7278         (aarch64_classify_symbol): Likewise.
7279         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7280         (ldr_got_tiny): Define.
7282 2013-07-13  Tobias Grosser  <tobias@grosser.es>
7284         PR tree-optimization/54094
7285         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7286           scheduling dimension for the parallelism check from the polyhedral
7287           information in the AST.
7288         * graphite-dependences.c (carries_deps): Do not assume the schedule is
7289           in 2D + 1 form.
7291 2013-07-13  Jason Merrill  <jason@redhat.com>
7293         * print-tree.c (debug_vec_tree): Use debug_raw.
7294         (debug_raw (vec<tree, va_gc> &)): New.
7295         (debug_raw (vec<tree, va_gc> *)): New.
7296         * tree.h: Declare them.
7298 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
7300         * ifcvt.c (ifcvt_after_combine): New static variable.
7301         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7302         for size.
7303         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
7304         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7305         rest_of_handle_if_after_reload): Pass new argument for if_convert.
7307 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
7309         * config/mips/mips.c (mips_expand_call): Remove empty statement.
7311 2013-07-12  Michael Matz  <matz@suse.de>
7313         PR middle-end/55771
7314         * convert.c (convert_to_real): Reject non-float inner types.
7316 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
7318         * config/aarch64/aarch64-protos.h
7319         (aarch64_simd_immediate_valid_for_move): Remove.
7320         * config/aarch64/aarch64.c (simd_immediate_info): New member.
7321         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7322         cases.
7323         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7325 2013-07-11  Steve Ellcey  <sellcey@mips.com>
7327         * config/mips/mips.c (mips_conditional_register_usage): Do not
7328         use t[0-7] registers in MIPS16 mode when optimizing for size.
7330 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
7332         * config/i386/i386.c (dispatch_function_versions): Fix array
7333         indexing of function_version_info to match actual_versions.
7335 2013-07-11  Teresa Johnson  <tejohnson@google.com>
7337         * vec.h (struct va_gc): Move release out-of-line.
7338         (va_gc::release): Call ggc_free on released vec.
7340 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7342         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7343         Require GOT register as additional operand in UNSPEC.
7344         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7345         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7346         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7347         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7348         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7349         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7350         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7352 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
7354         PR target/57631
7355         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7356         name seen by assembler/linker rather if available.
7358 2013-07-11  Andreas Schwab  <schwab@suse.de>
7360         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7362 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
7364         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7366 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
7368         * doc/tm.texi.in: Move hook documentation to ....
7369         * target.def: ... here.
7371         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7372         text on @hook line.
7373         * doc/tm.texi: Regenerate.
7375 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
7377         PR c++/57869
7378         * doc/invoke.texi: Document -Wconditionally-supported.
7380 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7382         PR target/57844
7383         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7384         of my_fp.
7386 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7388         PR target/57506
7389         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7390         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7391         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7392         Remove duplicate devices.
7393         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7394         * config/avr/t-multilib: Regenerate.
7395         * config/avr/avr-tables.opt: Regenerate.
7396         * doc/avr-mmcu.texi: Regenerate.
7398 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7400         PR target/56987
7401         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7403 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
7405         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7406         the cost of MULT when optimizing for size.
7408 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7410         * config/cr16/cr16-protos.h: Don't include target.h.
7412 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
7414         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7415         adjust register size for TDmode and TFmode for VSX registers.
7417 2013-07-08  Kai Tietz  <ktietz@redhat.com>
7419         PR target/56892
7420         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7421         hook_bool_const_tree_true.
7423 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7425         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7426         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7427         * config/s390/s390.md: Define FPR*_REGNUM constants.
7428         Fix FPR2_REGNUM constant (18 -> 17).
7429         ("*trunc<BFP:mode><DFP_ALL:mode>2")
7430         ("*trunc<DFP_ALL:mode><BFP:mode>2")
7431         ("trunc<BFP:mode><DFP_ALL:mode>2")
7432         ("trunc<DFP_ALL:mode><BFP:mode>2")
7433         ("*extend<BFP:mode><DFP_ALL:mode>2")
7434         ("*extend<DFP_ALL:mode><BFP:mode>2")
7435         ("extend<BFP:mode><DFP_ALL:mode>2")
7436         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7437         FPR4_REGNUM.
7439 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
7441         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7443 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7445         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7446         and cfun_fpr_bit_p to cfun_fpr_save_p.
7447         (s390_frame_area, s390_register_info, s390_frame_info)
7448         (s390_emit_prologue, s390_emit_epilogue)
7449         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7450         register numbers.
7451         * config/s390/s390.h: Define *_REGNUM macros for floating point
7452         register numbers.
7454 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
7456         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7458 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
7460         PR rtl-optimization/57786
7461         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7462         and break out of the loop when it is set to false.
7464 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
7466         PR target/57819
7467         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7468         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7469         (const_int 63)) 0)).
7470         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7471         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7472         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7474         PR rtl-optimization/57829
7475         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7476         mask bits outside of mode are just sign-extension from mode to HWI.
7478 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7480         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7481         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7482         adjust_address instead of change_address to keep info about alignment.
7483         (emit_strmov): Remove.
7484         (emit_memmov): New function.
7485         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7486         (expand_movmem_epilogue): Likewise and return updated rtx for
7487         destination.
7488         (expand_constant_movmem_prologue): Likewise and return updated rtx for
7489         destination and source.
7490         (decide_alignment): Refactor, handle vector_loop.
7491         (ix86_expand_movmem): Likewise.
7492         (ix86_expand_setmem): Likewise.
7493         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7495 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
7497         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7498         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7500 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7502         * config/i386/sse.md (sse_movlhps): Change alternative 3
7503         of operand 2 to "m".
7505 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7507         PR target/57807
7508         * config/i386/sse.md (iptr): New mode attribute.
7509         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7510         (<sse>_vm<plusminus_insn><mode>3): Ditto.
7511         (<sse>_vmmul<mode>3): Ditto.
7512         (<sse>_vmdiv<mode>3): Ditto.
7513         (sse_vmrcpv4sf2): Ditto.
7514         (<sse>_vmsqrt<mode>2): Ditto.
7515         (sse_vmrsqrtv4sf2): Ditto.
7516         (<sse>_vm<code><mode>3): Ditto.
7517         (avx_vmcmp<mode>3): Ditto.
7518         (<sse>_vmmaskcmp<mode>3): Ditto.
7519         (<sse>_comi): Ditto.
7520         (<sse>_ucomi): Ditto.
7521         (*xop_vmfrcz_<mode>): Ditto.
7522         (*fmai_fmadd_<mode>): Ditto.
7523         (*fmai_fmsub_<mode>): Ditto.
7524         (*fmai_fnmadd_<mode>): Ditto.
7525         (*fmai_fnmsub_<mode>): Ditto.
7526         (*fma4i_vmfmadd_<mode>): Ditto.
7527         (*fma4i_vmfmsub_<mode>): Ditto.
7528         (*fma4i_vmfnmadd_<mode>): Ditto.
7529         (*fma4i_vmfnmsub_<mode>): Ditto.
7530         (*xop_vmfrcz_<mode>): Ditto.
7531         (sse_cvtps2pi): Ditto.
7532         (sse_cvttps2pi): Ditto.
7533         (sse_cvtss2si): Ditto.
7534         (sse_cvtss2si_2): Ditto.
7535         (sse_cvtss2siq_2): Ditto.
7536         (sse_cvttss2si): Ditto.
7537         (sse_cvttss2siq): Ditto.
7538         (sse_cvtsd2si): Ditto.
7539         (sse_cvtsd2si_2): Ditto.
7540         (sse_cvtsd2siq_2): Ditto.
7541         (sse_cvttsd2si): Ditto.
7542         (sse_cvttsd2siq): Ditto.
7543         (sse_cvtsd2ss): Ditto.
7544         (sse_cvtss2sd): Ditto.
7545         (avx2_pbroadcast<mode>): Ditto.
7546         (avx2_pbroadcast<mode>_1): Ditto.
7547         (*avx_vperm_broadcast_v4sf): Ditto.
7549         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7550         (sse_movlhps): Ditto.
7551         (sse_storehps): Ditto.
7552         (sse_loadhps): Ditto.
7553         (sse_storelps): Ditto.
7554         (sse_loadlps): Ditto.
7555         (*vec_concatv4sf): Ditto.
7556         (*vec_interleave_highv2df): Ditto.
7557         (*vec_interleave_lowv2df): Ditto.
7558         (*vec_extractv2df_1_sse): Ditto.
7559         (*vec_extractv2df_0_sse): Ditto.
7560         (sse2_storelpd): Ditto.
7561         (sse2_loadlpd): Ditto.
7562         (sse2_movsd): Ditto.
7563         (*vec_concatv4si): Ditto.
7564         (vec_concatv2di): Ditto.
7566         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
7567         for Intel asm dialect.
7568         (mmx_punpcklwd): Ditto.
7569         (mmx_punpckldq): Ditto.
7571         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
7572         for intel assembler dialect.
7574 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
7576         PR target/29776
7577         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
7578         for BUILT_IN_C{LZ,LRSB}*.
7579         * tree.h (CASE_INT_FN): Add FN##IMAX case.
7580         * tree-vrp.c (extract_range_basic): Handle
7581         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
7582         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
7583         fall thru to code calling set_value*.
7584         * builtins.c (expand_builtin): Remove *IMAX cases.
7585         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
7586         if width is bigger than 2*HWI.
7588 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
7590         PR rtl-optimization/55342
7591         * lra-int.h (lra_subreg_reload_pseudos): New.
7592         * lra.c: Add undoing optional reloads to the block diagram.
7593         (lra_subreg_reload_pseudos): New.
7594         (lra_optional_reload_pseudos): Change comments.
7595         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
7596         lra_optional_reload_pseudos after undo transformations.
7597         * lra-assigns.c (pseudo_prefix_title): New.
7598         (lra_setup_reg_renumber): Use it.
7599         (spill_for): Ditto.  Check subreg reload pseudos too.
7600         (assign_by_spills): Consider subreg reload pseudos too.
7601         * lra-constraints.c (simplify_operand_subreg): Use
7602         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
7603         (curr_insn_transform): Recognize and do optional reloads.
7604         (undo_optional_reloads): New.
7605         (lra_undo_inheritance): Call undo_optional_reloads.
7607 2013-07-05  Thomas Quinot  <quinot@adacore.com>
7609         * tree-complex.c (expand_complex_operations_1): Fix typo.
7611 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
7613         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
7614         (tune_params): New member 'const vec_costs'.
7615         * config/aarch64/aarch64.c (generic_vector_cost): New.
7616         (generic_tunings): New member 'generic_vector_cost'.
7617         (aarch64_builtin_vectorization_cost): New.
7618         (aarch64_add_stmt_cost): New.
7619         (TARGET_VECTORIZE_ADD_STMT_COST): New.
7620         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
7622 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
7624         PR target/57777
7625         * config/i386/predicates.md (vsib_address_operand): Disallow
7626         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
7628 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
7630         PR middle-end/55030
7631         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
7632         expand_builtin_setjmp_receiver.
7633         (expand_label): Adjust, call expand_builtin_setjmp_receiver
7634         with NULL for the label parameter.
7635         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
7636         the frame-pointer.  Adjust comments.
7637         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
7638         only if LABEL is non-NULL.
7640 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
7642         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
7643         (ARM_ABI_AAPCS64): Ditto.
7644         (arm_abi): Ditto.
7645         (ARM_DEFAULT_ABI): Ditto.
7647 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
7649         * config/aarch64/aarch64-builtins.c
7650         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
7651         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
7652         (st1): Likewise.
7653         * config/aarch64/aarch64-simd.md
7654         (aarch64_ld1<VALL:mode>): New.
7655         (aarch64_st1<VALL:mode>): Likewise.
7656         * config/aarch64/arm_neon.h
7657         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
7659 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
7661         * config/i386/i386.c (gate_insert_vzeroupper): Check if
7662         target ISA is AVX.
7663         (ix86_option_override_internal):Turn on all -mavx target flags by
7664         default as they are dependent on AVX anyway.
7666 2013-07-02  Cary Coutant  <ccoutant@google.com>
7668         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
7669         deterministic hash.
7670         (loc_checksum_ordered): Likewise.
7671         (hash_loc_operands): Remove inline keyword.
7673 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
7675         PR tree-optimization/57741
7676         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
7677         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
7678         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
7679         Allow REAL_CST step_exprs if flag_associative_math.
7680         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
7682 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7684         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
7686 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7688         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
7690 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7692         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
7693         encoding.
7694         (iorsi3_insn): Likewise.
7695         (arm_xorsi3): Likewise.
7697 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
7699         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
7700         to ...
7701         (attribute "type"): ... here, and prefix with "wmmx_".
7702         (attribute "core_cycles"): Update for attribute changes.
7703         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
7704         (tbcstv4hi): Likewise.
7705         (tbcstv2si): Likewise.
7706         (iwmmxt_iordi3): Likewise.
7707         (iwmmxt_xordi3): Likewise.
7708         (iwmmxt_anddi3): Likewise.
7709         (iwmmxt_nanddi3): Likewise.
7710         (iwmmxt_arm_movdi): Likewise.
7711         (iwmmxt_movsi_insn): Likewise.
7712         (mov<mode>_internal): Likewise.
7713         (and<mode>3_iwmmxt): Likewise.
7714         (ior<mode>3_iwmmxt): Likewise.
7715         (xor<mode>3_iwmmxt): Likewise.
7716         (add<mode>3_iwmmxt): Likewise.
7717         (ssaddv8qi3): Likewise.
7718         (ssaddv4hi3): Likewise.
7719         (ssaddv2si3): Likewise.
7720         (usaddv8qi3): Likewise.
7721         (usaddv4hi3): Likewise.
7722         (usaddv2si3): Likewise.
7723         (sub<mode>3_iwmmxt): Likewise.
7724         (sssubv8qi3): Likewise.
7725         (sssubv4hi3): Likewise.
7726         (sssubv2si3): Likewise.
7727         (ussubv8qi3): Likewise.
7728         (ussubv4hi3): Likewise.
7729         (ussubv2si3): Likewise.
7730         (mulv4hi3_iwmmxt): Likewise.
7731         (smulv4hi3_highpart): Likewise.
7732         (umulv4hi3_highpart): Likewise.
7733         (iwmmxt_wmacs): Likewise.
7734         (iwmmxt_wmacsz): Likewise.
7735         (iwmmxt_wmacu): Likewise.
7736         (iwmmxt_wmacuz): Likewise.
7737         (iwmmxt_clrdi): Likewise.
7738         (iwmmxt_clrv8qi): Likewise.
7739         (iwmmxt_clr4hi): Likewise.
7740         (iwmmxt_clr2si): Likewise.
7741         (iwmmxt_uavgrndv8qi3): Likewise.
7742         (iwmmxt_uavgrndv4hi3): Likewise.
7743         (iwmmxt_uavgv8qi3): Likewise.
7744         (iwmmxt_uavgv4hi3): Likewise.
7745         (iwmmxt_tinsrb): Likewise.
7746         (iwmmxt_tinsrh): Likewise.
7747         (iwmmxt_tinsrw): Likewise.
7748         (iwmmxt_textrmub): Likewise.
7749         (iwmmxt_textrmsb): Likewise.
7750         (iwmmxt_textrmuh): Likewise.
7751         (iwmmxt_textrmsh): Likewise.
7752         (iwmmxt_textrmw): Likewise.
7753         (iwmxxt_wshufh): Likewise.
7754         (eqv8qi3): Likewise.
7755         (eqv4hi3): Likewise.
7756         (eqv2si3): Likewise.
7757         (gtuv8qi3): Likewise.
7758         (gtuv4hi3): Likewise.
7759         (gtuv2si3): Likewise.
7760         (gtv8qi3): Likewise.
7761         (gtv4hi3): Likewise.
7762         (gtv2si3): Likewise.
7763         (smax<mode>3_iwmmxt): Likewise.
7764         (umax<mode>3_iwmmxt): Likewise.
7765         (smin<mode>3_iwmmxt): Likewise.
7766         (umin<mode>3_iwmmxt): Likewise.
7767         (iwmmxt_wpackhss): Likewise.
7768         (iwmmxt_wpackwss): Likewise.
7769         (iwmmxt_wpackdss): Likewise.
7770         (iwmmxt_wpackhus): Likewise.
7771         (iwmmxt_wpackwus): Likewise.
7772         (iwmmxt_wpackdus): Likewise.
7773         (iwmmxt_wunpckihb): Likewise.
7774         (iwmmxt_wunpckihh): Likewise.
7775         (iwmmxt_wunpckihw): Likewise.
7776         (iwmmxt_wunpckilb): Likewise.
7777         (iwmmxt_wunpckilh): Likewise.
7778         (iwmmxt_wunpckilw): Likewise.
7779         (iwmmxt_wunpckehub): Likewise.
7780         (iwmmxt_wunpckehuh): Likewise.
7781         (iwmmxt_wunpckehuw): Likewise.
7782         (iwmmxt_wunpckehsb): Likewise.
7783         (iwmmxt_wunpckehsh): Likewise.
7784         (iwmmxt_wunpckehsw): Likewise.
7785         (iwmmxt_wunpckelub): Likewise.
7786         (iwmmxt_wunpckeluh): Likewise.
7787         (iwmmxt_wunpckeluw): Likewise.
7788         (iwmmxt_wunpckelsb): Likewise.
7789         (iwmmxt_wunpckelsh): Likewise.
7790         (iwmmxt_wunpckelsw): Likewise.
7791         (ror<mode>3): Likewise.
7792         (ashr<mode>3_iwmmxt): Likewise.
7793         (lshr<mode>3_iwmmxt): Likewise.
7794         (ashl<mode>3_iwmmxt): Likewise.
7795         (ror<mode>3_di): Likewise.
7796         (ashr<mode>3_di): Likewise.
7797         (lshr<mode>3_di): Likewise.
7798         (ashl<mode>3_di): Likewise.
7799         (iwmmxt_wmadds): Likewise.
7800         (iwmmxt_wmaddu): Likewise.
7801         (iwmmxt_tmia): Likewise.
7802         (iwmmxt_tmiaph): Likewise.
7803         (iwmmxt_tmiabb): Likewise.
7804         (iwmmxt_tmiatb): Likewise.
7805         (iwmmxt_tmiabt): Likewise.
7806         (iwmmxt_tmiatt): Likewise.
7807         (iwmmxt_tmovmskb): Likewise.
7808         (iwmmxt_tmovmskh): Likewise.
7809         (iwmmxt_tmovmskw): Likewise.
7810         (iwmmxt_waccb): Likewise.
7811         (iwmmxt_wacch): Likewise.
7812         (iwmmxt_waccw): Likewise.
7813         (iwmmxt_waligni): Likewise.
7814         (iwmmxt_walignr): Likewise.
7815         (iwmmxt_walignr0): Likewise.
7816         (iwmmxt_walignr1): Likewise.
7817         (iwmmxt_walignr2): Likewise.
7818         (iwmmxt_walignr3): Likewise.
7819         (iwmmxt_wsadb): Likewise.
7820         (iwmmxt_wsadh): Likewise.
7821         (iwmmxt_wsadbz): Likewise.
7822         (iwmmxt_wsadhz): Likewise.
7823         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
7824         (iwmmxt_wabsdiffb): Likewise.
7825         (iwmmxt_wabsdiffh): Likewise.
7826         (iwmmxt_wabsdiffw): Likewise.
7827         (iwmmxt_waddsubhx): Likewise
7828         (iwmmxt_wsubaddhx): Likewise.
7829         (addc<mode>3): Likewise.
7830         (iwmmxt_avg4): Likewise.
7831         (iwmmxt_avg4r): Likewise.
7832         (iwmmxt_wmaddsx): Likewise.
7833         (iwmmxt_wmaddux): Likewise.
7834         (iwmmxt_wmaddsn): Likewise.
7835         (iwmmxt_wmaddun): Likewise.
7836         (iwmmxt_wmulwsm): Likewise.
7837         (iwmmxt_wmulwum): Likewise.
7838         (iwmmxt_wmulsmr): Likewise.
7839         (iwmmxt_wmulumr): Likewise.
7840         (iwmmxt_wmulwsmr): Likewise.
7841         (iwmmxt_wmulwumr): Likewise.
7842         (iwmmxt_wmulwl): Likewise.
7843         (iwmmxt_wqmulm): Likewise.
7844         (iwmmxt_wqmulwm): Likewise.
7845         (iwmmxt_wqmulmr): Likewise.
7846         (iwmmxt_wqmulwmr): Likewise.
7847         (iwmmxt_waddbhusm): Likewise.
7848         (iwmmxt_waddbhusl): Likewise.
7849         (iwmmxt_wqmiabb): Likewise.
7850         (iwmmxt_wqmiabt): Likewise.
7851         (iwmmxt_wqmiatb): Likewise.
7852         (iwmmxt_wqmiatt): Likewise.
7853         (iwmmxt_wqmiabbn): Likewise.
7854         (iwmmxt_wqmiabtn): Likewise.
7855         (iwmmxt_wqmiatbn): Likewise.
7856         (iwmmxt_wqmiattn): Likewise.
7857         (iwmmxt_wmiabb): Likewise.
7858         (iwmmxt_wmiabt): Likewise.
7859         (iwmmxt_wmiatb): Likewise.
7860         (iwmmxt_wmiatt): Likewise.
7861         (iwmmxt_wmiabbn): Likewise.
7862         (iwmmxt_wmiabtn): Likewise.
7863         (iwmmxt_wmiatbn): Likewise.
7864         (iwmmxt_wmiattn): Likewise.
7865         (iwmmxt_wmiawbb): Likewise.
7866         (iwmmxt_wmiawbt): Likewise.
7867         (iwmmxt_wmiawtb): Likewise.
7868         (iwmmxt_wmiawtt): Likewise.
7869         (iwmmxt_wmiawbbn): Likewise.
7870         (iwmmxt_wmiawbtn): Likewise.
7871         (iwmmxt_wmiawtbn): Likewise.
7872         (iwmmxt_wmiawttn): Likewise.
7873         (iwmmxt_wmerge): Likewise.
7874         (iwmmxt_tandc<mode>3): Likewise.
7875         (iwmmxt_torc<mode>3): Likewise.
7876         (iwmmxt_torvsc<mode>3): Likewise.
7877         (iwmmxt_textrc<mode>3): Likewise.
7878         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
7879         (wmmxt_pack): Likewise.
7880         (wmmxt_mult_c1): Likewise.
7881         (wmmxt_mult_c2): Likewise.
7882         (wmmxt_alu_c1): Likewise.
7883         (wmmxt_alu_c2): Likewise.
7884         (wmmxt_alu_c3): Likewise.
7885         (wmmxt_transfer_c1): Likewise.
7886         (wmmxt_transfer_c2): Likewise.
7887         (wmmxt_transfer_c3): Likewise.
7888         (marvell_f_iwmmxt_wstr): Likewise.
7889         (marvell_f_iwmmxt_wldr): Likewise.
7891 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
7893         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
7895 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7897         Revert:
7898         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7899         * lra-constraints.c (need_for_split_p): Check call used hard regs
7900         living through calls.
7902         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
7903         call used regs for call insn.
7905 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
7907         PR target/57736
7908         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
7909         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
7911 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7913         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
7915 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7917         * lra-constraints.c (need_for_split_p): Check call used hard regs
7918         living through calls.
7920 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
7922         PR target/57744
7923         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
7924         to tie with any other modes.  Eliminate Altivec vector mode tests,
7925         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
7926         code, to return 0 if testing MODE2 for a condition, if we've
7927         already tested MODE1 for the same condition.
7929 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7931         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
7932         layout.
7934 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7936         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7937         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
7939 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7941         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
7942         Define.
7943         (aarch64_symbolic_constant_p): Remove.
7944         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
7945         static.  Fix line length and white space.
7946         (aarch64_symbolic_constant_p): Remove.
7947         * config/aarch64/predicates.md (aarch64_valid_symref):
7948         Use aarch64_classify_symbol_expression.
7950 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7952         * config/arm/constraints.md (Ts): New constraint.
7953         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
7954         16-bit encodings.
7955         (compare_scc): Use "Ts" constraint for operand 0.
7956         (ior_scc_scc): Likewise.
7957         (and_scc_scc): Likewise.
7958         (and_scc_scc_nodom): Likewise.
7959         (ior_scc_scc_cmp): Likewise for operand 7.
7960         (and_scc_scc_cmp): Likewise.
7961         * config/arm/thumb2.md (thumb2_movsi_insn):
7962         Add alternatives for 16-bit encodings.
7963         (thumb2_movhi_insn): Likewise.
7964         (thumb2_movsicc_insn): Likewise.
7965         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
7966         (thumb2_negscc): Use "Ts" constraint.
7967         Move mvn instruction outside cond_exec block.
7968         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
7969         for 16-bit encodings.
7971 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7973         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
7974         encoding.
7975         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
7976         (mulsi3subsi): Likewise.
7977         (mulsidi3adddi): Likewise.
7978         (mulsidi3_v6): Likewise.
7979         (umulsidi3_v6): Likewise.
7980         (umulsidi3adddi_v6): Likewise.
7981         (smulsi3_highpart_v6): Likewise.
7982         (umulsi3_highpart_v6): Likewise.
7983         (mulhisi3tb): Likewise.
7984         (mulhisi3bt): Likewise.
7985         (mulhisi3tt): Likewise.
7986         (maddhisi4): Likewise.
7987         (maddhisi4tb): Likewise.
7988         (maddhisi4tt): Likewise.
7989         (maddhidi4): Likewise.
7990         (maddhidi4tb): Likewise.
7991         (maddhidi4tt): Likewise.
7992         (zeroextractsi_compare0_scratch): Likewise.
7993         (insv_zero): Likewise.
7994         (insv_t2): Likewise.
7995         (anddi_notzesidi_di): Likewise.
7996         (anddi_notsesidi_di): Likewise.
7997         (andsi_notsi_si): Likewise.
7998         (iordi_zesidi_di): Likewise.
7999         (xordi_zesidi_di): Likewise.
8000         (andsi_iorsi3_notsi): Likewise.
8001         (smax_0): Likewise.
8002         (smax_m1): Likewise.
8003         (smin_0): Likewise.
8004         (not_shiftsi): Likewise.
8005         (unaligned_loadsi): Likewise.
8006         (unaligned_loadhis): Likewise.
8007         (unaligned_loadhiu): Likewise.
8008         (unaligned_storesi): Likewise.
8009         (unaligned_storehi): Likewise.
8010         (extv_reg): Likewise.
8011         (extzv_t2): Likewise.
8012         (divsi3): Likewise.
8013         (udivsi3): Likewise.
8014         (arm_zero_extendhisi2addsi): Likewise.
8015         (arm_zero_extendqisi2addsi): Likewise.
8016         (compareqi_eq0): Likewise.
8017         (arm_extendhisi2_v6): Likewise.
8018         (arm_extendqisi2addsi): Likewise.
8019         (arm_movt): Likewise.
8020         (thumb2_ldrd): Likewise.
8021         (thumb2_ldrd_base): Likewise.
8022         (thumb2_ldrd_base_neg): Likewise.
8023         (thumb2_strd): Likewise.
8024         (thumb2_strd_base): Likewise.
8025         (thumb2_strd_base_neg): Likewise.
8026         (arm_negsi2): Add alternative for 16-bit encoding.
8027         (arm_one_cmplsi2): Likewise.
8029 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8031         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8032         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8033         (movdfcc): Likewise.
8034         * config/arm/vfp.md (*thumb2_movsf_vfp):
8035         Disable predication for arm_restrict_it.
8036         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8037         (*thumb2_movdfcc_vfp): Likewise.
8038         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8039         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8040         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8041         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8042         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8043         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8044         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8045         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8046         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8047         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8048         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8049         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8050         Disable predication for arm_restrict_it.
8052 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8054         * config/i386/bmiintrin.h (_bextr_u32): New.
8055         (_bextr_u64): Ditto.
8057 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8059         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8060         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8061         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8062         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8063         * config/mips/n32-elf.h: ...this new file.
8065 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8067         PR target/57224
8068         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8069         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8071 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8073         * config/mips/mips-tables.opt: Regenerate.
8074         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8075         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8076         * doc/invoke.texi: Add -m14kc.
8078 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8080         PR target/57623
8081         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8082         constraints of operand 1 and 2.
8084         PR target/57623
8085         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8086         to match RTL canonicalization.  Swap predicates and
8087         constraints of operand 1 and 2.
8089 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8091         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8092         Process OP_INOUT regs for splitting too.
8094 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8096         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8097         decl before the loop, initialize to NULL.
8098         (vectorizable_load): Initialize ptr_incr to NULL.
8100 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8102         PR lto/57208
8103         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8104         * ipa-ref.c (ipa_maybe_record_reference): New function.
8105         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8106         * ipa-cp.c (create_specialized_node): Record potential references from
8107         aggvals.
8108         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8110 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8112         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8113         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8114         'mode' to force_reg.
8115         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8116         (aarch64_expand_mov_immediate): Likewise.
8118 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8120         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8121         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8123 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8125         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8126         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8127         (struct machine_function): Add tbegin_p.
8128         (s390_canonicalize_comparison): Fold CC mode compares to
8129         conditional jump if possible.
8130         (s390_emit_jump): Return the emitted jump.
8131         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8132         Handle CCRAWmode compares.
8133         (s390_option_override): Default to -mhtm if available.
8134         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8135         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8136         FPRs instead of df_regs_ever_live_p.
8137         (s390_optimize_nonescaping_tx): New function.
8138         (s390_init_frame_layout): Extend clobbered_regs array to cover
8139         FPRs as well.
8140         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8141         (s390_expand_tbegin): New function.
8142         (enum s390_builtin): New enum definition.
8143         (code_for_builtin): New array definition.
8144         (s390_init_builtins): New function.
8145         (s390_expand_builtin): New function.
8146         (TARGET_INIT_BUILTINS): Define.
8147         (TARGET_EXPAND_BUILTIN): Define.
8148         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8149         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8150         (s390_alc_comparison): Likewise.
8151         * config/s390/s390-modes.def: Add CCRAWmode.
8152         * config/s390/s390.h (processor_flags): Add PF_TX.
8153         (TARGET_CPU_HTM): Define macro.
8154         (TARGET_HTM): Define macro.
8155         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8156         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8157         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8158         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8159         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8160         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8161         ("*cc_to_int"): Move up.
8162         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8163         constants other than 0.
8164         ("*ccraw_to_int"): New insn and splitter definition.
8165         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8166         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8167         ("tx_assist"): New expander.
8168         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8169         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8170         * config/s390/s390.opt: Add -mhtm option.
8171         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8172         * config/s390/htmxlintrin.h: New file.
8173         * config/s390/htmintrin.h: New file.
8174         * config/s390/s390intrin.h: New file.
8175         * doc/extend.texi: Document htm builtins.
8176         * config.gcc: Add the new header files to extra_headers.
8178 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8180         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8181         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8183 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8184             Pat Haugen  <pthaugen@us.ibm.com>
8185             Peter Bergner  <bergner@vnet.ibm.com>
8187         * config/rs6000/power8.md: New.
8188         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8189         setting for power8 entry.
8190         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8191         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8192         test for Power4/Power5 only.
8193         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8194         support.
8195         (force_new_group): Adjust comment.
8196         * config/rs6000/rs6000.md: Include power8.md.
8198 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8200         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8201         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8202         declaration.
8203         (tune_params): Update comment.
8204         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8205         (arm_max_conditional_execute): New function.
8206         (thumb2_final_prescan_insn): Use max_insn_skipped and
8207         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8209 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8211         PR tree-optimization/57705
8212         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8213         SSA_NAME step, provided that it is not defined inside the loop.
8214         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8215         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8217 2013-06-25  Martin Jambor  <mjambor@suse.cz>
8219         PR middle-end/57670
8220         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8221         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8222         calls in the dump.
8223         (ipa_note_param_call): Initialize member_ptr flag.
8224         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8225         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8226         (ipa_write_indirect_edge_info): Stream member_ptr flag.
8227         (ipa_read_indirect_edge_info): Likewise.
8229 2013-06-25  Richard Biener  <rguenther@suse.de>
8231         PR middle-end/56977
8232         * passes.c (init_optimization_passes): Move pass_fold_builtins
8233         and pass_dce earlier with -Og.
8235 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
8237         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8238         <BIT_FIELD_REF>: Remove trailing TAB.
8239         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8240         remove blank line.
8242 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8244         PR tree-optimization/57358
8245         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8246         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8247         (ipa_analyze_params_uses): Generate pessimistic info when true.
8249 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8251         PR tree-optimization/57539
8252         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8253         global.inlined_to of the new node to it.  All callers changed.
8254         * ipa-inline-transform.c (clone_inlined_nodes): New variable
8255         inlining_into, pass it to cgraph_clone_node.
8256         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8257         ipa_free_edge_args_substructures.
8258         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8259         rdesc linked list.  Do not assert rdesc edges have inlined caller.
8260         Assert we have found an rdesc in the rdesc list.
8262 2013-06-24  Richard Biener  <rguenther@suse.de>
8264         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8265         (pointer_set_lookup): Declare.
8266         (class pointer_map): New template class implementing a
8267         generic pointer to T map.
8268         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8269         pointer_map<T>::contains, pointer_map<T>::insert,
8270         pointer_map<T>::traverse): New functions.
8271         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8272         (pointer_set_lookup): New function.
8273         (pointer_set_contains): Use pointer_set_lookup.
8274         (pointer_set_insert): Likewise.
8275         (insert_aux): Remove.
8276         (struct pointer_map_t): Embed a pointer_set_t.
8277         (pointer_map_create): Adjust.
8278         (pointer_map_destroy): Likewise.
8279         (pointer_map_contains): Likewise.
8280         (pointer_map_insert): Likewise.
8281         (pointer_map_traverse): Likewise.
8282         * tree-streamer.h (struct streamer_tree_cache_d): Use a
8283         pointer_map<unsigned> instead of a pointer_map_t.
8284         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8285         (streamer_tree_cache_lookup): Likewise.
8286         (streamer_tree_cache_create): Likewise.
8287         (streamer_tree_cache_delete): Likewise.
8288         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8289         pointer_map<unsigned> instead of a pointer_map_t.
8290         (lto_init_tree_ref_encoder): Adjust.
8291         (lto_destroy_tree_ref_encoder): Likewise.
8292         * lto-section-out.c (lto_output_decl_index): Likewise.
8293         (lto_record_function_out_decl_state): Likewise.
8294         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8296 2013-06-24  Richard Biener  <rguenther@suse.de>
8298         PR tree-optimization/57488
8299         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8301 2013-06-24  Alan Modra  <amodra@gmail.com>
8303         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8304         (gen_easy_altivec_constant): Likewise.
8305         * config/rs6000/predicates.md (easy_vector_constant_add_self,
8306         easy_vector_constant_msb): Likewise.
8308 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
8310         PR target/57688
8311         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8312         add missing return true.
8314 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
8316         PR target/52483
8317         * config/sh/predicates.md (general_extend_operand): Invoke
8318         general_movsrc_operand for memory operands.
8319         (general_movsrc_operand): Allow reg+reg addressing, do not use
8320         general_operand for memory operands.
8322 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
8324         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
8325         when current target options does not apply.
8326         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8327         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8328         * config/i386/bmiintrin.h: Pass appropriate target
8329         attributes to header.
8330         * config/i386/mmintrin.h: Ditto.
8331         * config/i386/nmmintrin.h: Ditto.
8332         * config/i386/avx2intrin.h: Ditto.
8333         * config/i386/fxsrintrin.h: Ditto.
8334         * config/i386/tbmintrin.h: Ditto.
8335         * config/i386/xsaveintrin.h: Ditto.
8336         * config/i386/f16cintrin.h: Ditto.
8337         * config/i386/xtestintrin.h: Ditto.
8338         * config/i386/xsaveoptintrin.h: Ditto.
8339         * config/i386/bmi2intrin.h: Ditto.
8340         * config/i386/lzcntintrin.h: Ditto.
8341         * config/i386/smmintrin.h: Ditto.
8342         * config/i386/wmmintrin.h: Ditto.
8343         * config/i386/x86intrin.h: Remove all header include guards.
8344         * config/i386/prfchwintrin.h: Ditto.
8345         * config/i386/pmmintrin.h: Ditto.
8346         * config/i386/tmmintrin.h: Ditto.
8347         * config/i386/xmmintrin.h: Ditto.
8348         * config/i386/popcntintrin.h: Ditto.
8349         * config/i386/rdseedintrin.h: Ditto.
8350         * config/i386/ammintrin.h: Ditto.
8351         * config/i386/emmintrin.h: Ditto.
8352         * config/i386/immintrin.h: Remove all header include guards.
8353         * config/i386/fma4intrin.h: Ditto.
8354         * config/i386/lwpintrin.h: Ditto.
8355         * config/i386/xopintrin.h: Ditto.
8356         * config/i386/ia32intrin.h: Ditto.
8357         * config/i386/avxintrin.h: Ditto.
8358         * config/i386/rtmintrin.h: Ditto.
8359         * config/i386/fmaintrin.h: Ditto.
8360         * config/i386/mm3dnow.h: Ditto.
8362 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
8364         * common/config/i386/i386-common.c: Handle LZCNT.
8366 2013-06-22  Andi Kleen  <ak@linux.intel.com>
8368         * doc/extend.texi: Use __atomic_store_n instead of
8369         __atomic_store in HLE example.
8371 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
8373         * config/sh/sh.c: Remove <cstdlib> workaround.
8375 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8377         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
8379 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8381         * doc/extend.texi: Document that __atomic_clear and
8382           __atomic_test_and_set should only be used with bool.
8384 2013-06-20  Jan Hubicka  <jh@suse.cz>
8386         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8387         types_same_for_odr.
8388         * tree.c (decls_same_for_odr): New function.
8389         (same_for_edr): New function.
8390         (types_same_for_odr): New function.
8391         (get_binfo_at_offset): Use it.
8392         * tree.h (types_same_for_odr): Declare.
8394 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
8395             Jason Merrill  <jason@redhat.com>
8397         * system.h: Include <cstdlib> as well as <stdlib.h>.
8399 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
8401         PR target/57655
8402         * config/i386/i386.c (construct_container): Report error if
8403         long double is used with disabled x87 float returns.
8405 2013-06-20  Jan Hubicka  <jh@suse.cz>
8407         * lto-cgraph.c (input_symtab): Do not set cgraph state.
8409 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
8411         PR rtl-optimization/57425
8412         PR rtl-optimization/57569
8413         * alias.c (write_dependence_p): Remove parameters mem_mode and
8414         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
8415         Changed all callers.
8416         (canon_anti_dependence): Get comments and semantics in sync.
8417         Add parameter mem_canonicalized.  Changed all callers.
8418         * rtl.h (canon_anti_dependence): Update prototype.
8420 2013-06-20  Richard Biener  <rguenther@suse.de>
8422         * data-streamer-in.c (streamer_read_uhwi): Optimize single
8423         byte case, inline streamer_read_uchar and defer section
8424         overrun check.
8426 2013-06-20  Richard Biener  <rguenther@suse.de>
8428         PR tree-optimization/57584
8429         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8430         SSA names into the expanded expression that take part in
8431         abnormal coalescing.
8433 2013-06-19  Sharad Singhai  <singhai@google.com>
8435         * gcov.c (print_usage): Handle new option.
8436         (process_args): Ditto.
8437         (get_gcov_intermediate_filename): New function.
8438         (output_intermediate_file): New function.
8439         (output_gcov_file): New function
8440         (generate_results): Handle new option.
8441         (release_function): Relase demangled name.
8442         (read_graph_file): Handle demangled name.
8443         (output_lines): Ditto.
8444         * doc/gcov.texi: Document gcov intermediate format.
8446 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
8448         PR bootstrap/57604
8449         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8450         (lra_emit_add): Use the functions.  Add comment about Y as an
8451         address segment.
8453 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
8455         PR driver/57652
8456         * collect2.c (collect_atexit): New.
8457         (collect_exit): Delete.
8458         (main): Register collect_atexit with atexit.
8459         (collect_wait): Change collect_exit to exit.
8460         (do_wait): Same.
8461         * collect2.h (collect_exit): Delete.
8462         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8464 2013-06-19  Wei Mi  <wmi@google.com>
8466         PR rtl-optimization/57518
8467         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8468         if regno is used in paradoxical subreg.
8469         (update_equiv_regs): Check pdx_subregs[regno] before
8470         set a reg to be equivalent with a mem.
8472 2013-06-19  Matthias Klose  <doko@ubuntu.com>
8474         PR driver/57651
8475         * file-find.h (find_a_file): Add a mode parameter.
8476         * file-find.c (find_a_file): Likewise.
8477         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8478         with X_OK for the executables.
8479         * collect2.c (main): Call find_a_file with X_OK.
8481 2013-06-19  Steve Ellcey  <sellcey@mips.com>
8483         PR target/56942
8484         * config/mips/mips.md (casesi_internal_mips16_<mode>):
8485         Use NEXT_INSN instead of next_real_insn.
8487 2013-06-19  Jan Hubicka  <jh@suse.cz>
8489         * cgraph.h (const_value_known_p): Replace by ...
8490         (ctor_for_folding): .. this one.
8491         * cgraphunit.c (process_function_and_variable_attributes): Use it.
8492         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8493         * expr.c (expand_expr_real_1): Likewise.
8494         (string_constant): Likewise.
8495         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8496         * ipa.c (process_references): Likewise.
8497         (symtab_remove_unreachable_nodes): Likewise.
8498         * ipa-inline-analysis.c (param_change_prob): Likewise.
8499         * gimple-fold.c (canonicalize_constructor_val): Likewise.
8500         (get_base_constructor): Likwise.
8501         * varpool.c (varpool_remove_node): Likewise.
8502         (varpool_remove_initializer): LIkewise.
8503         (dump_varpool_node): LIkwise.
8504         (const_value_known_p): Rewrite to ...
8505         (ctor_for_folding): ... this one.
8507 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
8509         PR driver/57651
8510         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8511         PERSONALITY in $PATH derived prefixes.
8513 2013-06-19  Jeff Law  <law@redhat.com>
8515         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8516         in comment.
8518         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8519         (simplify_bitwise_binary): Use it to simpify certain binary ops on
8520         booleans.
8522 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
8524         * config/arm/vfp.md: Move VFP instruction classification documentation
8525         to ...
8526         * config/arm/arm.md: ... here.  Update instruction classification
8527         documentation.
8529 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
8531         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8532         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8533         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
8535 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
8537         PR target/57609
8538         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8539         with NEXT_INSN.  Use tablejump_p to check for jump table data
8540         insns.
8542 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
8544         PR c++/56544
8545         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8546         that now in C++ the value is correct per the C++ standards.
8548 2013-06-19  Richard Biener  <rguenther@suse.de>
8550         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8551         for global context.
8553 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8555         Revert:
8556         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8558         PR target/57609
8559         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8560         with next_active_insn.
8562 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
8564         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
8565         functions are inlined during failures to flag an error.
8566         * tree-inline.c (expand_call_inline): Allow the error to be flagged
8567         in early inline pass.
8569 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
8571         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
8572         in comments.
8574 2013-06-18  Julian Brown  <julian@codesourcery.com>
8576         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
8577         Permit virtual register pre-reload if !strict.
8578         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
8579         change.
8580         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
8581         prototype.
8582         * config/arm/neon.md (movmisalign<mode>): Use
8583         neon_perm_struct_or_reg_operand instead of
8584         neon_struct_or_register_operand.
8585         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
8586         neon_permissive_struct_operand instead of neon_struct_operand.
8587         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
8588         neon_vector_mem_operand.
8589         * config/arm/predicates.md (neon_struct_operand): Adjust call to
8590         neon_vector_mem_operand.
8591         (neon_permissive_struct_operand): New.
8592         (neon_struct_or_register_operand): Rename to...
8593         (neon_perm_struct_or_reg_operand): This. Adjust call to
8594         neon_vector_mem_operand.
8596 2013-06-18  Richard Biener  <rguenther@suse.de>
8598         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
8599         * lto-streamer.h: Include pointer-set.h.
8600         (struct lto_decl_slot): Remove.
8601         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
8602         Remove next_index entry.
8603         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8604         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8605         (lto_init_tree_ref_encoder): Adjust.
8606         (lto_destroy_tree_ref_encoder): Likewise.
8607         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8608         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8609         (lto_output_decl_index): Adjust.
8610         (lto_new_out_decl_state): Likewise.
8611         (lto_record_function_out_decl_state): Likewise.
8612         * lto-streamer-out.c (copy_function): Likewise.
8614 2013-06-18  Richard Biener  <rguenther@suse.de>
8616         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
8617         * cgraphunit.c: Include cfgloop.h.
8618         (init_lowered_empty_function): Initialize the loop tree.
8619         (assemble_thunk): Insert new BBs into loops.
8621 2013-06-18  Richard Biener  <rguenther@suse.de>
8623         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
8624         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
8625         the map from cache entry to cache index optional.
8626         (streamer_tree_cache_replace_tree): Adjust accordingly.
8627         (streamer_tree_cache_append): Likewise.
8628         (streamer_tree_cache_delete): Likewise.
8629         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
8630         streamer cache map from cache entry to cache index.
8631         * lto-streamer-out.c (create_output_block): Adjust.
8633 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
8635         * config/arm/arm.md (attribute "insn"): Move multiplication and
8636         division attributes to...
8637         (attribute "type"): ... here.  Remove mult.
8638         (attribute "mul32"): New attribute.
8639         (attribute "mul64"): Add umaal.
8640         (*arm_mulsi3): Update attributes.
8641         (*arm_mulsi3_v6): Likewise.
8642         (*thumb_mulsi3): Likewise.
8643         (*thumb_mulsi3_v6): Likewise.
8644         (*mulsi3_compare0): Likewise.
8645         (*mulsi3_compare0_v6): Likewise.
8646         (*mulsi_compare0_scratch): Likewise.
8647         (*mulsi_compare0_scratch_v6): Likewise.
8648         (*mulsi3addsi): Likewise.
8649         (*mulsi3addsi_v6): Likewise.
8650         (*mulsi3addsi_compare0): Likewise.
8651         (*mulsi3addsi_compare0_v6): Likewise.
8652         (*mulsi3addsi_compare0_scratch): Likewise.
8653         (*mulsi3addsi_compare0_scratch_v6): Likewise.
8654         (*mulsi3subsi): Likewise.
8655         (*mulsidi3adddi): Likewise.
8656         (*mulsi3addsi_v6): Likewise.
8657         (*mulsidi3adddi_v6): Likewise.
8658         (*mulsidi3_nov6): Likewise.
8659         (*mulsidi3_v6): Likewise.
8660         (*umulsidi3_nov6): Likewise.
8661         (*umulsidi3_v6): Likewise.
8662         (*umulsidi3adddi): Likewise.
8663         (*umulsidi3adddi_v6): Likewise.
8664         (*smulsi3_highpart_nov6): Likewise.
8665         (*smulsi3_highpart_v6): Likewise.
8666         (*umulsi3_highpart_nov6): Likewise.
8667         (*umulsi3_highpart_v6): Likewise.
8668         (mulhisi3): Likewise.
8669         (*mulhisi3tb): Likewise.
8670         (*mulhisi3bt): Likewise.
8671         (*mulhisi3tt): Likewise.
8672         (maddhisi4): Likewise.
8673         (*maddhisi4tb): Likewise.
8674         (*maddhisi4tt): Likewise.
8675         (maddhidi4): Likewise.
8676         (*maddhidi4tb): Likewise.
8677         (*maddhidi4tt): Likewise.
8678         (divsi3): Likewise.
8679         (udivsi3): Likewise.
8680         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
8681         (thumb2_mulsi_short_compare0): Likewise.
8682         (thumb2_mulsi_short_compare0_scratch): Likewise.
8683         * config/arm/arm1020e.md (1020mult1): Update attribute change.
8684         (1020mult2): Likewise.
8685         (1020mult3): Likewise.
8686         (1020mult4): Likewise.
8687         (1020mult5): Likewise.
8688         (1020mult6): Likewise.
8689         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
8690         change.
8691         (cortex_a15_mult64): Likewise.
8692         (cortex_a15_sdiv): Likewise.
8693         (cortex_a15_udiv): Likewise.
8694         * config/arm/arm1026ejs.md (mult1): Update attribute change.
8695         (mult2): Likewise.
8696         (mult3): Likewise.
8697         (mult4): Likewise.
8698         (mult5): Likewise.
8699         (mult6): Likewise.
8700         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
8701         (pj4_ir_div): Likewise.
8702         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
8703         (11_mult2): Likewise.
8704         (11_mult3): Likewise.
8705         (11_mult4): Likewise.
8706         (11_mult5): Likewise.
8707         (11_mult6): Likewise.
8708         (11_mult7): Likewise.
8709         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
8710         (cortex_a8_mla): Likewise.
8711         (cortex_a8_mull): Likewise.
8712         (cortex_a8_smulwy): Likewise.
8713         (cortex_a8_smlald): Likewise.
8714         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
8715         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
8716         (cortex_r4_mul_3): Likewise.
8717         (cortex_r4_mla_4): Likewise.
8718         (cortex_r4_mla_3): Likewise.
8719         (cortex_r4_smlald): Likewise.
8720         (cortex_r4_mull): Likewise.
8721         (cortex_r4_sdiv): Likewise.
8722         (cortex_r4_udiv): Likewise.
8723         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
8724         (cortex_a7_idiv): Likewise.
8725         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
8726         (9_mult2): Likewise.
8727         (9_mult3): Likewise.
8728         (9_mult4): Likewise.
8729         (9_mult5): Likewise.
8730         (9_mult6): Likewise.
8731         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
8732         (cortex_a53_sdiv): Likewise.
8733         (cortex_a53_udiv): Likewise.
8734         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
8735         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
8736         (mp626_mult2): Likewise.
8737         (mp626_mult3): Likewise.
8738         (mp626_mult4): Likewise.
8739         * config/arm/fa526.md (526_mult1): Update attribute change.
8740         (526_mult2): Likewise.
8741         * config/arm/arm-generic.md (mult): Update attribute change.
8742         (mult_ldsched_strongarm): Likewise.
8743         (mult_ldsched): Likewise.
8744         (multi_cycle): Likewise.
8745         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
8746         * config/arm/fa606te.md (606te_mult1): Update attribute change.
8747         (606te_mult2): Likewise.
8748         (606te_mult3): Likewise.
8749         (606te_mult4): Likewise.
8750         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
8751         (cortex_a9_mac16): Likewise.
8752         (cortex_a9_multiply): Likewise.
8753         (cortex_a9_mac): Likewise.
8754         (cortex_a9_multiply_long): Likewise.
8755         * config/arm/fa626te.md (626te_mult1): Update attribute change.
8756         (626te_mult2): Likewise.
8757         (626te_mult3): Likewise.
8758         (626te_mult4): Likewise.
8760 2013-06-18  Richard Biener  <rguenther@suse.de>
8762         PR lto/57334
8763         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
8765 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8767         PR target/57609
8768         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8769         with next_active_insn.
8771 2013-06-18  Alan Modra  <amodra@gmail.com>
8773         * config/rs6000/rs6000.h (enum data_align): New.
8774         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
8775         (DATA_ABI_ALIGNMENT): Define.
8776         (CONSTANT_ALIGNMENT): Correct comment.
8777         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
8778         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
8780 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
8782         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
8783         ATTRIBUTE_UNUSED marking.
8785 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
8787         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
8788         alternative and update.
8789         (aarch64_dup_lanedi): Delete.
8790         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
8791         * config/aarch64/aarch64-simd-builtins.def: Update.
8793 2013-06-17  Richard Biener  <rguenther@suse.de>
8795         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
8796         (lto_input_scc): Declare.
8797         (lto_input_tree_1): Likewise.
8798         (struct lto_stats_d): Add num_tree_bodies_output and
8799         num_pickle_refs_output.
8800         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
8801         (lto_read_tree_1): Split out from ...
8802         (lto_read_tree): ... this.
8803         (lto_input_scc): New function.
8804         (lto_input_tree_1): Split out from ...
8805         (lto_input_tree): ... this.  Handle LTO_tree_scc.
8806         (lto_data_in_create): Create the streamer cache without hashes.
8807         * lto-streamer-out.c (create_output_block): Create the streamer
8808         cache with hashes when not doing WPA.
8809         (lto_write_tree_1): Split out from ...
8810         (lto_write_tree): ... this.
8811         (get_symbol_initial_value): New function.
8812         (lto_output_tree_1): Split out from ...
8813         (lto_output_tree): ... this.  Write trees as series of SCCs
8814         using a DFS walk via DFS_write_tree.
8815         (struct sccs, struct scc_entry): New types.
8816         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
8817         (DFS_write_tree_body): New function.
8818         (DFS_write_tree): Likewise.
8819         (hash_tree): Likewise.
8820         (scc_entry_compare): Likewise.
8821         (hash_scc): Likewise.
8822         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
8823         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
8824         TREE_CHAIN as regular reference.
8825         (streamer_read_integer_cst): Remove.
8826         (streamer_get_pickled_tree): Adjust.
8827         * tree-streamer-out.c (streamer_write_chain): Disable streaming
8828         of DECL_EXTERNALs in BLOCK_VARS for now.
8829         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
8830         reference.
8831         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
8832         Add hash value argument and record that if hashes are recorded
8833         in the cache.
8834         (streamer_tree_cache_insert_1): Adjust.
8835         (streamer_tree_cache_insert): Likewise.
8836         (streamer_tree_cache_insert_at): Rename to ...
8837         (streamer_tree_cache_replace_tree): ... this and adjust.
8838         (streamer_tree_cache_append): Adjust.
8839         (record_common_node): Likewise.
8840         (streamer_tree_cache_create): Add argument whether to
8841         record hash values together with trees.
8842         (streamer_tree_cache_delete): Adjust.
8843         * tree-streamer.h (struct streamer_tree_cache_d): Add
8844         vector of hashes.
8845         (streamer_read_integer_cst): Remove.
8846         (streamer_tree_cache_insert): Adjust.
8847         (streamer_tree_cache_append): Likewise.
8848         (streamer_tree_cache_insert_at): Rename to ...
8849         (streamer_tree_cache_replace_tree): ... this and adjust.
8850         (streamer_tree_cache_create): Add argument whether to record hashes.
8851         (streamer_tree_cache_get): Rename to ...
8852         (streamer_tree_cache_get_tree): ... this.
8853         (streamer_tree_cache_get_hash): New function.
8854         * tree.c (cache_integer_cst): New function.
8855         * tree.h (cache_integer_cst): Declare.
8856         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
8857         * lto-symtab.c (lto_varpool_replace_node): Only release
8858         DECL_INITIAL of non-prevailing decls.
8859         * varpool.c (varpool_remove_initializer): Do not release
8860         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
8862 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8864         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
8865         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
8866         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
8867         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
8868         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
8869         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
8870         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
8871         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
8872         instead of TARGET_64BIT.
8873         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
8874         Require ISA_HAS_<D>DIV.
8876 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
8878         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
8879         (mips*-*-linux*): Move default with_llsc setting to where other
8880         defaults are set.
8881         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
8882         with_arch block.
8883         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
8884         Likewise.  Remove default with_tune setting.  Move default float
8885         setting to its own block.  Handle with_llsc in the same block as above.
8887 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
8889         PR rtl-optimization/57425
8890         PR rtl-optimization/57569
8891         * alias.c (write_dependence_p): Add new parameters mem_mode,
8892         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
8893         Changed all callers.
8894         (canon_anti_dependence): New function.
8895         * cse.c (check_dependence): Use canon_anti_dependence.
8896         * cselib.c (cselib_invalidate_mem): Likewise.
8897         * rtl.h (canon_anti_dependence): Declare.
8899 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8901         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
8902         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
8903         ".set mips3" for 64-bit targets.
8905 2013-06-15  Dehao Chen  <dehao@google.com>
8907         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
8908         * gimple-low.c (gimple_check_call_matching_types): Likewise.
8909         (gimple_check_call_args): Likewise.
8910         * value-prof.c (check_ic_target): Likewise.
8911         * ipa-inline.c (early_inliner): Likewise.
8912         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
8913         * cgraph.c (cgraph_create_edge_1): Likewise.
8914         (cgraph_make_edge_direct): Likewise.
8916 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
8918         PR target/57615
8919         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
8920         rs6000_output_move_128bit to handle emitting quad memory
8921         operations.  Set attribute length to 8 bytes.
8923 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
8925         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
8926         New pattern.
8927         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
8928         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
8929         (aarch64_<su>mlsl<mode>): Likewise.
8931 2013-06-14  Mike Stump  <mikestump@comcast.net>
8933         * Makefile.in (TARGET_H): Add insn-codes.h.
8935 2013-06-14  Alan Modra  <amodra@gmail.com>
8937         PR middle-end/57134
8938         PR middle-end/57586
8939         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
8940         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
8941         bitfield expansion when EXPAND_MEMORY.
8942         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
8944 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
8946         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
8947         test for clearing quad memory on 32-bit later.
8949 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8951         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
8952         (fold_negate_expr): Likewise.
8953         (fold_real_zero_addition_p): Handle vectors.
8954         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
8956 2013-06-14  Alan Modra  <amodra@gmail.com>
8958         * varasm.c (force_const_mem): Revert 2013-06-07 change.
8960 2013-06-13  Jan Hubicka  <jh@suse.cz>
8962         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
8963         Local comdats are not externally visible.
8964         * symtab.c (dump_symtab_base): Dump externally visible.
8965         (verify_symtab_base): Verify back links in the symtab hash.
8967 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8969         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
8970         CONVERT_EXPR as equal nodes.
8972 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8974         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
8976 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8978         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
8979         Generalize to complex and vector.
8980         * tree.c (build_all_ones_cst): New function.
8981         * tree.h (build_all_ones_cst): Declare it.
8983 2013-06-13  Alan Modra  <amodra@gmail.com>
8985         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
8986         * config/rs6000/rs6000.md (signbittf2): New insn.
8987         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
8988         (abstf2_internal, cmptf_internal2): Likewise.
8989         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
8991 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
8992             Pat Haugen  <pthaugen@us.ibm.com>
8993             Peter Bergner  <bergner@vnet.ibm.com>
8995         * config/rs6000/rs6000.c (emit_load_locked): Add support for
8996         power8 byte, half-word, and quad-word atomic instructions.
8997         (emit_store_conditional): Likewise.
8998         (rs6000_expand_atomic_compare_and_swap): Likewise.
8999         (rs6000_expand_atomic_op): Likewise.
9001         * config/rs6000/sync.md (larx): Add new modes for power8.
9002         (stcx): Likewise.
9003         (AINT): New mode iterator to include TImode as well as normal
9004         integer modes on power8.
9005         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9006         that VSX registers are not considered.  Use AINT mode iterator
9007         instead of INT1 to allow inclusion of quad word atomic operations
9008         on power8.
9009         (load_locked<mode>): Likewise.
9010         (store_conditional<mode>): Likewise.
9011         (atomic_compare_and_swap<mode>): Likewise.
9012         (atomic_exchange<mode>): Likewise.
9013         (atomic_nand<mode>): Likewise.
9014         (atomic_fetch_<fetchop_name><mode>): Likewise.
9015         (atomic_nand_fetch<mode>): Likewise.
9016         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9017         each type.
9018         (ATOMIC): On power8, add QImode, HImode modes.
9019         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9020         modes that promote to SImode.
9021         (load_lockedti): Convert TImode arguments to PTImode, so that we
9022         get a guaranteed even/odd register pair.
9023         (load_lockedpti): Likewise.
9024         (store_conditionalti): Likewise.
9025         (store_conditionalpti): Likewise.
9027         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9028         atomic load/store instructions.
9029         (HSI): Likewise.
9031 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9033         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9034         loads.
9035         (insn_count): New attribute, with most cases extracted from...
9036         (length): ...here.  Redefine most cases in terms of insn_count.
9037         (single_insn): Delete.
9038         (can_delay): Use insn_count to check for single instructions.
9039         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9040         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9041         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9042         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9043         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9044         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9045         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9046         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9047         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9048         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9049         rather than "length".
9050         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9051         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9052         Use "insn_count" rather than "length".
9053         * config/mips/mips-dsp.md
9054         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9055         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9056         length attributes.
9058 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9060         PR tree-optimization/57361
9061         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9063 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9065         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9066         to split.
9067         (aarch64_simd_combine<mode>): New instruction expansion.
9068         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9069         function prototype.
9070         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9071         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9073 2013-06-12  Jan Hubicka  <jh@suse.cz>
9075         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9076         decl has when in streaming stage.
9077         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9078         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9080 2013-06-12  Roland Stigge  <stigge@antcom.de>
9082         PR target/57578
9083         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9085 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9087         PR tree-optimization/57537
9088         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9089         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9091 2013-06-12  Richard Biener  <rguenther@suse.de>
9093         * data-streamer.h (streamer_write_char_stream): CSE
9094         obs->current_pointer.
9095         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9096         streamer_write_char_stream manually and optimize the resulting loop.
9097         (streamer_write_hwi_stream): Likewise.
9099 2013-06-12  Jan Hubicka  <jh@suse.cz>
9101         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9102         * cgraph.h (varpool_create_empty_node): Declare.
9103         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9104         duplicated nodes.
9105         * symtab.c (symtab_unregister_node): Be lax about missin entries
9106         in node hash.
9107         (symtab_get_node): Update comment.
9108         * varpool.c (varpool_create_empty_node): Break out from ...
9109         (varpool_node_for_decl): ... here.
9110         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9112 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9114         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9115         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9116         part.  Use straight-line flow at the end.
9117         <COMPONENT_REF>: Remove superfluous else.
9118         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9120 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9122         PR target/56564
9123         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9124         target hook even for !TREE_PUBLIC decls.  If no resolution info
9125         is available, return false for common and external decls.
9127 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9129         * config/rl78/constraints.md (U): New constraint.
9130         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9131         valloc attribute.
9133 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9135         PR target/57589
9136         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9137         to allow returning address to AT_PLATFORM name.
9139 2013-06-11  Jan Hubicka  <jh@suse.cz>
9141         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9142         * cgraph.h (symtab_node_base): Add weakref flag.
9143         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9144         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9145         (output_weakrefs): Use weakref flag.
9146         * fold-const.c (simple_operand_p): Handle WEAK.
9147         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9148         * ipa.c (varpool_externally_visible_p): Drop weakref.
9149         (function_and_variable_visibility): Update comment; fix weakref
9150         sanity checks; do not clear DECL_WEAK on them.
9151         * lto-cgraph.c (lto_output_node): update.
9152         (lto_output_varpool_node): Update.
9153         (input_overwrite_node): Update.
9154         (input_node): Update.
9155         (input_varpool_node): Update.
9156         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9157         (lto_symtab_merge_symbols): Add sanity check.
9158         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9159         * passes.c (rest_of_decl_compilation): Set static flag, too.
9160         * symtab.c (dump_symtab_base): Dump weakref.
9161         (verify_symtab_base): Sanity check weakrefs.
9162         (symtab_make_decl_local): Remove duplicated code.
9163         (symtab_alias_ultimate_target): Simplify.
9164         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9166 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9168         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9169         sequence_vect == NULL.
9171 2013-06-11  DJ Delorie  <dj@redhat.com>
9173         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9174         (rl78_unwind_word_mode): New.
9176 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9178         * final.c (debug_prefix_maps): Make static.
9180 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9182         * function.c (initial_trampoline): Remove stray copy.
9184 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9186         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9188 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9190         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9191         within bounds at the beginning of the function.
9193 2013-06-11  Alan Modra  <amodra@gmail.com>
9195         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9196         reporting.
9197         (get_named_section): Don't NULL !DECL_P decl.
9199 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9201         * doc/invoke.texi (core-avx2): Document.
9202         (slm): Likewise.
9203         (atom): Updated with MOVBE.
9205 2013-06-11  Richard Biener  <rguenther@suse.de>
9207         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9209 2013-06-11  Anton Blanchard  <anton@samba.org>
9211         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9212         correct shift value in little-endian mode.
9214 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9216         PR target/56564
9217         * varasm.c (get_variable_align): Move #endif to the right place.
9219 2013-06-10  Cary Coutant  <ccoutant@google.com>
9221         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9222         for hash so that hash table traversal order is deterministic.
9224 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
9225             Pat Haugen  <pthaugen@us.ibm.com>
9226             Peter Bergner  <bergner@vnet.ibm.com>
9228         * config/rs6000/vector.md (GPR move splitter): Do not split moves
9229         of vectors in GPRS if they are direct moves or quad word load or
9230         store moves.
9232         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9233         declaration.
9234         (direct_move_p): Likewise.
9235         (quad_load_store_p): Likewise.
9237         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9238         classes into bins based on the physical register type.
9239         (reg_class_to_reg_type): Likewise.
9240         (IS_STD_REG_TYPE): Likewise.
9241         (IS_FP_VECT_REG_TYPE): Likewise.
9242         (reload_fpr_gpr): Arrays to determine what insn to use if we can
9243         use direct move instructions.
9244         (reload_gpr_vsx): Likewise.
9245         (reload_vsx_gpr): Likewise.
9246         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9247         information that is a simplification of register classes.  Also
9248         precalculate direct move reload helpers.
9249         (direct_move_p): New function to return true if the operation can
9250         be done as a direct move instruciton.
9251         (quad_load_store_p): New function to return true if the operation
9252         is a quad memory operation.
9253         (rs6000_legitimize_address): If quad memory, only allow register
9254         indirect for TImode addresses.
9255         (rs6000_legitimate_address_p): Likewise.
9256         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9257         (rs6000_reload_register_type): Likewise.
9258         (register_to_reg_type): Return register type.
9259         (rs6000_secondary_reload_simple_move): New helper function for
9260         secondary reload and secondary memory needed to identify anything
9261         that is a simple move, and does not need reloading.
9262         (rs6000_secondary_reload_direct_move): New helper function for
9263         secondary reload to identify cases that can be done with several
9264         instructions via the direct move instructions.
9265         (rs6000_secondary_reload_move): New helper function for secondary
9266         reload to identify moves between register types that can be done.
9267         (rs6000_secondary_reload): Add support for quad memory operations
9268         and for direct move.
9269         (rs6000_secondary_memory_needed): Likewise.
9270         (rs6000_debug_secondary_memory_needed): Change argument names.
9271         (rs6000_output_move_128bit): New function to return the move to
9272         use for 128-bit moves, including knowing about the various
9273         limitations of quad memory operations.
9275         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9276         memory operations.  call rs6000_output_move_128bit for the actual
9277         instruciton(s) to generate.
9278         (vsx_movti_64bit): Likewise.
9280         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9281         (UNSPEC_P8V_MTVSRWZ): Likewise.
9282         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9283         (UNSPEC_P8V_MTVSRD): Likewise.
9284         (UNSPEC_P8V_XXPERMDI): Likewise.
9285         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9286         (UNSPEC_FUSION_GPR): Likewise.
9287         (FMOVE128_GPR): New iterator for direct move.
9288         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9289         (f32_sv): Likewise.
9290         (f32_dm): Likewise.
9291         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9292         loads and direct move instructions.
9293         (zero_extendsidi2_lfiwzx): Likewise.
9294         (extendsidi2_lfiwax): Likewise.
9295         (extendsidi2_nocell): Likewise.
9296         (floatsi<mode>2_lfiwax): Likewise.
9297         (lfiwax): Likewise.
9298         (floatunssi<mode>2_lfiwzx): Likewise.
9299         (lfiwzx): Likewise.
9300         (fix_trunc<mode>_stfiwx): Likewise.
9301         (fixuns_trunc<mode>_stfiwx): Likewise.
9302         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9303         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9304         (parity<mode>2_cmpb): Set length/type attr.
9305         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9306         for 'mr.' to fast_compare.
9307         (bpermd_<mode>): Change type attr to popcnt.
9308         (p8_fmrgow_<mode>): New insns for power8 direct move support.
9309         (p8_mtvsrwz_1): Likewise.
9310         (p8_mtvsrwz_2): Likewise.
9311         (reload_fpr_from_gpr<mode>): Likewise.
9312         (p8_mtvsrd_1): Likewise.
9313         (p8_mtvsrd_2): Likewise.
9314         (p8_xxpermdi_<mode>): Likewise.
9315         (reload_vsx_from_gpr<mode>): Likewise.
9316         (reload_vsx_from_gprsf): Likewise.
9317         (p8_mfvsrd_3_<mode>): LIkewise.
9318         (reload_gpr_from_vsx<mode>): Likewise.
9319         (reload_gpr_from_vsxsf): Likewise.
9320         (p8_mfvsrd_4_disf): Likewise.
9321         (multi-word GPR splits): Do not split direct moves or quad memory
9322         operations.
9324 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
9326         * tree-into-ssa.c (interesting_blocks): Make static.
9328 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
9330         PR target/56564
9331         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9332         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9333         Use DATA_ABI_ALIGNMENT for that case instead if defined.
9334         (get_variable_align): New function.
9335         (get_variable_section, emit_bss, emit_common,
9336         assemble_variable_contents, place_block_symbol): Use
9337         get_variable_align instead of DECL_ALIGN.
9338         (assemble_noswitch_variable): Add align argument, use it
9339         instead of DECL_ALIGN.
9340         (assemble_variable): Adjust caller.  Use get_variable_align
9341         instead of DECL_ALIGN.
9342         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9343         caller.
9344         (DATA_ABI_ALIGNMENT): Define.
9345         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9346         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
9347         opt is false, only return the psABI mandated alignment increase.
9348         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9349         (DATA_ABI_ALIGNMENT): ... this.
9350         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9351         (DATA_ABI_ALIGNMENT): ... this.
9352         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9353         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9354         (DATA_ABI_ALIGNMENT): ... this.
9355         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9356         * doc/tm.texi: Regenerated.
9358 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
9360         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9361         cmp_code to construct REG_EQUAL note.
9363 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
9365         PR target/57568
9366         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9367         that operands[2] doesn't overlap with operands[0].
9369 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
9370             Jan Hubicka  <jh@suse.cz>
9372         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9373         hack to mark symbols as used.
9375 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
9377         PR rtl-optimization/57559
9378         * lra-constraints.c (process_alt_operands): Don't discourage
9379         memory with known offset for offsetable memory constraint.
9380         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9382 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
9384         * varasm.c (struct oc_local_state): Reorder fields.
9385         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9386         and adjust accordingly.
9387         (output_constructor): Reorder initialization code and adjust call to
9388         output_constructor_bitfield.
9390 2013-06-07  Jan Hubicka  <jh@suse.cz>
9392         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9394 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
9396         * tree-object-size.c (unknown): Make const.
9398 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9400         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9401         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9402         for last alternative in the cpu_facility attribute.
9404 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9406         PR target/56315
9407         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9408         (xordi3): Change operand 2 constraint to arm_xordi_operand.
9409         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9410         * config/arm/constraints.md (Dg): New constraint.
9411         * config/arm/neon.md (xordi3_neon): Remove.
9412         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9413         * config/arm/predicates.md (arm_xordi_operand): New predicate.
9415 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9417         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9418         Clean up alternatives.
9420 2013-06-07  Alan Modra  <amodra@gmail.com>
9422         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9423         va_list_gpr_size.
9425 2013-06-07  Alan Modra  <amodra@gmail.com>
9427         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9429 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9431         * config/arm/constraints.md (Df): New constraint.
9432         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9433         Correct length attribute for last two alternatives.
9435 2013-06-07  Alan Modra  <amodra@gmail.com>
9437         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9438         override user -mfp-in-toc.
9439         (offsettable_ok_by_alignment): Consider just the current access
9440         rather than the whole object, unless BLKmode.  Handle
9441         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9442         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9443         for -mcmodel=medium.
9444         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9445         override user -mfp-in-toc or -msum-in-toc.  Default to
9446         -mno-fp-in-toc for -mcmodel=medium.
9448 2013-06-06  DJ Delorie  <dj@redhat.com>
9450         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9451         TARGET_VALID_POINTER_MODE.
9453 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
9454             Pat Haugen  <pthaugen@us.ibm.com>
9455             Peter Bergner  <bergner@vnet.ibm.com>
9457         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9458         Document new power8 builtins.
9460         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9461         condition code register, to allow 128-bit logical operations to be
9462         done in the VSX or GPR registers.
9463         (nor<mode>3): Use the canonical form for nor.
9464         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9465         vclz*, and vpopcnt* vector instructions.
9466         (nand<mode>3): Likewise.
9467         (orc<mode>3): Likewise.
9468         (clz<mode>2): LIkewise.
9469         (popcount<mode>2): Likewise.
9471         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9472         that only the GPRs are recognized.
9474         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9475         support for new power8 builtins.
9477         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9478         builtin functions.
9479         (xscvdpspn): Likewise.
9480         (vclz): Likewise.
9481         (vclzb): Likewise.
9482         (vclzh): Likewise.
9483         (vclzw): Likewise.
9484         (vclzd): Likewise.
9485         (vpopcnt): Likewise.
9486         (vpopcntb): Likewise.
9487         (vpopcnth): Likewise.
9488         (vpopcntw): Likewise.
9489         (vpopcntd): Likewise.
9490         (vgbbd): Likewise.
9491         (vmrgew): Likewise.
9492         (vmrgow): Likewise.
9493         (eqv): Likewise.
9494         (eqv_v16qi3): Likewise.
9495         (eqv_v8hi3): Likewise.
9496         (eqv_v4si3): Likewise.
9497         (eqv_v2di3): Likewise.
9498         (eqv_v4sf3): Likewise.
9499         (eqv_v2df3): Likewise.
9500         (nand): Likewise.
9501         (nand_v16qi3): Likewise.
9502         (nand_v8hi3): Likewise.
9503         (nand_v4si3): Likewise.
9504         (nand_v2di3): Likewise.
9505         (nand_v4sf3): Likewise.
9506         (nand_v2df3): Likewise.
9507         (orc): Likewise.
9508         (orc_v16qi3): Likewise.
9509         (orc_v8hi3): Likewise.
9510         (orc_v4si3): Likewise.
9511         (orc_v2di3): Likewise.
9512         (orc_v4sf3): Likewise.
9513         (orc_v2df3): Likewise.
9515         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9516         allow power8 quad mode in 64-bit.
9517         (rs6000_builtin_vectorized_function): Add support to vectorize
9518         ISA 2.07 count leading zeros, population count builtins.
9519         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9520         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9521         (builtin_function_type): Add vgbbd builtin function which takes an
9522         unsigned argument.
9523         (altivec_expand_vec_perm_const): Add support for new power8 merge
9524         instructions.
9526         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9527         that does not include TImdoe for use with 32-bit.
9528         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9529         instructions.
9530         (UNSPEC_VSX_CVDPSPN): Likewise.
9531         (vsx_xscvdpspn): Likewise.
9532         (vsx_xscvspdpn): Likewise.
9533         (vsx_xscvdpspn_scalar): Likewise.
9534         (vsx_xscvspdpn_directmove): Likewise.
9535         (vsx_and<mode>3): Split logical operations into 32-bit and
9536         64-bit. Add support to do logical operations on TImode as well as
9537         VSX vector types.  Allow logical operations to be done in either
9538         VSX registers or in general purpose registers in 64-bit mode.  Add
9539         splitters if GPRs were used. For AND, add clobber of CCmode to
9540         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
9541         encoding.
9542         (vsx_and<mode>3_32bit): Likewise.
9543         (vsx_and<mode>3_64bit): Likewise.
9544         (vsx_ior<mode>3): Likewise.
9545         (vsx_ior<mode>3_32bit): Likewise.
9546         (vsx_ior<mode>3_64bit): Likewise.
9547         (vsx_xor<mode>3): Likewise.
9548         (vsx_xor<mode>3_32bit): Likewise.
9549         (vsx_xor<mode>3_64bit): Likewise.
9550         (vsx_one_cmpl<mode>2): Likewise.
9551         (vsx_one_cmpl<mode>2_32bit): Likewise.
9552         (vsx_one_cmpl<mode>2_64bit): Likewise.
9553         (vsx_nor<mode>3): Likewise.
9554         (vsx_nor<mode>3_32bit): Likewise.
9555         (vsx_nor<mode>3_64bit): Likewise.
9556         (vsx_andc<mode>3): Likewise.
9557         (vsx_andc<mode>3_32bit): Likewise.
9558         (vsx_andc<mode>3_64bit): Likewise.
9559         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
9560         and xxlorc instructions.
9561         (vsx_eqv<mode>3_64bit): Likewise.
9562         (vsx_nand<mode>3_32bit): Likewise.
9563         (vsx_nand<mode>3_64bit): Likewise.
9564         (vsx_orc<mode>3_32bit): Likewise.
9565         (vsx_orc<mode>3_64bit): Likewise.
9567         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
9569         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
9570         instruction.
9571         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
9572         (p8_vmrgow): Likewise.
9573         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
9574         GPRs to be split under VSX.
9575         (p8v_clz<mode>2): Add power8 count leading zero support.
9576         (p8v_popcount<mode>2): Add power8 population count support.
9577         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
9578         support.
9580         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
9581         instruction.
9583         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
9584         builtin functions.
9585         (vec_nand): Likewise.
9586         (vec_vclz): Likewise.
9587         (vec_vclzb): Likewise.
9588         (vec_vclzd): Likewise.
9589         (vec_vclzh): Likewise.
9590         (vec_vclzw): Likewise.
9591         (vec_vgbbd): Likewise.
9592         (vec_vmrgew): Likewise.
9593         (vec_vmrgow): Likewise.
9594         (vec_vpopcnt): Likewise.
9595         (vec_vpopcntb): Likewise.
9596         (vec_vpopcntd): Likewise.
9597         (vec_vpopcnth): Likewise.
9598         (vec_vpopcntw): Likewise.
9600 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9602         PR rtl-optimization/57468
9603         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
9604         spilled pseudos.
9606 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9608         PR rtl-optimization/57459
9609         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
9610         type when setting live regs.
9612 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9614         * config/s390/s390.opt (mlra): New option.
9615         * config/s390/s390.c (s390_decompose_address): Check displacement
9616         for all registers for LRA.
9617         (s390_secondary_reload): Don't used secondary reloads for LRA.
9618         (s390_lra_p): New function.
9619         (TARGET_LRA_P): Define.
9620         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
9621         of attribute cpu_facility to zarch for the last alternative.
9622         (*cmpmem_short): Ditto.
9624 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
9626         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
9627         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
9628         (arm_expand_prologue): Likewise.
9630 2013-06-06  Teresa Johnson  <tejohnson@google.com>
9632         PR c++/53743
9633         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
9634         as this is now done by redirect_edge_and_branch_force.
9635         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
9636         barriers, and fix interaction with splitting.
9637         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
9638         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
9639         reflect changes made in the routine.
9640         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
9641         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
9642         since this is called in cfglayout mode, and replace partition fixup
9643         with assert as that is now done by force_nonfallthru_and_redirect.
9644         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
9645         already be marked with region crossing note.
9646         (insert_section_boundary_note): Make non-static, gate on flag
9647         has_bb_partition, rewrite to also check for multiple partitions.
9648         (rest_of_handle_reorder_blocks): Remove call to
9649         insert_section_boundary_note, now done later during free_cfg.
9650         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
9651         * bb-reorder.h (insert_section_boundary_note): Declare.
9652         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
9653         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
9654         invoke insert_section_boundary_note.
9655         (try_redirect_by_replacing_jump): Remove unnecessary
9656         check for region crossing note.
9657         (fixup_partition_crossing): New function.
9658         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
9659         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
9660         in non-cfglayout mode.
9661         (force_nonfallthru_and_redirect): Fixup partition boundaries,
9662         remove old code that tried to do this. Emit barrier correctly
9663         when we are in cfglayout mode.
9664         (last_bb_in_partition): New function.
9665         (rtl_split_edge): Correctly fixup partition boundaries.
9666         (commit_one_edge_insertion): Remove old code that tried to
9667         fixup region crossing edge since this is now handled in
9668         split_block, and set up insertion point correctly since
9669         block may now end in a jump.
9670         (verify_hot_cold_block_grouping): Guard against checking when not in
9671         linearized RTL mode.
9672         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
9673         notes.
9674         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
9675         rtl_verify_flow_info, so not called in cfglayout mode.
9676         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
9677         (fixup_reorder_chain): Remove old code that attempted to fixup region
9678         crossing note as this is now handled in force_nonfallthru_and_redirect.
9679         (duplicate_insn_chain): Don't duplicate switch section notes.
9680         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
9681         note.
9682         * basic-block.h (emit_barrier_after_bb): Declare.
9684 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9686         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
9687         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
9688         arm_usatsihi): Adjust alternatives for arm_restrict_it.
9690 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9692         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
9693         where appropriate.
9694         * config/arm/ldmstm.md: Regenerate.
9696 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9698         * config/arm/sync.md (atomic_loaddi_1):
9699         Disable predication for arm_restrict_it.
9700         (arm_load_exclusive<mode>): Likewise.
9701         (arm_load_exclusivesi): Likewise.
9702         (arm_load_exclusivedi): Likewise.
9703         (arm_load_acquire_exclusive<mode>): Likewise.
9704         (arm_load_acquire_exclusivesi): Likewise.
9705         (arm_load_acquire_exclusivedi): Likewise.
9706         (arm_store_exclusive<mode>): Likewise.
9707         (arm_store_exclusive<mode>): Likewise.
9708         (arm_store_release_exclusivedi): Likewise.
9709         (arm_store_release_exclusive<mode>): Likewise.
9711 2013-06-06  Richard Biener  <rguenther@suse.de>
9713         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
9714         after LTO_null.
9715         (lto_tag_is_tree_code_p): Adjust.
9716         (lto_tag_is_gimple_code_p): Likewise.
9717         (lto_gimple_code_to_tag): Likewise.
9718         (lto_tag_to_gimple_code): Likewise.
9719         (lto_tree_code_to_tag): Likewise.
9720         (lto_tag_to_tree_code): Likewise.
9721         * data-streamer.h (streamer_write_hwi_in_range): Use
9722         uhwi streaming to stream the normalized range.
9723         (streamer_read_hwi_in_range): Likewise.
9725 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9727         * config/arm/arm.md (enabled_for_depr_it): New attribute.
9728         (predicable_short_it): Likewise.
9729         (predicated): Likewise.
9730         (enabled): Handle above.
9731         (define_cond_exec): Set predicated attribute to yes.
9733 2013-06-05  Mike Stump  <mikestump@comcast.net>
9735         * gdbinit.in (__FUNCTION__): Add.
9737 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
9739         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
9740         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
9742 2013-06-05  Jan Hubicka  <jh@suse.cz>
9744         * varasm.c (mark_decl_referenced): Revert the removal until targets
9745         are fixed.
9747 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
9749         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
9750         instead of mark_decl_referenced.
9752 2013-06-05  Jan Hubicka  <jh@suse.cz>
9754         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
9755         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
9756         and symtab_used_from_object_file_p.
9757         (cgraph_make_node_local_1): Clear forced_by_abi.
9758         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
9759         * cgraph.h (symtab_node_base): Add forced_by_abi.
9760         (decide_is_variable_needed): Remove.
9761         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
9762         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
9763         (decide_is_symbol_needed): ... this one; handle symbols in general;
9764         always analyze virtuals; honnor forced_by_abi.
9765         (cgraph_finalize_function): Update.
9766         (varpool_finalize_decl): Update.
9767         (symbol_defined_and_needed): Remove.
9768         (analyze_functions): Update.
9769         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9770         output_refs, input_overwrite_node): Handle forced_by_abi.
9771         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
9772         (address_taken_from_non_vtable_p): ... this one.
9773         (comdat_can_be_unshared_p_1): New function.
9774         (cgraph_comdat_can_be_unshared_p): Rename to ...
9775         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
9776         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
9777         (function_and_variable_visibility): Clear forced_by_abi as needed.
9778         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
9779         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
9780         * symtab.c (dump_symtab_base): Dump forced_by_abi.
9781         * varpool.c (decide_is_variable_needed): Remove.
9783 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9785         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
9786         (arm_option_override): Override arm_restrict_it where appropriate.
9787         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
9788         * config/arm/arm.opt (mrestrict-it): New command-line option.
9789         * doc/invoke.texi: Document -mrestrict-it.
9791 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9793         * tsan.c (tsan_atomic_table): Make const.
9795 2013-06-05  Richard Biener  <rguenther@suse.de>
9797         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
9798         index associated with the tree we are supposed to replace.
9799         * tree-streamer-out.c (pack_ts_base_value_fields): Output
9800         TREE_ASM_WRITTEN as zero for everything but SSA names.
9802 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9804         * tree-ssa-structalias.c (call_stmt_vars): Make static.
9806 2013-06-04  Jan Hubicka  <jh@suse.cz>
9808         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
9809         (input_node, input_varpool_node): Handle correctly external same
9810         body aliases.
9811         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
9812         nodes at ltrans stage.
9814 2013-06-04  Jan Hubicka  <jh@suse.cz>
9816         * ipa-inline.c (update_caller_keys): Fix availability test.
9817         (update_callee_keys): Likewise.
9818         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
9819         to follow ELF standard.
9821 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
9823         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
9824         (mips64r5900el-*-elf*): New configurations.
9825         * config/mips/mips-cpus.def (r5900): New processor.
9826         * config/mips/mips-tables.opt: Regenerate.
9827         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
9828         (mips_issue_rate): Handle PROCESSOR_R5900.
9829         (mips_reorg_process_insns): Force reorder mode for the R5900.
9830         * config/mips/mips.h (TARGET_MIPS5900): Define.
9831         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
9832         TARGET_MIPS5900.
9833         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
9834         TARGET_MIPS5900.
9835         * config/mips/mips.md (processor): Add r5900.
9836         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
9838 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9840         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
9841         into function to generate MOVI instruction.
9842         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
9843         (aarch64_preferred_simd_mode): Turn into wrapper.
9844         (aarch64_output_scalar_simd_mov_immediate): New function.
9845         * config/aarch64/aarch64-protos.h: Add prototype for above.
9847 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9849         * config/aarch64/aarch64.c (simd_immediate_info): Remove
9850         element_char member.
9851         (sizetochar): Return signed char.
9852         (aarch64_simd_valid_immediate): Remove elchar and other
9853         unnecessary variables.
9854         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
9855         Calculate element_char as required.
9856         * config/aarch64/aarch64-protos.h: Update and move prototype
9857         for aarch64_output_simd_mov_immediate.
9858         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9859         Update arguments.
9861 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9863         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
9864         information completed by aarch64_simd_valid_immediate.
9865         (aarch64_legitimate_constant_p): Update arguments.
9866         (aarch64_simd_valid_immediate): Work with struct rather than many
9867         pointers.
9868         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
9869         (aarch64_simd_make_constant): Update arguments.
9870         (aarch64_output_simd_mov_immediate): Work with struct rather than
9871         many pointers.  Output immediate directly rather than as operand.
9872         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9873         Update prototype.
9874         * config/aarch64/constraints.md (Dn): Update arguments.
9876 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9878         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
9879         longer static.
9880         (aarch64_simd_immediate_valid_for_move): Remove.
9881         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
9882         (aarch64_simd_make_constant): Update call.
9883         (aarch64_output_simd_mov_immediate): Update call.
9884         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9885         Add prototype.
9886         * config/aarch64/constraints.md (Dn): Update call.
9888 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9890         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
9891         return type to bool for prototype.
9892         (aarch64_legitimate_constant_p): Check for true instead of not -1.
9893         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
9894         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
9896 2013-06-04  Catherine Moore  <clm@codesourcery.com>
9898         * config/mips/mips.opt (meva): New.
9899         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
9900         (ASM_SPEC): Handle -meva.
9901         * doc/invoke.texi (meva):  Document.
9903 2013-06-04  Alan Modra  <amodra@gmail.com>
9905         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
9906         constant output.
9908 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9910         * rtl.def: Add extra fourth optional field to define_cond_exec.
9911         * gensupport.c (process_one_cond_exec): Process attributes from
9912         define_cond_exec.
9913         * doc/md.texi: Document fourth field in define_cond_exec.
9915 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
9917         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
9918         out the processing order as in store_bit_field_1.
9920 2013-06-04  Jan Hubicka  <jh@suse.cz>
9922         PR middle-end/57500
9923         * cgraphunit.c (cgraph_process_same_body_aliases): Create
9924         non-VAR_DECL node if it does not exist yet.
9926 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
9928         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
9929         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
9930         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
9931         target_cpu_default setting.
9933 2013-06-03  Teresa Johnson  <tejohnson@google.com>
9935         * dumpfile.c (opt_info_switch_p): Change -fopt-info
9936         default to -fopt-info=optimized instead of all.
9937         * doc/invoke.texi: Ditto.
9938         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
9939         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
9940         (execute_vect_slp): Emit BB vectorization success under
9941         MSG_OPTIMIZED_LOCATIONS.
9942         * tree-vect-slp.c (vect_slp_transform_bb): Change
9943         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
9944         * tree-vect-loop.c (vect_transform_loop): Ditto.
9946 2013-06-03  Jason Merrill  <jason@redhat.com>
9948         PR c++/57415
9949         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9950         Use TARGET_EXPR for C++.
9952 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
9954         PR rtl-optimization/57268
9955         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
9956         if DEBUG_INSN_P (insn).
9958         Reapply
9959         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
9961         PR rtl-optimization/57268
9962         * sched-deps.c (sched_analyze_2): Flush dependence lists if
9963         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
9965 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
9967         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
9968         (ix86_avoid_lea_for_addr): Likewise.
9969         (exact_dependency_1): Likewise.
9970         (ix86_adjust_cost): Likewise.
9971         (swap_top_of_ready_list): Fix formatting and !reload_completed check
9972         removed.
9973         (do_reorder_for_imul): Fix typo, formatting and
9974         !reload_completed check removed.
9975         (ix86_sched_reorder): Fix typo and formatting.
9976         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
9977         list.
9979 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
9981         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
9983 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
9985         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
9986         <STRING_CST>: Likewise.
9987         <VECTOR_CST>: Likewise.
9989 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
9990             Mikael Morin  <mikael@gcc.gnu.org>
9992         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
9993         * config.in: Regenerated.
9994         * configure: Regenerated.
9996 2013-06-01  Jan Hubicka  <jh@suse.cz>
9998         PR middle-end/57366
9999         * cgraphunit.c (compile): When weakref is not supported,
10000         set up transparent aliases before final output pass.
10001         * varasm.c (assemble_alias): Do not try to do it here.
10003 2013-06-01  Jan Hubicka  <jh@suse.cz>
10005         PR middle-end/57467
10006         * passes.c (for_per_function): Skip unanalyzed functions.
10008 2013-06-01  Jan Hubicka  <jh@suse.cz>
10010         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10011         (lto_symtab_merge_symbols_1): ... this one.
10012         (lto_symtab_merge_cgraph_nodes): Rename to ...
10013         (lto_symtab_merge_symbols): ... this one; simplify.
10014         * cgraph.c (same_body_aliases_done): Rename to ...
10015         (cpp_implicit_aliases_done): ... this one.
10016         (cgraph_create_function_alias): Update.
10017         (cgraph_same_body_alias): Update.
10018         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10019         (verify_edge_corresponds_to_fndecl): Simplify.
10020         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10021         (cgraph_node): Remove same_body_alias.
10022         (varpool_node): Remove alias_of and extra_name_alias.
10023         (same_body_aliases_done): Rename to ..
10024         (cpp_implicit_aliases_done): ... this one.
10025         (symtab_alias_ultimate_target): Add default parameter.
10026         (symtab_resolve_alias): New function.
10027         (fixup_same_cpp_alias_visibility): Declare.
10028         (cgraph_function_node): Add default parameter.
10029         (cgraph_node_asm_name): Likewise.
10030         (cgraph_function_or_thunk_node): Add default parameter; do
10031         not ICE when it is NULL.
10032         (varpool_variable_node): Likewise.
10033         * tree-emutls.c (create_emultls_var): Update.
10034         (ipa_lower_emutls): Update.
10035         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10036         (cgraph_reset_node): Reset alias info.
10037         (cgraph_finalize_function): Update.
10038         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10039         (analyze_function): Simplify.
10040         (cgraph_process_same_body_aliases): Simplify.
10041         (analyze_functions): Fixup same body aliases.
10042         (handle_alias_pairs): Simplify.
10043         (assemble_thunk): Update.
10044         (assemble_thunks_and_aliases): Update.
10045         (output_weakrefs): Rewrite.
10046         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10047         (lto_output_varpool_node): Likewise.
10048         (compute_ltrans_boundary): Remve assert.
10049         (get_alias_symbol): New functoin.
10050         (input_node): Rewrite alias handling.
10051         (input_varpool_node): Likewise.
10052         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10053         * ipa.c (process_references): Handle weakrefs correctly.
10054         (symtab_remove_unreachable_nodes): Likewise.
10055         * trans-mem.c (get_cg_data): Update.
10056         (ipa_tm_create_version_alias): Update.
10057         (ipa_tm_execute): Update.
10058         * symtab.c (dump_symtab_base): Dump aliases.
10059         (verify_symtab_base): Verify aliases.
10060         (symtab_node_availability): New function.
10061         (symtab_alias_ultimate_target): Simplify.
10062         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10063         handle all the fixup cases.
10064         (symtab_resolve_alias): New function.
10065         * passes.c (ipa_write_summaries): Handle weakrefs.
10066         * varpool.c (varpool_analyze_node): Simplify.
10067         (assemble_aliases): Update.
10068         (varpool_create_variable_alias): Simplify.
10069         (varpool_extra_name_alias): Simplify.
10070         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10071         (lto_symtab_merge_symbols): ... this one.
10073 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10075         Revert
10076         PR rtl-optimization/57268
10077         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10078         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10080 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10082         Partially reverted:
10083         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10085         PR middle-end/57073
10086         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10087         further up.
10089 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10091         PR rtl-optimization/57268
10092         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10093         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10095 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10097         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10098         unordered comparison operators when -fno-trapping-math is in effect
10099         on the e500.
10100         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10101         and implement unordered comparison operators properly on the e500.
10103 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10105         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10106         for constant scalar integers.
10107         (simplify_relational_operation_1): Likewise.
10109 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10111         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10112         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10113         Fix comment.
10115 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10116             Igor Zamyatin  <igor.zamyatin@intel.com>
10118         Silvermont (SLM) architecture performance tuning.
10119         * config/i386/i386.h (enum ix86_tune_indices): Add
10120         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10121         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10123         * config/i386/i386.c (initial_ix86_tune_features)
10124         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10125         (ix86_lea_outperforms): Handle Silvermont tuning.
10126         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10127         call.
10128         (ix86_use_lea_for_mov): Likewise.
10129         (ix86_avoid_lea_for_addr): Likewise.
10130         (ix86_lea_for_add_ok): Likewise.
10131         (exact_dependency_1): New function.
10132         (exact_store_load_dependency): Likewise.
10133         (ix86_adjust_cost): Handle Silvermont tuning.
10134         (do_reoder_for_imul): Likewise.
10135         (swap_top_of_ready_list): New function.
10136         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10138         * config/i386/i386.md (peepholes that split memory operand in fp
10139         converts): New.
10141 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10143         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10144         Remove un-necessary braces.
10146 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10148         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10149         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10151 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10153         PR middle-end/57073
10154         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10156 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10158         PR target/56315
10159         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10160         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10161         * config/arm/neon.md (iordi3_neon): Remove.
10162         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10163         * config/arm/predicates.md (imm_for_neon_logic_operand):
10164         Move to earlier in the file.
10165         (neon_logic_op2): Likewise.
10166         (arm_iordi_operand_neon): New predicate.
10168 2013-05-31  Richard Biener  <rguenther@suse.de>
10170         PR tree-optimization/57478
10171         PR tree-optimization/57453
10172         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10173         are life as well.
10175 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10177         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10178         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10180 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10181             Thomas Koenig  <tkoenig@gcc.gnu.org>
10183         PR middle-end/57073
10184         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10185         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10187 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10189         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10191 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10193         * target.def (register_usage_leveling_p): New hook.
10194         * targhooks.c (default_register_usage_leveling_p): New.
10195         * targhooks.h (default_register_usage_leveling_p): New prototype.
10196         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10197         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10198         * doc/tm.texi: Update.
10199         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10201 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10203         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10204         (*insv_reg<mode>): New define_insn.
10206 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10208         PR rtl-optimization/57439
10209         * postreload.c (move2add_valid_value_p): Check that we have
10210         a zero subreg_regno_offset when accessing the register in
10211         the requested mode.
10213 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10214             Igor Zamyatin  <igor.zamyatin@intel.com>
10216         Silvermont (SLM) architecture pipeline model, tuning and
10217         insn selection.
10218         * config.gcc: Add slm config options and target.
10220         * config/i386/slm.md: New.
10222         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10224         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
10225         PROCESSOR_SLM.
10226         (ix86_target_macros_internal): Likewise.
10228         * config/i386/i386.c (slm_cost): New cost.
10229         (m_SLM): New macro flag.
10230         (initial_ix86_tune_features): Set m_SLM.
10231         (x86_accumulate_outgoing_args): Likewise.
10232         (x86_arch_always_fancy_math_387): Likewise.
10233         (processor_target_table): Add slm cost.
10234         (cpu_names): Add slm cpu name.
10235         (x86_option_override_internal): Set SLM ISA.
10236         (ix86_issue_rate): New case PROCESSOR_SLM.
10237         (ia32_multipass_dfa_lookahead): Likewise.
10238         (fold_builtin_cpu): Add slm.
10240         * config/i386/i386.h (TARGET_SLM): New target macro.
10241         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10242         (processor_type): Add PROCESSOR_SLM.
10244         * config/i386/i386.md (cpu): Add new value "slm".
10245         (slm.md): Include slm.md.
10247 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
10248             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10250         * config/arm/arm-protos.h: Add and update function protos.
10251         * config/arm/arm.c (use_simple_return_p): New added.
10252         (thumb2_expand_return): Check simple_return flag.
10253         * config/arm/arm.md: Add simple_return and conditional simple_return.
10254         * config/arm/iterators.md: Add iterator for return and simple_return.
10256 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10258         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10259         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10260         (arm_emit_vfp_multi_reg_pop): Likewise.
10261         (thumb2_emit_ldrd_pop): Likewise.
10262         (arm_expand_epilogue): Add misc REG_CFA notes.
10263         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10265 2013-05-29  Lawrence Crowl  <crowl@google.com>
10267         * config/arm/t-arm: Update for below.
10269         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10270         Change type to hash_table.  Update dependent calls and types.
10272         * config/i386/t-cygming: Update for below.
10274         * config/i386/t-interix: Update for below.
10276         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10277         Change type to hash_table.  Update dependent calls and types.
10278         (i386_find_on_wrapper_list::wrappers): Likewise.
10280         * config/ia64/t-ia64: Update for below.
10282         * config/ia64/ia64.c (bundle_state_table):
10283         Change type to hash_table.  Update dependent calls and types.
10285         * config/mips/mips.c (mips_reorg_process_insns::htab):
10286         Change type to hash_table.  Update dependent calls and types.
10288         * config/sol2.c (solaris_comdat_htab):
10289         Change type to hash_table.  Update dependent calls and types.
10291         * config/t-sol2: Update for above.
10293 2013-05-29  Teresa Johnson  <tejohnson@google.com>
10295         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10296         functions are not yet marked as defined.
10298 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
10299             Pat Haugen  <pthaugen@us.ibm.com>
10300             Peter Bergner  <bergner@vnet.ibm.com>
10302         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10303         instructions.
10304         (VEC_A): Likewise.
10305         (VEC_C): Likewise.
10306         (vrotl<mode>3): Likewise.
10307         (vashl<mode>3): Likewise.
10308         (vlshr<mode>3): Likewise.
10309         (vashr<mode>3): Likewise.
10311         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10312         support for power8 V2DI builtins.
10314         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10315         power8 V2DI builtins.
10316         (vupkhsw): Likewise.
10317         (vupklsw): Likewise.
10318         (vaddudm): Likewise.
10319         (vminsd): Likewise.
10320         (vmaxsd): Likewise.
10321         (vminud): Likewise.
10322         (vmaxud): Likewise.
10323         (vpkudum): Likewise.
10324         (vpksdss): Likewise.
10325         (vpkudus): Likewise.
10326         (vpksdus): Likewise.
10327         (vrld): Likewise.
10328         (vsld): Likewise.
10329         (vsrd): Likewise.
10330         (vsrad): Likewise.
10331         (vsubudm): Likewise.
10332         (vcmpequd): Likewise.
10333         (vcmpgtsd): Likewise.
10334         (vcmpgtud): Likewise.
10335         (vcmpequd_p): Likewise.
10336         (vcmpgtsd_p): Likewise.
10337         (vcmpgtud_p): Likewise.
10338         (vupkhsw): Likewise.
10339         (vupklsw): Likewise.
10340         (vaddudm): Likewise.
10341         (vmaxsd): Likewise.
10342         (vmaxud): Likewise.
10343         (vminsd): Likewise.
10344         (vminud): Likewise.
10345         (vpksdss): Likewise.
10346         (vpksdus): Likewise.
10347         (vpkudum): Likewise.
10348         (vpkudus): Likewise.
10349         (vrld): Likewise.
10350         (vsld): Likewise.
10351         (vsrad): Likewise.
10352         (vsrd): Likewise.
10353         (vsubudm): Likewise.
10355         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10356         support for power8 V2DI instructions.
10358         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10359         power8 V2DI instructions.  Combine pack and unpack insns to use an
10360         iterator for each mode.  Check whether a particular mode supports
10361         Altivec instructions instead of just checking TARGET_ALTIVEC.
10362         (UNSPEC_VPKUWUM): Likewise.
10363         (UNSPEC_VPKSHSS): Likewise.
10364         (UNSPEC_VPKSWSS): Likewise.
10365         (UNSPEC_VPKUHUS): Likewise.
10366         (UNSPEC_VPKSHUS): Likewise.
10367         (UNSPEC_VPKUWUS): Likewise.
10368         (UNSPEC_VPKSWUS): Likewise.
10369         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10370         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10371         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10372         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10373         (UNSPEC_VUPKHSB): Likewise.
10374         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10375         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10376         (UNSPEC_VUPKHSH): Likewise.
10377         (UNSPEC_VUPKLSB): Likewise.
10378         (UNSPEC_VUPKLSH): Likewise.
10379         (VI2): Likewise.
10380         (VI_char): Likewise.
10381         (VI_scalar): Likewise.
10382         (VI_unit): Likewise.
10383         (VP): Likewise.
10384         (VP_small): Likewise.
10385         (VP_small_lc): Likewise.
10386         (VU_char): Likewise.
10387         (add<mode>3): Likewise.
10388         (altivec_vaddcuw): Likewise.
10389         (altivec_vaddu<VI_char>s): Likewise.
10390         (altivec_vadds<VI_char>s): Likewise.
10391         (sub<mode>3): Likewise.
10392         (altivec_vsubcuw): Likewise.
10393         (altivec_vsubu<VI_char>s): Likewise.
10394         (altivec_vsubs<VI_char>s): Likewise.
10395         (altivec_vavgs<VI_char>): Likewise.
10396         (altivec_vcmpbfp): Likewise.
10397         (altivec_eq<mode>): Likewise.
10398         (altivec_gt<mode>): Likewise.
10399         (altivec_gtu<mode>): Likewise.
10400         (umax<mode>3): Likewise.
10401         (smax<mode>3): Likewise.
10402         (umin<mode>3): Likewise.
10403         (smin<mode>3): Likewise.
10404         (altivec_vpkuhum): Likewise.
10405         (altivec_vpkuwum): Likewise.
10406         (altivec_vpkshss): Likewise.
10407         (altivec_vpkswss): Likewise.
10408         (altivec_vpkuhus): Likewise.
10409         (altivec_vpkshus): Likewise.
10410         (altivec_vpkuwus): Likewise.
10411         (altivec_vpkswus): Likewise.
10412         (altivec_vpks<VI_char>ss): Likewise.
10413         (altivec_vpks<VI_char>us): Likewise.
10414         (altivec_vpku<VI_char>us): Likewise.
10415         (altivec_vpku<VI_char>um): Likewise.
10416         (altivec_vrl<VI_char>): Likewise.
10417         (altivec_vsl<VI_char>): Likewise.
10418         (altivec_vsr<VI_char>): Likewise.
10419         (altivec_vsra<VI_char>): Likewise.
10420         (altivec_vsldoi_<mode>): Likewise.
10421         (altivec_vupkhsb): Likewise.
10422         (altivec_vupkhs<VU_char>): Likewise.
10423         (altivec_vupkls<VU_char>): Likewise.
10424         (altivec_vupkhsh): Likewise.
10425         (altivec_vupklsb): Likewise.
10426         (altivec_vupklsh): Likewise.
10427         (altivec_vcmpequ<VI_char>_p): Likewise.
10428         (altivec_vcmpgts<VI_char>_p): Likewise.
10429         (altivec_vcmpgtu<VI_char>_p): Likewise.
10430         (abs<mode>2): Likewise.
10431         (vec_unpacks_hi_v16qi): Likewise.
10432         (vec_unpacks_hi_v8hi): Likewise.
10433         (vec_unpacks_lo_v16qi): Likewise.
10434         (vec_unpacks_hi_<VP_small_lc>): Likewise.
10435         (vec_unpacks_lo_v8hi): Likewise.
10436         (vec_unpacks_lo_<VP_small_lc>): Likewise.
10437         (vec_pack_trunc_v8h): Likewise.
10438         (vec_pack_trunc_v4si): Likewise.
10439         (vec_pack_trunc_<mode>): Likewise.
10441         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10442         V2DI builtins.
10443         (vec_vmaxsd): Likewise.
10444         (vec_vmaxud): Likewise.
10445         (vec_vminsd): Likewise.
10446         (vec_vminud): Likewise.
10447         (vec_vpksdss): Likewise.
10448         (vec_vpksdus): Likewise.
10449         (vec_vpkudum): Likewise.
10450         (vec_vpkudus): Likewise.
10451         (vec_vrld): Likewise.
10452         (vec_vsld): Likewise.
10453         (vec_vsrad): Likewise.
10454         (vec_vsrd): Likewise.
10455         (vec_vsubudm): Likewise.
10456         (vec_vupkhsw): Likewise.
10457         (vec_vupklsw): Likewise.
10459 2013-05-29  Jan Hubicka  <jh@suse.cz>
10461         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10462         flags; reorder rest of fields in more consistent way.
10463         (varpool_node): Remove analyzed, finalized and alias.
10464         (cgraph_ndoe): Likewise.
10465         (symtab_alias_ultimate_target): New function.
10466         (cgraph_function_node): Move offline.
10467         (cgraph_reset_node): Declare.
10468         (cgraph_comdat_can_be_unshared_p): Remove.
10469         (varpool_remove_initializer): Declare.
10470         (varpool_first_defined_variable, varpool_next_defined_variable
10471         cgraph_first_defined_function, cgraph_next_defined_function): Update.
10472         (cgraph_function_with_gimple_body_p): Update.
10473         (varpool_all_refs_explicit_p): Update.
10474         (symtab_alias_target): New function.
10475         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10476         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10477         (cgraph_function_or_thunk_node): Simplify using
10478         symtab_alias_ultimate_target.
10479         (varpool_variable_node): Likewise.
10480         * cgraph.c (cgraph_create_function_alias): Update.
10481         (cgraph_add_thunk): Update.
10482         (cgraph_remove_node): Update.
10483         (dump_cgraph_node): Do not dump removed flags.
10484         (cgraph_function_body_availability): Update.
10485         (cgraph_propagate_frequency): Update.
10486         (verify_cgraph_node): Check sanity of local flag.
10487         (cgraph_function_node): Move here from cgraph.h; revamp for
10488         cgraph_function_or_thunk_node.
10489         * lto-symtab.c (lto_varpool_replace_node): Update.
10490         (lto_symtab_resolve_can_prevail_p): Update.
10491         (lto_symtab_merge_cgraph_nodes): Update.
10492         * ipa-cp.c (determine_versionability, initialize_node_lattices,
10493         propagate_constants_accross_call, devirtualization_time_bonus,
10494         ipcp_propagate_stage): Update.
10495         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10496         * ipa-inline-transform.c (clone_inlined_nodes,
10497         preserve_function_body_p): Update.
10498         * ipa-reference.c (propagate): Update.
10499         (write_node_summary_p): Update.
10500         * toplev.c (wrapup_global_declaration_2): Update.
10501         * cgraphunit.c (cgraph_analyze_function): Rename to ...
10502         (analyze_function) ... this one.
10503         (cgraph_process_new_functions): Update.
10504         (cgraph_reset_node): Export.
10505         (cgraph_finalize_function): Update.
10506         (cgraph_add_new_function): Update.
10507         (process_function_and_variable_attributes): Update.
10508         (varpool_finalize_decl): Update.
10509         (symbol_finalized): Remove.
10510         (symbol_finalized_and_needed): Rename to ...
10511         (symbol_defined_and_needed): ... update.
10512         (cgraph_analyze_functions): Update.
10513         (handle_alias_pairs): Update.
10514         (mark_functions_to_output): Update.
10515         (assemble_thunk): Update.
10516         (output_in_order): Update.
10517         (output_weakrefs): Update.
10518         (finalize_compilation_unit): Update.
10519         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10520         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10521         input_node, input_varpool_node): Update.
10522         * dbxout.c (dbxout_expand_expr): Update.
10523         * cgraphclones.c (cgraph_clone_node): Update.
10524         (cgraph_copy_node_for_versioning): Update.
10525         (cgraph_materialize_clone): Update.
10526         (cgraph_materialize_all_clones): Update.
10527         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10528         propagate_pure_const, propagate_nothrow): Update.
10529         * lto-streamer-out.c (lto_output, write_symbol): Update.
10530         * ipa-utils.c (ipa_reverse_postorder): Update.
10531         * ipa-inline.c (can_inline_edge_p): Update.
10532         (update_caller_keys, ipa_inline): Update.
10533         * dwarf2out.c (reference_to_unused,
10534         premark_types_used_by_global_vars_helper): Update.
10535         * tree-eh.c (tree_could_trap_p): Update.
10536         * ipa-split.c (consider_split, execute_split_functions): Update.
10537         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10538         has_addr_references_p): Update; move ahead in file for better
10539         readability.
10540         (process_references): Simplify.
10541         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10542         bodies are removed.
10543         (cgraph_comdat_can_be_unshared_p): Make static.
10544         (cgraph_externally_visible_p): Update.
10545         (varpool_externally_visible_p): Update.
10546         (function_and_variable_visibility): Update.
10547         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10548         ipa_tm_mark_force_output_node): Update.
10549         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10550         estimate_edge_devirt_benefit, inline_generate_summary,
10551         inline_write_summary): Update.
10552         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10553         * ipa-prop.c (ipa_compute_jump_functions): Update.
10554         (ipa_print_node_params, ipa_prop_read_section,
10555         ipa_update_after_lto_read, read_replacements_section): Update.
10556         * varasm.c (mark_decl_referenced): Update.
10557         (assemble_alias, dump_tm_clone_pairs): Update.
10558         * tree-inline.c (copy_bb): Update.
10559         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
10560         Update.
10561         * symtab.c (dump_symtab_base): Print new flags.
10562         (verify_symtab_base): Verify new flags.
10563         (symtab_alias_ultimate_target): New function.
10564         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
10565         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
10566         Update.
10567         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
10568         Update.
10569         * i386.c (ix86_get_function_versions_dispatcher,
10570         ix86_generate_version_dispatcher_body): Update.
10571         (fold_builtin_cpu): Use varpool_add_new_variable.
10572         * varpool.c (varpool_remove_initializer): Break out from ...
10573         (varpool_remove_node): ... this one.
10574         (dump_varpool_node, varpool_node_for_asm,
10575         cgraph_variable_initializer_availability, varpool_analyze_node,
10576         varpool_assemble_decl, varpool_remove_unreferenced_decls,
10577         varpool_finalize_named_section_flags, varpool_create_variable_alias):
10578         Update.
10580 2013-05-29  Jan Hubicka  <jh@suse.cz>
10582         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
10584 2013-05-29  Easwaran Raman  <eraman@google.com>
10586         PR tree-optimization/57442
10587         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
10588         when control exits the main loop.
10590 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
10592         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
10593         and RX600.
10594         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
10595         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
10596         * rx/t-rx: Add rx100 under multi library matches option for nofpu
10597         option.
10599 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10601         PR tree-optimization/57441
10602         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
10603         Don't limit size of incr_vec to number of candidates.
10605 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
10607         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
10608         and mips16 directories.
10609         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
10610         (MULTILIB_DIRNAMES): Ditto.
10611         (MULTILIB_EXCEPTIONS): Add new exceptions.
10612         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
10613         (MULTILIB_DIRNAMES): Ditto.
10614         (MULTILIB_EXCEPTIONS): Add new exceptions.
10616 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10617             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10619         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
10620         SYMBOL_TINY_ABSOLUTE.
10621         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
10622         SYMBOL_TINY_ABSOLUTE.
10623         (aarch64_expand_mov_immediate): Likewise.
10624         (aarch64_classify_symbol): Likewise.
10625         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
10626         Permit SYMBOL_TINY_ABSOLUTE.
10627         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
10629 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10630             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10632         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
10633         Refactor if/switch.  Replace gcc_assert with if.
10635 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10637         * config/i386/i386.c (initial_ix86_tune_features): Enable
10638         FP Reassociation for AMD bdver1 and bdver2.
10640 2013-05-29  Martin Jambor  <mjambor@suse.cz>
10642         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
10643         and IMAGPART_EXPR do not occur within other handled_components.
10645 2013-05-29  Richard Biener  <rguenther@suse.de>
10647         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
10648         access on whether the use is in the BB we currently try to
10649         vectorize.
10650         (vect_bb_vectorization_profitable_p): Pass the BB we currently
10651         vectorize to vect_bb_slp_scalar_cost.
10653 2013-05-29  Richard Biener  <rguenther@suse.de>
10655         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
10656         computing scalar cost offsetted by stmts that are kept live
10657         by scalar uses.
10658         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
10659         for computation of scalar cost.
10661 2013-05-28  Steve Ellcey  <sellcey@mips.com>
10663         * config/mips/mips-cpus.def (mips32r2): Change processor type.
10665 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
10667         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
10668         array notation built-in reduction functions.
10669         * doc/passes.texi (Passes): Added documentation about changes done
10670         for Cilk Plus.
10671         * doc/invoke.texi (C Dialect Options): Added documentation about
10672         the -fcilkplus flag.
10673         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
10674         (BUILTINS_DEF): Depend on cilkplus.def.
10675         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
10676         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
10677         * cilkplus.def: New file.
10679 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
10681         PR rtl-optimization/57439
10682         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
10684 2013-05-28  Easwaran Raman  <eraman@google.com>
10686         PR tree-optimization/57337
10687         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
10688         (find_insert_point): Correctly identify the insertion point
10689         when two statements with the same UID is compared.
10691 2013-05-28  Richard Biener  <rguenther@suse.de>
10693         PR tree-optimization/56787
10694         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
10695         from the list of data references.
10696         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
10697         clobbers.
10698         (vect_analyze_loop_operations): Likewise.
10699         (vect_transform_loop): Remove clobbers.
10701 2013-05-28  Martin Jambor  <mjambor@suse.cz>
10703         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
10704         and REALPART_EXPRs have scalar type.
10706 2013-05-28  Richard Biener  <rguenther@suse.de>
10708         PR tree-optimization/57411
10709         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
10710         virtual operands.
10711         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
10712         virtual operand propagation.
10714 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10716         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
10717         destination register for bmasksi_vis.
10718         (vector_init_bshuffle): Likewise.
10719         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
10721 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10723         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
10724         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
10725         mode if the instruction isn't available in the original mode.
10726         * config/sparc/sparc.opt (mfix-ut699): New option.
10727         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
10728         (divdf3): Turn into expander.
10729         (divdf3_nofix): New insn.
10730         (divdf3_fix): Likewise.
10731         (divsf3): Disable if -mfix-ut699.
10732         (sqrtdf2): Turn into expander.
10733         (sqrtdf2_nofix): New insn.
10734         (sqrtdf2_fix): Likewise.
10735         (sqrtsf2): Disable if -mfix-ut699.
10737 2013-05-27  Richard Biener  <rguenther@suse.de>
10739         PR middle-end/57412
10740         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
10741         block for the new loop.
10743 2013-05-27  Richard Biener  <rguenther@suse.de>
10745         PR tree-optimization/57343
10746         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
10747         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
10748         (number_of_iterations_cond): Do not build the folded tree.
10750 2013-05-27  Richard Biener  <rguenther@suse.de>
10752         Revert
10753         PR middle-end/57381
10754         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10755         OEP_CONSTANT_ADDRESS_OF retained.
10757         PR tree-optimization/57417
10758         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
10759         for unchanged base.
10760         (set_ssa_val_to): Compare addresses using
10761         get_addr_base_and_unit_offset.
10763 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
10765         PR rtl-optimization/56833
10766         * postreload.c (move2add_record_mode): New function.
10767         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
10768         (move2add_use_add2_insn): Use move2add_record_sym_value.
10769         (move2add_use_add3_insn): Likewise.
10770         (reload_cse_move2add): Use move2add_valid_value_p and
10771         move2add_record_mode.  Invalidate call-clobbered and REG_INC
10772         affected regs by setting reg_mode to VOIDmode.
10773         (move2add_note_store): Don't pretend the inside of a SUBREG is
10774         the actual destination.  Invalidate single/leading registers by
10775         setting reg_mode to VOIDmode.
10776         Use move2add_record_sym_value, move2add_valid_value_p and
10777         move2add_record_mode.
10779 2013-05-27  Richard Biener  <rguenther@suse.de>
10781         PR tree-optimization/57396
10782         * tree-affine.c (double_int_constant_multiple_p): Properly
10783         return false for val == 0 and div != 0.
10785 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10787         * config/mips/mips.h: Use #elif in preprocessor conditions.
10789 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10791         PR target/53916
10792         * config/mips/constraints.md (kl): New constraint.
10793         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
10794         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
10795         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
10796         from LO for MIPS16.
10797         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
10799 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10801         PR target/55777
10802         * config/mips/mips.c (mips_can_inline_p): New function.
10803         (TARGET_CAN_INLINE_P): Define.
10805 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
10807         * sched-int.h (ds_t, dw_t): Make unsigned int.
10808         Fix documentation that describes how all the ds_t bits are used.
10809         Reserve the last bit for delayed-branch scheduling.
10810         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
10811         (BITS_PER_DEP_WEAK): Fix definition and documentation.
10812         (gen_dep_weak_1): Remove prototype.
10813         * sched-deps.c (get_dep_weak_1): Make static.
10814         * target.def (speculate_insn, needs_block_p, gen_spec_check,
10815         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
10816         * doc/tm.texi: Regenerate.
10817         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
10819 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
10821         PR debug/56950
10822         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
10824 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
10825             Sandra Loosemore  <sandra@codesourcery.com>
10827         * config.gcc (powerpc-*): Allow native for with-cpu.
10829 2013-05-24  Jeff Law  <law@redhat.com>
10831         PR tree-optimization/57124
10832         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
10833         conversion feeding a condition if the range has an overflow
10834         if -fstrict-overflow.  Add warnings for when we do make the
10835         transformation.
10837 2013-05-24  Dehao Chen  <dehao@google.com>
10839         * tree-cfg.c (locus_discrim_map): Fix the typo.
10840         (locus_discrim_hasher): Likewise.
10841         (locus_discrim_hasher::hash): Likewise.
10842         (locus_discrim_hasher::equal): Likewise.
10844 2013-05-24  Martin Jambor  <mjambor@suse.cz>
10846         PR tree-optimization/57294
10847         * cgraph.h (ipa_record_stmt_references): Declare.
10848         * cgraphbuild.c (ipa_record_stmt_references): New function.
10849         (build_cgraph_edges): Use ipa_record_stmt_references.
10850         (rebuild_cgraph_edges): Likewise.
10851         (cgraph_rebuild_references): Likewise.
10852         * ipa-prop.c (ipa_modify_call_arguments): Discard references
10853         associated with the old statement and build references from the
10854         newly built statements.
10855         * ipa-ref.c (ipa_remove_stmt_references): New function.
10856         * ipa-ref.h (ipa_remove_stmt_references): Declare.
10858 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
10860         * lra-constraints.c (emit_spill_move): Use smaller mode for
10861         mem-mem moves.
10862         (check_and_process_move): Consider mem-reg moves for secondary
10863         too.
10864         (curr_insn_transform): Don't lose insns emitted before for
10865         secondary memory moves.
10866         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
10867         reg set up in the current insn.
10869 2013-05-24  Dehao Chen  <dehao@google.com>
10871         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
10872         hash function.
10873         (locus_descrim_hasher::equal): Likewise.
10874         (build_gimple_cfg): New discriminator assignment algorithm.
10875         (make_edges): Likewise.
10876         (next_discriminator_for_locus): Likewise.
10877         (same_line_p): Likewise.
10878         (assign_discriminators): Likewise.
10879         (make_cond_expr_edges): Likewise.
10880         (make_gimple_switch_edges): Likewise.
10881         (make_goto_expr_edges): Likewise.
10882         (make_gimple_asm_edges): Likewise.
10884 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
10886         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
10887         X format specifier to only display bottom 16 bits.
10888         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
10889         immediate to match for operand 2, since it will be masked.
10891 2013-05-24  Richard Biener  <rguenther@suse.de>
10893         PR tree-optimization/57287
10894         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
10895         all SSA names that occur in abnormal PHIs.
10897 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10899         PR tree-ssa/57385
10900         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
10901         that index is not negative.
10903 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
10905         PR rtl-optimization/55177
10906         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
10907         (simplify_byte_swapping_operation): New.
10908         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
10909         (simplify_relational_operation_1): Deal with BSWAP.
10911 2013-05-23  Richard Henderson  <rth@redhat.com>
10913         PR target/56742
10914         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
10915         (ix86_reorg): Call it.
10917 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
10919         PR target/57379
10920         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
10921         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
10922         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
10924 2013-05-23  Christian Bruel  <christian.bruel@st.com>
10926         PR debug/57351
10927         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
10929 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10930             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10932         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
10933         * config/aarch64/constraints.md (Usa): Remove.
10934         * doc/md.texi (AArch64 Usa): Remove.
10936 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10937             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10939         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
10940         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
10941         * config/aarch64/predicates.md (aarch64_const_address): Remove.
10942         (aarch64_mov_operand): Use aarch64_mov_operand_p.
10944 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
10946         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
10947         instruction (AdvSIMD).
10948         * config/aarch64/aarch64-builtins.c
10949         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
10950         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
10952 2013-05-23  Martin Jambor  <mjambor@suse.cz>
10954         PR middle-end/57347
10955         * tree.h (contains_bitfld_component_ref_p): Declare.
10956         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
10957         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
10958         caller.
10959         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
10960         not access a bit-field.  Assert all final offsets are byte-aligned.
10962 2013-05-23  Richard Biener  <rguenther@suse.de>
10964         PR tree-optimization/57380
10965         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
10966         least one invariant or re-used load.
10967         * passes.c (init_optimization_passes): Move pass_phiprop before
10968         pass_forwprop.
10970 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
10972         * config/aarch64/aarch64-simd.md
10973         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
10975 2013-05-23  Richard Biener  <rguenther@suse.de>
10977         PR middle-end/57381
10978         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10979         OEP_CONSTANT_ADDRESS_OF retained.
10981 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
10983         PR middle-end/57344
10984         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
10985         don't lower unit.  Handle unit not being always BITS_PER_WORD.
10987 2013-05-23  Richard Biener  <rguenther@suse.de>
10989         PR rtl-optimization/57341
10990         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
10991         instead of true_dependence.
10993 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
10995         * bb-reorder.c (branch_threshold): Make const.
10996         (exec_threshold): Ditto.
10998 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
10999             Pat Haugen  <pthaugen@us.ibm.com>
11000             Peter Bergner  <bergner@vnet.ibm.com>
11002         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11003         documentation for the power8 crypto builtins.
11005         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11007         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11008         macros for defining power8 builtin functions.
11009         (BU_P8V_AV_2): Likewise.
11010         (BU_P8V_AV_P): Likewise.
11011         (BU_P8V_VSX_1): Likewise.
11012         (BU_P8V_OVERLOAD_1): Likewise.
11013         (BU_P8V_OVERLOAD_2): Likewise.
11014         (BU_CRYPTO_1): Likewise.
11015         (BU_CRYPTO_2): Likewise.
11016         (BU_CRYPTO_3): Likewise.
11017         (BU_CRYPTO_OVERLOAD_1): Likewise.
11018         (BU_CRYPTO_OVERLOAD_2): Likewise.
11019         (XSCVSPDP): Fix typo, point to the correct instruction.
11020         (VCIPHER): Add power8 crypto builtins.
11021         (VCIPHERLAST): Likewise.
11022         (VNCIPHER): Likewise.
11023         (VNCIPHERLAST): Likewise.
11024         (VPMSUMB): Likewise.
11025         (VPMSUMH): Likewise.
11026         (VPMSUMW): Likewise.
11027         (VPERMXOR_V2DI): Likewise.
11028         (VPERMXOR_V4SI: Likewise.
11029         (VPERMXOR_V8HI: Likewise.
11030         (VPERMXOR_V16QI: Likewise.
11031         (VSHASIGMAW): Likewise.
11032         (VSHASIGMAD): Likewise.
11033         (VPMSUM): Likewise.
11034         (VPERMXOR): Likewise.
11035         (VSHASIGMA): Likewise.
11037         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11038         __CRYPTO__ if the crypto instructions are available.
11039         (altivec_overloaded_builtins): Add support for overloaded power8
11040         builtins.
11042         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11043         support for power8 crypto builtins.
11044         (builtin_function_type): Likewise.
11045         (altivec_init_builtins): Add support for builtins that take vector
11046         long long (V2DI) arguments.
11048         * config/rs6000/crypto.md: New file, define power8 crypto
11049         instructions.
11051 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11052             Pat Haugen  <pthaugen@us.ibm.com>
11053             Peter Bergner  <bergner@vnet.ibm.com>
11055         * doc/invoke.texi (Option Summary): Add power8 options.
11056         (RS/6000 and PowerPC Options): Likewise.
11058         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11059         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11060         wm, wn, wr documentation.
11062         * config/rs6000/constraints.md (wm): New constraint for VSX
11063         registers if direct move instructions are enabled.
11064         (wn): New constraint for no registers.
11065         (wq): New constraint for quad word even GPR registers.
11066         (wr): New constraint if 64-bit instructions are enabled.
11067         (wv): New constraint if power8 vector instructions are enabled.
11068         (wQ): New constraint for quad word memory locations.
11070         * config/rs6000/predicates.md (const_0_to_15_operand): New
11071         constraint for 0..15 for crypto instructions.
11072         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11073         as GPR and floating point registers.
11074         (int_reg_operand): New predicate to match only GPR registers.
11075         (base_reg_operand): New predicate to match base registers.
11076         (quad_int_reg_operand): New predicate to match even GPR registers
11077         for quad memory operations.
11078         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11079         operations in both GPR and VSX registers.
11080         (quad_memory_operand): New predicate for quad memory operations.
11081         (reg_or_indexed_operand): New predicate for direct move support.
11083         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11084         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11085         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11086         (POWERPC_MASKS): Add power8 options.
11087         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11088         various options.
11090         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11091         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11093         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11094         (-mpower8-fusion): New power8 options.
11095         (-mpower8-fusion-sign): Likewise.
11096         (-mpower8-vector): Likewise.
11097         (-mcrypto): Likewise.
11098         (-mdirect-move): Likewise.
11099         (-mquad-memory): Likewise.
11101         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11102         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11103         registers.
11104         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11105         (rs6000_debug_vector_unit): Add p8_vector.
11106         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11107         definitions.  Also print fusion state.
11108         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11109         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11110         (rs6000_option_override_internal): Add support for power8.
11111         (rs6000_common_init_builtins): Add debugging for skipped builtins
11112         if -mdebug=builtin.
11113         (rs6000_adjust_cost): Add power8 support.
11114         (rs6000_issue_rate): Likewise.
11115         (insn_must_be_first_in_group): Likewise.
11116         (insn_must_be_last_in_group): Likewise.
11117         (force_new_group): Likewise.
11118         (rs6000_register_move_cost): Likewise.
11119         (rs6000_opt_masks): Likewise.
11121         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11122         power8 capable assembler, default to power7 options.
11123         (TARGET_DIRECT_MOVE): Likewise.
11124         (TARGET_CRYPTO): Likewise.
11125         (TARGET_P8_VECTOR): Likewise.
11126         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11127         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11128         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11129         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11130         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11131         (TARGET_XSCVDPSPN): Likewise.
11132         (TARGET_XSCVSPDPN): Likewsie.
11133         (TARGET_SYNC_HI_QI): Likewise.
11134         (TARGET_SYNC_TI): Likewise.
11135         (MASK_CRYPTO): Likewise.
11136         (MASK_DIRECT_MOVE): Likewise.
11137         (MASK_P8_FUSION): Likewise.
11138         (MASK_P8_VECTOR): Likewise.
11139         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11140         temporary used by some of the direct move instructions to get two FP
11141         temporary registers does not force creation of a stack frame.
11142         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11143         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11144         that any VSX registers are tieable, even if they are also an
11145         Altivec vector mode.
11146         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11147         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11148         (RS6000_BTM_CRYPTO): Likewise.
11149         (RS6000_BTM_COMMON): Likewise.
11151         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11152         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11153         (enum rs6000_vector): Add power8 vector support.
11155 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11157         PR target/19599
11158         PR target/57340
11159         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11160         (any_sibcall_could_use_r3): this and handle indirect calls.
11161         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11163 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11165         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11167 2013-05-22  Richard Biener  <rguenther@suse.de>
11169         PR middle-end/57349
11170         * profile.c (branch_prob): Do not split blocks that are
11171         abnormally receiving from ECF_RETURNS_TWICE functions.
11173 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11175         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11176         address mode.  Move pointer mode initialization to the same place.
11178 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11180         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11181         while it has any effect.
11183 2013-05-21  Easwaran Raman  <eraman@google.com>
11185         PR tree-optimization/57322
11186         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11187         UID of the statement added to the BB to be 1.
11189 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11191         PR tree-optimization/57331
11192         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11193         of conversion from pointer type to integral type with integer.
11195 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11197         PR lto/57289
11198         * ipa-prop.c (ipa_read_node_info): Process param_used and
11199         controlled_uses in the same order as when writing.
11201 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11203         PR plugins/56754
11204         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11206 2013-05-21  Richard Biener  <rguenther@suse.de>
11208         PR tree-optimization/57318
11209         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11210         estimate stmts with side-effects as likely eliminated.
11212 2013-05-21  Richard Biener  <rguenther@suse.de>
11214         PR tree-optimization/57330
11215         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11216         preserve the call stmts fntype.
11218 2013-05-21  Richard Biener  <rguenther@suse.de>
11220         PR tree-optimization/57303
11221         * tree-ssa-sink.c (statement_sink_location): Improve killing
11222         stmt detection and properly handle self-assignments.
11224 2013-05-21  Christian Bruel  <christian.bruel@st.com>
11226         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11227         spanning registers. LEAF_REG_REMAP is supported only for contiguous
11228         registers. Set register size out of the PARALLEL loop.
11230 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
11232         PR target/56547
11233         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11234         (*fmasf4, *fmasf4_media): New insns.
11236 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11238         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11239         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11240         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11241         (mips_idiv_insns): Update the comments to say that the returned
11242         instruction counts are in units of BASE_INSN_LENGTH.
11243         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11244         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
11245         using 2 rather than 4 as the length of indirect MIPS16 and
11246         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
11247         length of a NOP.  Don't divide MIPS16 lengths by 2.
11248         (mips16_split_long_branches): Assume a branch is long if the
11249         length is greater than 4 rather than 8.
11250         * config/mips/mips.md (length): Give MIPS16 lengths directly,
11251         rather than multiplying them by 2.  Multiply instruction counts
11252         by BASE_INSN_LENGTH rather than 4.
11253         (*jump_mips16, tls_get_tp_mips16_<mode>)
11254         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11256 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11258         * config/mips/mips.md (extended_mips16): Remove branch case.
11259         (length): Remove duplicated extended_mips16 test.
11261 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11263         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11265 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
11267         * recog.h (Recog_data): Rename to...
11268         (recog_data_d): ...this.
11269         (recog_data): Update accordingly.
11270         * recog.c (recog_data): Likewise.
11271         * reload.c (save_recog_data): Likewise.
11272         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11273         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11275 2013-05-17  Julian Brown  <julian@codesourcery.com>
11277         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11278         found in a REG_EQUAL note, invalidate it.
11280 2013-05-17  Easwaran Raman  <eraman@google.com>
11282         * tree-ssa-reassoc.c (find_insert_point): New function.
11283         (insert_stmt_after): Likewise.
11284         (get_def_stmt): Likewise.
11285         (ensure_ops_are_available): Likewise.
11286         (not_dominated_by): Likewise.
11287         (rewrite_expr_tree): Do not move statements beyond what is
11288         necessary. Remove call to swap_ops_for_binary_stmt...
11289         (reassociate_bb): ... and move it here.
11290         (build_and_add_sum): Assign UIDs for new statements.
11291         (linearize_expr): Likewise.
11292         (do_reassoc): Renumber gimple statement UIDs.
11294 2013-05-17  Jan Hubicka  <jh@suse.cz>
11296         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11297         weakrefs.
11298         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11299         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11300         weakrefs.
11301         (output_weakrefs): Update.
11303 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
11304             Martin Jambor  <mjambor@suse.cz>
11306         PR middle-end/57276
11307         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11308         value that corresponds to the given aggval is found in values vector.
11310 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
11312         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11313         sse, sse2, sse3, ssse3 and sse4a flags to options.
11315 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
11317         * gengtype-state.c: (s_expr_writer): New class, to handle
11318         prettifying of output layout of s-expressions.
11319         (state_writer): New class, to write out gtype.state.
11320         (state_written_type_count): Move this variable into member data of
11321         state_writer.
11322         (s_expr_writer::s_expr_writer): New code: constructor for new class
11323         (state_writer::state_writer(): ditto
11324         (s_expr_writer::write_new_line): New function
11325         (s_expr_writer::write_any_indent): ditto
11326         (s_expr_writer::begin_s_expr): ditto
11327         (s_expr_writer::end_s_expr): ditto
11328         (write_state_fileloc): convert to method of state_writer...
11329         (state_writer:: write_state_fileloc): ...and use methods of
11330         s_expr_writer to write indentation into the gtype.state output file
11331         to visually represent the hierarchical structure of the list
11332         structures
11333         (write_state_fields): ditto, renaming to...
11334         (state_writer::write_state_fields)
11335         (write_state_a_string): ditto, renaming to...
11336         (state_writer::write_state_a_string)
11337         (write_state_string_option): ditto, renaming to...
11338         (state_writer::write_state_string_option)
11339         (write_state_type_option): ditto, renaming to...
11340         (state_writer::write_state_type_option)
11341         (write_state_nested_option): ditto, renaming to...
11342         (state_writer::write_state_nested_option)
11343         (write_state_option): ditto, renaming to...
11344         (state_writer::write_state_option)
11345         (write_state_options): ditto, renaming to...
11346         (state_writer::write_state_options)
11347         (write_state_lang_bitmap): ditto, renaming to...
11348         (state_writer::write_state_lang_bitmap)
11349         (write_state_version): ditto, renaming to...
11350         (state_writer::write_state_version)
11351         (write_state_scalar_type): ditto, renaming to...
11352         (state_writer::write_state_scalar_type)
11353         (write_state_string_type): ditto, renaming to...
11354         (state_writer::write_state_string_type)
11355         (write_state_undefined_type): ditto, renaming to...
11356         (state_writer::write_state_undefined_type)
11357         (write_state_struct_union_type): ditto, renaming to...
11358         (state_writer::write_state_struct_union_type)
11359         (write_state_struct_type): ditto, renaming to...
11360         (state_writer::write_state_struct_type)
11361         (write_state_user_struct_type): ditto, renaming to...
11362         (state_writer::write_state_user_struct_type)
11363         (write_state_lang_struct_type): ditto, renaming to...
11364         (state_writer::write_state_lang_struct_type)
11365         (write_state_param_struct_type): ditto, renaming to...
11366         (state_writer::write_state_param_struct_type)
11367         (write_state_pointer_type): ditto, renaming to...
11368         (state_writer::write_state_pointer_type)
11369         (write_state_array_type): ditto, renaming to...
11370         (state_writer::write_state_array_type)
11371         (write_state_gc_used): ditto, renaming to...
11372         (state_writer::write_state_gc_used)
11373         (write_state_common_type_content): ditto, renaming to...
11374         (state_writer::write_state_common_type_content)
11375         (write_state_type): ditto, renaming to...
11376         (state_writer::write_state_type)
11377         (write_state_pair_list): ditto, renaming to...
11378         (state_writer::write_state_pair_list)
11379         (write_state_pair): ditto, renaming to...
11380         (state_writer::write_state_pair)
11381         (write_state_typedefs): ditto, renaming to...
11382         (state_writer::write_state_typedefs)
11383         (write_state_structures): ditto, renaming to...
11384         (state_writer::write_state_structures)
11385         (write_state_param_structs): ditto, renaming to...
11386         (state_writer::write_state_param_structs)
11387         (write_state_variables): ditto, renaming to...
11388         (state_writer::write_state_variables)
11389         (write_state_srcdir): ditto, renaming to...
11390         (state_writer::write_state_srcdir)
11391         (write_state_files_list): ditto, renaming to...
11392         (state_writer::write_state_files_list)
11393         (write_state_languages): ditto, renaming to...
11394         (state_writer::write_state_languages)
11395         (write_state): create a state_writer instance and use it when
11396         writing out the state file
11398 2013-05-17  Mike Stump  <mikestump@comcast.net>
11400         PR rtl-optimization/57304
11401         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11402         accessing DF_REF_REAL_LOC.
11404 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11406         PR rtl-optimization/57281
11407         PR rtl-optimization/57300
11408         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11409         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11410         what the other splitter did if the registers are dead.
11412 2013-05-17  Richard Biener  <rguenther@suse.de>
11414         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11415         MEM_REF offsets.
11417 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11419         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11420         linking.
11422 2013-05-17  Marek Polacek  <polacek@redhat.com>
11424         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11425         length when doing non-zero store of storing '\0' to '\0'.
11427 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11429         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11430         vect_external_def oprnd1 with loop_vinfo, try to emit
11431         optional cast, negation and and stmts on the loop preheader
11432         edge instead of into the pattern def seq.
11434         PR tree-optimization/57051
11435         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11436         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11438 2013-05-16  Nick Clifton  <nickc@redhat.com>
11440         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11441         (rl78_is_naked_func): New function.
11442         (rl78_expand_prologue): Skip prologue generation for naked functions.
11443         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11444         * doc/extend.texi (naked): Add RL78 to the list of processors
11445         that supports this attribute.
11447 2013-05-16  Jeff Law  <law@redhat.com>
11449         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11451 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11453         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11454         cache parameters using detect_caches_amd also for CYRIX,
11455         NSC and TM2 signatures.
11457 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11458             Dzianis Kahanovich  <mahatma@eu.by>
11460         PR target/45359
11461         PR target/46396
11462         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11463         VIA/Centaur processors and determine their cache parameters
11464         using detect_caches_amd.
11466 2013-05-16  Teresa Johnson  <tejohnson@google.com>
11468         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11469         (rtl_verify_edges): New function.
11470         (rtl_verify_bb_insns): Ditto.
11471         (rtl_verify_bb_pointers): Ditto.
11472         (rtl_verify_bb_insn_chain): Ditto.
11473         (rtl_verify_fallthru): Ditto.
11474         (rtl_verify_bb_layout): Ditto.
11475         (rtl_verify_flow_info_1): Outline checks into new functions.
11476         (rtl_verify_flow_info): Ditto.
11478 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
11480         * cfghooks.c (copy_bbs): Add update_dominance argument.
11481         * cfghooks.h (copy_bbs): Update prototype.
11482         * tree-cfg.c (gimple_duplicate_sese_region):
11483         Add update_dominance argument.
11484         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11485         * tree-ssa-loop-ch.c (copy_loop_headers): Update
11486         gimple_duplicate_sese_region call.
11487         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11488         Update copy_bbs call.
11489         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11490         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11492 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11494         * tree-vectorizer.h (NUM_PATTERNS): Increment.
11495         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11496         vect_recog_rotate_pattern.
11497         (vect_recog_rotate_pattern): New function.
11499 2013-05-16  Jason Merrill  <jason@redhat.com>
11501         * Makefile.in (LLINKER): New variable.
11502         (mostlyclean): Remove link mutex.
11503         * configure.ac: Handle --enable-link-mutex.
11504         * lock-and-run.sh: New script.
11506 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11508         PR target/19599
11509         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11510         for NULL decl.
11512 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11514         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11516 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11518         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11519         * config/arm/arm.c (next_consecutive_mem): New function.
11520         (gen_movmem_ldrd_strd): Likewise.
11521         * config/arm/arm.md (movmemqi): Update condition and code.
11522         (unaligned_loaddi, unaligned_storedi): New patterns.
11524 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11526         * config.gcc: Obsolete *-*-solaris2.9*.
11527         * doc/install.texi (Specific, *-*-solaris2*): Document it.
11529 2013-05-16  Richard Biener  <rguenther@suse.de>
11531         * passes.c (init_optimization_passes): Move pass_parallelize_loops
11532         earlier, after GRAPHITE transforms and IV canonicalization.
11534 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11536         * omp-low.c (extract_omp_for_data): For collapsed loops,
11537         if at least one of the loops is known at compile time to
11538         iterate zero times, set count to 0.
11539         (expand_omp_regimplify_p): New function.
11540         (expand_omp_for_generic): For collapsed loops, if at least
11541         one of the loops isn't known to iterate at least once,
11542         add runtime check with setting count to 0.
11543         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11544         For unsigned types if it isn't known at compile time that
11545         the loop will iterate at least once, add runtime check to bypass
11546         the whole loop if initial condition isn't true.
11548 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
11550         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11552 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
11554         PR middle-end/57286
11555         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
11556         transformations to avoid an infinite loop.
11558 2013-05-16  Marek Polacek  <polacek@redhat.com>
11560         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
11562 2013-05-15  Leif Ekblad  <leif@rdos.net>
11564         * config/i386/i386.c (ix86_decompose_address): Use
11565         DEFAULT_TLS_SEG_REG to access TLS segment register.
11566         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
11567         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
11568         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
11570 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
11572         PR target/57260
11573         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
11574         sibling calls to functions that would normally be lazily bound,
11575         unless $gp is call-clobbered.
11577 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
11579         * config/i386/i386.c (ix86_option_override_internal): Update
11580         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
11581         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
11582         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
11583         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
11584         of TARGET_3DNOW.
11585         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
11587 2013-05-15  Andreas Schwab  <schwab@suse.de>
11589         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
11590         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
11591         third operand.
11593 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11595         * loop-unroll.c (report_unroll_peel): Check decision before
11596         emitting unroll/peel message.
11598 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11600         * function.h (has_bb_partition): New rtl_data flag.
11601         (bb_reorder_complete): Ditto.
11602         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
11603         instead of flag_reorder_blocks_and_partition.
11604         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
11605         with some enhancements.
11606         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
11607         * bb-reorder.c (connect_traces): Check for has_bb_partition
11608         instead of flag_reorder_blocks_and_partition.
11609         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
11610         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
11611         verify_hot_cold_block_grouping.
11612         (partition_hot_cold_basic_blocks): Set has_bb_partition.
11614 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11616         PR target/19599
11617         * config/arm/predicates.md (call_insn_operand): New predicate.
11618         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
11619         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
11620         if insn is not a tail call.
11621         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
11622         registers.
11623         * config/arm/arm.h (enum reg_class): New caller save register class.
11624         (REG_CLASS_NAMES): Likewise.
11625         (REG_CLASS_CONTENTS): Likewise.
11626         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
11627         without decls.
11629 2013-05-15  Richard Biener  <rguenther@suse.de>
11631         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
11632         of MSG_OPTIMIZED_LOCATIONS.
11633         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
11634         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
11635         message.
11636         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
11637         of MSG_OPTIMIZED_LOCATIONS.
11638         (execute_vect_slp): Likewise.
11639         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
11640         (vect_create_cond_for_alias_checks): Likewise.
11641         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11642         (vect_recog_widen_mult_pattern): Likewise.
11643         (vect_recog_widen_sum_pattern): Likewise.
11644         (vect_recog_over_widening_pattern): Likewise.
11645         (vect_recog_widen_shift_pattern): Likewise.
11646         (vect_recog_vector_vector_shift_pattern): Likewise.
11647         (vect_recog_divmod_pattern): Likewise.
11648         (vect_recog_mixed_size_cond_pattern): Likewise.
11649         (vect_recog_bool_pattern): Likewise.
11650         (vect_pattern_recog_1): Likewise.
11652 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11654         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
11655         non-functions to builtin_unreachable.
11656         * ipa-inline-transform.c (inline_call): Do not assert estimates were
11657         correct when new direct edges were discovered.
11659 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11661         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
11662         header, print symbol order instead of node uid, print more information
11663         about indirect edge targets.
11664         (ipa_make_edge_direct_to_target): Print symbol order instead of node
11665         uids.
11666         (ipa_make_edge_direct_to_target): Likewise.
11667         (remove_described_reference): Likewise.
11668         (propagate_controlled_uses): Likewise.
11669         (ipa_print_node_params): Also print symbol order.
11670         (ipcp_transform_function): Print symbol order instead of node uids.
11671         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
11672         (cgraph_get_create_real_symbol_node): Likewise.
11673         * ipa-cp.c (print_lattice): Likewise.
11674         (print_all_lattices): Likewise.
11675         (determine_versionability): Likewise.
11676         (initialize_node_lattices): Likewise.
11677         (estimate_local_effects): Likewise.
11678         (update_profiling_info): Likewise.
11679         (create_specialized_node): Likewise.
11680         (perhaps_add_new_callers): Likewise.
11681         (decide_about_value): Likewise.
11682         (decide_whether_version_node): Likewise.
11683         (identify_dead_nodes): Likewise.
11684         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
11685         (dump_inline_summary): Likewise.
11686         (estimate_node_size_and_time): Likewise.
11687         (inline_analyze_function): Likewise.
11688         * ipa-inline.c (report_inline_failed_reason): Likewise.
11689         (want_early_inline_function_p): Likewise.
11690         (edge_badness): Likewise.
11691         (update_edge_key): Likewise.
11692         (inline_small_functions): Likewise.  Add dumping of order to two other
11693         dumps.
11694         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
11695         instead of node uids.
11696         (propagate_pure_const): Likewise.
11697         (propagate_pure_const): Likewise.
11698         * ipa-utils.c (dump_cgraph_node_set): Likewise.
11699         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
11700         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
11701         of node uids.
11702         * tree-pretty-print.c (dump_function_header): Likewise.
11703         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
11704         Print symbol order instead of node uids.
11706 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11708         * config/s390/s390.c (s390_register_move_cost): Don't impose the
11709         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
11711 2013-05-15  Richard Biener  <rguenther@suse.de>
11713         PR tree-optimization/57275
11714         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
11715         return value for fail to do runtime alias checks for gather loads.
11717 2013-05-15  Jan Hubicka  <jh@suse.cz>
11719         PR lto/57038
11720         PR lto/47375
11721         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
11722         weakrefs are not external.
11723         (lto_symtab_merge_decls): Fix thinko when dealing with
11724         non-lto_symtab decls.
11725         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
11726         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
11727         * varpool.c (dump_varpool_node): Dump more flags.
11729 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11731         * config/i386/i386.c (processor_alias_table): Add instruction
11732         FSGSBASE for AMD bdver3 architecture.
11734 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11736         * tree.c (warn_deprecated_use): Print file:line using locus color.
11737         * diagnostic.c (diagnostic_report_current_module): Print file:line
11738         and file:line:column using locus color.
11740 2013-05-14  Mike Stump  <mikestump@comcast.net>
11742         * gdbinit.in: Add __null.
11744 2013-05-14  Mike Stump  <mikestump@comcast.net>
11746         * recog.h: Rename struct recog_data to Recog_data.
11747         * recog.c: Likewise.
11748         * reload.c (can_reload_into): Likewise.
11749         * config/picochip/picochip.c: Likewise.
11751 2013-05-14  Mike Stump  <mikestump@comcast.net>
11753         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
11755 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
11757         * resource.h (struct resources): Remove unch_memory member.
11758         (CLEAR_RESOURCE): Don't clear unch_memory.
11759         * resource.c (mark_referenced_resources): Don't set it.
11760         (mark_set_resources): Likewise.
11761         (mark_target_live_regs): Don't clear it.
11762         (init_resource_info): Likewise.
11763         * reorg.c (resource_conflicts_p): Don't compare it.
11764         (redundant_insn): Don't set it.
11766         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
11767         Remove prototypes.
11768         * emit-rtl.c (next_label): Remove unused function.
11769         (skip_consecutive_labels, link_cc0_insns): Move to ...
11770         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
11771         only place where these functions are used, and make them static.
11773 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11775         * fold-const.c (fold_negate_expr): Handle vectors.
11776         (fold_truth_not_expr): Make it static.
11777         (fold_invert_truthvalue): New static function.
11778         (invert_truthvalue_loc): Handle vectors. Do not call
11779         fold_truth_not_expr directly.
11780         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
11781         <TRUTH_NOT_EXPR>: Do not cast to boolean.
11782         (fold_comparison): Handle vector constants.
11783         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
11784         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
11785         * tree.h (fold_truth_not_expr): Remove declaration.
11787 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
11789         * config/aarch64/aarch64-simd.md
11790         (aarch64_vcond_internal<mode>): Rename to...
11791         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
11792         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
11793         float modes. Clarify all iterator modes.
11794         (vcond<mode><mode>): Use new name for vcond expanders.
11795         (vcond<v_cmp_result><mode>): Likewise.
11796         (vcondu<mode><mode>: Likewise.
11797         * config/aarch64/iterators.md (VDQF_COND): New.
11799 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11801         PR bootstrap/57266
11802         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
11803         variable for the shift amount. Check that we shift by non-negative
11804         amounts.
11806 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
11808         PR target/42017
11809         * config/arm/arm.h (EPILOGUE_USES): Only return true
11810         for LR_REGNUM after epilogue_completed.
11812 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
11814         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
11815         is error_mark_node.
11817 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11819         PR target/57261
11820         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
11821         and Solaris 11+/x86 with gld.
11822         * configure: Regenerate.
11824 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11826         * expmed.c (expand_shift_1): Canonicalize rotates by
11827         constant bitsize / 2 to bitsize - 1.
11828         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
11829         case ROTATERT>: Likewise.
11831         Revert:
11832         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
11834         * config/i386/i386.md (rotateinv): New code attr.
11835         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
11836         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
11837         roll $31, %eax, etc.
11839 2013-05-14  Richard Biener  <rguenther@suse.de>
11841         PR middle-end/57235
11842         * tree-eh.c (sink_clobbers): Give up for successors with
11843         multiple predecessors and no virtual uses.
11845 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
11847         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
11848         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
11850 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11852         PR middle-end/57251
11853         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
11854         the case when both op0 and op1 have VOIDmode.
11856 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11858         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
11859         in multiply-accumulate mode.
11861 2013-05-13  Guozhi Wei  <carrot@google.com>
11863         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
11865 2013-05-13  Kai Tietz  <ktietz@redhat.com>
11867         PR target/56975
11868         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
11869         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
11870         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
11871         * config/i386/i386.c (ix86_option_override_internal): Likewise.
11872         (ix86_expand_prologue): Likewise.
11873         (ix86_expand_split_stack_prologue): Likewise.
11874         (legitimate_pic_address_disp_p): Likewise.
11875         (legitimize_pic_address): Likewise.
11876         (legitimize_tls_address): Likewise.
11877         (legitimize_pe_coff_symbol): Likewise.
11878         (output_pic_addr_const): Likewise.
11879         (construct_plt_address): Likewise.
11880         (ix86_expand_call): Likewise.
11881         (x86_output_mi_thunk): Likewise.
11882         (x86_function_profiler): Likewise.
11884 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
11886         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
11887         similar switch cases.
11888         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
11889         (aarch64_simd_mov_to_<mode>low): Delete.
11890         (aarch64_simd_mov_to_<mode>high): Delete.
11891         (move_lo_quad_<mode>): Add w<-r alternative.
11892         (aarch64_simd_move_hi_quad_<mode>): Likewise.
11893         (aarch64_simd_mov_from_*): Update type attribute.
11894         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
11895         statement.
11897 2013-05-13  Jan Hubicka  <jh@suse.cz>
11899         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
11900         * config/i386/i386.c (ix86_compute_frame_layout,
11901         ix86_expand_epilogue, emit_i387_cw_initialization,
11902         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
11903         ix86_local_alignment): Fix use of size/speed predicates.
11905 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11907         PR tree-optimization/45216
11908         PR tree-optimization/57157
11909         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
11910         the (-Y) & (B - 1) variant if OP is |.
11911         * expmed.c (expand_shift_1): For rotations by const0_rtx just
11912         return shifted.  Use (-op1) & (prec - 1) as other_amount
11913         instead of prec - op1.
11915 2013-05-13  Martin Jambor  <mjambor@suse.cz>
11917         PR middle-end/42371
11918         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
11919         (ipa_constant_data): New type.
11920         (ipa_jump_func): Use ipa_constant_data to hold information about
11921         constant jump functions.
11922         (ipa_get_jf_constant): Adjust to jump function type changes.
11923         (ipa_get_jf_constant_rdesc): New function.
11924         (ipa_param_descriptor): New field controlled_uses.
11925         (ipa_get_controlled_uses): New function.
11926         (ipa_set_controlled_uses): Likewise.
11927         * ipa-ref.h (ipa_find_reference): Declare.
11928         * ipa-prop.c (ipa_cst_ref_desc): New type.
11929         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
11930         changes.
11931         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
11932         New parameter cs.  Adjust all callers.
11933         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
11934         (remove_described_reference): New function.
11935         (jfunc_rdesc_usable): Likewise.
11936         (try_make_edge_direct_simple_call): Decrement controlled use count,
11937         attempt to remove reference if it hits zero.
11938         (combine_controlled_uses_counters): New function.
11939         (propagate_controlled_uses): Likewise.
11940         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
11941         (ipa_edge_duplication_hook): Duplicate reference descriptions.
11942         (ipa_print_node_params): Print described use counter.
11943         (ipa_write_jump_function): Adjust to jump function type changes.
11944         (ipa_read_jump_function): New parameter CS, pass it to
11945         ipa_set_jf_constant.  Adjust caller.
11946         (ipa_write_node_info): Stream controlled use count
11947         (ipa_read_node_info): Likewise.
11948         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
11949         asserting.
11950         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
11951         count.  Remove cloning-added reference if it reaches zero.
11952         * ipa-ref.c (ipa_find_reference): New function.
11954 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11956         * config/i386/i386.c (processor_target_table): Modified default
11957         alignment values for AMD BD and BT architectures.
11959 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
11961         * tree-vect-generic.c (uniform_vector_p): Move ...
11962         * tree.c (uniform_vector_p): ... here.
11963         * tree.h (uniform_vector_p): Declare it.
11964         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
11965         into a scalar.
11967 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11969         PR tree-optimization/57230
11970         * tree-ssa-strlen.c (handle_char_store): Record length for
11971         array store from STRING_CST.
11973         PR tree-optimization/57230
11974         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
11975         check.
11977 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
11979         * config/epiphany/epiphany.c (epiphany_init): Check size of
11980         NUM_MODES_FOR_MODE_SWITCHING.
11981         (epiphany_expand_prologue):
11982         Remove CONFIG_REGNUM initial value handling code.
11983         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
11984         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
11985         (emit_set_fp_mode, epiphany_mode_after): Likewise.
11986         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
11987         Don't return 1 for FP_MODE_NONE.
11988         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
11989         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
11990         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
11991         * config/epiphany/epiphany.md (save_config): New pattern.
11993 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
11995         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
11997 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
11999         * config/i386/i386.md (memory): Handle sseishft1.
12000         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12001         (*vec_extractv2di_1): Ditto.
12003 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12005         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12006         saved registers.
12008 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12010         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12011         Add mthumb/march=armv7-a multilib.
12012         Add mthumb/march=armv7-r multilib.
12013         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12015 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12017         * config/v850/t-rtems: Add more multilibs.
12019 2013-05-10  Richard Biener  <rguenther@suse.de>
12021         PR tree-optimization/57214
12022         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12023         not propagate from SSA names that occur in abnormal PHI nodes.
12025 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12027         * stor-layout.c (element_precision): New function.
12028         * machmode.h (element_precision): Declare it.
12029         * tree.c (build_minus_one_cst): New function.
12030         (element_precision): Likewise.
12031         * tree.h (build_minus_one_cst): Declare new function.
12032         (element_precision): Likewise.
12033         * fold-const.c (operand_equal_p): Use element_precision.
12034         (fold_binary_loc): Handle vector types.
12035         * convert.c (convert_to_integer): Use element_precision.
12036         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12037         separately.
12039 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12041         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12042         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12043         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12044         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12045         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12046         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12047         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12048         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12049         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12050         (Uuw8): New constraints.
12051         (Usb4): Move into alphabetical order.
12052         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12053         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12054         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12055         previously unnamed patterns.
12056         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12057         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12058         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12059         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12060         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12061         of specific lengths.
12063 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12065         * config/i386/i386.md (rotateinv): New code attr.
12066         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12067         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12068         roll $31, %eax, etc.
12070         PR tree-optimization/45216
12071         PR tree-optimization/57157
12072         * tree-ssa-forwprop.c (simplify_rotate): New function.
12073         (ssa_forward_propagate_and_combine): Call it.
12075 2013-05-10  Richard Biener  <rguenther@suse.de>
12077         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12078         disable peeling when we version for aliasing.
12079         (vector_alignment_reachable_p): Honor explicit user alignment.
12080         (vect_supportable_dr_alignment): Likewise.
12081         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12082         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12083         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12084         then peeling to arrange for the cost-model check to come first.
12086 2013-05-10  Alan Modra  <amodra@gmail.com>
12088         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12089         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12090         * configure: Regenerate.
12092 2013-05-10  Alan Modra  <amodra@gmail.com>
12094         PR target/55033
12095         * varasm.c (default_elf_select_section): Move !DECL_P check..
12096         (get_named_section): ..to here before calling get_section_name.
12097         Adjust assertion.
12098         (default_section_type_flags): Add DECL_P check.
12099         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12100         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12102 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12104         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12105         When using gen_stack_adjust_str with a register offset, add a
12106         REG_FRAME_RELATED_EXPR note.
12108 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12110         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12111         (*vec_extractv4si_zext_mem): Ditto.
12112         (*vec_extractv2di): Add 0->x and x->x alternatives.
12113         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12114         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12116 2013-05-09  Jason Merrill  <jason@redhat.com>
12118         N3639 C++1y VLA support
12119         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12120         DECL_VALUE_EXPR.
12122         * tree.c (build_constructor_va): New.
12123         * tree.h: Declare it.
12125 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12127         PR lto/57084
12128         * gimple-fold.c (canonicalize_constructor_val): Call
12129         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12131 2013-05-09  Jan Hubicka  <jh@suse.cz>
12132             Richard Biener  <rguenther@suse.de>
12134         PR lto/54095
12135         * symtab.c (symtab_make_decl_local): Do not add private names.
12137 2013-05-09  Jan Hubicka  <jh@suse.cz>
12139         PR lto/54095
12140         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12141         (unlink_from_assembler_name_hash): Likewise.
12142         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12143         symtab_unregister_node, symtab_initialize_asm_name_hash,
12144         change_decl_assembler_name): Update.
12146 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12148         * config/aarch64/aarch64.md: New movtf split.
12149         (*movtf_aarch64): Update.
12150         (aarch64_movdi_tilow): Handle TF modes and rename to
12151         aarch64_movdi_<mode>low.
12152         (aarch64_movdi_tihigh): Handle TF modes and rename to
12153         aarch64_movdi_<mode>high
12154         (aarch64_movtihigh_di): Handle TF modes and rename to
12155         aarch64_mov<mode>high_di
12156         (aarch64_movtilow_di): Handle TF modes and rename to
12157         aarch64_mov<mode>low_di
12158         (aarch64_movtilow_tilow): Remove spurious whitespace.
12159         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12160         splits.
12161         (aarch64_print_operand): Update.
12163 2013-05-09  Alan Modra  <amodra@gmail.com>
12165         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12166         powerpc64le.
12167         * configure: Regenerate.
12169 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12171         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12172         splitter preparation statements.
12173         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12174         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12175         adjust_address_nv.
12177 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12179         * gimple-ssa-strength-reduction.c (count_candidates): Change
12180         return value to int.
12181         (analyze_candidates_and_replace): Change type of length to int.
12183 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12185         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12186         (*vec_extract<mode>): Use VI12_128 mode iterator.
12187         (*vec_extract<mode>_mem): Ditto.
12188         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12189         attribute.
12191 2013-05-08  Diego Novillo  <dnovillo@google.com>
12193         PR bootstrap/54659
12195         Revert:
12196         2012-08-17  Diego Novillo  <dnovillo@google.com>
12198         PR bootstrap/54281
12199         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12200         * config.in: Regenerate.
12201         * configure: Regenerate.
12202         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12204 2013-05-08  Jan Hubicka  <jh@suse.cz>
12206         PR lto/54095
12207         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12208         * cgraph.h (symtab_node_base): Add unique_name.
12209         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12210         input_overwrite_node, input_varpool_node): Stream unique_name.
12211         * cgraphclones.c (cgraph_create_virtual_clone,
12212         cgraph_function_versioning): Set unique_name.
12213         * ipa.c (function_and_variable_visibility): Set unique_name.
12215 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12217         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12218         (alloc_cand_and_find_basis): Restrict conditional candidate
12219         processing to CAND_MULTs.
12221 2013-05-08  Jan Hubicka  <jh@suse.cz>
12223         PR lto/54095
12224         lto-symtab.c (lto_symtab_symbol_p): New function.
12225         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12226         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12227         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12228         Skip static symbols.
12230 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
12232         PR tree-optimization/57200
12233         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12234         Only call inform if the preceding warning_at returns true.
12236 2013-05-07  Han Shen  <shenhan@google.com>
12238         * cfgexpand.c (record_or_union_type_has_array_p): New function.
12239         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12240         * common.opt (fstack-protector-strong): New option.
12241         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12242         * doc/invoke.texi (Optimization Options): Document
12243         "-fstack-protector-strong".
12244         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12246 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12248         * config/mips/mips.c (mips_machine_reorg2): Return 0.
12250 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
12252         * ira.c (update_equiv_regs): Add insn having equiv memory even if
12253         it is not lhs of the insn.
12254         (setup_reg_equiv): Remove insn having equiv memory which it is not
12255         lhs of the insn.
12256         * lra-constraints.c (process_address): Try to improve generation
12257         code for address base + disp.
12258         (lra_constraints): Make correct the code for checking insn setting
12259         up backward equivalence.  Remove insn only if it is in the init
12260         insn list.
12261         * lra-eliminations.c (update_reg_eliminate): Change return value.
12262         (lra_eliminate): Use the result.
12264 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
12266         * config/i386/sse.md (ssescalarnummask): New mode attribute.
12267         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12268         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12269         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12270         register target operands.
12271         (*vec_extractv8hi_sse2): New pattern.
12272         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12273         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12274         (*vec_extract<mode>_mem): New insn and split pattern.
12276 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
12278         * config/arm/arm.c (arm_asan_shadow_offset): New function.
12279         (TARGET_ASAN_SHADOW_OFFSET): Define.
12280         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12281         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12283 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12285         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12286         (incr_vec_index): Return -1 if increment not found.
12287         (create_add_on_incoming_edge): Assert if increment not found.
12288         (record_increment): Limit number of increments recorded.
12289         (all_phi_incrs_profitable): Return false if an increment not found.
12290         (replace_profitable_candidates): Don't process increments that were
12291         not recorded.
12292         (analyze_candidates_and_replace): Limit size of incr_vec.
12294 2013-05-07  Richard Biener  <rguenther@suse.de>
12296         * calls.c (special_function_p): setjmp-like functions are leaf.
12297         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12298         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12300 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
12302         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12303         (aarch64_simd_mov<mode>): New expander.
12304         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12305         (aarch64_simd_mov_to_<mode>high): Likewise.
12306         (aarch64_simd_mov_from_<mode>low): Likewise.
12307         (aarch64_simd_mov_from_<mode>high): Likewise.
12308         (aarch64_dup_lane<mode>): Update.
12309         (aarch64_dup_lanedi): New instruction pattern.
12310         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12311         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12313 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12315         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12316         (replace_mult_candidate): Remove unnecessary argument; remove
12317         unnecessary parameter from call to introduce_cast_before_cand.
12318         (replace_unconditional_candidate): Remove unnecessary parameter
12319         from call to replace_mult_candidate.
12320         (replace_conditional_candidate): Likewise.
12321         (insert_initializers): Use make_temp_ssa_name.
12322         (introduce_cast_before_cand): Remove unnecessary argument; use
12323         make_temp_ssa_name.
12324         (replace_one_candidate): Remove unnecessary argument; remove
12325         unnecessary parameter from calls to introduce_cast_before_cand.
12326         (replace_profitable_candidates): Remove unnecessary parameters
12327         from calls to replace_one_candidate.
12329 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12331         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12332         phi def as possibly hiding a basis for a CAND_ADD whose operands
12333         have been commuted in the analysis.
12334         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12336 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12338         * config/aarch64/aarch64.md
12339         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12340         shift value between 0-4.
12342 2013-05-07  Richard Biener  <rguenther@suse.de>
12344         * double-int.h (rshift): New overload.
12345         * double-int.c (rshift): New function.
12346         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12347         (create_reference_ops_from_ref): Remove.
12348         (vn_reference_insert): Use shared ops for constructing the
12349         reference and copy it.
12351 2013-05-07  Richard Biener  <rguenther@suse.de>
12353         PR middle-end/57190
12354         * tree-eh.c (sink_clobbers): Properly propagate
12355         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12357 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
12359         PR tree-optimization/57149
12360         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12361         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12362         collect_phi_def_edges, execute_late_warn_uninitialized): Use
12363         uninit_undefined_value_p instead of ssa_undefined_value_p.
12365         PR debug/57184
12366         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12367         for modifier == EXPAND_INITIALIZER.
12369 2013-05-07  Anton Blanchard  <anton@samba.org>
12371         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12372         for powerpc64 little endian.
12373         * configure: Regenerate.
12375 2013-05-06  Graham Stott  <grahams@btinternet.com>
12377         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12378         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12379         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12380         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12382 2013-05-06  Graham Stott  <grahams@btinternet.com>
12384         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12385         codes which allow non-lvalues.
12387 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
12389         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12390         components are all 1s.
12391         (integer_minus_onep): New function.
12392         * tree.h (integer_minus_onep): Declare it.
12393         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12394         integer_minus_onep instead of integer_all_onesp.
12396 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12398         PR target/52933
12399         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12400         variations of these patterns.
12402 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12404         * config/i386/i386.md (isa): Add x64_sse4 member.
12405         (enabled): Handle x64_sse4.
12406         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12407         instruction for 64bit SSE4_1 targets.  Update insn attributes.
12408         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12409         instruction for SSE4_1 targets.  Update insn attributes.
12410         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12411         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12412         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12413         const_1 selector.
12414         (*vec_extractv4si): Rename from *sse4_1_pextrd.
12415         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12416         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12418 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12420         PR target/57108
12421         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12423 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
12425         * final.c (do_assembler_dialects): Don't handle curly braces and
12426         vertical bar escaped by % as dialect delimiters.
12427         (output_asm_insn): Print curly braces and vertical bar if escaped
12428         by % and ASSEMBLER_DIALECT defined.
12429         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12430         * doc/tm.texi: Regenerated.
12432 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12434         * config/mips/mips.c: Include tree-pass.h.
12435         (mips_reorg): Split in pre- and post-dbr_schedule parts.
12436         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12437         (pass_mips_machine_reorg2): New machine specific pass.
12438         (insert_pass_mips_machine_reorg2): New pass plugin definition.
12439         (mips_option_override): Register the new pass.
12440         * rtl.h (cleanup_barriers): Remove prototype.
12441         (dbr_schedule): Likewise.
12442         * jump.c (cleanup_barriers): Make static.
12443         * reorg.c (dbr_schedule): Likewise.
12445 2013-05-06  Richard Biener  <rguenther@suse.de>
12447         PR tree-optimization/57185
12448         * tree-parloops.c (add_field_for_reduction): Handle anonymous
12449         SSA names properly.
12451 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12453         PR target/57106
12454         * config/i386/i386.c (add_parameter_dependencies): Add dependence
12455         between "first_arg" and "insn", not "last" and "insn".
12457 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12459         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12460         (find_candidates_in_block): Re-enable slsr_process_phi.
12461         (create_phi_basis): Fix double counting of candidate adjustment.
12463 2013-05-06  Richard Biener  <rguenther@suse.de>
12465         PR middle-end/57147
12466         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12467         the edge is also fallthru, preserve it and just clear the
12468         abnormal flag.
12469         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12470         also complex, preserve that and just clear the fallthru flag.
12471         * tree-inline.c (update_ssa_across_abnormal_edges): Also
12472         update virtual operands.
12474 2013-05-06  Alan Modra  <amodra@gmail.com>
12476         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12477         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12478         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12479         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12480         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12481         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12483 2013-05-06  Alan Modra  <amodra@gmail.com>
12485         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12486         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
12487         (DEFAULT_ASM_ENDIAN): Define.
12488         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12489         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12490         Update -K PIC clause from sysv4.h.
12491         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12492         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12494 2013-05-06  Alan Modra  <amodra@gmail.com>
12496         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12497         twice for little-endian.
12498         (ashrdi3_no_power, ashrdi3): Support little-endian.
12500 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12502         PR target/55303
12503         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12504         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12505         related expanders.
12506         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12507         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12508         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12509         New predicates.
12511 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
12512             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12514         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12515         * config/pa/pa.opt: Make mbig-switch a no-op.
12516         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12517         (CASE_VECTOR_MODE): Always return SImode.
12518         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12519         for the !TARGET_BIG_SWITCH case.
12520         * config/pa/pa-linux.h: Likewise.
12521         * config/pa/pa-openbsd.h: Likewise.
12522         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12523         * config/pa/pa.md (short_jump): Remove define_insn.
12524         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12525         (casesi0): Remove define_insn.
12526         (type): Remove btable_branch.
12527         (pa_combine_type): Likewise.
12528         (in_nullified_branch_delay): Likewise.
12529         (in_call_delay): Likewise.
12530         (define_delay): Likewise.
12531         (define_insn_reservation "Z3"): Likewise.
12532         (define_insn_reservation "Z4"): Likewise.
12533         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12534         (pa_adjust_insn_length): Remove adjustment for btable branches.
12535         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12536         and mno-big-switch
12538 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
12540         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12541         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12542         Add m->r,x alternatives.
12543         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12544         splitters using SWI48x mode iterator.
12545         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
12546         TARGET_64BIT.  Add m->x alternative.
12547         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12548         Add o->x alternative.  Enable for TARGET_SSE.
12549         (sse_storeq): Remove expander.
12550         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12551         with memory input operand.
12552         (*vec_extractv2di_1 splitter): New.
12553         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12554         * config/i386/i386.md (ssevecmodelower): New mode attribute.
12556 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12558         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
12559         (INT_LOWPART): Delete.
12560         (extract_MB): Adjust.
12561         (extract_ME): Adjust.
12562         (print_operand): Adjust.
12564 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12566         * config/rs6000/predicates.md (reg_or_add_cint_operand,
12567         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12568         (reg_or_logical_cint_operand, easy_fp_constant,
12569         logical_const_operand): Delete "CONST_DOUBLE" case.
12570         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
12571         "HOST_BITS_PER_WIDE_INT == 64" test.
12572         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
12573         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
12574         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
12575         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
12576         test.
12577         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
12578         CONST_DOUBLE DImode/VOIDmode case.
12579         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
12580         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12581         CONST_DOUBLE VOIDmode case.
12582         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12583         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
12584         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
12585         Delete CONST_DOUBLE case.
12586         (splitters for mov FMOVE64 const_double): Delete
12587         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12588         "HOST_BITS_PER_WIDE_INT >= 64" test.
12589         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
12590         case.
12591         (mov DI const_double): Delete.
12593 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
12595         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
12596         on op shows all bits zero in mode of a lowpart subreg, return zero.
12598 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
12600         PR target/57150
12601         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
12602         to save TFmode registers and DImode to save TImode registers for
12603         caller save operations.
12604         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
12605         mark being partially clobbered since they only use the first
12606         double word.
12608         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
12609         and TDmode only use the upper 64-bits of each VSX register.
12611 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12613         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
12614         (find_candidates_in_block): Disable slsr_process_phi.
12616 2013-05-03  Guozhi Wei  <carrot@google.com>
12618         * coverage.c (coverage_obj_init): Move the construction of gcov
12619         constructor to ...
12620         (build_init_ctor): ... here.
12622 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12624         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
12625         (slsr_cand_d): Redefine def_phi.
12626         (stride_status, phi_adjust_status, count_phis_status): New enums.
12627         (find_phi_def): New.
12628         (find_basis_for_base_expr): New.
12629         (find_basis_for_candidate): Handle hidden bases.
12630         (alloc_cand_and_find_basis): Handle phi candidates.
12631         (slsr_process_phi): New.
12632         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
12633         (create_mul_imm_cand): Likewise.
12634         (create_add_ssa_cand): Exclude phi base candidates.
12635         (create_add_imm_cand): Likewise.
12636         (slsr_process_cast): Likewise.
12637         (slsr_process_copy): Likewise.
12638         (find_candidates_in_block): Handle phi candidates.
12639         (dump_candidate): Likewise.
12640         (unconditional_cands): Delete.
12641         (unconditional_cands_with_known_stride_p): Delete.
12642         (phi_dependent_cand_p): New.
12643         (cand_increment): Handle phi-dependent candidates.
12644         (replace_dependent): Delete.
12645         (replace_mult_candidate): New.
12646         (replace_unconditional_candidate): New.
12647         (incr_vec_index): Move to avoid forward reference.
12648         (create_add_on_incoming_edge): New.
12649         (create_phi_basis): New.
12650         (replace_dependents): Delete.
12651         (replace_conditional_candidate): New.
12652         (phi_add_costs): New.
12653         (replace_uncond_cands_and_profitable_phis): New.
12654         (record_increment): Handle phi adjustments.
12655         (record_phi_increments): New.
12656         (record_increments): Handle phi adjustments.
12657         (phi_incr_cost): New.
12658         (lowest_cost_path): Handle phis.
12659         (total_savings): Likewise.
12660         (analyze_increments): Likewise.
12661         (ncd_with_phi): New.
12662         (ncd_of_cand_and_phis): New.
12663         (nearest_common_dominator_for_cands): Handle phi increments.
12664         (all_phi_incrs_profitable): New.
12665         (replace_profitable_candidates): Handle phi-dependent candidates.
12666         (analyze_candidates_and_replace): Likewise.
12668 2013-05-03  Teresa Johnson  <tejohnson@google.com>
12670         PR bootstrap/57154
12671         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
12672         do not exceed REG_BR_PROB_BASE.
12674 2013-05-03  Jeff Law  <law@redhat.com>
12676         PR tree-optimization/57144
12677         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
12678         operand of the condition will bit into the new type when eliminating
12679         a cast feeding a condition.
12681 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
12683         PR rtl-optimization/57130
12684         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
12685         of COMPARE as in_code to the recursive call if needed.
12687 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
12689         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
12690         (enabled): Handle new members.
12691         * config/i386/sse.md (*vec_concatv2si): Merge from
12692         *vec_concatv2si_sse2 and vec_concatv2si_sse.
12693         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
12695 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
12697         PR tree-optimization/57027
12698         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
12699         for fnms opportunity, check we got the prerequisite kind
12700         of tree / gimple before using accessor functions.
12702 2013-05-03  Richard Biener  <rguenther@suse.de>
12704         * double-int.h (lshift): New overload without precision
12705         and arith argument.
12706         (operator *=, operator +=, operator -=): Move ...
12707         * double-int.c (operator *=, operator +=, operator -=): ... here
12708         and implement more efficiently.
12709         (mul_double_with_sign): Remove.
12710         (lshift_double): Adjust to take unsinged shift argument, push
12711         dispatching code to callers.
12712         (mul_double_wide_with_sign): Add early out for callers that
12713         are not interested in high parts or overflow.
12714         (lshift): New function.
12715         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
12716         dispatch code here.
12717         (lrotate, rrotate): Use logical shifts.
12718         * expr.c (get_inner_reference): Use lshift.
12719         * fixed-value.c (do_fixed_divide): Likewise.
12720         * tree-dfa.c (get_ref_base_and_extent): Likewise.
12721         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
12722         (indirect_refs_may_alias_p): Likewise.
12723         (stmt_kills_ref_p_1): Likewise.
12725 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12727         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
12729 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12731         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
12732         scalar form of FABD instruction.
12734 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
12736         * lra-constraints.c (process_alt_operands): Add checking alt
12737         number to choose the best alternative.
12739 2013-05-02  Richard Biener  <rguenther@suse.de>
12741         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
12742         bitmap and its handling.
12743         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
12745 2013-05-02  Richard Biener  <rguenther@suse.de>
12747         PR middle-end/57140
12748         * tree-inline.c (copy_loops): Properly handle removed loops.
12749         (copy_cfg_body): Mark destination loops for fixup if source
12750         loops needed fixup.
12752 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
12754         PR target/56732
12755         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
12756         generating simple_return for naked functions.
12758 2013-05-02  Martin Jambor  <mjambor@suse.cz>
12760         PR middle-end/56988
12761         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
12762         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
12763         flags match.
12764         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
12765         ipa_agg_replacement_value structures.
12766         (known_aggs_to_agg_replacement_list): Likewise.
12767         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
12768         (read_agg_replacement_chain): Likewise.
12769         (ipcp_transform_function): Also check that by_ref flags match.
12771 2013-05-02  Richard Biener  <rguenther@suse.de>
12773         * graphds.h (struct graph): Add obstack member.
12774         * graphds.c (new_graph): Initialize obstack and allocate
12775         vertices from it.
12776         (add_edge): Allocate edge from the obstack.
12777         (free_graph): Free the obstack instead of all edges and vertices.
12779 2013-05-02  Teresa Johnson  <tejohnson@google.com>
12781         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
12782         divides.
12783         * cfg.c (update_bb_profile_for_threading): Ditto.
12784         * tree-inline.c (copy_bb): Ditto.
12785         (copy_edges_for_bb): Ditto.
12786         (initialize_cfun): Ditto.
12787         (copy_cfg_body): Ditto.
12788         (expand_call_inline): Ditto.
12789         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
12790         (estimate_node_size_and_time): Ditto.
12791         (inline_merge_summary): Ditto.
12792         * cgraphclones.c (cgraph_clone_edge): Ditto.
12793         (cgraph_clone_node): Ditto.
12794         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12795         (compute_trg_info): Ditto.
12797 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12799         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
12800         S reg when fp attribute set.
12801         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
12803 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12805         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
12806         New pattern.
12807         (*and_one_cmplsi3_compare0_uxtw): Likewise.
12808         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
12809         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
12811 2013-05-02  Richard Biener  <rguenther@suse.de>
12813         * tree-scalar-evolution.c (scev_info_hasher): Remove.
12814         (struct instantiate_cache_entry): New type.
12815         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
12816         (struct instantiate_cache_type): New type.
12817         (set_instantiated_value, get_instantiated_value): Remove.
12818         (get_instantiated_value_entry): New function.
12819         (instantiate_scev_name): Use the new cache and adjust.
12820         (instantiate_scev_poly): Adjust.
12821         (instantiate_scev_binary): Likewise.
12822         (instantiate_array_ref): Likewise.
12823         (instantiate_scev_convert): Likewise.
12824         (instantiate_scev_not): Likewise.
12825         (instantiate_scev_3): Likewise.
12826         (instantiate_scev_2): Likewise.
12827         (instantiate_scev_r): Likewise.
12828         (instantiate_scev): Likewise.
12829         (resolve_mixers): Likewise.
12831 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
12833         PR target/57091
12834         * lra-constraints.c (best_small_class_operands_num): Remove.
12835         (process_alt_operands): Remove small_class_operands_num.  Take
12836         small classes operands into losers and only if the operand is not
12837         matched.  Modify debugging output.
12838         (curr_insn_transform): Remove best_small_class_operands_num.
12839         Print insn name.
12841 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12843         * config/aarch64/aarch64-builtins.c
12844         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
12845         * config/aarch64/aarch64-simd-builtins.def
12846         (reduc_splus_): Add new modes.
12847         (reduc_uplus_): New.
12848         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
12849         (reduc_uplus_v4sf): Likewise.
12850         (reduc_splus_v4sf): Likewise.
12851         (aarch64_addv<mode>): Likewise.
12852         (reduc_uplus_<mode>): Likewise.
12853         (reduc_splus_<mode>): Likewise.
12854         (aarch64_addvv2di): Likewise.
12855         (reduc_uplus_v2di): Likewise.
12856         (reduc_splus_v2di): Likewise.
12857         (aarch64_addvv2si): Likewise.
12858         (reduc_uplus_v2si): Likewise.
12859         (reduc_splus_v2si): Likewise.
12860         (reduc_<sur>plus_<mode>): New.
12861         (reduc_<sur>plus_v2di): Likewise.
12862         (reduc_<sur>plus_v2si): Likewise.
12863         (reduc_<sur>plus_v4sf): Likewise.
12864         (aarch64_addpv4sf): Likewise.
12865         * config/aarch64/arm_neon.h
12866         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
12867         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
12868         add UNSPEC_SADDV, UNSPEC_UADDV.
12869         (SUADDV): New.
12870         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
12872 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12874         * config/aarch64/arm_neon.h
12875         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
12877 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12879         * config/aarch64/aarch64-builtins
12880         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
12882 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12884         * config/aarch64/aarch64-simd-builtins.def
12885         (reduc_smax_): New.
12886         (reduc_smin_): Likewise.
12887         (reduc_umax_): Likewise.
12888         (reduc_umin_): Likewise.
12889         (reduc_smax_nan_): Likewise.
12890         (reduc_smin_nan_): Likewise.
12891         (fmax): Remove.
12892         (fmin): Likewise.
12893         (smax): Update for V2SF, V4SF and V2DF modes.
12894         (smin): Likewise.
12895         (smax_nan): New.
12896         (smin_nan): Likewise.
12897         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
12898         (<su><maxmin><mode>3): ...This, refactor.
12899         (s<maxmin><mode>3): New.
12900         (<maxmin_uns><mode>3): Likewise.
12901         (reduc_<maxmin_uns>_<mode>): Refactor.
12902         (reduc_<maxmin_uns>_v4sf): Likewise.
12903         (reduc_<maxmin_uns>_v2si): Likewise.
12904         (aarch64_<fmaxmin><mode>: Remove.
12905         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
12906         new builtin names.
12907         (vmin<q>_f<32,64>): Likewise.
12908         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12909         (FMAXMIN): New.
12910         (su): Add mappings for smax, smin, umax, umin.
12911         (maxmin): New.
12912         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12913         (FMAXMIN): Rename as...
12914         (FMAXMIN_UNS): ...This.
12915         (maxminv): Remove.
12916         (fmaxminv): Likewise.
12917         (fmaxmin): Likewise.
12918         (maxmin_uns): New.
12919         (maxmin_uns_op): Likewise.
12921 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12923         * config/aarch64/arm_neon.h
12924         (vac<ge, gt><sd>_f<32, 64>): Rename to...
12925         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
12926         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
12928 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12930         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
12931         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
12933 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12935         * config/aarch64/aarch64-simd.md
12936         (vcond<mode>_internal): Handle special cases for constant masks.
12937         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
12938         (vcondu<mode><mode>): Likewise.
12939         (vcond<v_cmp_result><mode>): New.
12941 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12943         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
12944         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
12945         * config/aarch64/aarch64-simd-builtins.def
12946         (cmeq): Update to BUILTIN_VALLDI.
12947         (cmgt): Likewise.
12948         (cmge): Likewise.
12949         (cmle): Likewise.
12950         (cmlt): Likewise.
12951         * config/aarch64/arm_neon.h
12952         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
12953         to builtins or C as appropriate.
12955 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12957         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
12958         (cmgeu): ...This.
12959         (cmhi): Rename to...
12960         (cmgtu): ...This.
12961         * config/aarch64/aarch64-simd.md
12962         (simd_mode): Add SF.
12963         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
12964         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
12965         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
12966         (cstore<mode>_neg): ...This.
12967         * config/aarch64/iterators.md
12968         (VALLF): new.
12969         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
12970         (COMPARISONS): New.
12971         (UCOMPARISONS): Likewise.
12972         (optab): Add missing comparisons.
12973         (n_optab): New.
12974         (cmp_1): Likewise.
12975         (cmp_2): Likewise.
12976         (CMP): Likewise.
12977         (cmp): Remove.
12978         (VCMP_S): Likewise.
12979         (VCMP_U): Likewise.
12980         (V_cmp_result): Add DF, SF modes.
12981         (v_cmp_result): Likewise.
12982         (v): Likewise.
12983         (vmtype): Likewise.
12984         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
12986 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
12988         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
12989         define_insn to define_insn_and_split.
12990         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
12991         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
12992         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
12993         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
12994         (thumb2_negscc): Likewise.
12996 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
12998         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13000 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13002         * config/arm/thumb2.md: Remove trailing whitespaces.
13004 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13006         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13007         Use gen_int_mode rather than GEN_INT.
13009 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13011         * value-prof.c (stream_in_histogram_value): Remove the strayed
13012         debug_gimple_stmt.
13014 2013-04-30  Richard Biener  <rguenther@suse.de>
13016         PR middle-end/57122
13017         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13019 2013-04-30  Richard Biener  <rguenther@suse.de>
13021         PR middle-end/57107
13022         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13024 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13026         PR rtl-optimization/56957
13027         PR rtl-optimization/57105
13028         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13029         variable.  Use just INSN_UID for determining whether an insn
13030         should be only disconnected from the insn stream.
13031         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13033 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13035         PR tree-optimization/57104
13036         * tsan.c (instrument_expr): Don't instrument accesses to
13037         DECL_HARD_REGISTER VAR_DECLs.
13039 2013-04-30  Richard Biener  <rguenther@suse.de>
13041         * function.h (loops_for_fn): New inline function.
13042         (set_loops_for_fn): Likewise.
13043         * cfgloop.h (place_new_loop): Add struct function parameter.
13044         (get_loop): Likewise.
13045         (get_loops): Likewise.
13046         (number_of_loops): Likewise.
13047         (fel_next): Adjust.
13048         (fel_init): Likewise.
13049         * cfg.c (get_loop_copy): Adjust.
13050         * cfgloop.c (flow_loops_dump): Likewise.
13051         (record_loop_exits): Likewise.
13052         (verify_loop_structure): Likewise.
13053         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13054         (estimate_reg_pressure_cost): Likewise.
13055         (mark_loop_exit_edges): Likewise.
13056         * cfgloopmanip.c (place_new_loop): Likewise.
13057         (add_loop): Likewise.
13058         (duplicate_loop): Likewise.
13059         * graph.c (draw_cfg_nodes): Likewise.
13060         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13061         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13062         (extract_affine_chrec): Likewise.
13063         (build_scop_iteration_domain): Likewise.
13064         * graphite.c (graphite_initialize): Likewise.
13065         * ira-build.c (create_loop_tree_nodes): Likewise.
13066         (more_one_region_p): Likewise.
13067         (rebuild_regno_allocno_maps): Likewise.
13068         (mark_loops_for_removal): Likewise.
13069         (mark_all_loops_for_removal): Likewise.
13070         (remove_unnecessary_regions): Likewise.
13071         (ira_build): Likewise.
13072         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13073         * loop-init.c (fix_loop_structure): Likewise.
13074         (gate_rtl_move_loop_invariants): Likewise.
13075         (gate_rtl_unswitch): Likewise.
13076         (gate_rtl_unroll_and_peel_loops): Likewise.
13077         (rtl_doloop): Likewise.
13078         * lto-streamer-in.c (input_cfg): Likewise.
13079         * lto-streamer-out.c (output_cfg): Likewise.
13080         * modulo-sched.c (sms_schedule): Likewise.
13081         * predict.c (tree_estimate_probability): Likewise.
13082         (tree_estimate_probability_driver): Likewise.
13083         (estimate_loops): Likewise.
13084         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13085         (move_sese_region_to_fn): Likewise.
13086         (debug_loop_num): Likewise.
13087         * tree-chrec.c (chrec_evaluate): Likewise.
13088         (hide_evolution_in_other_loops_than_loop): Likewise.
13089         (chrec_component_in_loop_num): Likewise.
13090         (reset_evolution_in_loop): Likewise.
13091         (evolution_function_is_invariant_rec_p): Likewise.
13092         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13093         * tree-inline.c (copy_loops): Likewise.
13094         (copy_cfg_body): Likewise.
13095         (tree_function_versioning): Likewise.
13096         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13097         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13098         Likewise.
13099         (add_to_evolution_1): Likewise.
13100         (scev_const_prop): Likewise.
13101         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13102         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13103         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13104         (tree_ssa_lim_initialize): Likewise.
13105         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13106         (verify_loop_closed_ssa): Likewise.
13107         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13108         (tree_ssa_loop_im): Likewise.
13109         (tree_ssa_loop_unswitch): Likewise.
13110         (tree_vectorize): Likewise.
13111         (check_data_deps): Likewise.
13112         (tree_ssa_loop_ivcanon): Likewise.
13113         (tree_ssa_loop_bounds): Likewise.
13114         (tree_complete_unroll): Likewise.
13115         (tree_complete_unroll_inner): Likewise.
13116         (tree_parallelize_loops): Likewise.
13117         (tree_ssa_loop_prefetch): Likewise.
13118         (tree_ssa_loop_ivopts): Likewise.
13119         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13120         * tree-vectorizer.c (vectorize_loops): Likewise.
13122 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13124         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13125         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13126         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13127         with EABI_LINK_SPEC.
13129 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13131         PR target/44578
13132         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13133         alternative.
13135 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13137         PR target/57097
13138         * lra-constraints.c (process_alt_operands): Discourage a bit more
13139         using memory for pseudos.  Print cost dump for alternatives.
13140         Modify cost values for conflicts with early clobbers.
13141         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13143 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13145         PR target/57098
13146         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13148 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13150         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13151         from/to S register.
13152         (movdi_aarch64): Support LDR/STR from/to D register.
13154 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13156         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13157         or higher by default.
13159 2013-04-29  Richard Biener  <rguenther@suse.de>
13161         PR middle-end/57075
13162         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13163         even if not adding abnormal edges for calls that can make
13164         abnormal gotos.
13166 2013-04-29  Richard Biener  <rguenther@suse.de>
13168         PR middle-end/57103
13169         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13170         TREE_BLOCK.
13171         (move_stmt_r): Remove redundant checking.
13173 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13175         PR bootstrap/57077
13176         * basic-block.h (apply_scale): New function.
13177         (apply_probability): Use apply_scale.
13178         * gimple-streamer-in.c (input_bb): Ditto.
13179         * lto-streamer-in.c (input_cfg): Ditto.
13180         * lto-cgraph.c (merge_profile_summaries): Ditto.
13181         * tree-optimize.c (execute_fixup_cfg): Ditto.
13182         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13183         (copy_edges_for_bb): Ditto.
13184         (copy_cfg_body): Ditto.
13186 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13188         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13189         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13190         (tail_merge_optimize): Handle current_loops == NULL.
13192 2013-04-26  Jeff Law  <law@redhat.com>
13194         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13195         (simplify_cond_using_ranges): Generalize code to simplify
13196         COND_EXPRs where one argument is a constant and the other
13197         is an SSA_NAME created by an integral type conversion.
13199 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13201         * config/arm/arm.md (store_minmaxsi): Use only when
13202         optimize_insn_for_size_p.
13204 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13206         PR target/57108
13207         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13209 2013-04-29  Richard Biener  <rguenther@suse.de>
13211         PR middle-end/57089
13212         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13213         loop tree make sure to schedule a fixup for the child as well.
13214         (expand_omp_for_generic): Properly add loops.
13215         (expand_omp_for_static_nochunk): Likewise.
13216         (expand_omp_for_static_chunk): Likewise.
13217         (expand_omp_for): For the degenerate case fixup loops.
13218         (expand_omp_sections): Fix default bb placement in loops.
13219         (expand_omp_atomic_pipeline): Properly add loops.
13221 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13223         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13225 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13227         * tree-ssa-tail-merge.c: Update header comment.
13229 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13231         * config/aarch64/arm_neon.h
13232         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13233         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13234         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13235         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13236         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13237         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13238         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13239         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13241 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13243         * config/aarch64/aarch64-simd.md
13244         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13245         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13246         fix_trunc, fixuns_trunc.
13247         (ftrunc<VDQF:mode>2): New.
13248         * config/aarch64/iterators.md (optab): Add fix, fixuns.
13249         (fix_trunc_optab): New.
13251 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13253         * config/aarch64/aarch64-builtins.c
13254         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13255         iceilf, lround, iroundf.
13257 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13259         PR target/54349
13260         * config/i386/i386.h (enum ix86_tune_indices)
13261         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13262         New, split from X86_TUNE_INTER_UNIT_MOVES.
13263         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13264         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13265         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13266         (TARGET_INTER_UNIT_MOVES): Remove.
13267         * config/i386/i386.c (initial_ix86_tune_features): Update.
13268         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13269         (ix86_expand_convert_uns_didf_sse): Use
13270         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13271         (ix86_expand_vector_init_one_nonzero): Ditto.
13272         (ix86_expand_vector_init_interleave): Ditto.
13273         (inline_secondary_memory_needed): Return true for moves from SSE class
13274         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13275         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13276         * config/i386/constraints.md (Yi, Ym): Depend on
13277         TARGET_INTER_UNIT_MOVES_TO_VEC.
13278         (Yj, Yn): New constraints.
13279         * config/i386/i386.md (*movdi_internal): Change constraints of
13280         operand 1 from Yi to Yj and from Ym to Yn.
13281         (*movsi_internal): Ditto.
13282         (*movdf_internal): Ditto.
13283         (*movsf_internal): Ditto.
13284         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13285         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13286         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13287         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13288         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13289         * config/i386/sse.md (movdi_to_sse): Ditto.
13290         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13291         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13292         TARGET_INTER_UNIT_MOVES.
13293         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13294         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13295         instead of TARGET_INTER_UNIT_MOVES.
13296         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13297         operand 1 from Yi to Yj and from Ym to Yn.
13299 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13301         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13302         (float_truncate_hi_): Likewise.
13303         (float_extend_lo_): Likewise.
13304         (float_truncate_lo_): Likewise.
13305         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13306         (aarch64_float_extend_lo_v2df): Likewise.
13307         (vec_unpacks_hi_v4sf): Likewise.
13308         (aarch64_float_truncate_lo_v2sf): Likewise.
13309         (aarch64_float_truncate_hi_v4sf): Likewise.
13310         (vec_pack_trunc_v2df): Likewise.
13311         (vec_pack_trunc_df): Likewise.
13313 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13315         * config/aarch64/aarch64-builtins.c
13316         (aarch64_fold_builtin): Fold float conversions.
13317         * config/aarch64/aarch64-simd-builtins.def
13318         (floatv2si, floatv4si, floatv2di): New.
13319         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13320         * config/aarch64/aarch64-simd.md
13321         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13322         * config/aarch64/iterators.md (FLOATUORS): New.
13323         (optab): Add float, floatuns.
13324         (su_optab): Likewise.
13326 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13328         * config/aarch64/aarch64-builtins.c
13329         (aarch64_builtin_vectorized_function): Use new names for
13330         fcvt builtins.
13331         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13332         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13333         (fcvtzu): Split as...
13334         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13335         (fcvtas): Split as...
13336         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13337         (fcvtau): Split as...
13338         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13339         (fcvtps): Split as...
13340         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13341         (fcvtpu): Split as...
13342         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13343         (fcvtms): Split as...
13344         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13345         (fcvtmu): Split as...
13346         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13347         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13348         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13349         (lfrintnusf, lfrintnudf): Likewise.
13350         * config/aarch64/aarch64-simd.md
13351         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13352         define_insn.
13353         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13354         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13355         (fcvt_pattern): Likewise.
13357 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13359         * config/aarch64/aarch64-simd.md
13360         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13361         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13363 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13365         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13366         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13367         (vrnd<a,m,n,p>_f32): Implement using builtins.
13368         (vrnd<i,x><q>_f<32, 64>): New.
13370 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13372         * config/aarch64/aarch64-builtins.c
13373         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13374         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13375         (frintz): Rename to...
13376         (btrunc): ...this.
13377         (frintp): Rename to...
13378         (ceil): ...this.
13379         (frintm): Rename to...
13380         (floor): ...this.
13381         (frinti): Rename to...
13382         (nearbyint): ...this.
13383         (frintx): Rename to...
13384         (rint): ...this.
13385         (frinta): Rename to...
13386         (round): ...this.
13387         * config/aarch64/aarch64-simd.md
13388         (aarch64_frint<frint_suffix><mode>): Delete.
13389         (<frint_pattern><mode>2): Convert to insn.
13390         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13391         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13392         (frint_pattern): Likewise.
13393         (frint_suffix): Likewise.
13395 2013-04-29  Richard Biener  <rguenther@suse.de>
13397         PR tree-optimization/57081
13398         * loop-init.c: Include tree-flow.h.
13399         (loop_optimizer_finalize): Free number of iteration estimates.
13400         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13402 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
13404         PR tree-optimization/57083
13405         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13406         non-singleton shift count range, zero extend low_bound for uns case.
13408         * config/i386/predicates.md (general_vector_operand): New predicate.
13409         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13410         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13411         if they aren't nonimmediate operands.  If their original values
13412         satisfy const_vector_equal_evenodd_p, don't shift them.
13413         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13414         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
13415         if not nonimmediate_operand.
13416         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13417         instead of register_operand.
13418         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13420 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
13422         * stor-layout.c (finalize_size_functions): Allocate a structure and
13423         reset cfun before dumping the functions.
13425 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
13427         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13429         PR target/56866
13430         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13431         use xop_pmacsdqh if uns_p.
13432         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13433         the immediate rotate count.
13435 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
13437         * rtl.h (struct rtx_def): Add comment for field jump.
13438         (LRA_SUBREG_P): New macro.
13439         * recog.c (register_operand): Check LRA_SUBREG_P.
13440         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13441         stack frame.
13442         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13443         (lra_final_code_change): Skip subreg change for operators.
13444         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13445         if there are no operand changes.
13446         * lra-constraints.c (curr_insn_set): New.
13447         (match_reload): Set LRA_SUBREG_P.
13448         (emit_spill_move): Ditto.
13449         (check_and_process_move): Use curr_insn_set. Process only single
13450         set insns.  Don't initialize sec_mem_p and change_p.
13451         (simplify_operand_subreg): Use LRA_SUBREG_P.
13452         (reg_in_class_p): New function.
13453         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
13454         of #ifdef.  Add code to remove cycling.
13455         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13456         non-null disp.  Reload inner instead of disp when base and index
13457         are null.  Try to put lo_sum into register.
13458         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13459         (check_and_process_move): Move code for move cost check to
13460         simple_move_p.  Remove equiv_substitution.
13461         (simple_move_p): New function.
13462         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
13463         curr_insn_set.  Call check_and_process_move only for single set
13464         insns.  Use the new function.  Move call of check_and_process_move
13465         after operand equiv substitution and address process.
13467 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13469         PR go/57045
13470         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13471         with nonlocal goto receivers or returns twice calls, ignore
13472         unininitialized values from abnormal edges to nl goto receiver
13473         or returns twice call.
13475 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13477         PR tree-optimization/57051
13478         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13479         and VEC_RSHIFT_EXPR if shift count is a multiple of element
13480         bitsize.
13482 2013-04-26  Richard Biener  <rguenther@suse.de>
13484         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13485         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
13486         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13487         (fixup_loop_arrays_after_move): New function.
13488         (move_sese_region_to_fn): Properly outline the loop tree parts
13489         of the SESE region.
13491 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
13493         * config/i386/i386.md (type, unit): Fix long lines.
13495 2013-04-26  Richard Biener  <rguenther@suse.de>
13497         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13498         (lto-streamer-out.o): Likewise.
13499         * cfgloop.c (init_loops_structure): Export, add struct function
13500         argument and adjust.
13501         (flow_loops_find): Adjust.
13502         * cfgloop.h (enum loop_estimation): Add EST_LAST.
13503         (init_loops_structure): Declare.
13504         * lto-streamer-in.c: Include cfgloop.h.
13505         (input_cfg): Input the loop tree.
13506         * lto-streamer-out.c: Include cfgloop.h.
13507         (output_cfg): Output the loop tree.
13508         (output_struct_function_base): Do not drop PROP_loops.
13510 2013-03-26  Richard Biener  <rguenther@suse.de>
13512         * tree-cfg.c (execute_build_cfg): Build the loop tree.
13513         (pass_build_cfg): Provide PROP_loops.
13514         (move_sese_region_to_fn): Remove loops that are outlined into fn
13515         for now.
13516         * tree-inline.c: Include cfgloop.h.
13517         (initialize_cfun): Do not drop PROP_loops.
13518         (copy_loops): New function.
13519         (copy_cfg_body): Copy loop structure.
13520         (tree_function_versioning): Initialize destination loop tree.
13521         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13522         (pass_parallelize_loops): Do IL verification.
13523         * loop-init.c (loop_optimizer_init): Fixup loops if required.
13524         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13525         the CFG make sure we fixup loops as well.
13526         * tree-ssa-tail-merge.c: Include cfgloop.h.
13527         (replace_block_by): When merging loop latches mark loops for fixup.
13528         * lto-streamer-out.c (output_struct_function_base): Drop
13529         PROP_loops for now.
13530         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13531         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13532         * ipa-split.c: Include cfgloop.h.
13533         (split_function): Add the new return block to the loop tree root.
13534         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13535         whether we have removed the forwarder block.
13536         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13537         * cfgloop.h (place_new_loop): Declare.
13538         * cfgloopmanip.c (place_new_loop): Export.
13539         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13540         (tree-switch-conversion.o): Likewise.
13541         (tree-complex.o): Likewise.
13542         (tree-inline.o): Likewise.
13543         (tree-ssa-tailmerge.o): Likewise.
13544         (ipa-split.o): Likewise.
13545         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13546         (tree-ssa-copy.o): Likewise.
13547         * tree-switch-conversion.c: Include cfgloop.h
13548         (process_switch): If we emit a bit-test cascade, schedule loops
13549         for fixup.
13550         * tree-complex.c: Include cfgloop.h.
13551         (expand_complex_div_wide): Properly add new basic-blocks to loops.
13552         * asan.c: Include cfgloop.h.
13553         (create_cond_insert_point): Properly add new basic-blocks to
13554         loops, schedule loop fixup.
13555         * cfgloop.c (verify_loop_structure): Check that looks are not
13556         marked for fixup.
13557         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
13558         to loops.
13559         (expand_omp_for_generic): Likewise.
13560         (expand_omp_sections): Likewise.
13561         (expand_omp_atomic_pipeline): Schedule loops for fixup.
13562         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
13563         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
13564         is initialized, not when loops are present.
13565         * tree-parloops.c (parallelize_loops): Remove checking here.
13566         * passes.c (init_optimization_passes): Schedule a copy-propagation
13567         pass before complete unrolling of inner loops.
13569 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13571         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
13572         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
13573         (colorize_init): Add argument to _WIN32 version.
13574         * toplev.c: Include diagnostic-color.h.
13575         (process_options): Default to -fdiagnostics-color=auto if
13576         GCC_COLORS env var is in the environment.
13577         * common.opt (fdiagnostics-color=): Add Var and Init.
13578         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
13579         env var is in the environment, the default is auto rather than never.
13581         * diagnostic.h (file_name_as_prefix): Add context argument.
13582         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
13583         the string as locus.
13584         * langhooks.c (lhd_print_error_function): Adjust caller.
13586 2013-04-25  Lawrence Crowl  <crowl@google.com>
13588         * var-tracking.c (shared_hash_def::htab):
13589         Change type to hash_table.  Update dependent calls and types.
13591 2013-04-25  Lawrence Crowl  <crowl@google.com>
13593         * Makefile.in: Update as needed below.
13595         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
13596         Move declaration to after the type's method definitons.
13598         * attribs.c (htab_t scoped_attributes::attribute_hash):
13599         Change type to hash_table.  Update dependent calls and types.
13601         * bitmap.c (htab_t bitmap_desc_hash):
13602         Change type to hash_table.  Update dependent calls and types.
13604         * cselib.c (htab_t cselib_hash_table):
13605         Change type to hash_table.  Update dependent calls and types.
13607         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
13608         (hash_string_slot_node): Move implementation into lto-streamer.h
13609         struct string_slot_hasher.
13610         (eq_string_slot_node): Likewise.
13612         * data-streamer-out.c: Update output_block::string_hash_table
13613         dependent calls and types.
13615         * dwarf2cfi.c (htab_t trace_index):
13616         Change type to hash_table.  Update dependent calls and types.
13618         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
13619         Change type to hash_table.  Update dependent calls and types.
13620         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
13621         (htab_t optimize_external_refs::map): Likewise.
13622         (htab_t output_comp_unit::extern_map): Likewise.
13623         (htab_t output_comdat_type_unit::extern_map): Likewise.
13624         (htab_t output_macinfo::macinfo_htab): Likewise.
13625         (htab_t optimize_location_lists::htab): Likewise.
13626         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
13628         * except.c (htab_t ehspec_hash_type):
13629         Change type to hash_table.  Update dependent calls and types.
13630         (assign_filter_values::ttypes): Likewise.
13631         (assign_filter_values::ehspec): Likewise.
13632         (sjlj_assign_call_site_values::ar_hash): Likewise.
13633         (convert_to_eh_region_ranges::ar_hash): Likewise.
13635         * gcse.c (htab_t pre_ldst_table):
13636         Change type to hash_table.  Update dependent calls and types.
13638         * ggc-common.c (htab_t saving_htab):
13639         Change type to hash_table.  Update dependent calls and types.
13640         (htab_t loc_hash): Likewise.
13641         (htab_t ptr_hash): Likewise.
13642         (call_count): Rename ggc_call_count.
13643         (call_alloc): Rename ggc_call_alloc.
13644         (loc_descriptor): Rename make_loc_descriptor.
13645         (add_statistics): Rename ggc_add_statistics.
13647         * ggc-common.c (saving_htab):
13648         Change type to hash_table.  Update dependent calls and types.
13650         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
13651         (push_gimplify_context): Likewise.
13652         (pop_gimplify_context): Likewise.
13653         (struct gimple_temp_hash_elt): Added.
13654         (struct gimplify_hasher): Likewise.
13655         (struct gimplify_ctx.temp_htab):
13656         Change type to hash_table.  Update dependent calls and types.
13658         * gimple-fold.c: Include gimplify-ctx.h.
13660         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
13661         Change type to hash_table.  Update dependent calls and types.
13662         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
13663         avoid potential global name collision.
13665         * gimplify.c: Include gimplify-ctx.h.
13666         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
13667         (htab_t gimplify_ctx::temp_htab):
13668         Update dependent calls and types for new type hash_table.
13669         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
13670         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
13672         * gimplify-ctx.h: New.
13673         (struct gimple_temp_hash_elt): Move from gimplify.c.
13674         (class gimplify_hasher): New.
13675         (struct gimplify_ctx): Move from gimple.h.
13676         (htab_t gimplify_ctx::temp_htab):
13677         Change type to hash_table.  Update dependent calls and types.
13679         * graphite-clast-to-gimple.c: Include graphite-htab.h.
13680         (htab_t ivs_params::newivs_index):
13681         Change type to hash_table.  Update dependent calls and types.
13682         (htab_t ivs_params::params_index): Likewise.
13683         (htab_t print_generated_program::params_index): Likewise.
13684         (htab_t gloog::newivs_index): Likewise.
13685         (htab_t gloog::params_index): Likewise.
13687         * graphite.c: Include graphite-htab.h.
13688         4htab_t graphite_transform_loops::bb_pbb_mapping):
13689         Change type to hash_table.  Update dependent calls and types.
13691         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
13692         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
13693         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
13695         * graphite-dependences.c: Include graphite-htab.h.
13696         (loop_is_parallel_p): Change hash table type of parameter.
13698         * graphite-htab.h: New.
13699         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
13700         (extern find_pbb_via_hash): Move from graphite-poly.h.
13701         (extern loop_is_parallel_p): Move from graphite-poly.h.
13702         (extern get_loop_body_pbbs): Move from graphite-poly.h.
13704         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
13705         (extern loop_is_parallel_p): Move to graphite-htab.h.
13706         (extern get_loop_body_pbbs): Move to graphite-htab.h.
13708         * haifa-sched.c (htab_t delay_htab):
13709         Change type to hash_table.  Update dependent calls and types.
13710         (htab_t delay_htab_i2): Likewise.
13712         * ira-color.c (htab_t allocno_hard_regs_htab):
13713         Change type to hash_table.  Update dependent calls and types.
13715         * ira-costs.c (htab_t cost_classes_htab):
13716         Change type to hash_table.  Update dependent calls and types.
13718         * loop-invariant.c (htab_t merge_identical_invariants::eq):
13719         Change type to hash_table.  Update dependent calls and types.
13721         * loop-iv.c (htab_t bivs):
13722         Change type to hash_table.  Update dependent calls and types.
13724         * loop-unroll.c (htab_t opt_info::insns_to_split):
13725         Change type to hash_table.  Update dependent calls and types.
13726         (htab_t opt_info::insns_with_var_to_expand): Likewise.
13728         * lto-streamer.h (struct string_slot): Move from data-streamer.h
13729         (struct string_slot_hasher): New.
13730         (htab_t output_block::string_hash_table):
13731         Change type to hash_table.  Update dependent calls and types.
13733         * lto-streamer-in.c (freeing_string_slot_hasher): New.
13734         (htab_t file_name_hash_table):
13735         Change type to hash_table.  Update dependent calls and types.
13737         * lto-streamer-out.c: Update output_block::string_hash_table dependent
13738         calls and types.
13740         * lto-streamer.c (htab_t tree_htab):
13741         Change type to hash_table.  Update dependent calls and types.
13743         * omp-low.c: Include gimplify-ctx.h.
13745         * passes.c (htab_t name_to_pass_map):
13746         Change type to hash_table.  Update dependent calls and types.
13747         (pass_traverse): Rename to passes_pass_traverse.
13749         * plugin.c (htab_t event_tab):
13750         Change type to hash_table.  Update dependent calls and types.
13752         * postreload-gcse.c (htab_t expr_table):
13753         Change type to hash_table.  Update dependent calls and types.
13754         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
13756         * sese.c (debug_rename_map_1): Make extern.
13757         (htab_t copy_bb_and_scalar_dependences::rename_map):
13758         Change type to hash_table.  Update dependent calls and types.
13760         * sese.h (extern debug_rename_map): Move to .c file.
13762         * store-motion.c (htab_t store_motion_mems_table):
13763         Change type to hash_table.  Update dependent calls and types.
13765         * trans-mem.c (htab_t tm_new_mem_hash):
13766         Change type to hash_table.  Update dependent calls and types.
13768         * tree-browser.c (htab_t TB_up_ht):
13769         Change type to hash_table.  Update dependent calls and types.
13771         * tree-cfg.c (htab_t discriminator_per_locus):
13772         Change type to hash_table.  Update dependent calls and types.
13774         * tree-complex.c: Include tree-hasher.h
13775         (htab_t complex_variable_components):
13776         Change type to hash_table.  Update dependent calls and types.
13778         * tree-eh.c (htab_t finally_tree):
13779         Change type to hash_table.  Update dependent calls and types.
13781         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
13782         struct int_tree_hasher.
13783         (extern int_tree_map_eq): Likewise.
13784         (uid_decl_map_hash): Removed.
13785         (extern decl_tree_map_eq): Likewise.
13787         * tree-hasher.h: New.
13788         (struct int_tree_hasher): New.
13789         (typedef int_tree_htab_type): New.
13791         * tree-inline.c: Include gimplify-ctx.h.
13793         * tree-mudflap.c: Include gimplify-ctx.h.
13795         * tree-parloops.c: Include tree-hasher.h.
13796         (htab_t eliminate_local_variables_stmt::decl_address):
13797         Change type to hash_table.  Update dependent calls and types.
13798         (htab_t separate_decls_in_region::decl_copies): Likewise.
13800         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
13801         Change type to hash_table.  Update dependent calls and types.
13803         * tree-sra.c (candidates):
13804         Change type to hash_table.  Update dependent calls and types.
13806         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
13807         in tree-flow.h.
13808         (int_tree_map_hash): Likewise.
13810         * tree-ssa-dom.c (htab_t avail_exprs):
13811         Change type to hash_table.  Update dependent calls and types.
13813         * tree-ssa-live.c (var_map_base_init::tree_to_index):
13814         Change type to hash_table.  Update dependent calls and types.
13816         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
13817         Change type to hash_table.  Update dependent calls and types.
13819         * tree-ssa-phiopt.c (seen_ssa_names):
13820         Change type to hash_table.  Update dependent calls and types.
13822         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
13823         Change type to hash_table.  Update dependent calls and types.
13825         * tree-ssa-uncprop.c (equiv):
13826         Change type to hash_table.  Update dependent calls and types.
13828 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
13830         PR rtl-optimization/57003
13831         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
13832         call note_stores with kill_clobbered_value callback again after
13833         killing regs_invalidated_by_call.
13835 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13837         * config/aarch64/aarch64-simd.md
13838         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
13839         (aarch64_simd_bsl<mode>): Likewise.
13840         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
13842 2013-04-25  Marek Polacek  <polacek@redhat.com>
13844         PR tree-optimization/57066
13845         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
13847 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
13849         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
13851 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13853         * config/aarch64/aarch64-builtins.c
13854         (aarch64_fold_builtin): New.
13855         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
13856         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
13857         * config/aarch64/aarch64-simd-builtins.def (abs): New.
13858         * config/aarch64/arm_neon.h
13859         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
13861 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13862             Tejas Belagod  <tejas.belagod@arm.com>
13864         * config/aarch64/aarch64-builtins.c
13865         (aarch64_gimple_fold_builtin): New.
13866         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
13867         * config/aarch64/aarch64-simd-builtins.def (addv): New.
13868         * config/aarch64/aarch64-simd.md (addpv4sf): New.
13869         (addvv4sf): Update.
13870         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
13872 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13874         * config/aarch64/aarch64.md
13875         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
13877 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13879         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
13880         (*ngcsi_uxtw): New pattern.
13882 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13883             Julian Brown  <julian@codesourcery.com>
13885         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
13886         (TB_DREG): Add T_V4HF.
13887         (v4hf_UP): New macro.
13888         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13889         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13890         Handle initialisation of V4HF. Adjust initialisation of reinterpret
13891         built-ins.
13892         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13893         (arm_vector_mode_supported_p): Handle V4HF.
13894         (arm_mangle_map): Handle V4HFmode.
13895         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
13896         * config/arm/arm_neon_builtins.def: Add entries for
13897         vcvtv4hfv4sf, vcvtv4sfv4hf.
13898         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
13899         (neon_vcvtv4hfv4sf): Likewise.
13900         * config/arm/neon-gen.ml: Handle half-precision floating point
13901         features.
13902         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
13903         * config/arm/arm_neon.h: Regenerate.
13904         * config/arm/neon.ml (type elts): Add F16.
13905         (type vectype): Add T_float16x4, T_floatHF.
13906         (type vecmode): Add V4HF.
13907         (type features): Add Requires_FP_bit feature.
13908         (elt_width): Handle F16.
13909         (elt_class): Likewise.
13910         (elt_of_class_width): Likewise.
13911         (mode_of_elt): Refactor.
13912         (type_for_elt): Handle F16, fix error messages.
13913         (vectype_size): Handle T_float16x4.
13914         (vcvt_sh): New function.
13915         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
13916         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
13917         (string_of_mode): Handle V4HF.
13918         * doc/arm-neon-intrinsics.texi: Regenerate.
13920 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13922         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
13923         format specifier in 'X' case.
13925 2013-04-25  Alan Modra  <amodra@gmail.com>
13927         PR target/57052
13928         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
13929         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
13930         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
13931         Repeat for many other rotate/shift and mask patterns using subregs.
13932         Name lshiftrt insns.
13933         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
13934         on WORDS_BIG_ENDIAN.
13936 2013-04-25  Alan Modra  <amodra@gmail.com>
13938         * config.gcc: Support little-endian powerpc-linux targets.
13939         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
13940         (LINK_OS_LINUX_SPEC): Define.
13941         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
13942         Preserve MASK_LITTLE_ENDIAN.
13943         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
13944         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
13945         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
13946         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
13947         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
13948         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
13949         Correct fp word order for little-endian.  Don't shift toc entries
13950         smaller than a word for little-endian.
13951         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
13952         (bswapdi2 splits): Correct low-part subreg for little-endian.
13953         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
13954         low/high where such is correct only for be.
13955         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
13956         little-endian for -mcall-aixdesc.
13958 2013-04-25  Alan Modra  <amodra@gmail.com>
13960         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
13961         replace_equiv_address_nv.
13963 2013-04-25  Alan Modra  <amodra@gmail.com>
13965         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
13967 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13969         Revert:
13970         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13971         * rtl.h (struct rtx_def): ...
13973 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13975         PR rtl-optimizations/57046
13976         * lra-constraints (split_reg): Set up lra_risky_transformations_p
13977         for multi-reg splits.
13979 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
13981         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
13983 2013-04-24  Sterling Augustine  <saugustine@google.com>
13985         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
13986         (comp_dir_string, debug_str_dwo_section): New.
13987         (DEBUG_STR_DWO_SECTION): Rename to ...
13988         (DEBUG_DWO_STR_SECTION): ... this.
13989         (DEBUG_NORM_STR_SECTION): Delete.
13990         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
13991         (DEBUG_STR_DWO_SECTION_FLAGS): New.
13992         (find_AT_string): Move most logic to ...
13993         (find_AT_string_in_table): ... here.  New.
13994         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
13995         add_skeleton_AT_string.  Delete logic.
13996         (output_skeleton_debug_sections): Remove call to
13997         add_top_level_skeleton_die_attrs.
13998         (add_comp_dir_attribute): Move logic to comp_dir_string.
13999         (dwarf2out_init): Initialize debug_str_dwo_section.
14000         (output_indirect_string): Call find_string_form.
14001         (output_indirect_strings): Rewrite.
14002         (prune_unused_types): Empty skeleton_debug_str_hash.
14003         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14004         (dwarf2out_finish):  Call output_indirect_strings.
14006 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14008         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14010 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14012         * rtl.h (struct rtx_def): Add comment for field jump.
14013         (LRA_SUBREG_P): New macro.
14014         * recog.c (register_operand): Check LRA_SUBREG_P.
14015         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14016         stack frame.
14017         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14018         (lra_final_code_change): Skip subreg change for operators.
14019         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14020         if there are no operand changes.
14021         * lra-constraints.c (curr_insn_set): New.
14022         (match_reload): Set LRA_SUBREG_P.
14023         (emit_spill_move): Ditto.
14024         (check_and_process_move): Use curr_insn_set. Process only single
14025         set insns.  Don't initialize sec_mem_p and change_p.
14026         (simplify_operand_subreg): Use LRA_SUBREG_P.
14027         (reg_in_class_p): New function.
14028         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14029         of #ifdef.  Add code to remove cycling.
14030         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14031         non-null disp.  Reload inner instead of disp when base and index
14032         are null.  Try to put lo_sum into register.
14033         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14034         (check_and_process_move): Move code for move cost check to
14035         simple_move_p.  Remove equiv_substitution.
14036         (simple_move_p): New function.
14037         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14038         curr_insn_set.  Call check_and_process_move only for single set
14039         insns.  Use the new function.  Move call of check_and_process_move
14040         after operand equiv substitution and address process.
14042 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14044         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14045         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14046         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14048 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14050         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14052 2013-04-24  Marek Polacek  <polacek@redhat.com>
14054         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14055         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14056         (select_loops_exit_conditions): Likewise.
14057         (number_of_iterations_for_all_loops): Likewise.
14058         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14059         (scev_analysis): Likewise.
14061 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14062             Chao-ying Fu  <fu@mips.com>
14064         * config/mips/micromips.md (jraddiusp): New pattern.
14065         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14066         instruction if possible.
14068 2013-04-24  Alan Modra  <amodra@gmail.com>
14070         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14072 2013-04-24  Julian Brown  <julian@codesourcery.com>
14073             Chung-Lin Tang  <cltang@codesourcery.com>
14075         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14076         dependency behavior in enumeration type DIE generation. Add TODO note
14077         to comments about future DW_FORM_sdata/udata re-work of related code.
14079 2013-04-23  Lawrence Crowl  <crowl@google.com>
14081         * Makefile.in: Update as needed below.
14083         * hash-table.h (class hash_table):
14084         Correct many methods with parameter types compare_type to the correct
14085         value_type.  (Correct code was unlikely to notice the change.)
14086         (hash_table::elements_with_deleted) New.
14087         (class hashtable::iterator): New.
14088         (hashtable::begin()): New.
14089         (hashtable::end()): New.
14090         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14092         * statistics.c (statistics_hashes):
14093         Change type to hash_table.  Update dependent calls and types.
14095         * tree-into-ssa.c (var_infos):
14096         Change type to hash_table.  Update dependent calls and types.
14098         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14099         Change type to hash_table.  Update dependent calls and types.
14101         * tree-ssa-loop-im.c (struct mem_ref.refs):
14102         Change type to hash_table.  Update dependent calls and types.
14104         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14105         Change type to hash_table.  Update dependent calls and types.
14107         * tree-ssa-sccvn.c (vn_tables_s::nary):
14108         Change type to hash_table.  Update dependent calls and types.
14109         (vn_tables_s::phis): Likewise.
14110         (vn_tables_s::references): Likewise.
14112         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14113         (vn_reference_eq): Update parameter and return types.
14115         * tree-ssa-structalias.c (pointer_equiv_class_table):
14116         Change type to hash_table.  Update dependent calls and types.
14117         (location_equiv_class_table): Likewise.
14119         * tree-vect-data-refs.c: Consequential changes for making
14120         peeling a hash_table.
14122         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14123         (destroy_loop_vec_info): Dependent hash_table update.
14125         * tree-vectorizer.h (peeling_htab):
14126         Change type to hash_table.  Update dependent calls and types.
14128 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14130         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14131         to check the register content is equal or not.
14132         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14133         to assign register content record.
14134         * lra-eliminations.c (update_reg_eliminate): Use
14135         lra_update_reg_val_offset to update register content offset.
14136         * lra-int.h (struct lra_reg): Add offset member.
14137         (lra_reg_val_equal_p): New static inline function.
14138         (lra_update_reg_val_offset): New static inline function.
14139         (lra_assign_reg_val): New static inline function.
14140         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14141         to assign register content record.
14142         (initialize_lra_reg_info_element): Initial offset to zero.
14144 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14146         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14147         operands.  Record compression.
14149 2013-04-23  Xinliang David Li  <davidxl@google.com>
14151         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14153 2013-04-23  Richard Biener  <rguenther@suse.de>
14155         PR middle-end/57036
14156         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14157         parameter, only add abnormal goto edges from the copied body
14158         if the call could perform abnormal gotos.
14159         (copy_cfg_body): Adjust.
14161 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14163         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14165 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14167         * coretypes.h (gimple_stmt_iterator): Add struct to make
14168         compatible with C.
14170 2013-04-23  Richard Biener  <rguenther@suse.de>
14172         PR tree-optimization/57026
14173         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14174         from SSA names occuring in abnormal PHI nodes.
14176 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14178         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14180 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14182         * lto/lto.c (print_lto_report_1): Declare early.
14183         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14185 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14187         * common.opt (-flto-report-wpa): Add.
14188         * doc/invoke.texi (-flto-report-wpa): Add.
14189         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14190         (lto_main): dito.
14192 2013-04-22  Xinliang David Li  <davidxl@google.com>
14194         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14195         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14196         * Makefile.in: New dependency
14198             David Daney <ddaney.cavm@gmail.com>
14200         * configure.ac (gcc_cv_as_micromips_support):  Use the
14201         --fatal-warnings option.
14202         * configure: Regenerate.
14204 2013-04-22  Marek Polacek  <polacek@redhat.com>
14206         PR sanitizer/56990
14207         * tsan.c (instrument_expr): Don't instrument expression
14208         in case its size is zero.
14210 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14212         PR target/57032
14213         Revert:
14214         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14216         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14218 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14220         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14221         (gimple_stmt_iterator): New typedef.
14222         * gimple.h (gimple_stmt_iterator): Rename to...
14223         (gimple_stmt_iterator_d): ... This.
14224         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14225         trees be valid for GIMPLE and GENERIC.
14226         (TARGET_GIMPLE_FOLD_BUILTIN): New.
14227         * gimple-fold.c (gimple_fold_call): Call target hook
14228         gimple_fold_builtin.
14229         * hooks.c (hook_bool_gsiptr_false): New.
14230         * hooks.h (hook_bool_gsiptr_false): New.
14231         * target.def (fold_stmt): New.
14232         * doc/tm.texi: Regenerate.
14234 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
14236         PR target/57018
14237         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14238         a set sp if no stack realignment.
14240 2013-04-22  Nick Clifton  <nickc@redhat.com>
14242         * config.gcc (tilegx-linux): Extend extra_objs rather than
14243         overwriting it.
14244         (tilepro-linux): Likewise.
14246 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14248         * config/aarch64/aarch64-builtins.c
14249         (CF): Remove.
14250         (CF0, CF1, CF2, CF3, CF4, CF10): New.
14251         (VAR<1-12>): Add MAP parameter.
14252         (BUILTIN_*): Likewise.
14253         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14254         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14255         (aarch64_ushl_n<mode>): Likewise.
14256         (aarch64_sshr_n<mode>): Likewise.
14257         (aarch64_ushr_n<mode>): Likewise.
14258         (aarch64_<maxmin><mode>): Likewise.
14259         (aarch64_sqrt<mode>): Likewise.
14260         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14261         (vshr<q>_n_*): Likewise.
14263 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14265         * config/aarch64/aarch64-builtins.c
14266         (aarch64_simd_builtin_type_mode): Handle SF types.
14267         (sf_UP): Define.
14268         (BUILTIN_GPF): Define.
14269         (aarch64_init_simd_builtins): Handle SF types.
14270         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14271         (frecps): Likewise.
14272         (frecpx): Likewise.
14273         * config/aarch64/aarch64-simd.md
14274         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14275         (aarch64_frecpe<mode>): New.
14276         (aarch64_frecps<mode>): Likewise.
14277         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14278         (v8type): Add frecp<esx>.
14279         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14280         (aarch64_frecps<mode>): Likewise.
14281         * config/aarch64/iterators.md (FRECP): New.
14282         (frecp_suffix): Likewise.
14283         * config/aarch64/arm_neon.h
14284         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14286 2013-04-22  Christian Bruel  <christian.bruel@st.com>
14288         PR target/56995
14289         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14290         (REG_CLASS_NAMES): Idem.
14291         (REG_CLASS_CONTENTS): Idem.
14292         (REGCLASS_HAS_FP_REG): Idem.
14293         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14294         (sh_conditional_register_usage): Idem.
14296 2013-04-21  Jeff Law  <law@redhat.com>
14298         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14299         (ssa_forward_propagate_and_combine): Use it.
14301 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14303         * lra.c: Update the flow chart diagram.
14305 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14307         PR rtl-optimization/56847
14308         * lra-constraints.c (process_alt_operands): Discourage alternative
14309         with non-matche doffsettable memory constraint fro memory with
14310         known offset.
14312 2013-04-19  Richard Biener  <rguenther@suse.de>
14314         PR tree-optimization/56982
14315         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14316         function.
14317         * gimplify.c (gimplify_call_expr): Notice special calls.
14318         (gimplify_modify_expr): Likewise.
14319         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14320         abnormal control flow receivers.
14321         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14322         in the same way as cfun->has_nonlocal_labels.
14323         (gimple_purge_dead_abnormal_call_edges): Likewise.
14324         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14325         receivers start a basic-block.
14327 2013-04-19  Richard Biener  <rguenther@suse.de>
14329         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14330         member ...
14331         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
14332         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14333         (SLP_TREE_LOAD_PERMUTATION): Add.
14334         (vect_transform_slp_perm_load): Adjust prototype.
14335         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14336         (vect_free_slp_instance): Likewise.
14337         (vect_create_new_slp_node): Likewise.
14338         (vect_supported_slp_permutation_p): Remove.
14339         (vect_slp_rearrange_stmts): Adjust.
14340         (vect_supported_load_permutation_p): Likewise.  Inline
14341         vect_supported_slp_permutation_p here.
14342         (vect_analyze_slp_instance): Compute load permutations per
14343         slp node instead of per instance.
14344         (vect_get_slp_defs): Adjust.
14345         (vect_transform_slp_perm_load): Likewise.
14346         (vect_schedule_slp_instance): Remove redundant code.
14347         (vect_schedule_slp): Remove hack for PR56270, add it ...
14348         * tree-vect-stmts.c (vectorizable_load): ... here, do not
14349         CSE loads for SLP.  Adjust.
14351 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14353         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14354         spelling in two comments.
14356 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14358         PR target/56797
14359         * config/arm/arm.c (load_multiple_sequence): Require SP
14360         as base register for loads if SP is in the register list.
14362 2013-04-19  Martin Jambor  <mjambor@suse.cz>
14364         PR tree-optimization/56718
14365         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14366         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14367         and made public.  Adjusted all callers.
14368         (ipa_intraprocedural_devirtualization): New function.
14369         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14370         (ipa_intraprocedural_devirtualization): Likewise.
14371         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14373 2013-04-19  Richard Biener  <rguenther@suse.de>
14375         PR tree-optimization/57000
14376         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14378 2013-04-19  Terry Guo  <terry.guo@arm.com>
14380         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14381         Replace with ...
14382         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
14383         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14384         (cortex_m4_fmacs): Use new reservations.
14385         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14387 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
14389         PR rtl-optimization/56999
14390         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14391         related code.
14392         (lra_coalesce): Remove split_origin_bitmap and related code.
14393         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
14394         ranges if necessary.
14396 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
14398         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14399         New array.
14400         (ix86_expand_call): Remove clobbered_registers array and use
14401         x86_64_ms_sysv_extra_clobbered_registers instead.
14402         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14403         Declare here.
14404         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14405         predicate.
14406         * config/i386/i386.md (*call_rex64_ms_sysv): Use
14407         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
14408         (*call_value_rex64_ms_sysv): Ditto.
14410 2013-04-18  Cary Coutant  <ccoutant@google.com>
14412         * dwarf2out.c (output_pubnames): Check die_perennial_p of
14413         parent instead of die_mark.
14415 2013-04-18  Diego Novillo  <dnovillo@google.com>
14417         * gimple.c (create_gimple_tmp): New.
14418         (get_expr_type): New.
14419         (build_assign): New.
14420         (build_type_cast): New.
14421         * gimple.h (enum ssa_mode): Define.
14422         (gimple_seq_set_location): New.
14423         * asan.c (build_check_stmt): Change some gimple_build_* calls
14424         to use build_assign and build_type_cast.
14426 2013-04-18  Richard Biener  <rguenther@suse.de>
14428         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14429         handle negative step.  Remove redundant checks.
14430         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14431         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14432         for negative step and grouped loads fail to vectorize.
14434 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14436         * emit-rtl.c (reset_insn_used_flags): New function.
14437         (reset_all_used_flags): Use it.
14438         (verify_insn_sharing): New function.
14439         (verify_rtl_sharing): Fix verification for SEQUENCEs.
14441 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
14443         PR tree-optimization/56984
14444         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14445         and (x >> M) >= N don't register any assertion if N << M is the
14446         minimum value.
14448 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14450         * lower-subreg.c (resolve_simple_move): If called self-recursive,
14451         do not delete_insn insns that have not yet been emitted, only
14452         unlink them with remove_insn.
14453         * df-scan.c (df_insn_delete): Revert r197492.
14455 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
14457         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14458         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14460 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14462         * config/arm/arm.md (movsicc_insn): Convert define_insn into
14463         define_insn_and_split.
14464         (and_scc,ior_scc,negscc): Likewise.
14465         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14467 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14469         * config/arm/arm.c (use_return_insn): Return 0 for targets that
14470         can benefit from using a sequence of LDRD instructions in epilogue
14471         instead of a single LDM instruction.
14473 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14475         PR 45688
14476         * doc/extend.texi: Fix typo.
14478 2013-04-17  Richard Biener  <rguenther@suse.de>
14480         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14481         (vect_build_slp_tree): ... here.
14482         (vect_build_slp_tree_1): Compute which stmts of the SLP group
14483         match.  Remove special-casing of mismatched complex loads.
14484         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14485         re-try the match with swapped commutative operands.
14486         (vect_supported_load_permutation_p): Remove special-casing of
14487         mismatched complex loads.
14488         (vect_analyze_slp_instance): Adjust.
14490 2013-04-17  Richard Biener  <rguenther@suse.de>
14492         PR rtl-optimization/56921
14493         * cfgloop.h (struct loop): Add simple_loop_desc member.
14494         (struct niter_desc): Mark with GTY(()).
14495         (simple_loop_desc): Do not use aux field but simple_loop_desc.
14496         * loop-iv.c (get_simple_loop_desc): Likewise.
14497         (free_simple_loop_desc): Likewise.
14499         Revert
14500         2013-04-16  Richard Biener  <rguenther@suse.de>
14502         PR rtl-optimization/56921
14503         * loop-init.c (pass_rtl_move_loop_invariants): Add
14504         TODO_do_not_ggc_collect to todo_flags_finish.
14505         (pass_rtl_unswitch): Same.
14506         (pass_rtl_unroll_and_peel_loops): Same.
14507         (pass_rtl_doloop): Same.
14509 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
14511         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14512         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14513         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14514         references.
14515         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14516         * tree-streamer.c (record_common_node): Adjust reference in comment.
14518 2013-04-17  Terry Guo  <terry.guo@arm.com>
14520         * config/arm/cortex-m4.md: Add a new bypass.
14522 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14524         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14525         New pattern.
14526         (*subs_<optab><mode>_multp2): New pattern.
14527         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14528         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14530 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14532         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14533         (*subs_mul_imm_<mode>): New pattern.
14535 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
14537         PR target/56948
14538         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14539         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
14540         (vsx_movti_32bit): Change j->wa to O->wa.
14542 2013-04-16  Richard Biener  <rguenther@suse.de>
14544         PR rtl-optimization/56921
14545         * loop-init.c (pass_rtl_move_loop_invariants): Add
14546         TODO_do_not_ggc_collect to todo_flags_finish.
14547         (pass_rtl_unswitch): Same.
14548         (pass_rtl_unroll_and_peel_loops): Same.
14549         (pass_rtl_doloop): Same.
14551 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
14553         * config/arm/arm.c (emit_multi_reg_push): New declaration
14554         for an existing function.
14555         (arm_emit_strd_push): New function.
14556         (arm_expand_prologue): Used here.
14557         (arm_emit_ldrd_pop): New function.
14558         (arm_expand_epilogue): Used here.
14559         (arm_get_frame_offsets): Update condition.
14560         (arm_emit_multi_reg_pop): Add a special case for load of a single
14561         register with writeback.
14563 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
14565         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
14566         description.
14568 2013-04-16  Richard Biener  <rguenther@suse.de>
14570         PR tree-optimization/56756
14571         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
14572         (first_mem_ref_loc): New.
14573         (execute_sm): Place the load temporarily before a previous
14574         access instead of in the latch edge to ensure its SSA dependencies
14575         are defined at points dominating the load.
14577 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
14579         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
14580         correct fix by moving header and footer insn to the footer of
14581         the merged basic block.  Clear BB_END of the merged-away block.
14583         PR middle-end/43631
14584         * emit-rtl.c (make_note_raw): New function.
14585         (link_insn_into_chain): New static inline function.
14586         (add_insn): Use it.
14587         (add_insn_before, add_insn_after): Factor insn chain linking code...
14588         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
14589         using link_insn_into_chain.
14590         (note_outside_basic_block_p): New helper function for emit_note_after
14591         and emit_note_before.
14592         (emit_note_after): Use nobb variant of add_insn_after if the note
14593         should not be contained in a basic block.
14594         (emit_note_before): Use nobb variant of add_insn_before if the note
14595         should not be contained in a basic block.
14596         (emit_note_copy): Use make_note_raw.
14597         (emit_note): Likewise.
14598         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
14599         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
14600         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
14601         the moved barrier the tail of the basic block it follows.
14602         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
14604 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
14606         PR tree-optimization/56962
14607         * gimple-ssa-strength-reduction.c (record_increment): Only set
14608         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
14609         either rhs1 or rhs2 is equal to c->base_expr.
14611 2013-04-15  Richard Biener  <rguenther@suse.de>
14613         PR tree-optimization/56933
14614         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
14615         member.
14616         (GROUP_READ_WRITE_DEPENDENCE): Remove.
14617         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
14618         * tree-vect-data-refs.c (vect_analyze_group_access): Move
14619         dependence check ...
14620         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
14621         ... here.
14622         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
14623         GROUP_READ_WRITE_DEPENDENCE.
14625 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14627         * emit-rtl.c (reset_all_used_flags): New function.
14628         (verify_rtl_sharing): Call reset_all_used_flags before and after
14629         performing the checks.
14631 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14633         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
14634         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
14635         * config/arm/constraints.md (De): New constraint.
14636         * config/arm/neon.md (anddi3_neon): Delete.
14637         (neon_vand<mode>): Expand to standard anddi3 pattern.
14638         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
14639         Move earlier in the file.
14640         (neon_inv_logic_op2): Likewise.
14641         (arm_anddi_operand_neon): New predicate.
14643 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14645         * configure.ac (gcc_cv_ld_as_needed): Set
14646         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
14647         Use -z ignore, -z record on *-*-solaris2*.
14648         (HAVE_LD_AS_NEEDED): Update comment.
14649         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
14650         * configure: Regenerate.
14651         * config.in: Regenerate.
14652         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
14653         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
14654         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
14655         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
14656         equivalents.  Fix markup.
14657         * doc/tm.texi: Regenerate.
14659 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
14661         * config/i386/i386.opt: New option mstack-protector-guard=.
14662         * config/i386/i386-opts.h: Add enum stack_protector_guard.
14663         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
14664         TARGET_SSP_TLS_GUARD.
14665         * config/i386/i386.c (ix86_option_override_internal): Set
14666         ix86_stack_protector_guard.
14667         * config/i386/i386.md (stack_protect_set): Enable for
14668         TARGET_SSP_TLS_GUARD only.
14669         (stack_protect_set_<mode>): Ditto.
14670         (stack_protect_test): Ditto.
14671         (stack_protect_test_<mode>): Ditto.
14672         * doc/invoke.texi (i386 Option): Document.
14674 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
14676         PR target/56890
14677         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
14678         (S_MODES): Set H_MODE bit.
14679         (SF_MODES): Set only S_MODE and SF_MODE bits.
14680         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
14681         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
14682         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
14683         <MODE_FLOAT>: Likewise.
14685 2013-04-15  Joey Ye  <joey.ye@arm.com>
14687         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
14689 2013-04-15  Joey Ye  <joey.ye@arm.com>
14691         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
14692         for real far jump.
14693         (thumb_far_jump_used_p): Count instruction size and set
14694         far_jump_used.
14696 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
14698         * reorg.c (fill_simple_delay_slots): Reindent block of code.
14699         * resource.c (mark_target_live_regs): Reformat conditional block.
14701 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14703         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
14704         notes, they are emitted only just before final.
14705         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
14707 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14709         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
14710         * cfgrtl.c (delete_insn): Call it here instead.
14711         * lra-spills.c (lra_final_code_change): Use delete_insn.
14712         * haifa-sched.c (sched_remove_insn): Likewise.
14713         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
14714         returning to the nop pool.
14715         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
14716         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
14718 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14720         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
14721         * doc/tm.texi: Regenerated.
14723 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
14725         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
14726         QImode checks.
14728 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14730         * df-core.c (df_find_def): Compare register numbers.
14731         (df_find_use): Likewise.
14733 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
14735         PR target/56903
14736         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
14737         lra_in_progress for return.
14739 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14741         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
14742         define_insn into define_insn_and_split and emit movsicc patterns.
14744 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14746         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
14748 2013-04-12  Richard Biener  <rguenther@suse.de>
14750         * tree-pass.h (TODO_do_not_ggc_collect): New.
14751         * passes.c (execute_one_ipa_transform_pass): Honor
14752         TODO_do_not_ggc_collect.
14753         (execute_one_pass): Likewise.
14755         Revert
14756         2013-04-10  Richard Biener  <rguenther@suse.de>
14758         * passes.c (init_optimization_passes): Remove reload pass.
14759         * ira.c (do_reload): Merge into ...
14760         (ira): ... this.
14761         (rest_of_handle_reload): Remove.
14762         (pass_reload): Likewise.
14763         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14764         of reload for vzeroupper pass placement.
14766 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
14768         PR tree-optimization/56918
14769         PR tree-optimization/56920
14770         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
14771         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
14772         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
14773         use wide_mul_with_sign method.
14775 2013-04-12  Richard Biener  <rguenther@suse.de>
14777         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
14778         not be considered a gimple constant.
14780 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
14782         * fold-const.c (const_binop): Handle vector shifts by a scalar.
14783         (fold_binary_loc): Call const_binop also for mixed vector-scalar
14784         operations.
14786 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14787             Jakub Jelinek  <jakub@redhat.com>
14789         * opts.c: Include diagnostic-color.h.
14790         (common_handle_option): Handle OPT_fdiagnostics_color_.
14791         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
14792         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
14793         (diagnostic-color.o): New.
14794         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
14795         (diagnostic_color_rule): New enum.
14796         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
14797         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
14798         the location string.
14799         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
14800         either NULL, or color kind.
14801         * diagnostic-color.c: New file.
14802         * diagnostic-color.h: New file.
14803         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
14804         arguments.
14805         * doc/invoke.texi (-fdiagnostics-color): Document.
14806         * pretty-print.h (pp_show_color): Define.
14807         (struct pretty_print_info): Add show_color field.
14808         * diagnostic.c: Include diagnostic-color.h.
14809         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
14810         macros.  Colorize error:, warning: etc. strings and also the location
14811         string.
14812         (diagnostic_show_locus): Colorize the caret line.
14813         * pretty-print.c: Include diagnostic-color.h.
14814         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
14815         inside of %< %> quotes or quoted through q format modifier.
14817 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14819         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
14821 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14823         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
14824         code in CC_NZ mode.
14825         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
14826         pattern.
14828 2013-04-11  Marek Polacek  <polacek@redhat.com>
14830         PR tree-optimization/48184
14831         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
14833 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
14835         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
14836         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
14837         (skip_simple_arithmetic): Tidy up.
14838         * tree.h (skip_simple_constant_arithmetic): Declare.
14840 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14842         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
14844 2013-04-11  Richard Biener  <rguenther@suse.de>
14846         * tree-vect-loop.c (get_initial_def_for_induction): Properly
14847         generate vector constants.
14849 2013-04-11  Richard Biener  <rguenther@suse.de>
14851         PR tree-optimization/56878
14852         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
14853         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
14854         New function.
14855         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14856         Prefer to align the DR with the most invariant base address.
14858 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14860         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
14861         comment.
14863 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
14865         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
14866         floating-point vector comparisons against 0.
14868 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
14870         PR tree-optimization/56899
14871         * fold-const.c (extract_muldiv_1): Apply distributive law
14872         only if TYPE_OVERFLOW_WRAPS (ctype).
14874 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
14876         PR target/56124
14877         * ira-costs.c (scan_one_insn): Check whether the source rtx of
14878         loading has side effect.
14880 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
14882         * config/sparc/sparc.c: Include tree-pass.h.
14883         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
14884         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
14885         head of file.  Change return type.  Split off gate function.
14886         (sparc_gate_work_around_errata): New function.
14887         (pass_work_around_errata): New pass definition.
14888         (insert_pass_work_around_errata) New pass insert definition to
14889         insert pass_work_around_errata just after delayed-branch scheduling.
14890         (sparc_option_override): Insert the pass.
14891         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
14893 2013-04-10  David S. Miller  <davem@davemloft.net>
14895         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
14896         or -mcpu=hypersparc.
14898         * target.def (cstore_mode): New hook.
14899         * target.h: Include insn-codes.h
14900         * targhooks.c: Likewise.
14901         (default_cstore_mode): New function.
14902         * targhooks.h: Declare it.
14903         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
14904         * doc/tm.texi: Rebuild.
14905         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
14906         target hook, rather than inspecting the insn_data.
14907         * config/sparc/sparc.c (sparc_cstore_mode): New function.
14908         (TARGET_CSTORE_MODE): Redefine.
14909         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
14910         result patterns.
14911         * config/sparc/predicates.md (cstore_result_operand): New special
14912         predicate.
14913         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
14914         Use it for operand 0.
14915         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
14916         (*snesi_special): Likewise.
14917         (*snesi_zero): Likewise.
14918         (*seqsi_zero): Likewise.
14919         (*sltu_insn): Likewise.
14920         (*sgeu_insn): Likewise.
14921         (*seqdi_special): Make operand 0 and comparison operation be of
14922         DImode.
14923         (*snedi_special): Likewise.
14924         (*snedi_special_vis3): Likewise.
14925         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
14926         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
14927         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
14928         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
14929         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
14930         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
14931         (*sltu_extend_sp64): Likewise.
14932         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
14933         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
14934         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
14935         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
14936         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
14938 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
14940         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
14941         (aarch64_start_file): Use the new function.
14943 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14944             Jason Merrill  <jason@redhat.com>
14946         * common.opt: Add -gdwarf.
14947         * opts.c (common_handle_option): Handle it.
14948         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
14950 2013-04-10  Richard Biener  <rguenther@suse.de>
14952         * passes.c (execute_todo): Do not call ggc_collect conditional here.
14953         (execute_one_ipa_transform_pass): But unconditionally here.
14954         (execute_one_pass): And here.
14955         (init_optimization_passes): Remove reload pass.
14956         * tree-pass.h (TODO_ggc_collect): Remove.
14957         (pass_reload): Likewise.
14958         * ira.c (do_reload): Merge into ...
14959         (ira): ... this.
14960         (rest_of_handle_reload): Remove.
14961         (pass_reload): Likewise.
14962         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14963         of reload for vzeroupper pass placement.
14964         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
14965         and todo_flags_finish of all passes.
14967 2013-04-10  Richard Biener  <rguenther@suse.de>
14969         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
14970         first_const_oprnd field, rename first_def_type to first_op_type.
14971         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
14972         (vect_get_and_check_slp_defs): Always use the type of the
14973         operand.  Allow mixed vect_external_def, vect_constant_def types.
14974         (vect_get_constant_vectors): Handle mixed vect_external_def,
14975         vect_constant_def types.
14977 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
14979         PR tree-optimization/55524
14980         * tree-ssa-math-opts.c
14981         (convert_mult_to_fma): Don't use an fms construct
14982         when we don't have an fms operation, but fnma, and it looks
14983         likely that we'll be able to use the latter.
14985 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
14987         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
14988         function.
14989         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
14990         inline fail caused by overwritable functions.
14992 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
14994         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
14995         unnecessary bits in the constant power of two case.
14997 2013-04-10  Richard Biener  <rguenther@suse.de>
14999         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15000         broken code swapping operands.
15001         (vect_build_slp_tree): Do not compute load permutations here.
15002         (vect_analyze_slp_instance): Compute load permutations here,
15003         after building the SLP tree.
15005 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15007         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15008         of next/prev_real_insn.
15010 2013-04-09  Jan Hubicka  <jh@suse.cz>
15012         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15013         Drop aliased parameter.
15014         (function_and_variable_visibility): Do not handle alias pairs.
15015         * cgraph.c (varpool_externally_visible_p): Update prototype.
15016         * varpool.c (varpool_add_new_variable): Update.
15018 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15020         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15022 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15024         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15026         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15028 2013-04-09  Marek Polacek  <polacek@redhat.com>
15030         PR tree-optimization/48762
15031         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15033 2013-04-09  Richard Biener  <rguenther@suse.de>
15035         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15036         dealing with cost.
15037         (vect_build_slp_tree): Likewise.
15038         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15039         calculating the cost of a SLP instance.
15040         (vect_analyze_slp_instance): Use it from here, after building
15041         the SLP tree.
15043 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15045         PR middle-end/56883
15046         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15047         expand_omp_for_static_chunk): Use simple_p = true in
15048         force_gimple_operand_gsi calls when assigning to addressable decls.
15050 2013-04-09  Jeff Law  <law@redhat.com>
15052         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15053         when the boolean was created by converting a wider object which
15054         had a boolean range.
15056 2013-04-09  Richard Biener  <rguenther@suse.de>
15058         * tree-vectorizer.h (slp_void_p): Remove.
15059         (slp_tree): Typedef before _slp_tree declaration.
15060         (struct _slp_tree): Use a vector of slp_tree as children.
15061         (vect_get_place_in_interleaving_chain): Remove.
15062         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15063         Move ...
15064         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15065         and make static.
15066         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15067         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15068         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15069         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15070         Use slp_node instead of slp_void_p and adjust.
15072 2013-04-09  Richard Biener  <rguenther@suse.de>
15074         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15075         work that is not necessary.
15077 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15079         PR tree-optimization/56854
15080         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15081         forward into clobber stmts if it would change MEM_REF lhs into
15082         non-MEM_REF.
15084 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15086         * tree.c (type_hash_lookup, type_hash_add): Make static.
15087         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15089 2013-04-09  Richard Biener  <rguenther@suse.de>
15091         * tree.h (unsave_expr_now): Remove.
15092         * tree-inline.c (mark_local_for_remap_r): Remove.
15093         (unsave_expr_1): Likewise.
15094         (unsave_r): Likewise.
15095         (unsave_expr_now): Likewise.
15096         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15097         (propagate_tree_value): Likewise.
15099 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15101         * doc/rtl.texi (sequence): Rewrite documentation to match the
15102         current use of SEQUENCE rtl objects.
15103         * rtl.def (SEQUENCE): Likewise.
15105         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15106         Update documentation.
15107         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15108         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15110         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15112 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15114         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15115         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15116         methods.
15117         (estimate_edge_size_and_time): Add comment to suggest using rounding
15118         methods.
15119         (estimate_node_size_and_time): Ditto.
15120         (remap_edge_change_prob): Use helper rounding divide methods.
15121         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15122         (gimple_mod_pow2_value_transform): Ditto.
15123         (gimple_mod_subtract_transform): Ditto.
15124         (gimple_ic_transform): Ditto.
15125         (gimple_stringops_transform): Ditto.
15126         * stmt.c (conditional_probability): Ditto.
15127         (emit_case_dispatch_table): Ditto.
15128         * lto-cgraph.c (merge_profile_summaries): Ditto.
15129         * tree-optimize.c (execute_fixup_cfg): Ditto.
15130         * cfgcleanup.c (try_forward_edges): Ditto.
15131         * cfgloopmanip.c (scale_loop_profile): Ditto.
15132         (loopify): Ditto.
15133         (duplicate_loop_to_header_edge): Ditto.
15134         (lv_adjust_loop_entry_edge): Ditto.
15135         * tree-vect-loop.c (vect_transform_loop): Ditto.
15136         * profile.c (compute_branch_probabilities): Ditto.
15137         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15138         * lto-streamer-in.c (input_cfg): Ditto.
15139         * gimple-streamer-in.c (input_bb): Ditto.
15140         * ipa-cp.c (update_profiling_info): Ditto.
15141         (update_specialized_profile): Ditto.
15142         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15143         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15144         rounding methods.
15145         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15146         (compute_trg_info): Ditto.
15147         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15148         (purge_dead_edges): Ditto.
15149         * loop-unswitch.c (unswitch_loop): Ditto.
15150         * cgraphclones.c (cgraph_clone_edge): Ditto.
15151         (cgraph_clone_node): Ditto.
15152         * tree-inline.c (copy_bb): Ditto.
15153         (copy_edges_for_bb): Ditto.
15154         (initialize_cfun): Ditto.
15155         (copy_cfg_body): Ditto.
15156         (expand_call_inline): Ditto.
15158 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15160         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15161         TARGET_CYGWIN64 by TARGET_64BIT.
15163 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15165         * config/epiphany/epiphany.md (GPR_1): New constant.
15166         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15167         * config/epiphany/epiphany.c (gen_compare_reg):
15168         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15169         is already in place.
15170         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15171         Don't require being called during rtl expansion; If y operlaps r0,
15172         return 0.
15173         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15174         (epiphany_expand_epilogue): Likewise.
15176         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15177         Don't use CC_FPmode for ORDERED / UNORDERED.
15178         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15180         * config/epiphany/constraints.md (CnL): New constraint.
15181         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15182         * config/epiphany/predicates.md (add_operand): Allow 1024.
15184         * config/epiphany/epiphany.md (logical_op): New code iterator.
15185         (op_mnc): New code attribute.
15186         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15187         (mov_f+1, mov_f+2): New peephole2 patterns.
15189         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15190         (cstoresi4): Also allow re-use of zero result when doing a NE
15191         comparison to a non-zero operand.
15192         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15194         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15195         Use gen_addsi3_i / gen_subsi3_i.
15197 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15199         PR c++/34949
15200         PR c++/50243
15201         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15202         contain anything but clobbers, at most one __builtin_stack_restore,
15203         optionally debug stmts and final resx, and if it has at least one
15204         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15205         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15206         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15207         which isn't defaut definition, remove them.
15208         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15209         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15210         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15211         with MEM_REF LHS with SSA_NAME address.
15213 2013-04-08  Jeff Law  <law@redhat.com>
15215         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15217 2013-04-08  Richard Biener  <rguenther@suse.de>
15219         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15220         extra newline.
15221         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15222         determined vector type.
15223         (vect_analyze_data_refs): Likewise.
15224         (vect_get_new_vect_var): Adjust.
15225         (vect_create_destination_var): Preserve SSA name versions.
15226         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15227         not dump anything here.
15229 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15231         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15232         Add member lr_slot_known.
15233         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15234         if necessary.
15235         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15236         Remove code that sets lr_slot_offset according to what a previous
15237         version of epiphany_emit_save_restore used to do.
15238         (epiphany_emit_save_restore): When doing an lr save or restore,
15239         set/verify lr_slot_known and lr_slot_offset.
15241 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
15243         PR target/54338
15244         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15245         in ALL_REGS.
15247 2013-04-08  Richard Biener  <rguenther@suse.de>
15249         * alias.c (find_base_term): Fix thinko in previous change.
15251 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15253         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15254         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15255         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
15256         if possible to compute val.
15257         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15258         For QImode integers don't require anything about precision.  Use
15259         const_with_all_bytes_same to find out if the constant doesn't have
15260         repeated bytes in it.
15262 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15264         * config/s390/s390.c (s390_expand_insv): Only accept insertions
15265         within mode size.
15267 2013-04-08  Marek Polacek  <polacek@redhat.com>
15269         PR rtl-optimization/48182
15270         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15271         value to 1.
15273 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15275         PR target/55487
15276         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15277         nuses, make sure we have a label.
15279 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15281         PR target/56843
15282         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15283         (rs6000_emit_swdiv_low_precision): Remove.
15284         (rs6000_emit_swdiv): Rewrite to handle between one and four
15285         iterations of Newton-Raphson generally; modify required number of
15286         iterations for some cases.
15287         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15289 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
15291         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15292         set-but-unused variable.
15294         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15295         basic blocks of released function bodies garbage-collectable.
15297         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15298         (struct rtl_opt_pass): Add TODO_df_finish.
15300         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15302 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15304         * config/arm/constraints.md (q): New constraint.
15305         * config/arm/ldrdstrd.md: New file.
15306         * config/arm/arm.md (ldrdstrd.md) New include.
15307         (arm_movdi): Use "q" instead of "r" constraint
15308         for double-word memory access.
15309         (movdf_soft_insn): Likewise.
15310         * config/arm/vfp.md (movdi_vfp): Likewise.
15311         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15312         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15313         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15314         (mem_ok_for_ldrd_strd): Likewise.
15315         (output_move_double): Update assertion.
15317 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15319         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15321 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15323         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15324         define_insn_and_split.
15325         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15327 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15329         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15330         define_insn_and_split.
15331         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15332         (shiftsi3_compare): New pattern.
15333         (rrx): New pattern.
15334         * config/arm/unspecs.md (UNSPEC_RRX): New.
15336 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15338         * config/arm/arm.md (negdi_extendsidi): New pattern.
15339         (negdi_zero_extendsidi): Likewise.
15341 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15343         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15344         define_insn_and_split.
15345         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15346         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15348 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15350         * config/arm/arm.md (arm_subdi3): Convert define_insn into
15351         define_insn_and_split.
15352         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15353         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15355 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15357         * config/arm/arm.md (subsi3_carryin): New pattern.
15358         (subsi3_carryin_const): Likewise.
15359         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15360         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15362 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15364         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15366 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15368         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15369         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15371 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15373         * config/arm/arm.c (arm_expand_builtin): Change fcode
15374         type to unsigned int.
15376 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15378         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15380 2013-04-04  Ian Lance Taylor  <iant@google.com>
15382         * doc/standards.texi (Standards): The Go frontend supports the Go 1
15383         language standard.
15385 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
15387         PR middle-end/56729
15388         * df-scan.c (df_insn_delete): Disable failing assert.
15390 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15392         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15393         New function prototype.
15394         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15395         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15396         (arm_builtin_vectorized_function): New function.
15398 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15400         * config/arm/arm_neon_builtins.def: New file.
15401         * config/arm/arm.c (neon_builtin_data): Move contents to
15402         arm_neon_builtins.def.
15403         (enum arm_builtins): Include neon builtin definitions.
15404         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15405         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15407 2013-04-04  Marek Polacek  <polacek@redhat.com>
15409         PR tree-optimization/48186
15410         * predict.c (maybe_hot_frequency_p): Return false if
15411         HOT_BB_FREQUENCY_FRACTION is 0.
15412         (cgraph_maybe_hot_edge_p): Likewise.
15414 2013-04-04  Richard Biener  <rguenther@suse.de>
15416         PR tree-optimization/56826
15417         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15418         more accurately.
15420 2013-04-04  Richard Biener  <rguenther@suse.de>
15422         PR tree-optimization/56213
15423         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15424         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15426 2013-04-04  Richard Biener  <rguenther@suse.de>
15428         PR tree-optimization/56837
15429         * tree-loop-distribution.c (classify_partition): For non-zero
15430         values require that the value has the same precision as its
15431         mode to be useful as memset value.
15433 2013-04-03  Nick Clifton  <nickc@redhat.com>
15435         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15436         (fmssf4): Use fmsf.s on E3V5 architectures.
15437         (fnmasf4): Use fnmaf.s on E3V5 architectures.
15438         (fnmssf4): Use fnmsf.s on E3V5 architectures.
15440 2013-04-03  Jeff Law  <law@redhat.com>
15442         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15443         (lra-eliminations.o): Likewise.
15445 2013-04-03  Teresa Johnson  <tejohnson@google.com>
15447         * gcov-io.c (compute_working_sets): Moved most of body of old
15448         compute_working_sets here from profile.c.
15449         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15450         (gcov_working_set_t): Moved typedef here from basic-block.h
15451         (compute_working_set): Declare.
15452         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15453         (get_working_sets): Renamed from compute_working_set,
15454         replace most of body with call to new compute_working_sets.
15455         (get_exec_counts): Replace call to compute_working_sets
15456         to get_working_sets.
15457         * profile.h (get_working_sets): Renamed from compute_working_set.
15458         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15459         to get_working_sets.
15460         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15461         * gcov-dump.c (dump_working_sets): New function.
15463 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
15465         * hwint.c (sext_hwi, zext_hwi): New functions.
15466         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15467         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15468         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15469         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15470         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15471         (sext_hwi, zext_hwi): New functions.
15473 2013-04-03  Jeff Law  <law@redhat.com>
15475         PR tree-optimization/56799
15476         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15477         back test for widening conversion erroneously dropped in prior change.
15479 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15481         PR target/56809
15482         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15483         instead of next_real_insn.
15485 2013-04-03  Marek Polacek  <polacek@redhat.com>
15487         PR sanitizer/55702
15488         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15490 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15492         PR target/56809
15493         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15494         next_real_insn.
15495         (thumb1_output_casesi): Likewise.
15496         (thumb2_output_casesi): Likewise.
15498 2013-04-03  Richard Biener  <rguenther@suse.de>
15500         PR tree-optimization/56817
15501         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15502         Split out ...
15503         (tree_unroll_loops_completely_1): ... new function to manually
15504         walk the loop tree, properly defering outer loops of unrolled
15505         loops to later iterations.
15507 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15509         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15510         (vectorizable_load): Likewise.
15511         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15512         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15514 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15516         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15517         BIT_FIELD_REF.
15519 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15521         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15523 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
15525         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15527 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15529         PR tree-optimization/56790
15530         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15531         folding.
15533 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15535         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15536         Handle VEC_MERGE.
15537         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15538         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15539         equal arguments.
15541 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
15543         PR c/19449
15544         * tree.h (force_folding_builtin_constant_p): New decl.
15545         * builtins.c (force_folding_builtin_constant_p): New variable.
15546         (fold_builtin_constant_p): Fold immediately also if
15547         force_folding_builtin_constant_p.
15549 2013-04-03  Richard Biener  <rguenther@suse.de>
15551         PR tree-optimization/56812
15552         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15553         DRs of the same interleaving chain are independent.
15555 2013-04-02  Jason Merrill  <jason@redhat.com>
15557         * gdbinit.in (pbb): Use debug fn.
15559 2013-04-02  Lawrence Crowl  <crowl@google.com>
15561         * sese.h (struct ivtype_map_elt_s): Remove unused.
15562         (extern debug_ivtype_map): Remove unused.
15563         (extern eq_ivtype_map_elts): Remove unused.
15564         * sese.c (debug_ivtype_map): Removed unused.
15565         (debug_ivtype_map_1): Removed unused.
15566         (debug_ivtype_elt): Remove unused.
15567         (eq_ivtype_map_elts): Remove unused.
15570 2013-04-02  Kai Tietz  <ktietz@redhat.com>
15572         PR target/52790
15573         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
15574         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
15575         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
15576         function.
15577         (legitimize_pe_coff_symbol): Likewise.
15578         (is_imported_p): New helper-function.
15579         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
15580         for Windows x64 targets.
15581         (ix86_expand_prologue): Optimize for pe-coff targets.
15582         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
15583         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
15584         medium/large code-model.
15585         (legitimize_pic_address): Likewise.
15586         (legitimize_tls_address): Likewise.
15587         (ix86_expand_call): Likewise.
15588         (x86_output_mi_thunk): Likewise.
15589         (get_dllimport_decl): Add new beimport argument.
15590         (construct_plt_address): Don't assert for x64 pe-coff targets.
15591         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
15592         targets.
15593         (SYMBOL_FLAG_STUBVAR): New macro.
15594         (SYMBOL_REF_STUBVAR_P): Likewise.
15595         * config/i386/winnt.c (stub_list): New structure.
15596         (stub_head): New local variable.
15597         (i386_pe_record_stub): New function.
15598         (i386_pe_file_end): Emit refptr-stubs.
15600 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
15602         PR rtl-optimization/56745
15603         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
15604         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
15606         PR c++/34949
15607         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
15608         and both of them are MEM_REFs, just compare first argument for
15609         equality and attempt to deal even with differing offsets.
15611         PR c++/34949
15612         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
15613         of gimple_clobber_p to be MEM_REF.
15614         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
15615         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
15616         after gimplification.
15617         * asan.c (get_mem_ref_of_assignment): Don't instrument
15618         gimple_clobber_p stmts.
15619         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
15620         gimple_clobber_p stmt if they have MEM_REF lhs and
15621         are dead because of another gimple_clobber_p stmt.
15622         * tree-ssa-live.c (clear_unused_block_pointer): Treat
15623         gimple_clobber_p stmts like debug stmts.
15624         (remove_unused_locals): Remove clobbers with MEM_REF lhs
15625         that refer to unused VAR_DECLs or uninitialized values.
15626         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
15627         gimple_clobber_p stmts if they refer to removed parameters.
15628         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
15629         formatting.
15631 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
15633         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
15634         using SWI48 mode attribute.
15636 2013-04-02  Wei Mi  <wmi@google.com>
15638         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
15639         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
15640         *<rotate_insn><mode>3_mask in i386.md.
15642 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
15644         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
15646 2013-04-02  Richard Biener  <rguenther@suse.de>
15648         PR tree-optimization/56778
15649         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15650         Runtime alias tests are not supported for gather loads.
15651         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
15652         stmts referenced from SSA operands before updating SSA form.
15654 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
15655             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15657         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
15658         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
15659         * config/arm/cortex-a53.md: New file.
15660         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
15661         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
15662         * config/arm/arm.c (arm_issue_rate): Likewise.
15663         * config/arm/arm-tune.md: Regenerate
15664         * config/arm/arm-tables.opt: Regenerate.
15665         * config/arm/arm-cores.def: Add cortex-a53.
15667 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
15669         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
15670         non-static link.
15672 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15674         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
15675         scalar load/store operations using B/H registers.
15676         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
15678 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15680         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
15681         scalar move.
15682         * config/aarch64/aarch64.c
15683         (aarch64_simd_scalar_immediate_valid_for_move): New.
15684         * config/aarch64/aarch64-protos.h
15685         (aarch64_simd_scalar_immediate_valid_for_move): New.
15686         * config/aarch64/constraints.md (Dh, Dq): New.
15687         * config/aarch64/iterators.md (hq): New.
15689 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
15691         * reorg.c (get_branch_condition): Deal with conditional returns.
15692         (fill_simple_delay_slots): Remove dead code dealing with jumps.
15694 2013-04-01  Wei Mi  <wmi@google.com>
15696         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
15697         Truncate operand 2 using %b asm operand modifier.
15698         (*<shift_insn><mode>3_mask): Ditto.
15699         (*<rotate_insn><mode>3_mask): Ditto.
15701 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
15703         PR middle-end/56798
15704         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
15706 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
15708         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
15709         of next_real_insn.
15710         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
15712 2013-03-30  Lawrence Crowl  <crowl@google.com>
15714         * dse.c (clear_alias_sets): Remove never set.
15715         (disqualified_clear_alias_sets): Remove never set.
15716         (clear_alias_mode_pool): Remove never set.
15717         (dse_step0): Remove condition that is never true.
15718         (canon_address): Remove condition that is never true.
15719         (dse_step7): Remove condition that is never true.
15720         (rest_of_handle_dse): Remove condition that is never true.
15721         (rest_of_handle_dse::did_global): Remove never read from above.
15722         (dse_step2_spill): Remove never called from above.
15723         (dse_step5_spill): Remove never called from above.
15725 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
15727         * doc/md.texi (Standard Names) <casesi>: Update documentation for
15728         JUMP_TABLE_DATA changes.
15729         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
15730         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
15731         (Insns) <jump_table_data>: New entry.
15732         * doc/tm.texi: Regenerate.
15734         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
15736         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
15737         for table jump at the end of a basic block using tablejump_p.
15738         * targhooks.c (default_invalid_within_doloop): Likewise.
15739         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
15740         target hook implementation that is identical to the default hook.
15741         (rs6000_invalid_within_doloop): Remove.
15743         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
15744         unused variable from tablejump_p call.
15746         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
15747         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
15748         (INSN_DELETED_P): Likewise.
15749         (emit_jump_table_data): New prototype.
15750         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
15751         after 4th as unused.
15752         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
15753         * sched-vis.c (print_insn): Likewise.
15754         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
15755         insn for compatibility with back ends that use next_active_insn to
15756         identify jump table data.
15757         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
15758         (remove_insn): Likewise.
15759         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
15760         to be emitted.
15761         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
15762         (emit_jump_table_data): New function.
15764         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
15765         basic block, a JUMP_TABLE_DATA never is.
15766         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
15767         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
15768         off from code handling real insns.
15769         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
15770         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
15771         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
15772         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
15773         is not a NONDEBUG_INSN_P.
15774         * ira-costs.c (scan_one_insn): Likewise.
15775         * jump.c (mark_all_labels): Likewise.
15776         (mark_jump_label_1): Likewise.
15777         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15778         * lra.c (get_insn_freq): Expect all insns reaching here to be in
15779         a basic block.
15780         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
15781         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
15782         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
15783         JUMP_TABLE_DATA_P insns.
15784         (calculate_elim_costs_all_insns): Likewise.
15785         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
15786         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
15787         (delete_output_reload): Code style fixups.
15788         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
15789         insn flags on this non-insn.
15790         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
15791         as scheduling barriers, for pre-change compatibility.
15792         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
15793         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
15795         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
15796         redundant JUMP_TABLE_DATA_P test.
15797         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
15798         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15799         (frv_for_each_packet): Likewise.
15800         * config/i386/i386.c (min_insn_size): Likewise.
15801         (ix86_avoid_jump_mispredicts): Likewise.
15802         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15803         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
15804         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15805         (mips16_insn_length): Robustify.
15806         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
15807         (mips16_split_long_branches): Likewise.
15808         * config/pa/pa.c (pa_combine_instructions): Likewise.
15809         * config/rs6000/rs6000.c (get_next_active_insn): Treat
15810         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
15811         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
15812         as contributing to pool range lengths.
15813         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
15814         Remove redundant JUMP_TABLE_DATA_P test.
15815         (sh_loop_align): Likewise.
15816         (split_branches): Likewise.
15817         (sh_insn_length_adjustment): Likewise.
15818         * config/spu/spu.c (get_branch_target): Likewise.
15820 2013-03-29  Jan Hubicka  <jh@suse.cz>
15822         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
15823         gcov streaming; stream hot bb threshold to ltrans.
15824         * predict.c (get_hot_bb_threshold): Break out from ....
15825         (maybe_hot_count_p): ... here.
15826         (set_hot_bb_threshold): New function.
15827         * lto-section-in.c (lto_section_name): Add profile.
15828         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
15829         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
15830         and data-streamer.h
15831         (histogram_entry): New structure.
15832         (histogram, histogram_pool): New global vars.
15833         (histogram_hash): New structure.
15834         (histogram_hash::hash): New method.
15835         (histogram_hash::equal): Likewise.
15836         (account_time_size): New function.
15837         (cmp_counts): New function.
15838         (dump_histogram): New function.
15839         (ipa_profile_generate_summary): New function.
15840         (ipa_profile_write_summary): New function.
15841         (ipa_profile_read_summary): New function.
15842         (ipa_profile): Decide on threshold.
15843         (pass_ipa_profile): Add ipa_profile_write_summary and
15844         ipa_profile_read_summary.
15845         * Makefile.in (ipa.o): Update dependencies.
15846         * lto-streamer.h (LTO_section_ipa_profile): New section.
15848 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
15850         * tree.h (VAR_P): New.
15852 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
15854         PR lto/56777
15855         * doc/invoke.texi ([-fwhole-program]): Fix typo.
15857 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
15859         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
15860         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15861         (control_flow_insn_p): Likewise.
15862         * cfgrtl.c (duplicate_insn_chain): Likewise.
15863         * final.c (get_attr_length_1): Likewise.
15864         (shorten_branches): Likewise.
15865         (final_scan_insn): Likewise.
15866         * function.c (instantiate_virtual_regs): Likewise.
15867         * gcse.c (insert_insn_end_basic_block): Likewise.
15868         * ira-costs.c (scan_one_insn): Likewise.
15869         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15870         * lra.c (check_rtl): Likewise.
15871         * reload1.c (elimination_costs_in_insn): Likewise.
15872         * reorg.c (follow_jumps): Likewise.
15874         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
15875         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15876         (thumb_far_jump_used_p): Likewise.
15877         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
15878         (workaround_speculation): Likewise.
15879         (add_sched_insns_for_speculation): Likewise.
15880         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
15881         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15882         (frv_for_each_packet): Likewise.
15883         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
15884         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
15885         (final_emit_insn_group_barriers): Likewise.
15886         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15887         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15888         (mips16_insn_length): Likewise.
15889         * config/pa/pa.c (pa_reorg): Likewise.
15890         (pa_combine_instructions): Likewise.
15891         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
15892         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
15893         (sh_reorg): Likewise.
15894         (split_branches): Likewise.
15895         * config/spu/spu.c (get_branch_target): Likewise.
15897         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
15898         JUMP_TABLE_DATA_P.
15900 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
15902         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
15903         Fix declaration name.
15905 2013-03-28  Lawrence Crowl  <crowl@google.com>
15907         * graphds.h (struct graph.indicies): Remove unused.
15908         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
15909         (SCOP_ORIGINAL_PDDRS): Remove unused.
15910         * sese.h (extern insert_loop_close_phis): Removed unused.
15911         (extern insert_guard_phis): Removed unused.
15912         (extern ivtype_map_elt_info): Removed unused.
15913         (new_ivtype_map_elt): Removed unused.
15914         * sese.c (ivtype_map_elt_info): Removed unused.
15916 2013-03-28  Lawrence Crowl  <crowl@google.com>
15918         * Makefile.in: Add several missing include dependences.
15919         (DUMPFILE_H): New.
15920         (test-dump.o): New.  This object is not added to any executable,
15921         but is present for ad-hoc testing.
15922         * bitmap.c
15923         (debug (const bitmap_head_def &)): New.
15924         (debug (const bitmap_head_def *)): New.
15925         * bitmap.h
15926         (extern debug (const bitmap_head_def &)): New.
15927         (extern debug (const bitmap_head_def *)): New.
15928         * cfg.c
15929         (debug (edge_def &)): New.
15930         (debug (edge_def *)): New.
15931         * cfghooks.c
15932         (debug (basic_block_def &)): New.
15933         (debug (basic_block_def *)): New.
15934         * dumpfile.h
15935         (dump_node (const_tree, int, FILE *)): Correct source file.
15936         * dwarf2out.c
15937         (debug (die_struct &)): New.
15938         (debug (die_struct *)): New.
15939         * dwarf2out.h
15940         (extern debug (die_struct &)): New.
15941         (extern debug (die_struct *)): New.
15942         * gimple-pretty-print.c
15943         (debug (gimple_statement_d &)): New.
15944         (debug (gimple_statement_d *)): New.
15945         * gimple-pretty-print.h
15946         (extern debug (gimple_statement_d &)): New.
15947         (extern debug (gimple_statement_d *)): New.
15948         * ira-build.c
15949         (debug (ira_allocno_copy &)): New.
15950         (debug (ira_allocno_copy *)): New.
15951         (debug (ira_allocno &)): New.
15952         (debug (ira_allocno *)): New.
15953         * ira-int.h
15954         (extern debug (ira_allocno_copy &)): New.
15955         (extern debug (ira_allocno_copy *)): New.
15956         (extern debug (ira_allocno &)): New.
15957         (extern debug (ira_allocno *)): New.
15958         * ira-lives.c
15959         (debug (live_range &)): New.
15960         (debug (live_range *)): New.
15961         * lra-int.h
15962         (debug (lra_live_range &)): New.
15963         (debug (lra_live_range *)): New.
15964         * lra-lives.c
15965         (debug (lra_live_range &)): New.
15966         (debug (lra_live_range *)): New.
15967         * omega.c
15968         (debug (omega_pb_d &)): New.
15969         (debug (omega_pb_d *)): New.
15970         * omega.h
15971         (extern debug (omega_pb_d &)): New.
15972         (extern debug (omega_pb_d *)): New.
15973         * print-rtl.c
15974         (debug (const rtx_def &)): New.
15975         (debug (const rtx_def *)): New.
15976         * print-tree.c
15977         (debug_tree (tree): Move within file.
15978         (debug_raw (const tree_node &)): New.
15979         (debug_raw (const tree_node *)): New.
15980         (dump_tree_via_hooks (const tree_node *, int)): New.
15981         (debug (const tree_node &)): New.
15982         (debug (const tree_node *)): New.
15983         (debug_verbose (const tree_node &)): New.
15984         (debug_verbose (const tree_node *)): New.
15985         (debug_head (const tree_node &)): New.
15986         (debug_head (const tree_node *)): New.
15987         (debug_body (const tree_node &)): New.
15988         (debug_body (const tree_node *)): New.
15989         (debug_vec_tree (tree): Move and reimplement in terms of dump.
15990         (debug (vec<tree, va_gc> &)): New.
15991         (debug (vec<tree, va_gc> *)): New.
15992         * rtl.h
15993         (extern debug (const rtx_def &)): New.
15994         (extern debug (const rtx_def *)): New.
15995         * sbitmap.c
15996         (debug_raw (simple_bitmap_def &)): New.
15997         (debug_raw (simple_bitmap_def *)): New.
15998         (debug (simple_bitmap_def &)): New.
15999         (debug (simple_bitmap_def *)): New.
16000         * sbitmap.h
16001         (extern debug (simple_bitmap_def &)): New.
16002         (extern debug (simple_bitmap_def *)): New.
16003         (extern debug_raw (simple_bitmap_def &)): New.
16004         (extern debug_raw (simple_bitmap_def *)): New.
16005         * sel-sched-dump.c
16006         (debug (vinsn_def &)): New.
16007         (debug (vinsn_def *)): New.
16008         (debug_verbose (vinsn_def &)): New.
16009         (debug_verbose (vinsn_def *)): New.
16010         (debug (expr_def &)): New.
16011         (debug (expr_def *)): New.
16012         (debug_verbose (expr_def &)): New.
16013         (debug_verbose (expr_def *)): New.
16014         (debug (vec<rtx> &)): New.
16015         (debug (vec<rtx> *)): New.
16016         * sel-sched-dump.h
16017         (extern debug (vinsn_def &)): New.
16018         (extern debug (vinsn_def *)): New.
16019         (extern debug_verbose (vinsn_def &)): New.
16020         (extern debug_verbose (vinsn_def *)): New.
16021         (extern debug (expr_def &)): New.
16022         (extern debug (expr_def *)): New.
16023         (extern debug_verbose (expr_def &)): New.
16024         (extern debug_verbose (expr_def *)): New.
16025         (extern debug (vec<rtx> &)): New.
16026         (extern debug (vec<rtx> *)): New.
16027         * sel-sched-ir.h
16028         (_list_iter_cond_expr): Make inline instead of static.
16029         * sreal.c
16030         (debug (sreal &)): New.
16031         (debug (sreal *)): New.
16032         * sreal.h
16033         (extern debug (sreal &)): New.
16034         (extern debug (sreal *)): New.
16035         * tree.h
16036         (extern debug_raw (const tree_node &)): New.
16037         (extern debug_raw (const tree_node *)): New.
16038         (extern debug (const tree_node &)): New.
16039         (extern debug (const tree_node *)): New.
16040         (extern debug_verbose (const tree_node &)): New.
16041         (extern debug_verbose (const tree_node *)): New.
16042         (extern debug_head (const tree_node &)): New.
16043         (extern debug_head (const tree_node *)): New.
16044         (extern debug_body (const tree_node &)): New.
16045         (extern debug_body (const tree_node *)): New.
16046         (extern debug (vec<tree, va_gc> &)): New.
16047         (extern debug (vec<tree, va_gc> *)): New.
16048         * tree-cfg.c
16049         (debug (struct loop &)): New.
16050         (debug (struct loop *)): New.
16051         (debug_verbose (struct loop &)): New.
16052         (debug_verbose (struct loop *)): New.
16053         * tree-dump.c: Add header dependence.
16054         * tree-flow.h
16055         (extern debug (struct loop &)): New.
16056         (extern debug (struct loop *)): New.
16057         (extern debug_verbose (struct loop &)): New.
16058         (extern debug_verbose (struct loop *)): New.
16059         * tree-data-ref.c
16060         (debug (data_reference &)): New.
16061         (debug (data_reference *)): New.
16062         (debug (vec<data_reference_p> &)): New.
16063         (debug (vec<data_reference_p> *)): New.
16064         (debug (vec<ddr_p> &)): New.
16065         (debug (vec<ddr_p> *)): New.
16066         * tree-data-ref.h
16067         (extern debug (data_reference &)): New.
16068         (extern debug (data_reference *)): New.
16069         (extern debug (vec<data_reference_p> &)): New.
16070         (extern debug (vec<data_reference_p> *)): New.
16071         (extern debug (vec<ddr_p> &)): New.
16072         (extern debug (vec<ddr_p> *)): New.
16073         * tree-ssa-alias.c
16074         (debug (pt_solution &)): New.
16075         (debug (pt_solution *)): New.
16076         * tree-ssa-alias.h
16077         (extern debug (pt_solution &)): New.
16078         (extern debug (pt_solution *)): New.
16079         * tree-ssa-alias.c
16080         (debug (_var_map &)): New.
16081         (debug (_var_map *)): New.
16082         (debug (tree_live_info_d &)): New.
16083         (debug (tree_live_info_d *)): New.
16084         * tree-ssa-alias.h
16085         (extern debug (_var_map &)): New.
16086         (extern debug (_var_map *)): New.
16087         (extern debug (tree_live_info_d &)): New.
16088         (extern debug (tree_live_info_d *)): New.
16090 2013-03-28  Jan Hubicka  <jh@suse.cz>
16092         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16094 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16096         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16097         record only when desired or required.
16099 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16101         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16102         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16104 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16106         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16107         (*andsi3_compare0_uxtw): New pattern.
16108         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16109         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16111 2013-03-28  Jan Hubicka  <jh@suse.cz>
16113         * data-streamer-in.c (streamer_read_gcov_count): New function.
16114         * gimple-streamer-out.c: Include value-prof.h.
16115         (output_gimple_stmt): Output histogram.
16116         (output_bb): Use streamer_write_gcov_count.
16117         * value-prof.c: Include data-streamer.h
16118         (dump_histogram_value): Add HIST_TYPE_MAX.
16119         (stream_out_histogram_value): New function.
16120         (stream_in_histogram_value): New function.
16121         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16122         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16123         * data-streamer-out.c (streamer_write_gcov_count): New function.
16124         (streamer_write_gcov_count_stream): New function.
16125         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16126         (lto_output_node): Likewise.
16127         (input_node, input_edge): Likewise.
16128         * lto-streamer-out.c (output_cfg): Update streaming.
16129         * lto-streamer-in.c (input_cfg): Likewise.
16130         * data-streamer.h (streamer_write_gcov_count,
16131         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16132         * gimple-streamer-in.c: Include value-prof.h
16133         (input_gimple_stmt): Input histograms.
16134         (input_bb): Update profile streaming.
16136 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16138         * genmodes.c (emit_max_int): New function.
16139         (emit_insn_modes_h): Added call to emit_max_function.
16140         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16141         Added doc.
16142         * machmode.def: Fixed comment.
16144 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16146         * combine.c (try_combine): Removed useless assert.
16147         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16149 2013-03-28  Marek Polacek  <polacek@redhat.com>
16150             Richard Biener  <rguenther@suse.de>
16152         PR tree-optimization/56695
16153         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16154         build signed result of a vector comparison.
16155         * tree-cfg.c (verify_gimple_comparison): Check that a result
16156         of a vector comparison has signed type.
16158 2013-03-28  Richard Biener  <rguenther@suse.de>
16160         PR tree-optimization/37021
16161         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16162         do not restrict gaps between groups.
16163         * tree-vect-stmts.c (vectorizable_load): Properly account for
16164         a gap between groups.
16166 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16168         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16169         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16170         is not enabled.
16172 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16174         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16175         * doc/extend.texi (Named Address Spaces): Ditto.
16176         (Variable Attributes): Ditto.
16178 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16180         * config.build: Add support for cygwin x64 target.
16181         * config.gcc: Likewise.
16182         * config.host: Likewise.
16183         * configure.ac: Likewise
16184         * configure: Regenerated.
16186 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16188         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16189         * config/i386/t-cygwin-w64: New file.
16190         * config/i386/cygwin-w64.h: New file.
16191         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16192         and add support for x64-cygwin target.
16193         (CPP_SPEC): Likewise.
16194         (CXX_WRAP_SPEC_LIST): Undefine before define.
16195         (LIBGCJ_SONAME): Use 15 as version.
16197 2013-03-27  Richard Biener  <rguenther@suse.de>
16199         PR tree-optimization/56716
16200         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16201         dumping for ref nodes.
16203 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16205         PR tree-optimization/55334
16206         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16207         restricted pointers to arrays.
16209 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16211         * Makefile.in (.SUFFIXES): Add .cc.
16212         (.c.o): Apply same recipe for implicit rule .cc.o.
16214 2013-03-27  Richard Biener  <rguenther@suse.de>
16216         PR tree-optimization/37021
16217         * tree-vect-data-refs.c (vect_check_strided_load): Allow
16218         REALPART/IMAGPART_EXPRs around the supported refs.
16219         * tree-ssa-structalias.c (find_func_aliases): Assume that
16220         floating-point values are not used to transfer pointers.
16222 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16224         * target.def (TARGET_HAS_IFUNC_P): New target hook.
16225         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16226         * doc/tm.texi: Regenerate.
16227         * targhooks.h (default_has_ifunc_p): New.
16228         * targhooks.c (default_has_ifunc_p): Ditto.
16229         * config/linux-protos.h: New file.
16230         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16231         hook for linux which disables support of indirect functions in android.
16232         * config/linux-android.c: New file.
16233         * config/t-linux-android.c: Ditto.
16234         * config.gcc: Added new object file linux-android.o.
16235         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16236         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16237         * varasm.c (do_assemble_alias): Likewise.
16238         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16239         doesn't support indirect functions.
16240         * configure: Regenerate.
16242 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
16244         PR target/56102
16245         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16246         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16247         mult-word mode.
16249 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16251         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16253 2013-03-27  Terry Guo  <terry.guo@arm.com>
16255         * config/arm/arm-cores.def: Added core cortex-r7.
16256         * config/arm/arm-tune.md: Regenerated.
16257         * config/arm/arm-tables.opt: Regenerated.
16258         * doc/invoke.texi: Added entry for core cortex-r7.
16260 2013-03-27  Walter Lee  <walt@tilera.com>
16262         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16263         double-decrement of next_scratch_regno.
16265 2013-03-27  Walter Lee  <walt@tilera.com>
16267         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16268         input operands.
16269         (insn_v1mulus): Ditto.
16270         (insn_v2muls): Ditto.
16272 2013-03-27  Walter Lee  <walt@tilera.com>
16274         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16275         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16277 2013-03-27  Walter Lee  <walt@tilera.com>
16279         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16280         (*sibcall_value): Ditto.
16282 2013-03-27  Walter Lee  <walt@tilera.com>
16284         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16285         (insn_mnz_v8qi): ... this ...
16286         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
16287         vector equivalent.
16288         (insn_v<n>mnz): Replaced by ...
16289         (insn_v1mnz): ... this ...
16290         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
16291         equivalent.
16292         (insn_mz_<mode>): Replaced by ...
16293         (insn_mz_v8qi): ... this ...
16294         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
16295         vector equivalent.
16296         (insn_v<n>mz): Replaced by ...
16297         (insn_v1mz): ... this ...
16298         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
16299         equivalent.
16301 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
16303         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16305 2013-03-26  Roland McGrath  <mcgrathr@google.com>
16307         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16308         than fprintf with a non-constant, non-format string.
16310 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16312         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16313         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
16314         operand 0 predicate.
16315         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16316         attribute.  Use general_x64nomem_operand as operand 1 predicate.
16317         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16318         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16319         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16320         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16321         (mov<mode>_insv_1): Remove expander.  Merge insn with
16322         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16323         Use general_x64nomem_operand as operand 1 predicate.
16324         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16325         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16326         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16327         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16328         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
16329         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16330         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16331         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16332         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
16333         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16334         (general_x64nomem_operand): Ditto.
16336 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
16338         * config/rtems.opt: Add -pthread option.
16340 2013-03-26  Richard Biener  <rguenther@suse.de>
16342         * alias.c (find_base_term): Avoid redundant and not used recursion.
16343         (base_alias_check): Get the initial base term from the caller.
16344         (true_dependence_1): Compute and pass base terms to base_alias_check.
16345         (write_dependence_p): Likewise.
16346         (may_alias_p): Likewise.
16348 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
16350         * config/aarch64/aarch64.c (aarch64_classify_address): Support
16351         PC-relative load in SI modes and above only.
16353 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
16355         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16356         * config/arm/iwmmxt.md (WCGR0): Update.
16357         (WCGR1, WCGR2, WCGR3): Likewise.
16359 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16361         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16362         Use x64 and nox64 isa attributes.
16364 2013-03-26  Richard Biener  <rguenther@suse.de>
16366         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16367         alignment computations and rely on get_object_alignment_1
16368         for the !TYPE_P case.
16369         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16371 2013-03-26  Walter Lee  <walt@tilera.com>
16373         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16374         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16376 2013-03-25  Jeff Law  <law@redhat.com>
16378         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16379         check for INTEGRAL_TYPE_P that was missing due to checking in
16380         wrong version of prior patch.
16382 2013-03-25  Walter Lee  <walt@tilera.com>
16384         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16385         TILEGX_INSN_SHUFFLEBYTES1.
16386         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16387         shufflebytes1.
16388         (tilegx_builtins): Ditto.
16389         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16391 2013-03-25  Walter Lee  <walt@tilera.com>
16393         * config/tilegx/tilegx.md (floatsisf2): New pattern.
16394         (floatunssisf2): New pattern.
16395         (floatsidf2): New pattern.
16396         (floatunssidf2): New pattern.
16398 2013-03-25  Walter Lee  <walt@tilera.com>
16400         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16401         tests for constraint J, K, N, P.
16403 2013-03-25  Walter Lee  <walt@tilera.com>
16405         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16406         Use indirect/pcrel encoding.
16407         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16408         Ditto.
16410 2013-03-25  Steve Ellcey  <sellcey@mips.com>
16412         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16413         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16414         * config/mips/mips.c (mips_option_override): Set IMADD default.
16415         * config/mips/mips.h (PTF_AVOID_IMADD): New.
16416         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16417         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16418         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16419         * doc/invoke.texi (-mimadd/-mno-imadd): New.
16421 2013-03-25  Jeff Law  <law@redhat.com>
16423         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16424         slightly to avoid creating and folding useless trees.  Simplify
16425         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16427 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
16429         * config/i386/i386.md (*zero_extendsidi2): Merge with
16430         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
16431         * config/i386/predicates.md (x86_64_zext_operand): Rename from
16432         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
16433         targets.  Clarify comment.
16435 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16437         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16438         pass-through jump functions differently.
16439         (ipa_read_jump_function): Likewise.  Also use setter functions to set
16440         up jump functions.
16442 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16444         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16445         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16446         process it.
16447         (ipa_get_indirect_edge_target): New function.
16448         (devirtualization_time_bonus): New parameter known_aggs, pass it to
16449         ipa_get_indirect_edge_target.  Update all callers.
16450         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
16451         ipa_get_indirect_edge_target_1 instead of calling
16452         ipa_get_indirect_edge_target.
16453         (create_specialized_node): Pass aggvlas to
16454         ipcp_discover_new_direct_edges.
16456 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16458         * config/arm/arm.md (f_sels, f_seld): New types.
16459         (*cmov<mode>): New pattern.
16460         * config/arm/predicates.md (arm_vsel_comparison_operator): New
16461         predicate.
16463 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16465         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16466         POSIX-printf for mingw-hosted  builds.
16468 2013-03-25  Richard Biener  <rguenther@suse.de>
16470         PR middle-end/56694
16471         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16472         must-not-throw stmt location.
16474 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16476         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16477         Emit load-acquire versions when acq is true.
16478         (arm_emit_store_exclusive): Add rel parameter.
16479         Emit store-release versions when rel is true.
16480         (arm_split_compare_and_swap): Use acquire-release instructions
16481         instead.
16482         of barriers when appropriate.
16483         (arm_split_atomic_op): Likewise.
16484         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16485         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16486         (VUNSPEC_SLX): Likewise.
16487         (VUNSPEC_LDA): Likewise.
16488         (VUNSPEC_STL): Likewise.
16489         * config/arm/sync.md (atomic_load<mode>): New pattern.
16490         (atomic_store<mode>): Likewise.
16491         (arm_load_acquire_exclusive<mode>): Likewise.
16492         (arm_load_acquire_exclusivesi): Likewise.
16493         (arm_load_acquire_exclusivedi): Likewise.
16494         (arm_store_release_exclusive<mode>): Likewise.
16496 2013-03-25  Catherine Moore  <clm@codesourcery.com>
16498         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16499         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16500         * config/mip/predicates.md (lwsp_swsp_operand,
16501         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16502         sb16_operand, db4_operand, db7_operand, ib3_operand,
16503         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16504         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16505         andi16_operand): New predicates.
16506         * config/mips/mips.md (compression): New attribute.
16507         (enabled): New attribute.
16508         (length): Consider compression in computing length.
16509         (shift_compression): New code attribute.
16510         (*add<mode>3): New operands. Record compression.
16511         (sub<mode>3): Likewise.
16512         (one_cmpl<mode>2): Likewise.
16513         (*and<mode>3): Likewise.
16514         (*ior<mode>3): Likewise.
16515         (unnamed pattern for xor): Likewise.
16516         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16517         (*<optab><mode>3): Likewise.
16518         (*mov<mode>_internal: Likewise.
16519         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16520         (mips_unsigned_immediate_p): New.
16521         (umips_lwsp_swsp_address_p): New.
16522         (m16_based_address_p): New.
16523         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16524         (mips_unsigned_immediate_p): New prototype.
16525         (lwsp_swsp_address_p): New prototype.
16526         (m16_based_address_p): New prototype.
16527         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16528         (mips_signed_immediate_p): New function.
16529         (m16_based_address_p): New function.
16530         (lwsp_swsp_address_p): New function.
16531         (mips_print_operand_punctuation): Recognize short delay slot insns
16532         for microMIPS.add<mode>3"
16534 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16536         PR target/56720
16537         * config/arm/iterators.md (v_cmp_result): New mode attribute.
16538         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16540 2013-03-25  Richard Biener  <rguenther@suse.de>
16542         PR tree-optimization/56689
16543         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16544         any edge.
16546 2013-03-25  Richard Biener  <rguenther@suse.de>
16548         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16549         of bitmap.
16550         (memory_references): Likewise.
16551         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16552         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16553         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16554         (gather_mem_refs_in_loops): Fold into ...
16555         (analyze_memory_references): ... this.  Move initialization
16556         to tree_ssa_lim_initialize.
16557         (fill_always_executed_in): Rename to ...
16558         (fill_always_executed_in_1): ... this.
16559         (fill_always_executed_in): Move contains_call computation to
16560         this new function from ...
16561         (tree_ssa_lim_initialize): ... here.
16562         (tree_ssa_lim): Call fill_always_executed_in.
16564 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16566         * postreload.c (reload_combine): Fix code detecting returns.
16568 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16570         * function.c (emit_use_return_register_into_block): On cc0 targets,
16571         do not emit the sequence between cc0 setter and user.
16573 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16575         * config/i386/predicates.md (local_symbolic_operand): Interpret
16576         dll-imported symbols as none-local.
16578 2013-03-25  Richard Biener  <rguenther@suse.de>
16580         * tree-ssa-loop-im.c (struct depend): Remove.
16581         (struct lim_aux_data): Make depends a vec of gimples.
16582         (free_lim_aux_data): Adjust.
16583         (add_dependency): Likewise.
16584         (set_level): Likewise.
16586 2013-03-25  Richard Biener  <rguenther@suse.de>
16588         PR middle-end/56434
16589         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
16590         the pointer returned by calls with ECF_MALLOC set.
16592 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16594         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
16596 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16598         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
16599         using MMXMODE mode iterator.
16600         (*move<mode>_internal): Merge with *movv2sf_internal and
16601         *movv2sf_internal_rex64 using MMXMODE mode iterator.
16603 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16605         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
16606         (record_last_mem_set_info): Likewise.
16608         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
16609         of XNEWVEC followed by memset.
16610         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
16612 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16614         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
16615         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
16616         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
16617         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
16618         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
16619         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
16620         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
16621         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
16622         BARRIER_P instead of GET_CODE.
16624 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16626         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
16627         inaccuracy in the probing code.
16629         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
16630         (ctrapdi4): Likewise.
16632 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16634         * calls.c (expand_call): Add missing guard to code handling return
16635         of non-BLKmode structures in MSB.
16636         * function.c (expand_function_end): Likewise.
16638 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16640         * combine.c (try_combine): Adjust comment.  Do not add the set of
16641         insn #0 if the destination indirectly is set or dies in insn #2.
16642         Tidy up code to distribute a new note.
16644 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16646         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
16647         also for alternatives 16 and 17.
16649 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16651         * config/i386/sse.md (*mov<mode>_internal): Merge with
16652         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
16653         Emit insn template depending on type attribute.  Use
16654         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16655         movd instead of movq mnemonic for interunit moves.  Rewrite mode
16656         attribute calculation.  Remove unit attribute calculation.
16657         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
16658         Set prefix_data16 attribute for DImode ssemov types.
16659         Use Ym instead of y for SSE-MMX conversion alternatives.
16660         Reorder operand constraints.
16662 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
16664         * df.h (df_insn_delete): Adjust prototype.
16665         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
16666         and let it decide whether mark the basic block dirty.
16667         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
16668         * df-scan.c (df_insn_info_delete): New helper function, split
16669         off from df_insn_delete.
16670         (df_scan_free_bb_info): Use it.
16671         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
16672         Likewise.
16673         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
16674         that the insn is actually an insn and it has a non-NULL basic block.
16675         Do not mark basic block dirty if only deleting a DEBUG_INSN.
16677 2013-03-22  Richard Biener  <rguenther@suse.de>
16679         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
16680         dep_ref members.
16681         (mem_ref_alloc): Do not allocate them.
16682         (refs_independent_p): Do not query or maintain a cache.
16684 2013-03-22  Richard Biener  <rguenther@suse.de>
16686         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
16687         (gather_mem_refs_in_loops): Do not compute it.
16688         (analyze_memory_references): Do not allocate it.
16689         (tree_ssa_lim_finalize): Do not free it.
16690         (for_all_locs_in_loop): Do not query all_refs_in_loop.
16692 2013-03-22  Richard Biener  <rguenther@suse.de>
16694         * is-a.h (as_a): Use gcc_checking_assert.
16696 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
16698         * config/aarch64/aarch64.c (aarch64_print_operand): New
16699         format specifier for printing a constant in hex.
16700         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
16701         format specifier for printing second operand.
16703 2013-03-22  Richard Biener  <rguenther@suse.de>
16705         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
16706         bitmaps.
16707         (gather_mem_refs_in_loops): Perform store accumulation here.
16708         (create_vop_ref_mapping_loop): Remove.
16709         (create_vop_ref_mapping): Likewise.
16710         (analyze_memory_references): Initialize refs_stored_in_loop.
16711         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
16712         (record_indep_loop): Remove.
16713         (record_dep_loop): New function.
16714         (ref_indep_loop_p_1): Adjust to only walk over references
16715         in the loop, not its subloops.
16716         (ref_indep_loop_p): Rename to ...
16717         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
16718         maintaining a more fine-grained cache.
16719         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
16720         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
16722 2013-03-22  Richard Biener  <rguenther@suse.de>
16724         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
16725         (struct mem_ref): Make accesses_in_loop a vec of a vec of
16726         aggregate mem_ref_loc.
16727         (free_mem_ref_locs): Inline into ...
16728         (memref_free): ... this and adjust.
16729         (mem_ref_alloc): Adjust.
16730         (mem_ref_locs_alloc): Remove.
16731         (record_mem_ref_loc): Adjust.
16732         (get_all_locs_in_loop): Rewrite into ...
16733         (for_all_locs_in_loop): ... this iterator.
16734         (rewrite_mem_ref_loc): New functor.
16735         (rewrite_mem_refs): Use for_all_locs_in_loop.
16736         (sm_set_flag_if_changed): New functor.
16737         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
16738         (ref_always_accessed): New functor.
16739         (ref_always_accessed_p): Use for_all_locs_in_loop.
16741 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16743         * tree-pass.h (PROP_gimple_lvec): New.
16744         * passes.c (dump_properties): Handle PROP_gimple_lvec.
16745         (init_optimization_passes): Move pass_lower_vector.
16746         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
16747         PROP_gimple_lvec.
16748         (pass_lower_vector): Provide PROP_gimple_lvec.
16749         (pass_lower_vector_ssa): Likewise.
16750         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
16752 2013-03-21  Mark Wielaard  <mjw@redhat.com>
16754         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
16756 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16758         * config/i386/i386.md (*movdi_internal): Disparage slightly
16759         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
16760         conversion alternatives.
16762 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
16764         PR middle-end/48087
16765         * diagnostic.def (DK_WERROR): New kind.
16766         * diagnostic.h (werrorcount): Define.
16767         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
16768         promoted to DK_ERROR, increment DK_WERROR counter instead of
16769         DK_ERROR counter.
16770         * toplev.c (toplev_main): Call print_ignored_options even if
16771         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
16772         even if just werrorcount is non-zero.
16774         PR debug/55608
16775         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
16776         on failure.
16777         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
16778         (string_cst_pool_decl): New function.
16779         (optimize_one_addr_into_implicit_ptr): New function.
16780         (resolve_addr_in_expr): Optimize DWARF location expression
16781         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
16782         which doesn't live in memory, but has DW_AT_location or
16783         DW_AT_const_value, or refers to a string literal, into
16784         DW_OP_GNU_implicit_pointer.
16785         (optimize_location_into_implicit_ptr): New function.
16786         (resolve_addr): If removing DW_AT_location of a variable because
16787         it was DW_OP_addr of address of the variable, but the variable doesn't
16788         live in memory, try to emit const value attribute for the initializer.
16790 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16792         * tree.h (VECTOR_TYPE_P): New macro.
16793         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
16794         TYPE_MODE): Use it.
16795         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
16796         VEC_COND_EXPR cannot be lvalues.
16797         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
16799 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16801         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
16802         Restrict the transformation to equal modes.
16804 2013-03-21  Richard Biener  <rguenther@suse.de>
16806         PR tree-optimization/39326
16807         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
16808         (MEM_ANALYZABLE): Adjust.
16809         (record_mem_ref_loc): Move bitmap ops ...
16810         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
16811         unanalyzable refs, do not record locations for it.
16812         (analyze_memory_references): Allocate ref zero as shared
16813         unanalyzable ref.
16814         (refs_independent_p): Do not test for unanalyzed mems here.
16815         (ref_indep_loop_p_1): Special-case disambiguation against
16816         the unanalyzed ref.
16817         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
16819 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
16821         * config/arm/arm-protos.h (tune_params): Add
16822         prefer_neon_for_64bits field.
16823         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
16824         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
16825         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
16826         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
16827         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
16828         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
16829         (arm_option_override): Handle -mneon-for-64bits new option.
16830         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
16831         (prefer_neon_for_64bits): Declare new variable.
16832         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
16833         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
16834         (arch_enabled): Handle new arch types. Remove support for onlya8
16835         and nota8.
16836         (one_cmpldi2): Use new arch names.
16837         (zero_extend<mode>di2, extend<mode>di2): Ditto.
16838         * config/arm/arm.opt (mneon-for-64bits): Add option.
16839         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
16840         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
16841         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
16842         of onlya8.
16843         * doc/invoke.texi (-mneon-for-64bits): Document.
16845 2013-03-21  Richard Biener  <rguenther@suse.de>
16847         PR tree-optimization/39326
16848         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
16849         (sort_bbs_in_loop_postorder_cmp): New function.
16850         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
16852 2013-03-21  Richard Biener  <rguenther@suse.de>
16854         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
16855         (vect_insert_into_interleaving_chain): Likewise.
16856         (vect_drs_dependent_in_basic_block): Inline ...
16857         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
16858         split out from ...
16859         (vect_analyze_data_ref_dependence): ... here.  Simplify.
16860         (vect_check_interleaving): Simplify.
16861         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
16862         (vect_slp_analyze_data_ref_dependences): ... this new function.
16863         (dr_group_sort_cmp): New function.
16864         (vect_analyze_data_ref_accesses): Compute data-reference groups
16865         here instead of in vect_analyze_data_ref_dependence.  Use
16866         a more efficient algorithm.
16867         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
16868         vect_slp_analyze_data_ref_dependences.  Call
16869         vect_analyze_data_ref_accesses earlier.
16870         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
16871         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
16872         (vect_slp_analyze_data_ref_dependences): New prototype.
16874 2013-03-21  Richard Biener  <rguenther@suse.de>
16876         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
16877         ref is stored in the loop.
16878         (find_refs_for_sm): Walk only over all stores.
16879         (store_motion_loop): Allocate from lim_bitmap_obstack.
16880         (store_motion): Likewise.
16882 2013-03-21  Richard Biener  <rguenther@suse.de>
16884         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
16885         Update virtual SSA form.
16887 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16889         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
16890         * configure: Regenerate.
16891         * config.in: Regenerate.
16892         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
16893         if !HAVE_LD_EH_FRAME_CIEV3.
16895 2013-03-21  Richard Biener  <rguenther@suse.de>
16897         * tree-cfg.c (verify_expr_no_block): New function.
16898         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
16899         nor DECL_VALUE_EXPR have locations with associated blocks.
16900         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
16901         (clear_unused_block_pointer): Remove code dealing with
16902         blocks in DECL_DEBUG_EXPR locations.
16904 2013-03-21  Richard Biener  <rguenther@suse.de>
16906         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
16907         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
16908         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
16909         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
16910         instead of DECL_DEBUG_EXPR_IS_FROM.
16911         * gimplify.c (gimplify_modify_expr): Likewise.
16912         * tree-cfg.c (verify_expr_location_1): Likewise.
16913         * tree-complex.c (create_one_component_var): Likewise.
16914         * tree-sra.c (create_access_replacement): Likewise.
16915         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
16916         (clear_unused_block_pointer): Likewise.
16917         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
16918         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
16919         * var-tracking.c (var_debug_decl): Likewise.
16920         (track_expr_p): Likewise.
16921         * tree-inline.c (add_local_variables): Likewise.  Set
16922         DECL_HAS_DEBUG_EXPR_P after copying it.
16923         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
16924         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
16926 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16928         PR bootstrap/56656
16929         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
16930         * configure: Regenerate.
16931         * config.in: Regenerate.
16932         * config/i386/i386.md (*movdf_internal): Use
16933         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16934         movd instead of movq mnemonic for interunit moves.
16935         (*movdi_internal): Ditto.
16937 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16939         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
16940         (abd<mode>_3): New pattern.
16941         (aba<mode>_3): New pattern.
16942         (fabd<mode>_3): New pattern.
16944 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16946         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
16947         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
16948         occurrence of REGISTER_PREFIX as its empty string.
16950 2013-03-20  Jeff Law  <law@redhat.com>
16952         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
16953         addititional equivalences for equality comparisons between an SSA_NAME
16954         and a constant where the SSA_NAME was set from a widening conversion.
16956 2013-03-20  Walter Lee  <walt@tilera.com>
16958         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
16960 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16962         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
16963         depending on type attribute.
16964         (*movti_internal): Ditto.
16965         (*movtf_internal): Ditto.
16966         (*movxf_internal): Ditto.
16967         (*movdf_internal): Ditto.
16968         (*movsf_internal): Ditto.
16970 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16972         * config/i386/i386.md (*movti_internal): Set prefix attribute to
16973         maybe_vex for sselog1 and ssemov types.
16974         (*movdi_internal): Reorder operand constraints.
16975         (*movsi_internal): Ditto.  Set prefix attribute to
16976         maybe_vex for sselog1 and ssemov types.
16977         (*movtf_internal): Set prefix attribute to maybe_vex
16978         for sselog1 and ssemov types.
16979         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
16980         DImode ssemov types.  Reorder operand constraints.
16981         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
16982         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
16983         attribute for SImode ssemov types.  Reorder operand constraints.
16985 2013-03-20  Martin Jambor  <mjambor@suse.cz>
16987         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
16988         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
16990 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
16992         * config/rs6000/predicates.md (indexed_address, update_address_mem
16993         update_indexed_address_mem): New predicates.
16994         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
16995         attribute for load/store instructions.
16996         * config/rs6000/dfp.md (movsd_store): Likewise.
16997         (movsd_load): Likewise.
16998         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
16999         (unnamed HI->DI extend define_insn): Likewise.
17000         (unnamed SI->DI extend define_insn): Likewise.
17001         (unnamed QI->SI extend define_insn): Likewise.
17002         (unnamed QI->HI extend define_insn): Likewise.
17003         (unnamed HI->SI extend define_insn): Likewise.
17004         (unnamed HI->SI extend define_insn): Likewise.
17005         (extendsfdf2_fpr): Likewise.
17006         (movsi_internal1): Likewise.
17007         (movsi_internal1_single): Likewise.
17008         (movhi_internal): Likewise.
17009         (movqi_internal): Likewise.
17010         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17011         attribute for load/store instructions.
17012         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17013         instructions.
17014         (mov<mode>_softfloat): Likewise.
17015         (mov<mode>_hardfloat32): Likewise.
17016         (mov<mode>_hardfloat64): Likewise.
17017         (mov<mode>_softfloat64): Likewise.
17018         (movdi_internal32): Likewise.
17019         (movdi_internal64): Likewise.
17020         (probe_stack_<mode>): Likewise.
17022 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17024         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17025         floating point, and decimal floating point to reload iterator.
17027         * config/rs6000/constraints.md (wl constraint): New constraints to
17028         return FLOAT_REGS if certain options are used to reduce the number
17029         of separate patterns that exist in the file.
17030         (wx constraint): Likewise.
17031         (wz constraint): Likewise.
17033         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17034         -mdebug=reg, print wg, wl, wx, and wz constraints.
17035         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17036         Initialize the reload functions for 64-bit binary/decimal floating
17037         point types.
17038         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17039         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17040         create the buffer on the stack to overcome not having a 32-bit
17041         load and store.
17042         (rs6000_emit_move): Likewise.
17043         (rs6000_secondary_memory_needed_rtx): Likewise.
17044         (rs6000_alloc_sdmode_stack_slot): Likewise.
17045         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17046         via xxlxor, just like DFmode 0.0.
17048         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17049         define as 1 if we are running on a power7 or newer.
17050         (enum r6000_reg_class_enum): Add new constraints.
17052         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17053         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17054         with other moves by using conditional constraits (wg).  Use LFIWZX
17055         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17056         (movsd splitter): Likewise.
17057         (movsd_hardfloat): Likewise.
17058         (movsd_softfloat): Likewise.
17060         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17061         binary and decimal floating point moves.
17062         (fmove_ok): New attributes to combine binary and decimal floating
17063         point moves, and to combine power6x (mfpgpr) moves along normal
17064         floating moves.
17065         (real_value_to_target): Likewise.
17066         (f32_lr): Likewise.
17067         (f32_lm): Likewise.
17068         (f32_li): Likewise.
17069         (f32_sr): Likewise.
17070         (f32_sm): Likewise.
17071         (f32_si): Likewise.
17072         (movsf): Combine binary and decimal floating point moves.  Combine
17073         power6x (mfpgpr) moves with other moves by using conditional
17074         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17075         (mov<mode> for SFmode/SDmode); Likewise.
17076         (SFmode/SDmode splitters): Likewise.
17077         (movsf_hardfloat): Likewise.
17078         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17079         (movsf_softfloat): Likewise.
17080         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17082         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17083         wx and wz constraints.
17085         * config/rs6000/constraints.md (wg constraint): New constraint to
17086         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17088         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17089         constraint.
17091         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17092         -mdebug=reg, print wg, wl, wx, and wz constraints.
17093         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17094         Initialize the reload functions for 64-bit binary/decimal floating
17095         point types.
17096         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17097         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17098         create the buffer on the stack to overcome not having a 32-bit
17099         load and store.
17100         (rs6000_emit_move): Likewise.
17101         (rs6000_secondary_memory_needed_rtx): Likewise.
17102         (rs6000_alloc_sdmode_stack_slot): Likewise.
17103         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17104         via xxlxor, just like DFmode 0.0.
17106         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17107         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17108         with other moves by using conditional constraits (wg).  Use LFIWZX
17109         and STFIWX for loading SDmode on power7.
17110         (movdd splitters): Likewise.
17111         (movdd_hardfloat32): Likewise.
17112         (movdd_softfloat32): Likewise.
17113         (movdd_hardfloat64_mfpgpr): Likewise.
17114         (movdd_hardfloat64): Likewise.
17115         (movdd_softfloat64): Likewise.
17117         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17118         64-bit binary and decimal floating point moves.
17119         (FMOVE64X): Likewise.
17120         (movdf): Combine 64-bit binary and decimal floating point moves.
17121         Combine power6x (mfpgpr) moves with other moves by using
17122         conditional constraits (wg).
17123         (mov<mode> for DFmode/DDmode): Likewise.
17124         (DFmode/DDmode splitters): Likewise.
17125         (movdf_hardfloat32): Likewise.
17126         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17127         (movdf_softfloat32): Likewise.
17128         (movdf_hardfloat64_mfpgpr): Likewise.
17129         (movdf_hardfloat64): Likewise.
17130         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17131         (movdf_softfloat64): Likewise.
17132         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17133         (reload_<mode>_load): Move to later in the file so they aren't in
17134         the middle of the floating point move insns.
17135         (reload_<mode>_store): Likewise.
17137         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17138         constraint.
17140         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17141         constraint if -mdebug=reg.
17142         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17143         Enable using dd reload support if needed.
17145         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17146         binary and decimal floating point moves in rs6000.md.
17147         (movtd_internal): Likewise.
17149         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17150         decimal floating point moves.
17151         (movtf): Likewise.
17152         (movtf_internal): Likewise.
17153         (mov<mode>_internal, TDmode/TFmode): Likewise.
17154         (movtf_softfloat): Likewise.
17155         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17157         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17158         movdi_internal64, using wg constraint for move direct operations.
17159         (movdi_internal64): Likewise.
17161         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17162         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17163         the various virtual registers. Use GPR/FPR first/last values,
17164         instead of hard coding the register numbers.  Print which modes
17165         have reload functions registered.
17166         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17167         settings before/after setting cpu, target and subtarget settings.
17168         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17169         and for secondary reload failures in rs6000_secondary_reload_inner.
17170         (rs6000_secondary_reload_fail): Likewise.
17171         (rs6000_secondary_reload_inner): Likewise.
17173         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17174         macros for first/last GPR and FPR registers.
17175         (LAST_GPR_REGNO): Likewise.
17176         (FIRST_FPR_REGNO): Likewise.
17177         (LAST_FPR_REGNO): Likewise.
17179         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17180         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17181         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17182         (vcond<mode><mode>): Likewise.
17183         (vcondu<mode><mode>): Likewise.
17184         (vector_gtu<mode>): Likewise.
17185         (vector_gte<mode>): Likewise.
17186         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17187         to prevent the compiler from converting DImode operations to TImode.
17188         (ior<mode>3): Likewise.
17189         (and<mode>3): Likewise.
17190         (one_cmpl<mode>2): Likewise.
17191         (nor<mode>3): Likewise.
17192         (andc<mode>3): Likewise.
17194         * config/rs6000/constraints.md (wt constraint): New constraint
17195         that returns VSX_REGS if TImode is allowed in VSX registers.
17197         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17198         constant under VSX.
17200         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17201         similar to TImode, but it is restricted to being in the GPRs.
17203         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17204         TImode to occupy a single VSX register.
17206         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17207         -mvsx-timode for power7/power8.
17208         (power7 cpu): Likewise.
17209         (power8 cpu): Likewise.
17211         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17212         sure that TFmode/TDmode take up two registers if they are ever
17213         allowed in the upper VSX registers.
17214         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17215         registers.
17216         (rs6000_init_hard_regno_mode_ok): Likewise.
17217         (rs6000_debug_reg_global): Add debugging for PTImode and wt
17218         constraint.  Print if LRA is turned on.
17219         (rs6000_option_override_internal): Give an error if -mvsx-timode
17220         and VSX is not enabled.
17221         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
17222         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17223         to reg+offset addressing.  Use PTImode when checking offset
17224         addresses for validity.
17225         (reg_offset_addressing_ok_p): Likewise.
17226         (rs6000_legitimate_offset_address_p): Likewise.
17227         (rs6000_legitimize_address): Likewise.
17228         (rs6000_legitimize_reload_address): Likewise.
17229         (rs6000_legitimate_address_p): Likewise.
17230         (rs6000_eliminate_indexed_memrefs): Likewise.
17231         (rs6000_emit_move): Likewise.
17232         (rs6000_secondary_reload): Likewise.
17233         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
17234         reloads to fpr registers to continue to use reg+offset addressing,
17235         but 64-bit reloads to altivec registers need reg+reg addressing.
17236         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17237         it.  Treat LO_SUM like a PLUS operation.
17238         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17239         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17240         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17241         registers to share a register with a smaller sized type, since VSX
17242         puts scalars in the upper 64-bits.
17243         (print_operand): Add support for PTImode.
17244         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17245         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17246         registers, but don't have arithmetic support.
17247         (rs6000_memory_move_cost): Add test for VSX.
17248         (rs6000_opt_masks): Add -mvsx-timode.
17250         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17251         for TImode.
17252         (VSs): Likewise.
17253         (VSr): Use wt constraint for TImode.
17254         (VSv): Drop TImode support.
17255         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17256         (vsx_movti_64bit): Likewise.
17257         (vsx_movti_32bit): Likewise.
17258         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17259         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17260         one '?' on the appropriate output constraint.  Do not allow TImode
17261         logical operations on 32-bit systems.
17262         (vsx_ior<mode>3): Likewise.
17263         (vsx_xor<mode>3): Likewise.
17264         (vsx_one_cmpl<mode>2): Likewise.
17265         (vsx_nor<mode>3): Likewise.
17266         (vsx_andc<mode>3): Likewise.
17267         (vsx_concat_<mode>): Likewise.
17268         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17270         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17271         OPTION_MASK_VSX_TIMODE.
17272         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17273         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17275         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17276         (TI2 iterator): New iterator for TImode, PTImode.
17277         (wd mode attribute): Add values for vector types.
17278         (movti_string): Replace TI move operations with operations for TImode
17279         and PTImode.  Add support for TImode being allowed in VSX registers.
17280         (mov<mode>_string, TImode/PTImode): Likewise.
17281         (movti_ppc64): Likewise.
17282         (mov<mode>_ppc64, TImode/PTImode): Likewise.
17283         (TI mode splitters): Likewise.
17285         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17286         constraint.
17288 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
17290         PR tree-optimization/56355
17291         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17292         Also handle integers with undefined overflow.
17294 2013-03-20  Catherine Moore  <clm@codesourcery.com>
17295             Maciej W. Rozycki  <macro@codesourcery.com>
17296             Tom de Vries  <tom@codesourcery.com>
17297             Nathan Sidwell  <nathan@codesourcery.com>
17298             Iain Sandoe  <iain@codesourcery.com>
17299             Nathan Froyd  <froydnj@codesourcery.com>
17300             Chao-ying Fu  <fu@mips.com>
17302         * doc/extend.texi: (micromips, nomicromips, nocompression):
17303         Document new function attributes.
17304         * doc/invoke.texi (minterlink-compressed, mmicromips,
17305         m14k, m14ke, m14kec): Document new options.
17306         (minterlink-mips16): Update documentation.
17307         * doc/md.texi (ZC, ZD): Document new constraints.
17308         * configure.ac (gcc_cv_as_micromips): Check if linker
17309         supports the .set micromips directive.
17310         * configure: Regenerate.
17311         * config.in: Regenerate.
17312         * config/mips/mips-tables.opt: Regenerate.
17313         * config/mips/micromips.md: New file.
17314         * constraints.md (ZC, ZD): New constraints.
17315         * config/mips/predicates.md (movep_src_register): New predicate.
17316         (movep_src_operand): New predicate.
17317         (non_volatile_mem_operand): New predicate.
17318         * config/mips/mips.md (multimem): New type.
17319         (length): Differentiate between 17-bit and 18-bit branch offsets.
17320         (MOVEP1, MOVEP2): New mode iterator.
17321         (mov_<load>l): Use ZC constraint.
17322         (mov_<load>r): Likewise.
17323         (mov_<store>l): Likewise.
17324         (mov_<store>r): Likewise.
17325         (*branch_equality<mode>_inverted): Add microMIPS support.
17326         (*branch_equality<mode>): Likewise.
17327         (*jump_absolute): Likewise.
17328         (indirect_jump_<mode>): Likewise.
17329         (tablejump_<mode>): Likewise.
17330         (<optab>_internal): Likewise.
17331         (sibcall_internal): Likewise.
17332         (sibcall_value_internal): Likewise.
17333         (prefetch): Use constraint ZD.
17334         * config/mips/mips.opt (minterlink-compressed): New option.
17335         (minterlink-mips16): Now an alias for minterlink-compressed.
17336         (mmicromips): New option.
17337         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17338         (compare_and_swap_12): Likewise.
17339         (sync_add<mode>): Likewise.
17340         (sync_<optab>_12): Likewise.
17341         (sync_old_<optab>_12): Likewise.
17342         (sync_new_<optab>_12): Likewise.
17343         (sync_nand_12): Likewise.
17344         (sync_old_nand_12): Likewise.
17345         (sync_new_nand_12): Likewise.
17346         (sync_sub<mode>): Likewise.
17347         (sync_old_add<mode>): Likewise.
17348         (sync_old_sub<mode>): Likewise.
17349         (sync_new_add<mode>): Likewise.
17350         (sync_new_sub<mode>): Likewise.
17351         (sync_<optab><mode>): Likewise.
17352         (sync_old_<optab><mode>): Likewise.
17353         (sync_new_<optab><mode>): Likewise.
17354         (sync_nand<mode>): Likewise.
17355         (sync_old_nand<mode>): Likewise.
17356         (sync_new_nand<mode>): Likewise.
17357         (sync_lock_test_and_set<mode>): Likewise.
17358         (test_and_set_12): Likewise.
17359         (atomic_compare_and_swap<mode>): Likewise.
17360         (atomic_exchange<mode>_llsc): Likewise.
17361         (atomic_fetch_add<mode>_llsc): Likewise.
17362         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17363         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17364         (umips_save_restore_pattern_p): Likewise.
17365         (umips_load_store_pair_p): Likewise.
17366         (umips_output_load_store_pair): Likewise.
17367         (umips_movep_target_p): Likewise.
17368         (umips_12bit_offset_address_p): Likewise.
17369         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17370         (mips_base_mips16): Rename this...
17371         (mips_base_compression_flags): ...to this. Update all uses.
17372         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17373         (mips_mips16_decl_p): Delete.
17374         (mips_nomips16_decl_p): Delete.
17375         (mips_get_compress_on_flags): New function.
17376         (mips_get_compress_off_flags): New function.
17377         (mips_get_compress_mode): New function.
17378         (mips_get_compress_on_name): New function.
17379         (mips_get_compress_off_name): New function.
17380         (mips_insert_attributes): Support multiple compression types.
17381         (mips_merge_decl_attributes): Likewise.
17382         (umips_12bit_offset_address_p): New function.
17383         (mips_start_function_definition): Emit .set micromips directive.
17384         (mips_call_may_need_jalx_p): New function.
17385         (mips_function_ok_for_sibcall): Add microMIPS support.
17386         (mips_print_operand_punctuation): Support short delay slots and
17387         compact jumps.
17388         (umips_swm_mask, umips_swm_encoding): New.
17389         (umips_build_save_restore): New function.
17390         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17391         (was_mips16_p): Remove.
17392         (old_compression_mode): New.
17393         (mips_set_compression_mode): New function.
17394         (mips_set_current_function): Add microMIPS support.
17395         (mips_option_override): Likewise.
17396         (umips_save_restore_pattern_p): New function.
17397         (umips_output_save_restore): New function.
17398         (umips_load_store_pair_p_1): New function.
17399         (umips_load_store_pair_p): New function.
17400         (umips_output_load_store_pair_1): New function.
17401         (umips_output_load_store_pair): New function.
17402         (umips_movep_target_p) New function.
17403         (mips_prepare_pch_save): Add microMIPS support.
17404         * config/mips/mips.h (TARGET_COMPRESSION): New.
17405         (TARGET_CPU_CPP_BUILTINS): Update macro
17406         to use new compression flags and to support microMIPS.
17407         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17408         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17409         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17410         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17411         (ASM_SPEC): Support mmicromips and mno-micromips.
17412         (M16STORE_REG_P): New macro.
17413         (MIPS_CALL): Support TARGET_MICROMIPS.
17414         (MICROMIPS_J): New macro.
17415         (mips_base_mips16): Rename this...
17416         (mips_base_compression_flags): ...to this.
17417         (UMIPS_12BIT_OFFSET_P): New macro.
17418         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17419         (MULTILIB_DIRNAMES): Likewise.
17420 2013-03-20  Richard Biener  <rguenther@suse.de>
17422         PR tree-optimization/56661
17423         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17424         the result does not have to be distinct.
17426 2013-03-20  Richard Biener  <rguenther@suse.de>
17428         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17429         remap_gimple_op_r.
17431 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17432             Steven Bosscher  <steven@gcc.gnu.org>
17434         PR rtl-optimization/56605
17435         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17437 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17439         PR bootstrap/56656
17440         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17441         that require movd instead of movq.
17443 2013-03-20  Richard Biener  <rguenther@suse.de>
17445         * tree-ssa-structalias.c (struct variable_info): Add pointer
17446         to the first field of an aggregate with sub-vars.  Make
17447         this and the pointer to the next subfield its ID.
17448         (vi_next): New function.
17449         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17450         storedanything_id, integer_id): Increment by one.
17451         (new_var_info, get_call_vi, lookup_call_clobber_vi,
17452         get_call_clobber_vi): Adjust.
17453         (solution_set_expand): Simplify and speedup.
17454         (solution_set_add): Inline into ...
17455         (set_union_with_increment): ... this.  Adjust accordingly.
17456         (do_sd_constraint): Likewise.
17457         (do_ds_constraint): Likewise.
17458         (do_complex_constraint): Simplify.
17459         (build_pred_graph): Adjust.
17460         (solve_graph): Likewise.  Simplify and speedup.
17461         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17462         get_constraint_for_component_ref, get_constraint_for_1,
17463         first_vi_for_offset, first_or_preceding_vi_for_offset,
17464         create_function_info_for, create_variable_info_for_1,
17465         create_variable_info_for, intra_create_variable_infos): Adjust.
17466         (init_base_vars): Push NULL for ID zero.
17467         (compute_points_to_sets): Adjust.
17469 2013-03-20  Richard Biener  <rguenther@suse.de>
17471         * cfgloop.c (verify_loop_structure): Streamline and avoid
17472         ICEing on corrupt loop tree.
17473         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17474         loop tree.
17476 2013-03-20  Richard Biener  <rguenther@suse.de>
17478         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17479         check whether an SSA update is needed.
17481 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
17483         * config/mips/constraints.md (T): Rename to...
17484         (Yf): ...this.
17485         (U): Rename to...
17486         (Yd): ...this.
17487         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17488         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17490 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17492         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17493         (*subsi3_carryin_uxtw): Likewise.
17495 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17497         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17498         (*rorsi3_insn_uxtw): Likewise.
17500 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17502         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17503         (*extrsi5_insn_uxtw): Likewise.
17505 2013-03-19  Richard Biener  <rguenther@suse.de>
17507         PR tree-optimization/56273
17508         * passes.c (init_optimization_passes): Move second VRP after DOM.
17510 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
17512         * config/i386/i386.md (*movti_internal): Merge from
17513         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
17514         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
17515         nox64 isa attributes.
17517 2013-03-18  Richard Biener  <rguenther@suse.de>
17519         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17520         (unite): Likewise.
17521         (merge_node_constraints): Likewise.
17522         (build_succ_graph): Likewise.
17523         (valid_graph_edge): Inline into single caller.
17524         (unify_nodes): Likewise.  Use bitmap_set_bit return value
17525         and cache varinfo.
17526         (scc_visit): Fix formatting and variable use.
17527         (do_sd_constraint): Use gcc_checking_assert.
17528         (do_ds_constraint): Likewise.
17529         (do_complex_constraint): Likewise.
17530         (condense_visit): Likewise.  Cleanup.
17531         (dump_pred_graph): New function.
17532         (perform_var_substitution): Dump the pred-graph before
17533         variable substitution.
17534         (find_equivalent_node): Use gcc_checking_assert.
17535         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17537 2013-03-18  Richard Biener  <rguenther@suse.de>
17539         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17540         Remove cond_expr_stmt_list argument and do not gimplify the
17541         built expression.
17542         (vect_loop_versioning): Adjust.
17543         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17544         Cleanup to use less temporaries.
17545         (vect_create_data_ref_ptr): Cleanup.
17547 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17549         PR tree-optimization/56635
17550         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17551         require types_compatible_p types.
17553 2013-03-18  Nick Clifton  <nickc@redhat.com>
17555         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
17556         spurious backslash.
17558         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
17559         Add missing line to comment describing function.
17561 2013-03-18  Richard Biener  <rguenther@suse.de>
17563         PR tree-optimization/56210
17564         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17565         Handle string / character search functions.
17566         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
17568 2013-03-18  Richard Biener  <rguenther@suse.de>
17570         PR middle-end/56483
17571         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
17572         and implement properly.
17573         * gimple.h (gimple_cond_single_var_p): Remove.
17575 2013-03-18  Richard Biener  <rguenther@suse.de>
17577         * tree-data-ref.h (find_data_references_in_loop): Declare.
17578         * tree-data-ref.c (get_references_in_stmt): Use a stack
17579         vector pre-allocated in the callers.
17580         (find_data_references_in_stmt): Adjust.
17581         (graphite_find_data_references_in_stmt): Likewise.
17582         (create_rdg_vertices): Likewise.
17583         (find_data_references_in_loop): Export.
17584         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17585         Compute dependences here...
17586         (vect_analyze_data_refs): ...not here.  When we encounter
17587         a non-vectorizable data reference in basic-block vectorization
17588         truncate the data reference vector.  Do not bother to
17589         fixup data-dependence information for gather loads.
17590         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
17591         of data references, as reported.
17593 2013-03-18  Richard Biener  <rguenther@suse.de>
17595         PR tree-optimization/3713
17596         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
17597         has_constants and expr.
17598         (stmt_has_constants): Properly valueize SSA names when deciding
17599         whether the stmt has constants.
17601 2013-03-18  Richard Biener  <rguenther@suse.de>
17603         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
17604         whole function when there is nothing to do.
17605         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
17606         * tree-vectorizer.c (vectorize_loops): Update virtual and
17607         loop-closed SSA once.
17608         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
17610 2013-03-18  Richard Biener  <rguenther@suse.de>
17612         PR middle-end/56113
17613         * domwalk.c (bb_postorder): New global static.
17614         (cmp_bb_postorder): New function.
17615         (walk_dominator_tree): Replace scheme imposing an order for
17616         visiting dominator sons by one sorting them at the time they
17617         are pushed on the stack.
17619 2013-03-18  Richard Biener  <rguenther@suse.de>
17621         PR tree-optimization/39326
17622         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
17623         (struct mem_ref): Replace mem member with ao_ref typed member.
17624         (MEM_ANALYZABLE): Adjust.
17625         (memref_eq): Likewise.
17626         (mem_ref_alloc): Likewise.
17627         (gather_mem_refs_stmt): Likewise.
17628         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
17629         (execute_sm_if_changed_flag_set): Adjust.
17630         (execute_sm): Likewise.
17631         (ref_always_accessed_p): Likewise.
17632         (refs_independent_p): Likewise.
17633         (can_sm_ref_p): Likewise.
17635 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17637         PR c/56566
17638         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
17639         return 1 even for !unsignedp.
17641 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17643         * config/i386/i386.md (isa): Add x64 and nox64.
17644         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
17645         (*pushtf): Enable *roF alternative for x64 isa only.
17646         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
17647         mode attribute of integer alternatives to DImode for TARGET_64BIT.
17648         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
17649         (*movtf_internal): Merge from *movtf_internal_rex64 and
17650         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
17651         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
17652         nox64 isa attributes.
17653         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
17654         nox64 isa attributes.
17655         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
17657 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17659         * config/alpha/alpha.c (TARGET_LRA_P): New define.
17661 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
17663         PR target/56640
17664         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
17665         class names.  Remove trailing comma after "ALL_REGS".
17667 2013-03-16  Jan Hubicka  <jh@suse.cz>
17669         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
17670         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
17671         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
17672         of cgraph_get_create_node.
17673         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17675 2013-03-16  Jason Merrill  <jason@redhat.com>
17677         PR debug/49090
17678         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
17679         with DW_AT_default_value.
17681 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
17683         * BASE-VER: Set to 4.9.0.
17685 2013-03-14  Andi Kleen  <ak@linux.intel.com>
17687         PR target/56619
17688         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
17689         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
17690         Document _x* TSX intrinsics.
17692 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
17693             David Holsgrove  <david.holsgrove@xilinx.com>
17695         * configure.ac: Add MicroBlaze TLS support detection.
17696         * configure: Regenerate.
17697         * config/microblaze/microblaze-protos.h
17698         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
17699         symbol_mentioned_p, label_mentioned_p): Add prototypes.
17700         * config/microblaze/microblaze.c (microblaze_address_type): Add
17701         ADDRESS_TLS and tls_reloc address types.
17702         (microblaze_address_info): Add tls_reloc.
17703         (TARGET_HAVE_TLS): Define.
17704         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
17705          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
17706          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
17707         load_tls_operand,  microblaze_call_tls_get_addr,
17708         microblaze_legitimize_tls_address): New functions.
17709         (microblaze_classify_unspec): Handle UNSPEC_TLS.
17710         (get_base_reg): Use microblaze_tls_symbol_p.
17711         (microblaze_classify_address): Handle TLS.
17712         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
17713         label_mentioned_p and microblaze_tls_referenced_p.
17714         (microblaze_legitimize_address): Handle TLS.
17715         (microblaze_address_insns): Handle ADDRESS_TLS.
17716         (pic_address_needs_scratch): Handle TLS.
17717         (print_operand_address): Handle TLS.
17718         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
17719         (microblaze_expand_move): Handle TLS.
17720         (microblaze_legitimate_constant_p): Check
17721         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
17722         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
17723         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
17724         (PIC_OFFSET_TABLE_REGNUM): Set.
17725         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
17726         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
17727         (addsi3, movsi_internal2, movdf_internal): Update constraints
17728         * config/microblaze/predicates.md (arith_plus_operand): Define
17729         (move_operand): Redefine as move_src_operand,
17730         check microblaze_tls_referenced_p.
17732 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17734         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
17735         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
17737 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17739         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
17740         CC mode for AND.
17742 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
17744         PR tree-optimization/53265
17745         * common.opt (Waggressive-loop-optimizations): New option.
17746         * tree-ssa-loop-niter.c: Include tree-pass.h.
17747         (do_warn_aggressive_loop_optimizations): New function.
17748         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
17749         if number_of_latch_executions returned constant.
17750         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
17751         early.  If number_of_latch_executions returned constant, set
17752         nb_iterations_upper_bound back to it.
17753         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
17754         field.
17755         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
17756         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
17758         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
17759         (MULTILIB_OSDIRNAMES): Set.
17760         * genmultilib: If defaultosdirname doesn't start with :: , set
17761         defaultosdirname2 instead, clear it and emit two . multilib_raw
17762         entries instead of just one.
17764 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
17766         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17767         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
17768         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17769         (SUBTARGET_OVERRIDE_OPTIONS): New.
17771 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
17773         PR target/49880
17774         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
17775         (musermode): Convert to Var(TARGET_USERMODE).
17776         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
17777         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
17778         * config/sh/sh.c (sh_option_override): Use
17779         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
17780         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
17781         condition.
17782         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
17783         TARGET_SH4.
17784         (udivsi3_i4_single, divsi3_i4_single): Use
17785         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
17787 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
17789         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
17790         default setting.
17792 2013-03-13  Richard Biener  <rguenther@suse.de>
17794         PR tree-optimization/56608
17795         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
17796         calls when vectorizing basic-blocks.
17798 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
17800         PR plugins/45078
17801         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
17802         tm_file.
17804 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
17806         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
17808 2013-03-11  Jan Hubicka  <jh@suse.cz>
17810         PR lto/56557
17811         * lto-streamer-out.c (output_symbol_p): Skip references from
17812         constructors of external variables.
17814 2013-03-11  Jan Hubicka  <jh@suse.cz>
17816         PR middle-end/56571
17817         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
17818         from pseudos.
17819         * emit-rtl.c (verify_rtx_sharing): Likewise.
17820         (copy_insn_1): Likewise.
17821         * rtl.c (copy_rtx): Likewise.
17823 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
17825         PR target/56591
17826         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
17827         output_operand_lossage message.
17829 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
17831         PR target/56470
17832         * arm.c (shift_op): Validate RTL pattern on the fly.
17833         (arm_print_operand, case 'S'): Don't use shift_operator to validate
17834         the RTL.
17836 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17838         PR target/56347
17839         * config/pa/pa.md (call_value): Check for calls to powf and direct to
17840         new call patterns that clobber %fr12.
17841         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
17842         split and postreload patterns.
17843         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
17844         registers %fr12 and %fr12R as call used.
17846 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
17848         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
17849         (canon_address, record_store, replace_read, check_mem_read_rtx,
17850         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
17851         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
17852         rest_of_handle_dse): Likewise.
17854 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
17856         PR middle-end/56524
17857         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
17858         Add base_optabs.
17859         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
17860         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
17861         (save_optabs_if_changed): Replace with...
17862         (init_tree_optimization_optabs): ...this.
17863         * optabs.c (save_optabs_if_changed): Rename to...
17864         (init_tree_optimization_optabs): ...this.  Take the optimization node
17865         as argument.  Do nothing if the base optabs are already correct.
17866         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
17867         to recompute optabs.
17868         * function.h (function): Remove optabs field.
17869         * function.c (invoke_set_current_function_hook): Call
17870         init_tree_optimization_optabs.  Use the result to initialize
17871         this_fn_optabs.
17873 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
17875         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
17876         if GTMA_HAS_NO_INSTRUMENTATION.
17877         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
17878         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
17879         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
17880         * gimple-pretty-print.c (dump_gimple_transaction): Handle
17881         GTMA_HAS_NO_INSTRUMENTATION.
17883 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
17885         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
17886         libasan_preinit.o.
17888 2013-03-08  Marek Polacek  <polacek@redhat.com>
17889             Jakub Jelinek  <jakub@redhat.com>
17891         PR tree-optimization/56478
17892         * predict.c (is_comparison_with_loop_invariant_p): Change the
17893         type of loop_step to tree.
17894         (predict_loops): Adjust.
17895         (predict_iv_comparison): Perform the computations on double_ints.
17897 2013-03-08  Richard Biener  <rguenther@suse.de>
17899         PR tree-optimization/56570
17900         * tree-cfg.c (verify_expr_location_1): Verify locations for
17901         DECL_DEBUG_EXPR.
17902         * tree-sra.c (create_access_replacement): Strip locations
17903         from DECL_DEBUG_EXPRs.
17905 2013-03-08  Richard Biener  <rguenther@suse.de>
17907         * tree-inline.c (expand_call_inline): Do not associate
17908         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
17909         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
17911 2013-03-08  Richard Biener  <rguenther@suse.de>
17913         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
17914         or block changes with -Og.  Fix for location / block encoding
17915         changes and PHI arguments with locations.
17917 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
17919         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
17920         for all counters.
17921         (struct output_info): Likewise.
17922         (register_overhead): Remove bad gcc_assert.
17923         (bitmap_find_bit): If there is only a single bitmap element, do not
17924         count a miss as a search.
17925         (print_statistics): Update for counter type changes.
17926         (dump_bitmap_statistics): Likewise.  Print headers such that they
17927         are properly lined up with the printed counters.
17929 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
17931         PR tree-optimization/56559
17932         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
17933         check that it has only a single use.
17935 2013-03-07  Richard Biener  <rguenther@suse.de>
17937         * doc/invoke.texi (fwhole-program): Discourage use in combination
17938         with -flto.
17940 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
17942         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
17944         PR tree-optimization/56539
17945         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
17946         instead of GSI_CONTINUE_LINKING as last argument to
17947         force_gimple_operand_gsi.  Adjust function comment.
17949         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
17950         aarch64-cores.def.
17952         PR middle-end/56548
17953         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
17954         promoted mode, convert the result back to the original mode.
17956 2013-03-06  Richard Biener  <rguenther@suse.de>
17958         PR middle-end/56294
17959         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
17960         (insert_updated_phi_nodes_compare_uids): New function.
17961         (update_ssa): Sort symbols_to_rename after UID before
17962         traversing it to insert PHI nodes.
17964 2013-03-06  Richard Biener  <rguenther@suse.de>
17966         PR middle-end/50494
17967         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
17968         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
17970         Revert
17971         2013-02-13  Richard Biener  <rguenther@suse.de>
17973         PR lto/50494
17974         * varasm.c (output_constant_def_1): Get the decl representing
17975         the constant as argument.
17976         (output_constant_def): Wrap output_constant_def_1.
17977         (make_decl_rtl): Use output_constant_def_1 with the decl
17978         representing the constant.
17979         (build_constant_desc): Optionally re-use a decl already
17980         representing the constant.
17981         (tree_output_constant_def): Adjust.
17983 2013-03-06  Joey Ye  <joey.ye@arm.com>
17985         PR lto/50293
17986         * gcc.c (convert_white_space): New function.
17987         (main): Handles white space in function name.
17989 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
17991         PR target/56529
17992         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
17993         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
17994         to SH_DIV_CALL_TABLE for TARGET_SH2.
17995         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
17996         list.
17997         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
17998         call-table options.
18000 2013-03-05  Sterling Augustine  <saugustine@google.com>
18001             Cary Coutant  <ccoutant@google.com>
18003         PR debug/55364
18004         * dwarf2out.c (resolve_addr): Don't call
18005         remove_loc_list_addr_table_entries a second time for the same
18006         expression.
18008 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18010         PR debug/56510
18011         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18012         (avoid_complex_debug_insns): New function.
18013         (expand_debug_locations): Call it.
18015         PR rtl-optimization/56484
18016         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18017         lifetimes of hard registers on small register class machines.
18019 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18021         * config/microblaze/microblaze-protos.h: Rename
18022         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18023         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18024         fast_interrupt.
18025         (microblaze_fast_interrupt_function_p): New function.
18026         (microblaze_is_interrupt_handler): Rename to
18027         microblaze_is_interrupt_variant and add fast_interrupt check.
18028         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18029         (save_restore_insns): Likewise.
18030         (compute_frame_size): Likewise.
18031         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18032         (microblaze_globalize_label): Likewise.
18033         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18034         * config/microblaze/microblaze.md: Use wrapper
18035         microblaze_is_interrupt_variant.
18037 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18039         * sdbout.c (sdbout_one_type): Switch to current function's section
18040         supporting cold/hot.
18042 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18044         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18045         -mxl-reorder.
18047 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18049         PR middle-end/56461
18050         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18051         if VALGRIND_GET_VBITS is defined, temporarily make object
18052         memory all defined, and restore previous valgrind addressability
18053         and definability afterwards.  Free this_object at the end.
18055         PR middle-end/56461
18056         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18057         right before calling lra_create_live_ranges, also call it
18058         when clearing live_p.  Only call lra_clear_live_ranges
18059         at the end if live_p.
18061         PR middle-end/56461
18062         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18064 2013-03-05  Richard Biener  <rguenther@suse.de>
18066         PR tree-optimization/56521
18067         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18068         value-id.
18070 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18072         PR c++/55135
18073         * except.h (remove_unreachable_eh_regions): New prototype.
18074         * except.c (remove_eh_handler_splicer): New function, split out
18075         of remove_eh_handler.
18076         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18077         warning about running it on many EH regions one at a time.
18078         (remove_unreachable_eh_regions_worker): New function, walk the
18079         EH tree in depth-first order and remove non-marked regions.
18080         (remove_unreachable_eh_regions): New function.
18081         * tree-eh.c (mark_reachable_handlers): New function, split out
18082         from remove_unreachable_handlers.
18083         (remove_unreachable_handlers): Use mark_reachable_handlers and
18084         remove_unreachable_eh_regions.
18085         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18086         and remove_unreachable_eh_regions.
18088 2013-03-05  Richard Biener  <rguenther@suse.de>
18090         PR middle-end/56525
18091         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18092         not freeing them until the end.
18094 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18096         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18098 2013-03-05  Richard Biener  <rguenther@suse.de>
18100         PR tree-optimization/56270
18101         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18102         of loads after scheduling an SLP instance.
18104 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18106         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18107         tic6x.exp.
18108         (check_gcc_parallelize): Run guality.exp as a separate job from
18109         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18110         struct-layout-1.exp with stackalign.exp.
18112         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18114         PR middle-end/56461
18115         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18116         load_index sbitmap even if some bit in it isn't set.
18118         PR middle-end/56461
18119         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18120         (discover_iteration_bound_by_body_walk): Change queues to
18121         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18122         spelling in comment.  Call safe_push on queues[bound_index] directly.
18123         Release queues[queue_index] in every iteration unconditionally.
18124         Release bounds vector.
18126         PR middle-end/56461
18127         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18128         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18129         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18130         inner_phis vector.
18132 2013-03-05  Richard Biener  <rguenther@suse.de>
18134         PR lto/56515
18135         * tree-inline.c (remap_blocks_to_null): New function.
18136         (expand_call_inline): When expanding a call stmt without
18137         an associated block inline remap all callee blocks to NULL.
18139 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18141         PR rtl-optimization/56494
18142         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18143         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18144         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18146         PR middle-end/56461
18147         * sel-sched-ir.c (free_sched_pools): Release
18148         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18149         if succs_info_pool.max_top isn't -1.
18151         PR bootstrap/56509
18152         * opts.c (opts_obstack, opts_concat): Moved to...
18153         * opts-common.c (opts_obstack, opts_concat): ... here.
18155 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18157         PR middle-end/56461
18158         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18160 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18162         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18163         all appropriate places.
18165 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18167         PR tree-optimization/56424
18168         * ipa-split.c (split_function): Do not set the RSO flag if result is
18169         not by reference and its type is a register type.
18171 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18173         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18174         (microblaze_legitimate_pic_operand): Likewise
18175         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18176         new function microblaze_legitimate_pic_operand
18177         * config/microblaze/microblaze-protos.h
18178         (microblaze_legitimate_pic_operand): Declare.
18180 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18182         * config/microblaze/predicates.md (call_insn_simple_operand):
18183         New predicate for supported rtx code types.
18184         * config/microblaze/microblaze.md (call_internal1): Use
18185         call_insn_simple_operand predicate.
18187 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18189         PR middle-end/56461
18190         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18191         partitions.ordered_remove.
18193         PR middle-end/56461
18194         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18195         vec_oprnds0.create (1) for modifier == NONE.
18197         PR middle-end/56461
18198         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18199         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18200         vec_oprnds1 right before pushing anything to it for
18201         scalar_shift_arg.
18203         PR middle-end/56461
18204         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18205         set nbbs to 0 instead of having separate code path.
18206         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18207         instead of false as last argument if returning NULL.
18209 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18211         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18212         the attribute is now called "target" instead of "option".
18213         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18214         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18215         attribute/pragma name for TARGET_OPTION_VALID_P and
18216         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18217         * doc/tm.texi: Regenerated.
18219 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
18221         * config/microblaze/microblaze.c:
18222         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18223         * config/microblaze/microblaze.h: Add -mxl-reorder to
18224         DRIVER_SELF_SPECS.
18225         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18226         instructions emitted if TARGET_REORDER.
18227         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18228         or 0 for -m/-mno case, but initialises as 2 to detect default use case
18229         separately.
18231 2013-03-01  Xinliang David Li  <davidxl@google.com>
18233         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18234         walk length.
18236 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
18238         PR middle-end/56461
18239         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18240         vector even when returning true.  Fix up function comment formatting.
18242         PR middle-end/56461
18243         * ira-build.c (ira_loop_nodes_count): New variable.
18244         (create_loop_tree_nodes): Initialize it.
18245         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18247         PR middle-end/56461
18248         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18249         method on dr_chain and result_chain.
18250         * tree-vect-stmts.c (vectorizable_store): Only call
18251         result_chain.create if j == 0.
18253         PR middle-end/56461
18254         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18255         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18256         before overwriting it.
18258 2013-03-01  Tobias Burnus  <burnus@net-b.de>
18260         * doc/extended.texi (C Extensions): Change order in @menu
18261         to match @node.
18262         (Other MIPS Built-in Functions): Move last MIPS entry before
18263         "picoChip Built-in Functions".
18264         (SH Built-in Functions): Move after RX Built-in Functions.
18265         * doc/gcc.texi (Introduction): Change order in @menu
18266         to match @node.
18267         * doc/md.texi (Constraints): Ditto.
18268         * gty.texi (Type Information): Ditto.
18269         (User-provided marking routines for template types): Make
18270         subsection.
18271         * doc/invoke.texi (AArch64 Options): Move before
18272         "Adapteva Epiphany Options".
18274 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
18275             Jakub Jelinek  <jakub@redhat.com>
18277         PR sanitizer/56454
18278         * asan.c (gate_asan): Lookup no_sanitize_address instead of
18279         no_address_safety_analysis attribute.
18280         * doc/extend.texi (no_address_safety_attribute): Rename to
18281         no_sanitize_address attribute, mention no_address_safety_analysis
18282         attribute as deprecated alias.
18284 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
18286         PR middle-end/56461
18287         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18288         type to vec<vec<tree> > *.
18289         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
18290         to be vec<tree> instead of vec<tree> *, set vec_defs
18291         to vNULL and call vec_defs.create (number_of_vects), adjust other
18292         uses of vec_defs.
18293         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18294         vectorizable_condition): Adjust vect_get_slp_defs callers.
18296 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18298         * config/aarch64/aarch64.c
18299         (aarch64_float_const_representable): Remove unused variable.
18301 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18303         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18305 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18307         * config/aarch64/aarch64-builtins.c
18308         (aarch64_init_simd_builtins): Make static.
18310 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18312         * config/aarch64/aarch64.c
18313         (aarch64_simd_make_constant): Make static.
18315 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18317         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18318         with no initialization to the RHS of debug statements.
18320 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18322         PR tree-optimization/56294
18323         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18324         Adjust dumping.
18325         (get_access_replacement): Do not call create_access_replacement.
18326         Assert a replacement exists.
18327         (get_repl_default_def_ssa_name): Create the replacement declaration
18328         itself.
18330 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18332         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18333         final_end_function.
18335 2013-02-28  Marek Polacek  <polacek@redhat.com>
18337         PR rtl-optimization/56466
18338         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18339         if we're changing a loop.
18340         (peel_loops_completely): Likewise.
18342 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
18344         PR c++/55813
18345         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18347 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18349         PR target/56445
18350         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18351         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18352         INTX_FTYPE_FX, FX_FTYPE_INTX.
18353         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18355 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18357         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18358         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18359         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18360         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18361         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18362         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18363         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18364         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18365         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18366         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18367         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18368         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18369         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18370         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18371         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18372         (avrxmega6): Increase max flash segments from 5 to 6.
18373         * config/avr/t-multilib: Regenerate.
18374         * config/avr/avr-tables.opt: Regenerate.
18375         * doc/avr-mmcu.texi: Regenerate.
18377 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18379         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18380         (avr_device_to_arch): Rename to avr_device_to_ld.
18381         (avr_device_to_as): New prototype.
18382         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18383         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18384         * config/avr/driver-avr.c (avr_device_to_as): New.
18385         (avr_device_to_arch): Rename to avr_device_to_ld.
18387 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18389         PR middle-end/56461
18390         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18391         method on dr_chain and result_chain.
18393         PR middle-end/56461
18394         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18395         pointer_set_destroy on not_executed_last_iteration.
18397         PR middle-end/56461
18398         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18400         PR middle-end/56461
18401         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18402         FOR_EACH_DEFINED_FUNCTION when freeing state.
18404         PR middle-end/56461
18405         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18406         pool_free.
18407         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18408         overwriting it.
18410         PR middle-end/56461
18411         * ipa-cp.c (decide_whether_version_node): Call vec_free on
18412         known_aggs[i].items and release known_aggs vector.
18414         PR middle-end/56461
18415         * ipa-reference.c (propagate): Free node_info even for alias nodes.
18417 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18419         * config/microblaze/microblaze.c (microblaze_emit_compare):
18420         Use xor for EQ/NE comparisions.
18421         * config/microblaze/microblaze.md (cstoresf4): Add constraints
18422         (cbranchsf4): Adjust operator to comparison_operator.
18424 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18426         PR middle-end/56461
18427         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18428         vector.
18429         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18430         vec_safe_push, always update *slot.
18431         (redirect_edge_var_map_clear): Use vec_free.
18432         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18433         (free_var_map_entry): Use vec_free.
18434         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18435         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18437 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
18439         PR middle-end/45472
18440         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18441         when the may_trap_p bit of the exprs being merged differs.
18442         Reorder tests for speculativeness in the logical and operator.
18444 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18446         * incpath.c (add_standard_paths): Use reconcat instead of concat
18447         where appropriate and avoid leaking memory.
18449         * opts.h: Include obstack.h.
18450         (opts_concat): New prototype.
18451         (opts_obstack): New declaration.
18452         * opts.c (opts_concat): New function.
18453         (opts_obstack): New variable.
18454         (init_options_struct): Call gcc_init_obstack on opts_obstack.
18455         (finish_options): Use opts_concat instead of concat
18456         and XOBNEWVEC instead of XNEWVEC.
18457         * opts-common.c (generate_canonical_option, decode_cmdline_option,
18458         generate_option): Likewise.
18459         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18460         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18462         PR target/56455
18463         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18464         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18466 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18468         PR middle-end/56461
18469         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18471 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18473         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18474         (arm_block_move_unaligned_straight): Likewise.
18475         (arm_adjust_block_mem): Likewise.
18477 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18479         PR target/48901
18480         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18481         temp, cond and label.
18482         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18484         PR target/52500
18485         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18486         * config/c6x/c6x.h (dbx_register_map): Update declaration.
18488         PR target/52501
18489         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18490         of prologue/epilogue functions.
18492         PR target/52550
18493         * config/tilegx/tilegx.c (tilegx_expand_prologue):
18494         Remove unused variable cfa_offset.
18495         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18497         PR target/54639
18498         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18499         type promotion to unsigned.
18501         PR target/54640
18502         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18503         for HOST_WIDE_INT of 32 bit / same size as int.
18504         (arm_block_move_unaligned_straight): Likewise.
18505         (arm_adjust_block_mem): Likewise.
18507         PR target/54662
18508         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18509         ALL_CFLAGS.
18511 2013-02-26  Marek Polacek  <polacek@redhat.com>
18513         PR tree-optimization/56426
18514         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18516 2013-02-26  Richard Biener  <rguenther@suse.de>
18518         PR target/56444
18519         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18520         unused variable loops.
18522 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18524         PR tree-optimization/56448
18525         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18526         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18527         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18528         later operands of the references, or even first operand for
18529         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18531         PR tree-optimization/56443
18532         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18533         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18534         to type_for_mode langhook.
18536 2013-02-25  Matt Turner  <mattst88@gmail.com>
18538         * doc/invoke.texi: Document r4700.
18540 2013-02-25  Richard Biener  <rguenther@suse.de>
18542         PR tree-optimization/56175
18543         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18544         split out from ...
18545         (simplify_bitwise_binary): ... here.  Also guard the conversion
18546         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18548 2013-02-25  Catherine Moore  <clm@codesourcery.com>
18550         Revert:
18551         2013-02-24  Catherine Moore  <clm@codesourcery.com>
18552             Maciej W. Rozycki  <macro@codesourcery.com>
18553             Tom de Vries  <tom@codesourcery.com>
18554             Nathan Sidwell  <nathan@codesourcery.com>
18555             Iain Sandoe  <iain@codesourcery.com>
18556             Nathan Froyd  <froydnj@codesourcery.com>
18557             Chao-ying Fu  <fu@mips.com>
18559         * doc/extend.texi: (micromips, nomicromips, nocompression):
18560         Document new function attributes.
18561         * doc/invoke.texi (minterlink-compressed, mmicromips,
18562         m14k, m14ke, m14kec): Document new options.
18563         (minterlink-mips16): Update documentation.
18564         * doc/md.texi (ZC, ZD): Document new constraints.
18565         * configure.ac (gcc_cv_as_micromips): Check if linker
18566         supports the .set micromips directive.
18567         * configure: Regenerate.
18568         * config.in: Regenerate.
18569         * config/mips/mips-tables.opt: Regenerate.
18570         * config/mips/micromips.md: New file.
18571         * constraints.md (ZC, AD): New constraints.
18572         * config/mips/predicates.md (movep_src_register): New predicate.
18573         (movep_src_operand): New predicate.
18574         (non_volatile_mem_operand): New predicate.
18575         * config/mips/mips.md (multimem): New type.
18576         (length): Differentiate between 17-bit and 18-bit branch offsets.
18577         (MOVEP1, MOVEP2): New mode iterator.
18578         (mov_<load>l): Use ZC constraint.
18579         (mov_<load>r): Likewise.
18580         (mov_<store>l): Likewise.
18581         (mov_<store>r): Likewise.
18582         (*branch_equality<mode>_inverted): Add microMIPS support.
18583         (*branch_equality<mode>): Likewise.
18584         (*jump_absolute): Likewise.
18585         (indirect_jump_<mode>): Likewise.
18586         (tablejump_<mode>): Likewise.
18587         (<optab>_internal): Likewise.
18588         (sibcall_internal): Likewise.
18589         (sibcall_value_internal): Likewise.
18590         (prefetch): Use constraint ZD.
18591         * config/mips/mips.opt (minterlink-compressed): New option.
18592         (minterlink-mips16): Now an alias for minterlink-compressed.
18593         (mmicromips): New option.
18594         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18595         (compare_and_swap_12): Likewise.
18596         (sync_add<mode>): Likewise.
18597         (sync_<optab>_12): Likewise.
18598         (sync_old_<optab>_12): Likewise.
18599         (sync_new_<optab>_12): Likewise.
18600         (sync_nand_12): Likewise.
18601         (sync_old_nand_12): Likewise.
18602         (sync_new_nand_12): Likewise.
18603         (sync_sub<mode>): Likewise.
18604         (sync_old_add<mode>): Likewise.
18605         (sync_old_sub<mode>): Likewise.
18606         (sync_new_add<mode>): Likewise.
18607         (sync_new_sub<mode>): Likewise.
18608         (sync_<optab><mode>): Likewise.
18609         (sync_old_<optab><mode>): Likewise.
18610         (sync_new_<optab><mode>): Likewise.
18611         (sync_nand<mode>): Likewise.
18612         (sync_old_nand<mode>): Likewise.
18613         (sync_new_nand<mode>): Likewise.
18614         (sync_lock_test_and_set<mode>): Likewise.
18615         (test_and_set_12): Likewise.
18616         (atomic_compare_and_swap<mode>): Likewise.
18617         (atomic_exchange<mode>_llsc): Likewise.
18618         (atomic_fetch_add<mode>_llsc): Likewise.
18619         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18620         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18621         (umips_save_restore_pattern_p): Likewise.
18622         (umips_load_store_pair_p): Likewise.
18623         (umips_output_load_store_pair): Likewise.
18624         (umips_movep_target_p): Likewise.
18625         (umips_12bit_offset_address_p): Likewise.
18626         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18627         (mips_base_mips16): Rename this...
18628         (mips_base_compression_flags): ...to this. Update all uses.
18629         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18630         (mips_mips16_decl_p): Delete.
18631         (mips_nomips16_decl_p): Delete.
18632         (mips_get_compress_on_flags): New function.
18633         (mips_get_compress_off_flags): New function.
18634         (mips_get_compress_mode): New function.
18635         (mips_get_compress_on_name): New function.
18636         (mips_get_compress_off_name): New function.
18637         (mips_insert_attributes): Support multiple compression types.
18638         (mips_merge_decl_attributes): Likewise.
18639         (umips_12bit_offset_address_p): New function.
18640         (mips_start_function_definition): Emit .set micromips directive.
18641         (mips_call_may_need_jalx_p): New function.
18642         (mips_function_ok_for_sibcall): Add microMIPS support.
18643         (mips_print_operand_punctuation): Support short delay slots and
18644         compact jumps.
18645         (umips_swm_mask, umips_swm_encoding): New.
18646         (umips_build_save_restore): New function.
18647         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18648         (was_mips16_p): Remove.
18649         (old_compression_mode): New.
18650         (mips_set_compression_mode): New function.
18651         (mips_set_current_function): Add microMIPS support.
18652         (mips_option_override): Likewise.
18653         (umips_save_restore_pattern_p): New function.
18654         (umips_output_save_restore): New function.
18655         (umips_load_store_pair_p_1): New function.
18656         (umips_load_store_pair_p): New function.
18657         (umips_output_load_store_pair_1): New function.
18658         (umips_output_load_store_pair): New function.
18659         (umips_movep_target_p) New function.
18660         (mips_prepare_pch_save): Add microMIPS support.
18661         * config/mips/mips.h (TARGET_COMPRESSION): New.
18662         (TARGET_CPU_CPP_BUILTINS): Update macro
18663         to use new compression flags and to support microMIPS.
18664         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18665         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18666         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18667         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18668         (ASM_SPEC): Support mmicromips and mno-micromips.
18669         (M16STORE_REG_P): New macro.
18670         (MIPS_CALL): Support TARGET_MICROMIPS.
18671         (MICROMIPS_J): New macro.
18672         (mips_base_mips16): Rename this...
18673         (mips_base_compression_flags): ...to this.
18674         (UMIPS_12BIT_OFFSET_P): New macro.
18675         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18676         (MULTILIB_DIRNAMES): Likewise.
18678 2013-02-25  Tom de Vries  <tom@codesourcery.com>
18680         PR rtl-optimization/56131
18681         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
18682         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
18683         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
18685 2013-02-25  Tobias Burnus  <burnus@net-b.de>
18687         * doc/invoke.texi (-fsanitize=): Move from optimization
18688         to debugging options.
18690 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18692         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
18694 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18695             Alexander Monakov  <amonakov@ispras.ru>
18697         PR middle-end/56077
18698         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
18699         flush pending lists also on non-jumps.  Adjust comment.
18701 2013-02-24  Catherine Moore  <clm@codesourcery.com>
18702             Maciej W. Rozycki  <macro@codesourcery.com>
18703             Tom de Vries  <tom@codesourcery.com>
18704             Nathan Sidwell  <nathan@codesourcery.com>
18705             Iain Sandoe  <iain@codesourcery.com>
18706             Nathan Froyd  <froydnj@codesourcery.com>
18707             Chao-ying Fu  <fu@mips.com>
18709         * doc/extend.texi: (micromips, nomicromips, nocompression):
18710         Document new function attributes.
18711         * doc/invoke.texi (minterlink-compressed, mmicromips,
18712         m14k, m14ke, m14kec): Document new options.
18713         (minterlink-mips16): Update documentation.
18714         * doc/md.texi (ZC, ZD): Document new constraints.
18715         * configure.ac (gcc_cv_as_micromips): Check if linker
18716         supports the .set micromips directive.
18717         * configure: Regenerate.
18718         * config.in: Regenerate.
18719         * config/mips/mips-tables.opt: Regenerate.
18720         * config/mips/micromips.md: New file.
18721         * constraints.md (ZC, AD): New constraints.
18722         * config/mips/predicates.md (movep_src_register): New predicate.
18723         (movep_src_operand): New predicate.
18724         (non_volatile_mem_operand): New predicate.
18725         * config/mips/mips.md (multimem): New type.
18726         (length): Differentiate between 17-bit and 18-bit branch offsets.
18727         (MOVEP1, MOVEP2): New mode iterator.
18728         (mov_<load>l): Use ZC constraint.
18729         (mov_<load>r): Likewise.
18730         (mov_<store>l): Likewise.
18731         (mov_<store>r): Likewise.
18732         (*branch_equality<mode>_inverted): Add microMIPS support.
18733         (*branch_equality<mode>): Likewise.
18734         (*jump_absolute): Likewise.
18735         (indirect_jump_<mode>): Likewise.
18736         (tablejump_<mode>): Likewise.
18737         (<optab>_internal): Likewise.
18738         (sibcall_internal): Likewise.
18739         (sibcall_value_internal): Likewise.
18740         (prefetch): Use constraint ZD.
18741         * config/mips/mips.opt (minterlink-compressed): New option.
18742         (minterlink-mips16): Now an alias for minterlink-compressed.
18743         (mmicromips): New option.
18744         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18745         (compare_and_swap_12): Likewise.
18746         (sync_add<mode>): Likewise.
18747         (sync_<optab>_12): Likewise.
18748         (sync_old_<optab>_12): Likewise.
18749         (sync_new_<optab>_12): Likewise.
18750         (sync_nand_12): Likewise.
18751         (sync_old_nand_12): Likewise.
18752         (sync_new_nand_12): Likewise.
18753         (sync_sub<mode>): Likewise.
18754         (sync_old_add<mode>): Likewise.
18755         (sync_old_sub<mode>): Likewise.
18756         (sync_new_add<mode>): Likewise.
18757         (sync_new_sub<mode>): Likewise.
18758         (sync_<optab><mode>): Likewise.
18759         (sync_old_<optab><mode>): Likewise.
18760         (sync_new_<optab><mode>): Likewise.
18761         (sync_nand<mode>): Likewise.
18762         (sync_old_nand<mode>): Likewise.
18763         (sync_new_nand<mode>): Likewise.
18764         (sync_lock_test_and_set<mode>): Likewise.
18765         (test_and_set_12): Likewise.
18766         (atomic_compare_and_swap<mode>): Likewise.
18767         (atomic_exchange<mode>_llsc): Likewise.
18768         (atomic_fetch_add<mode>_llsc): Likewise.
18769         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18770         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18771         (umips_save_restore_pattern_p): Likewise.
18772         (umips_load_store_pair_p): Likewise.
18773         (umips_output_load_store_pair): Likewise.
18774         (umips_movep_target_p): Likewise.
18775         (umips_12bit_offset_address_p): Likewise.
18776         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18777         (mips_base_mips16): Rename this...
18778         (mips_base_compression_flags): ...to this. Update all uses.
18779         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18780         (mips_mips16_decl_p): Delete.
18781         (mips_nomips16_decl_p): Delete.
18782         (mips_get_compress_on_flags): New function.
18783         (mips_get_compress_off_flags): New function.
18784         (mips_get_compress_mode): New function.
18785         (mips_get_compress_on_name): New function.
18786         (mips_get_compress_off_name): New function.
18787         (mips_insert_attributes): Support multiple compression types.
18788         (mips_merge_decl_attributes): Likewise.
18789         (umips_12bit_offset_address_p): New function.
18790         (mips_start_function_definition): Emit .set micromips directive.
18791         (mips_call_may_need_jalx_p): New function.
18792         (mips_function_ok_for_sibcall): Add microMIPS support.
18793         (mips_print_operand_punctuation): Support short delay slots and
18794         compact jumps.
18795         (umips_swm_mask, umips_swm_encoding): New.
18796         (umips_build_save_restore): New function.
18797         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18798         (was_mips16_p): Remove.
18799         (old_compression_mode): New.
18800         (mips_set_compression_mode): New function.
18801         (mips_set_current_function): Add microMIPS support.
18802         (mips_option_override): Likewise.
18803         (umips_save_restore_pattern_p): New function.
18804         (umips_output_save_restore): New function.
18805         (umips_load_store_pair_p_1): New function.
18806         (umips_load_store_pair_p): New function.
18807         (umips_output_load_store_pair_1): New function.
18808         (umips_output_load_store_pair): New function.
18809         (umips_movep_target_p) New function.
18810         (mips_prepare_pch_save): Add microMIPS support.
18811         * config/mips/mips.h (TARGET_COMPRESSION): New.
18812         (TARGET_CPU_CPP_BUILTINS): Update macro
18813         to use new compression flags and to support microMIPS.
18814         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18815         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18816         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18817         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18818         (ASM_SPEC): Support mmicromips and mno-micromips.
18819         (M16STORE_REG_P): New macro.
18820         (MIPS_CALL): Support TARGET_MICROMIPS.
18821         (MICROMIPS_J): New macro.
18822         (mips_base_mips16): Rename this...
18823         (mips_base_compression_flags): ...to this.
18824         (UMIPS_12BIT_OFFSET_P): New macro.
18825         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18826         (MULTILIB_DIRNAMES): Likewise.
18828 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
18830         PR target/52555
18831         * target-globals.c (save_target_globals): For init_reg_sets and
18832         target_reinit remporarily set this_fn_optabs to this_target_optabs.
18834 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
18836         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
18837         * config/aarch64/t-aarch64
18838         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
18840 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
18842         PR inline-asm/56148
18843         * lra-constraints.c (process_alt_operands): Reload operand
18844         conflicting with earlier clobber only if no more other conflicting
18845         operands.
18847 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
18849         PR sanitizer/56393
18850         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
18851         if not linking a shared library.
18853 2013-02-22  Seth LaForge  <sethml@google.com>
18855         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
18857 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
18859         * config/arm/arm.md (split for extendsidi): Update condition.
18860         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
18861         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
18862         (qhs_zextenddi_cstr): Likewise.
18864 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18866         PR middle-end/56420
18867         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
18868         avoid signed wrapping.
18869         (expand_mult): Handle properly multiplication by
18870         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
18871         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
18872         in the compiler if coeff is HOST_WIDE_INT_MIN.
18873         (expand_divmod): Don't make ext_op1 static, change it's type to
18874         uhwi.  Avoid undefined behavior in -INTVAL (op1).
18876         PR rtl-optimization/50339
18877         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
18878         field.
18879         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
18880         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
18881         into splitting_ashiftrt field.
18882         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
18883         ASHIFTRT.
18884         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
18885         choices.
18887 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
18889         PR middle-end/56108
18890         * trans-mem.c (execute_tm_mark): Do not expand transactions that
18891         are sure to go irrevocable.
18893 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
18895         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
18896         scalars are valid operands.
18898 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18900         PR tree-optimization/56310
18901         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
18902         only matching indices and non-negative final offsets.
18903         (intersect_aggregates_with_edge): Pass src_idx to
18904         agg_replacements_to_vector.  Pass src_idx insstead of index to
18905         intersect_with_agg_replacements.
18907 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18909         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
18910         instead of hard-wired defaults.
18912 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
18914         * doc/invoke.texi (MIPS Options): Update documentation of the
18915         floating-point multiply-accumulate instruction restrictions.
18917 2013-02-21  Kostya Serebryany  <kcc@google.com>
18919         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
18920         asan_shadow_offset on x86_64 linux.
18922 2013-02-21  Richard Biener  <rguenther@suse.de>
18924         PR tree-optimization/56415
18925         Revert
18926         2013-02-11  Richard Biener  <rguenther@suse.de>
18928         PR tree-optimization/56273
18929         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
18930         first VRP run.
18932 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18934         PR bootstrap/56258
18935         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
18936         instead of @itemx.
18938         PR inline-asm/56405
18939         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
18940         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
18942 2013-02-20  Jan Hubicka  <jh@suse.cz>
18944         PR tree-optimization/56265
18945         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
18946         when target is referenced for first time.
18948 2013-02-20  Richard Biener  <rguenther@suse.de>
18950         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
18951         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
18952         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
18953         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
18954         not return anything.
18955         (rename_ssa_copies): Do not remove unused locals.
18956         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
18957         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
18958         * passes.c (execute_function_todo): Do not schedule unused locals
18959         removal if cleanup_tree_cfg did something.
18960         * tree-ssa-live.c (remove_unused_locals): Dump statistics
18961         about the number of removed locals.
18963 2013-02-20  Richard Biener  <rguenther@suse.de>
18965         PR tree-optimization/56398
18966         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
18968 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18970         PR tree-optimization/55334
18971         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
18972         restricted pointers to arrays.
18974 2013-02-20  Richard Biener  <rguenther@suse.de>
18975             Jakub Jelinek  <jakub@redhat.com>
18977         PR tree-optimization/56396
18978         * tree-ssa-ccp.c (n_const_val): New static variable.
18979         (get_value): Return NULL for SSA names we don't have a lattice
18980         entry for.
18981         (ccp_initialize): Initialize n_const_val.
18982         * tree-ssa-copy.c (n_copy_of): New static variable.
18983         (init_copy_prop): Initialize n_copy_of.
18984         (get_value): Return NULL_TREE for SSA names we don't have a
18985         lattice entry for.
18987 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18989         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
18991 2013-02-20  Richard Biener  <rguenther@suse.de>
18993         * genpreds.c (write_lookup_constraint): Do not compare first
18994         letter of the constraint again.
18996 2013-02-20  Richard Biener  <rguenther@suse.de>
18998         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
18999         and ceil_log2.
19000         (get_use_iv_cost): Terminate hashtable walk when coming across
19001         an empty entry.
19003 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19005         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19006         reassociation for avx2 targets.
19008 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19010         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19011         Add version check for v8.10.a to enable microblaze_has_clz
19012         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19013         version and TARGET_PATTERN_COMPARE check
19014         * config/microblaze/microblaze.md: New clzsi2 instruction
19016 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19018         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19019         function before branching.
19021 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19023         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19024         DUMP_INSN_RTX_UID.
19025         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19027 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19029         PR middle-end/55889
19030         * sel-sched.c: Include ira.h.
19031         (implicit_clobber_conflict_p): New function.
19032         (moveup_expr): Use it.
19033         * Makefile.in (sel-sched.o): Depend on ira.h.
19035 2013-02-19  Richard Biener  <rguenther@suse.de>
19037         PR tree-optimization/56384
19038         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19039         (vn_hash_type): Split out from ...
19040         (vn_hash_constant_with_type): ... here.
19041         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19042         (vn_phi_eq): Compare types from vn_phi_s structure.
19043         (vn_phi_lookup): Populate vn_phi_s type.
19044         (vn_phi_insert): Likewise.
19046 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19048         PR tree-optimization/56350
19049         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19050         if haven't found reduction or nested cycle operand, rather than
19051         asserting we must find it.
19053         PR tree-optimization/56381
19054         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19055         to fold_build3.
19057 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19058             Jakub Jelinek  <jakub@redhat.com>
19060         PR target/52555
19061         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19062         (swap_optab_enable): Same.
19063         (init_all_optabs): Use argument instead of global.
19064         * tree.h (struct tree_optimization_option): New field target_optabs.
19065         * expr.h (init_all_optabs): Add argument to prototype.
19066         (TREE_OPTIMIZATION_OPTABS): New.
19067         (save_optabs_if_changed): Protoize.
19068         * optabs.h: Declare this_fn_optabs.
19069         * optabs.c (save_optabs_if_changed): New.
19070         Declare this_fn_optabs.
19071         (init_optabs): Add argument to init_all_optabs() call.
19072         * function.c (invoke_set_current_function_hook): Handle per
19073         function optabs.
19074         * function.h (struct function): New field optabs.
19075         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19076         optimization_current_node has changed.
19077         * target-globals.h (save_target_globals_default_opts): Protoize.
19078         * target-globals.c (save_target_globals_default_opts): New.
19080 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19082         PR target/56347
19083         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19084         registers %fr12 and %fr12R as call used.
19086         PR target/56214
19087         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19088         and HImode, require all displacements to be an integer multiple of
19089         their mode size.
19090         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19091         only allow QImode and HImode when reload is in progress and strict is
19092         true.  Likewise for symbolic addresses.  Use base14_operand to check
19093         displacements in REG+BASE addresses.
19095 2013-02-18  Richard Biener  <rguenther@suse.de>
19097         PR tree-optimization/56366
19098         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19099         handle sign-conversion of outer-loop initial induction value.
19101 2013-02-18  Richard Biener  <rguenther@suse.de>
19103         PR middle-end/56349
19104         * cfghooks.c (merge_blocks): If we merge a latch into another
19105         block adjust references to it.
19106         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19107         (verify_loop_structure): Verify that a recorded latch is in fact
19108         a latch.
19110 2013-02-18  Richard Biener  <rguenther@suse.de>
19112         PR tree-optimization/56321
19113         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19114         order SSA name release and virtual operand unlinking.
19116 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19118         * config/microblaze/microblaze.md (save_stack_block): Define.
19119         (restore_stack_block): Likewise.
19121 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19123         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19124         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19125         * config/microblaze/microblaze.c (microblaze_option_override):
19126         Bail out early for PIC modes when target does not support PIC.
19128 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19130         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19131         Replace with a microblaze version.
19132         (microblaze_trampoline_init): Adapt for microblaze.
19133         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19134         microblaze.
19136 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19137             Dodji Seketeli  <dodji@redhat.com>
19139         PR asan/56330
19140         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19141         (instrument_mem_region_access): Do not forget to always put
19142         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19143         0) statement, even for cases where either 'base' or 'base + len'
19144         are not instrumented -- because they have been previously
19145         instrumented.  Simplify the logic by putting all the statements
19146         instrument 'base + len' inside a sequence, and then insert that
19147         sequence right before the current insertion point.  Then, to
19148         instrument 'base + len', just get an iterator on that statement.
19149         And do not forget to update the pointer to iterator the function
19150         received as argument.
19152 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19154         PR rtl-optimization/56348
19155         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19157 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19159         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19160         (clean_graph_dump_file): Pass base to start_graph_dump.
19162 2013-02-14  Richard Henderson  <rth@redhat.com>
19164         PR target/55941
19165         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19167 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19169         * collect2-aix.h: Define F_LOADONLY.
19171 2013-02-14  Richard Biener  <rguenther@suse.de>
19173         PR lto/50494
19174         * varasm.c (output_constant_def_1): Get the decl representing
19175         the constant as argument.
19176         (output_constant_def): Wrap output_constant_def_1.
19177         (make_decl_rtl): Use output_constant_def_1 with the decl
19178         representing the constant.
19179         (build_constant_desc): Optionally re-use a decl already
19180         representing the constant.
19181         (tree_output_constant_def): Adjust.
19183 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19185         Fix an asan crash
19186         * asan.c (instrument_builtin_call):  Really put the length of the
19187         second source argument into src1_len.
19189 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19191         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19192         argument.  If it is false, don't create edge from then_bb to
19193         fallthru_bb.
19194         (insert_if_then_before_iter): Pass true to it.
19195         (build_check_stmt): Pass false to it.
19196         (transform_statements): Flush hash table only on extended basic
19197         block boundaries, rather than at the beginning of every bb.
19198         Don't flush hash table on nonfreeing_call_p calls.
19199         * tree-flow.h (nonfreeing_call_p): New prototype.
19200         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19202 2013-02-13  David S. Miller  <davem@davemloft.net>
19204         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19206 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19208         PR target/56184
19209         * ira.c (max_regno_before_ira): Move from ...
19210         (ira): ... here.
19211         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19212         vec_safe_length.
19214 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19216         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19218 2013-02-13  Richard Biener  <rguenther@suse.de>
19220         PR lto/56295
19221         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19222         globals in MEM_REFs.
19224 2013-02-13  Richard Biener  <rguenther@suse.de>
19226         * loop-init.c (loop_optimizer_init): Clear loop state when
19227         re-initializing preserved loops.
19228         * loop-unswitch.c (unswitch_single_loop): Return whether
19229         we unswitched the loop.  Do not verify loop state here.
19230         (unswitch_loops): When we unswitched a loop discover new loops.
19232 2013-02-13  Kostya Serebryany  <kcc@google.com>
19234         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19235         on x86_64 linux.
19236         * sanitizer.def: Rename __asan_init to __asan_init_v1.
19238 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
19240         Avoid instrumenting duplicated memory access in the same basic block
19241         * Makefile.in (asan.o): Add new dependency on hash-table.h
19242         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19243         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19244         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19245         (free_mem_ref_resources, has_mem_ref_been_instrumented)
19246         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19247         (get_mem_ref_of_assignment): New functions.
19248         (get_mem_refs_of_builtin_call): Extract from
19249         instrument_builtin_call and tweak a little bit to make it fit with
19250         the new signature.
19251         (instrument_builtin_call): Use the new
19252         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
19253         of is_gimple_builtin_call.
19254         (instrument_derefs, instrument_mem_region_access): Insert the
19255         instrumented memory reference into the hash table.
19256         (maybe_instrument_assignment): Renamed instrument_assignment into
19257         this, and change it to advance the iterator when instrumentation
19258         actually happened and return true in that case.  This makes it
19259         homogeneous with maybe_instrument_assignment, and thus give a
19260         chance to callers to be more 'regular'.
19261         (transform_statements): Clear the memory reference hash table
19262         whenever we enter a new BB, when we cross a function call, or when
19263         we are done transforming statements.  Use
19264         maybe_instrument_assignment instead of instrumentation.  No more
19265         need to special case maybe_instrument_assignment and advance the
19266         iterator after calling it; it's now handled just like
19267         maybe_instrument_call.  Update comment.
19269 2013-02-13  Richard Biener  <rguenther@suse.de>
19271         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19272         Fix loop discovery code.
19274 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
19276         PR inline-asm/56148
19277         * lra-constraints.c (process_alt_operands): Match early clobber
19278         operand with itself.  Check conflicts with earlyclobber only if
19279         the operand is not reloaded.  Prefer to reload conflicting operand
19280         if earlyclobber and matching operands are the same.
19282 2013-02-12  Richard Biener  <rguenther@suse.de>
19284         PR lto/56297
19285         * lto-streamer-out.c (write_symbol): Do not output symbols
19286         for hard register variables.
19288 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
19290         PR target/54222
19291         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19292         (umulsidi3_insn, mulsidi3_insn): New insns.
19294 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
19296         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19297         (struct tune_params): Add vec_costs field.
19298         * config/arm/arm.c (arm_builtin_vectorization_cost)
19299         (arm_add_stmt_cost): New functions.
19300         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19301         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19302         (arm_default_vec_cost): New struct of type cpu_vec_costs.
19303         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19304         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19305         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19306         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19308 2013-02-12  Richard Biener  <rguenther@suse.de>
19310         PR lto/56295
19311         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19312         decls again if possible.
19314 2013-02-12  Richard Biener  <rguenther@suse.de>
19316         PR middle-end/56288
19317         * tree-ssa.c (verify_ssa_name): Fix check, move
19318         SSA_NAME_IN_FREE_LIST check up.
19320 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
19321             Steven Bosscher   <steven@gcc.gnu.org>
19323         PR rtl-optimization/56151
19324         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19325         equal to op0 or op1, and last_insn pattern is CODE operation
19326         with MEM dest and one of the operands matches that MEM.
19328 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
19330         * doc/extend.texi: Document Function Multiversioning and "default"
19331         parameter string to target attribute.
19332         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19333         target attribute parameter is "default".
19334         (ix86_compare_version_priority): Remove checks for target attribute.
19335         (ix86_mangle_function_version_assembler_name): Change error to sorry.
19336         Remove check for target attribute equal to NULL. Add assert.
19337         (ix86_generate_version_dispatcher_body): Change error to sorry.
19339 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
19340             Jack Howarth  <howarth@bromo.med.uc.edu>
19341             Patrick Marlier  <patrick.marlier@gmail.com>
19343         PR libitm/55693
19344         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19345         define ENDFILE_SPEC as TM_DESTRUCTOR.
19346         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19348 2013-02-11  Alexander Potapenko  <glider@google.com>
19349             Jack Howarth  <howarth@bromo.med.uc.edu>
19350             Jakub Jelinek  <jakub@redhat.com>
19352         PR sanitizer/55617
19353         * config/darwin.c (cdtor_record): Rename ctor_record.
19354         (sort_cdtor_records): Rename sort_ctor_records.
19355         (finalize_dtors): New routine to sort destructors by
19356         priority before use in assemble_integer.
19357         (machopic_asm_out_destructor): Use finalize_dtors if needed.
19359 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
19361         PR rtl-optimization/56275
19362         * simplify-rtx.c (avoid_constant_pool_reference): Check that
19363         offset is non-negative and less than cmode size before
19364         calling simplify_subreg.
19366 2013-02-11  Richard Biener  <rguenther@suse.de>
19368         PR tree-optimization/56264
19369         * cfgloop.h (fix_loop_structure): Adjust prototype.
19370         * loop-init.c (fix_loop_structure): Return the number of
19371         newly discovered loops.
19372         * tree-cfgcleanup.c (repair_loop_structures): When new loops
19373         are discovered, do a full loop-closed SSA rewrite.
19375 2013-02-11  Richard Biener  <rguenther@suse.de>
19377         PR tree-optimization/56273
19378         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19379         first VRP run.
19380         (check_array_ref): Fix missing newline in dumps.
19381         (search_for_addr_array): Likewise.
19383 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
19385         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19387 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
19389         PR target/56256
19390         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19392 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
19394         PR rtl-optimization/56246
19395         * lra-constraints.c (simplify_operand_subreg): Try to reuse
19396         reload pseudo.
19397         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19398         constraints are satisfied.
19400 2013-02-08  Jeff Law  <law@redhat.com>
19402         PR debug/53948
19403         * emit-rtl.c (reg_is_parm_p): New function.
19404         * regs.h (reg_is_parm_p): New prototype.
19405         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19406         callee-clobbered registers.
19408 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
19410         PR target/56043
19411         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19412         If there is no implicit builtin declaration, just return NULL.
19414 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19416         * config/i386/sse.md (FMAMODEM): New mode iterator.
19417         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19418         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19420 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19422         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19423         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19424         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19426 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19428         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19429         (microblaze*-*-elf): Likewise.
19430         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19431         LINK_SPEC.
19432         * config/microblaze/microblaze-c.c: Add builtin defines for
19433         _LITTLE_ENDIAN and _BIG_ENDIAN.
19434         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19435         add to TARGET_DEFAULT flags.
19436         Expand ASM_SPEC and LINK_SPEC.
19437         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19438         * config/microblaze/microblaze.md: Update extendsidi2 and
19439         movdi_internal instructions to use low-order / high-order reg
19440         print_operands.
19441         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19442         options and inversemask / mask of LITTLE_ENDIAN.
19443         * config/microblaze/t-microblaze: Expand multilib options to
19444         include mlittle-endian (le) and update exceptions patterns.
19446 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
19448         PR rtl-optimization/56195
19449         * lra-constraints.c (get_reload_reg): Don't reuse regs
19450         if they have smaller mode than requested, if they have
19451         wider mode than requested, try to return a SUBREG.
19453         PR tree-optimization/56250
19454         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19455         if type is unsigned and code isn't MULT_EXPR.
19457 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19459         PR tree-optimization/56064
19460         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19461         bits according to mode.
19462         * fixed-value.h (fixed_from_double_int)
19463         (const_fixed_from_double_int): Adjust comments.
19465 2013-02-08  Richard Biener  <rguenther@suse.de>
19467         PR lto/56231
19468         * lto-streamer.h (struct data_in): Remove current_file, current_line
19469         and current_col members.
19470         * lto-streamer-out.c (lto_output_location): Stream changed bits
19471         en-block for efficiency.
19472         * lto-streamer-in.c (clear_line_info): Remove.
19473         (lto_input_location): Cache current file, line and column
19474         globally via local statics.  Read changed bits en-block.
19475         (input_function): Do not call clear_line_info.
19476         (lto_read_body): Likewise.
19477         (lto_input_toplevel_asms): Likewise.
19479 2013-02-08  Michael Matz  <matz@suse.de>
19481         PR tree-optimization/52448
19482         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19483         (nt_call_phase): New static.
19484         (add_or_mark_expr): Only mark accesses with newer phase than any
19485         call seen.
19486         (nonfreeing_call_p): New.
19487         (nt_init_block): Update nt_call_phase, mark blocks as visited.
19488         (nt_fini_block): Keep blocks marked as visited.
19489         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19491 2013-02-08  Richard Biener  <rguenther@suse.de>
19493         * ira.c (ira): Free broken dominator information.
19495 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19497         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19499 2013-02-08  Marek Polacek  <polacek@redhat.com>
19501         * cfgloop.c (verify_loop_structure): Add more checking of headers.
19503 2013-02-08  Richard Biener  <rguenther@suse.de>
19505         PR middle-end/56181
19506         * cfgloop.h (flow_loops_find): Adjust.
19507         (bb_loop_header_p): Declare.
19508         * cfgloop.c (bb_loop_header_p): New function split out from ...
19509         (flow_loops_find): ... here.  Adjust function signature,
19510         support incremental loop structure update.
19511         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
19512         * cfgloopmanip.c (fix_loop_structure): Move ...
19513         * loop-init.c (fix_loop_structure): ... here.
19514         (apply_loop_flags): Split out from ...
19515         (loop_optimizer_init): ... here.
19516         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
19517         in incremental mode, only remove dead loops here.
19519 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19521         PR target/54222
19522         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19523         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19524         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19525         (*round<mode>3.libgcc): New insns for fixed-modes.
19526         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19527         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19528         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19529         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19530         implementations.  Define to __builtin_avr_absFX,
19531         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19532         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19533         __builtin_avr_countlsFX, respectively.
19534         * config/avr/avr-c.c (target.h): Include it.
19535         (enum avr_builtin_id): New enum.
19536         (avr_resolve_overloaded_builtin): New static function.
19537         (avr_register_target_pragmas): Use it to set
19538         targetm.resolve_overloaded_builtin.
19539         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19540         tree nodes used by DEF_BUILTIN.
19541         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19542         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19543         <AVR_BUILTIN_xxBITS>: Same.
19545 2013-02-08  Richard Biener  <rguenther@suse.de>
19547         * cfgloop.c (verify_loop_structure): Properly handle
19548         a loop exiting to another loop header.
19549         * ira-int.h (ira_loops): Remove.
19550         * ira.c (ira_loops): Remove.
19551         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19552         (do_reload): Use loop_optimizer_finalize.
19553         * ira-build.c (create_loop_tree_nodes): Use get_loops and
19554         number_of_loops to access the loop tree.
19555         (more_one_region_p): Likewise.
19556         (finish_loop_tree_nodes): Likewise.
19557         (rebuild_regno_allocno_maps): Likewise.
19558         (mark_loops_for_removal): Likewise.
19559         (mark_all_loops_for_removal): Likewise.
19560         (remove_unnecessary_regions): Likewise.
19561         (ira_build): Likewise.
19562         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19564 2013-02-08  Richard Biener  <rguenther@suse.de>
19566         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
19567         * ipa-pure-const.c (analyze_function): Avoid calling
19568         mark_irreducible_loops twice.
19569         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
19571 2013-02-07  David S. Miller  <davem@davemloft.net>
19573         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
19574         on 'reg'.
19575         * var-tracking.c (vt_add_function_parameter): Test the presence of
19576         HAVE_window_save properly and do not remap argument registers when
19577         we have a leaf function.
19579 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
19581         PR bootstrap/56227
19582         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
19583         instead of "ll".
19584         * config/i386/i386.c (ix86_print_operand): Ditto.
19586 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19588         * lra-constraints.c (process_alt_operands): Fix recently added comment.
19590 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19592         PR rtl-optimization/56225
19593         * lra-constraints.c (process_alt_operands): Check that reload hard
19594         reg can hold value for strict_low_part.
19596 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19598         PR debug/56154
19599         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
19600         dwarf2out_end_function.
19601         (in_first_function_p, maybe_at_text_label_p,
19602         first_loclabel_num_not_at_text_label): New variables.
19603         (dwarf2out_var_location): In the first function find out
19604         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
19605         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
19606         functions.
19608 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
19610         PR rtl-optimization/56178
19611         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
19612         SUBREG of a register.  Tidy up related block of code.
19613         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
19614         note if the source is a register or a SUBREG of a register.
19616 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19618         PR target/56228
19619         * config/rs6000/rs6000.md (ptrm): New mode attr.
19620         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
19621         call_value_indirect_aix<pttrsize>,
19622         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
19623         m in constraints.
19625 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
19627         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
19628         if -bnortl. Convert to strcmp and strncmp.
19630 2013-02-07  Alan Modra  <amodra@gmail.com>
19632         PR target/54009
19633         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
19634         addresses won't wrap when offsetting.
19635         (rs6000_secondary_reload): Provide secondary reloads needed for
19636         wrapping LO_SUM addresses.
19638 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
19640         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
19641         MACH, just __MACH__.
19643 2013-02-06  Richard Biener  <rguenther@suse.de>
19645         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
19646         instead of calling fix_loop_structure.
19648 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
19650         PR middle-end/56217
19651         * omp-low.c (use_pointer_for_field): Return false if
19652         lower_send_shared_vars doesn't generate any copy-out code.
19654 2013-02-06  Tom de Vries  <tom@codesourcery.com>
19656         PR rtl-optimization/56131
19657         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
19658         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
19659         of the label is NULL.  Add comment.
19661 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
19663         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
19665         PR sanitizer/55374
19666         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
19667         (STATIC_LIBTSAN_LIBS): Likewise.
19668         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
19669         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
19670         is defined, don't add anything else beyond that.
19671         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
19672         (LINK_COMMAND_SPEC): Use them.
19674         PR tree-optimization/56205
19675         * tree-stdarg.c (check_all_va_list_escapes): Return true if
19676         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
19677         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
19679 2013-02-05  Richard Biener  <rguenther@suse.de>
19681         PR tree-optimization/53342
19682         PR tree-optimization/53185
19683         * tree-vectorizer.h (vect_check_strided_load): Remove.
19684         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
19685         not disallow peeling for vectorized strided loads.
19686         (vect_check_strided_load): Make static and simplify.
19687         (vect_analyze_data_refs): Adjust.
19688         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
19689         correctly when vectorizing strided loads.
19691 2013-02-05  Richard Biener  <rguenther@suse.de>
19693         * doc/install.texi: Refer to ISL, not PPL.
19695 2013-02-05  Jan Hubicka  <jh@suse.cz>
19697         PR tree-optimization/55789
19698         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
19700 2013-02-05  Jan Hubicka  <jh@suse.cz>
19702         PR tree-optimization/55789
19703         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
19704         the dead call anyway.
19706 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
19708         PR sanitizer/55374
19709         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
19711 2013-02-04  Alexander Potapenko  <glider@google.com>
19712             Jack Howarth  <howarth@bromo.med.uc.edu>
19713             Jakub Jelinek  <jakub@redhat.com>
19715         PR sanitizer/55617
19716         * config/darwin.c (sort_ctor_records): Stabilized qsort
19717         on constructor priority by using original position.
19718         (finalize_ctors): New routine to sort constructors by
19719         priority before use in assemble_integer.
19720         (machopic_asm_out_constructor): Use finalize_ctors if needed.
19722 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
19724         PR libstdc++/54314
19725         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
19726         about visibility on artificial decls.
19727         * config/sol2.c (solaris_assemble_visibility): Likewise.
19729 2013-02-04  Kai Tietz  <ktietz@redhat.com>
19731         PR target/56186
19732         * config/i386/i386.c (function_value_ms_64): Add additional valtype
19733         argument and improve checking of return-argument types for 16-byte
19734         modes.
19735         (ix86_function_value_1): Add additional valtype argument on call
19736         of function_value_64.
19737         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
19738         handling infunction_value_64 function.
19740 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
19742         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
19744 2013-02-04  Richard Biener  <rguenther@suse.de>
19746         PR tree-optimization/56188
19747         * tree-ssa-structalias.c (label_visit): Consider case with
19748         initially non-empty points-to set.
19749         (perform_var_substitution): Dump node mapping and clean up.
19751 2013-02-04  Richard Guenther  <rguenther@suse.de>
19753         PR lto/56168
19754         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
19755         node prevail as last resort.
19756         (lto_symtab_merge_decls): Remove guard on LTRANS here.
19757         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
19759 2013-02-04  Richard Biener  <rguenther@suse.de>
19761         PR tree-optimization/56113
19762         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
19763         Merge into ...
19764         (equiv_class_lookup_or_add): ... this.
19765         (label_visit): Adjust and fix error in previous patch.
19766         (perform_var_substitution): Adjust.
19768 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
19770         * config/sh/divtab.c: Fix formatting and comments throughout the file.
19771         * config/sh/sh4-300.md: Likewise.
19772         * config/sh/sh4a.md: Likewise.
19773         * config/sh/constraints.md: Likewise.
19774         * config/sh/sh.md: Likewise.
19775         * config/sh/netbsd-elf.h: Likewise.
19776         * config/sh/predicates.md: Likewise.
19777         * config/sh/sh-protos.h: Likewise.
19778         * config/sh/ushmedia.h: Likewise.
19779         * config/sh/linux.h: Likewise.
19780         * config/sh/sh.c: Likewise.
19781         * config/sh/superh.h: Likewise.
19782         * config/sh/elf.h: Likewise.
19783         * config/sh/sh4.md: Likewise.
19784         * config/sh/sh.h: Likewise.
19786 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19788         * config/pa/constraints.md: Adjust unused letters.  Change "T"
19789         constraint to match_test floating_point_store_memory_operand().
19790         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
19791         (base14_operand): New.
19792         (floating_point_store_memory_operand): New.
19793         (integer_store_memory_operand): Revise to use base14_operand and
19794         reg_plus_base_memory_operand.
19795         (move_dest_operand): Allow symbolic_memory_operands.
19796         (symbolic_memory_operand): Check for LO_SOM.
19797         (symbolic_operand): Change default case to break.
19798         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
19799         CONST_DOUBLE values to be reloaded by putting them into memory when
19800         the destination is a floating point register.
19801         (movdf): Remove code to handle CONST_DOUBLE.
19802         (movsf): Likewise.
19803         (reload_indf_r1): New.
19804         (reload_insf_r1): New.
19805         Consistently use "Q" and "T" constraints with integer and floating
19806         point move instructions, respectively.
19807         (movdi): Remove FAIL.
19808         Change predicate for source operand unamed DImode move from
19809         general_operand to move_src_operand.
19810         (umulsidi3): Change predicate for destination operand to
19811         register_operand.
19812         Likewise for similar unamed patterns.
19813         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
19814         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
19815         (hppa_legitimize_address): Simplify mask calculation.
19816         (pa_emit_move_sequence): Revised handling of secondary reloads from
19817         REG+D addresses for floating point loads and stores.  Directly handle
19818         loading CONST0_RTX (mode) to a floating point register.
19819         (pa_secondary_reload): Handle reloading DF and SFmode constant values
19820         to floating point registers.  Don't restrict secondary reloads to
19821         floating point registers to integer modes.  Revise some comments and
19822         cleanup some code.
19823         (TARGET_LEGITIMATE_ADDRESS_P): Define.
19824         (pa_legitimate_address_p): New.
19825         (pa_legitimize_reload_address): New.
19826         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
19827         (STRICT_REG_OK_FOR_BASE_P): New.
19828         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
19829         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
19831 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
19832             Andrew Dixie  <andrewd@gentrack.com>
19834         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
19835         flag set.
19837 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
19839         * expmed.c (extract_bit_field_1): Pass the full width of the
19840         structure to get_best_reg_extraction_insn.
19842 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19844         PR target/54601
19845         * configure.ac (use_cxa_atexit): Add AIX.
19846         * configure: Regenerate.
19848         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
19850 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
19852         PR debug/54793
19853         * final.c (need_profile_function): New variable.
19854         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
19855         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
19856         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
19857         notes, targetm.asm_out.function_prologue doesn't emit anything,
19858         HAVE_prologue and profiler should be emitted before prologue,
19859         set need_profile_function instead of emitting it.
19860         (final_scan_insn): If need_profile_function, emit
19861         profile_function on the first NOTE_INSN_BASIC_BLOCK or
19862         NOTE_INSN_FUNCTION_BEG note.
19864 2013-02-01  Richard Henderson  <rth@redhat.com>
19866         * config/rs6000/rs6000.md (smulditi3): New.
19867         (umulditi3): New.
19869         * config/alpha/alpha.md (umulditi3): New.
19871 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19873         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
19874         (ASM_OUTPUT_ALIGNED_LOCAL): New.
19876 2013-02-01  Richard Biener  <rguenther@suse.de>
19878         PR tree-optimization/56113
19879         * tree-ssa-structalias.c (label_visit): Reduce work for
19880         single-predecessor nodes.
19882 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
19884         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
19885         range isn't testing for zero.
19887 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
19889         PR middle-end/56113
19890         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
19892 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
19893             Nick Clifton  <nickc@redhat.com>
19895         * config/v850/constraints.md (Q): Define as a memory constraint.
19896         * config/v850/predicates.md (label_ref_operand): New predicate.
19897         (e3v5_shift_operand): New predicate.
19898         (ior_operator): New predicate.
19899         * config/v850/t-v850: Add e3v5 multilib.
19900         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
19901         (v850_gen_movdi): Prototype.
19902         * config/v850/v850.c: Add support for e3v5 architecture.
19903         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
19904         TARGET_V850E_UP.
19905         (construct_save_jarl): Add e3v5 long JARL support.
19906         (v850_adjust_insn_length): New function.  Adjust length of call
19907         insns when using e3v5 instructions.
19908         (v850_gen_movdi): New function: Generate instructions to move a
19909         DImode value.
19910         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
19911         (CPP_SPEC): Define __v850e3v5__ as appropriate.
19912         (TARGET_USE_FPU): Enable for e3v5.
19913         (CONST_OK_FOR_W): New macro.
19914         (ADJUST_INSN_LENGTH): Define.
19915         * config/v850/v850.md (UNSPEC_LOOP): Define.
19916         (attr cpu): Add v850e3v5.
19917         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
19918         (movdi): New pattern.
19919         (movdi_internal): New pattern.
19920         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
19921         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
19922         (cstoresf4): Likewise.
19923         (cstoredf4): Likewise.
19924         (insv): New pattern.
19925         (rotlso3_a): New pattern.
19926         (rotlsi3_b): New pattern
19927         (rotlsi3_v850e3v5): New pattern.
19928         (doloop_begin): New pattern.
19929         (fix_loop_counter): New pattern.
19930         (doloop_end): New pattern.
19931         (branch_normal): Add e3v5 long branch support.
19932         (branch_invert): Likewise.
19933         (branch_z_normal): Likewise.
19934         (branch_z_invert): Likewise.
19935         (branch_nz_normal): Likewise.
19936         (branch_nz_invert): Likewise.
19937         (call_internal_short): Add e3v5 register-indirect JARL support.
19938         (call_internal_long): Likewise.
19939         (call_value_internal_short): Likewise.
19940         (call_value_internal_long): Likewise.
19941         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
19942         (mloop): New option.
19943         * config.gcc: Add support for configuring v840e3v5 target.
19944         * doc/invoke.texi: Document new v850 specific command line options.
19946 2013-01-31  Paul Koning  <ni1d@arrl.net>
19948         PR debug/55059
19949         PR debug/54508
19950         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
19951         children if parent is a class.
19952         (prune_unused_types_prune): Don't add DW_AT_declaration.
19954 2013-01-31  Richard Biener  <rguenther@suse.de>
19956         PR tree-optimization/56157
19957         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
19958         match up operand with SLP child.
19960 2013-01-31  Jason Merrill  <jason@redhat.com>
19962         PR debug/54410
19963         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
19964         parameters the first time.
19965         (gen_scheduled_generic_parms_dies): Check completeness here.
19967 2013-01-31  Richard Biener  <rguenther@suse.de>
19969         PR middle-end/53073
19970         * common.opt (faggressive-loop-optimizations): New flag,
19971         enabled by default.
19972         * doc/invoke.texi (faggressive-loop-optimizations): Document.
19973         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
19974         infer_loop_bounds_from_undefined by it.
19976 2013-01-31  Richard Biener  <rguenther@suse.de>
19978         PR tree-optimization/56150
19979         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
19980         visit virtual operands.
19981         (find_uses_to_rename_bb): Likewise.
19983 2013-01-31  Richard Biener  <rguenther@suse.de>
19985         PR tree-optimization/56150
19986         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
19987         mixed store non-store stmts.
19989 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
19991         PR sanitizer/55374
19992         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
19993         LIBASAN_EARLY_SPEC is defined.
19994         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
19995         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
19996         before %o.
19997         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
19999         PR c++/55742
20000         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20001         invalid args instead of ICEing on it.
20002         (ix86_valid_target_attribute_tree): Return error_mark_node if
20003         ix86_valid_target_attribute_inner_p failed.
20004         (ix86_valid_target_attribute_p): Return false only if
20005         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20006         target("default") attribute.
20007         (sorted_attr_string): Change argument from const char * to tree,
20008         merge in all target attribute arguments rather than just one.
20009         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20010         instead of free.  Avoid using strcat.
20011         (ix86_mangle_function_version_assembler_name): Mangle
20012         target("default") as if no target attribute is present.  Adjust
20013         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20014         instead of xmalloc and XDELETEVEC instead of free.
20015         (ix86_function_versions): Don't return true if one of the decls
20016         doesn't have target attribute.  If they don't and one of the decls
20017         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20018         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20019         (ix86_supports_function_versions): Remove.
20020         (make_name): Fix up formatting.
20021         (make_dispatcher_decl): Remove resolver_name and its initialization.
20022         Avoid leaking memory.
20023         (is_function_default_version): Return true if there is
20024         target("default") attribute rather than no target attribute at all.
20025         (make_resolver_func): Avoid leaking memory.
20026         (ix86_generate_version_dispatcher_body): Likewise.
20027         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20028         * target.def (supports_function_versions): Remove.
20029         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20030         * doc/tm.texi: Regenerated.
20032 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20034         PR rtl-optimization/56144
20035         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20036         for values with side effects.
20038 2013-01-30  Richard Biener  <rguenther@suse.de>
20040         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20041         (sparseset_pop): Likewise.
20042         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20043         to be able to use quick_push in the worker loop.
20045 2013-01-30  Marek Polacek  <polacek@redhat.com>
20047         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20049 2013-01-30  Richard Biener  <rguenther@suse.de>
20051         PR lto/56147
20052         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20054 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20056         PR tree-optimization/56064
20057         * fixed-value.c (fixed_from_double_int): New function.
20058         * fixed-value.h (fixed_from_double_int): New prototype.
20059         (const_fixed_from_double_int): New static inline function.
20060         * fold-const.c (native_interpret_fixed): New static function.
20061         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20062         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20063         (native_encode_fixed): New static function.
20064         (native_encode_expr) <FIXED_CST>: Use it.
20065         (native_interpret_int): Move double_int worker code to...
20066         * double-int.c (double_int::from_buffer): ...this new static method.
20067         * double-int.h (double_int::from_buffer): Prototype it.
20069 2013-01-30  Richard Biener  <rguenther@suse.de>
20071         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20072         New pointer-map and obstack.
20073         (init_alias_vars): Allocate pointer-map and obstack.
20074         (delete_points_to_sets): Free them.
20075         (find_what_var_points_to): Cache result.
20076         (find_what_p_points_to): Adjust for changed interface of
20077         find_what_var_points_to.
20078         (compute_points_to_sets): Likewise.
20079         (ipa_pta_execute): Likewise.
20081 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20083         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20084         * configure: Regenerate.
20085         * config.in: Regenerate.
20086         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20087         #nobits/#progbits if supported.
20089 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20091         PR target/56121
20092         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20093         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20094         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20096 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20098         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20099         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20101 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20103         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20104         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20106 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20108         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20109         declaration.
20110         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20111         * config/arm/cortex-a7.md: New bypasses using
20112         arm_mac_accumulator_is_result.
20114 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20116         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20117         (cortex_a7_neon_mla): Likewise.
20118         (cortex_a7_fpfmad): New reservation.
20119         (cortex_a7_fpmacs): Use ffmas and update required units.
20120         (cortex_a7_fpmuld): Update required units and latency.
20121         (cortex_a7_fpmacd): Likewise.
20122         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20123         (cortex_a7_neon). Likewise.
20124         (bypass) Update participating units.
20126 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20128         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20129         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20130         from fmac to ffma.
20131         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20132         (vfp_fmul): Use ffmad.
20133         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20134         (cortex_r4_fmacd): Use ffmad.
20135         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20136         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20137         (cortex_a9_fmacd): Use ffmad.
20138         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20139         (cortex_a8_vfp_macd): Use ffmad.
20140         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20141         (cortex_a5_fpmacd): Use ffmad.
20142         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20143         (cortex_a15_vfp_macd): Use ffmad.
20144         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20146 2013-01-29  Jason Merrill  <jason@redhat.com>
20148         PR libstdc++/54314
20149         * varasm.c (default_assemble_visibility): Don't warn about
20150         visibility on artificial decls.
20152 2013-01-29  Richard Biener  <rguenther@suse.de>
20154         PR tree-optimization/56113
20155         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20156         the bitmap leader.
20157         (label_visit): Free duplicate bitmaps and record the leader instead.
20158         (perform_var_substitution): Adjust.
20160 2013-01-29  Richard Biener  <rguenther@suse.de>
20162         PR tree-optimization/55270
20163         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20164         the CFG, schedule loops for fixup.
20166 2013-01-29  Nick Clifton  <nickc@redhat.com>
20168         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20169         SP_REG.
20171 2013-01-28  Leif Ekblad  <leif@rdos.net>
20173         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20174         * config/i386/i386.h (TARGET_RDOS): New macro.
20175         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20176         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20177         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20178         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20179         DEFAULT_LARGE_SECTION_THRESHOLD.
20180         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20181         * config/i386/rdos.h: New file.
20182         * config/i386/rdos64.h: New file.
20184 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20186         PR other/54814
20187         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20188         TEST_HARD_REG_BIT.
20190 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20192         PR rtl-optimization/56117
20193         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20194         call cselib_lookup_from_insn on the MEM before calling
20195         add_insn_mem_dependence.
20197 2013-01-28  Richard Biener  <rguenther@suse.de>
20199         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20200         to a stmt that didn't have one.
20201         (copy_phis_for_bb): Likewise for PHI arguments.
20202         (copy_debug_stmt): Likewise for debug stmts.
20204 2013-01-28  Richard Biener  <rguenther@suse.de>
20206         PR tree-optimization/56034
20207         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20208         (partition_builtin_p): Adjust.
20209         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20210         it is the last partition.
20211         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20212         up the vertex for the definition.
20213         (classify_partition): Classify whether a partition is a
20214         PKIND_REDUCTION, thus has uses outside of the loop.
20215         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20216         Merge all PKIND_REDUCTION partitions into the last partition.
20217         (tree_loop_distribution): Seed partitions from reductions as well.
20219 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20221         PR tree-optimization/56125
20222         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20223         pow(x,c) into sqrt(x) * powi(x, n/2) or
20224         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20225         optimizing for size.
20226         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20227         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20228         integer.
20230         PR tree-optimization/56094
20231         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20232         to UNKNOWN_LOCATION while gimplifying expr.
20234 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
20236         PR target/56114
20237         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20238         operand 0 in movabs insn template for -masm=intel asm alternative.
20239         (*movabs<mode>_2): Ditto for operand 1.
20241 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
20243         PR target/54663
20244         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20245         of microblaze-c.o
20247 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20249         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20250         tm_file.
20252 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20254         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20255         Undef to avoid warning.
20257 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20259         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20260         * configure: Regenerate.
20262 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
20264         PR tree-optimization/56098
20265         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20266         for stmts with volatile ops.
20267         (cond_store_replacement): Don't optimize if assign has volatile ops.
20268         (cond_if_else_store_replacement_1): Don't optimize if either
20269         then_assign or else_assign have volatile ops.
20270         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20271         volatile ops.
20273 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20275         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20277 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20279         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20280         missing ':' in asm example.
20282 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
20284         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20285         entries into lane and laneq entries.
20286         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20287         Remove AdvSIMD scalar modes.
20288         (aarch64_sq<r>dmulh_laneq<mode>): New.
20289         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20290         modes.
20291         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20292         builtin implementations to relfect changes in RTL in aarch64-simd.md.
20293         * config/aarch64/iterators.md (VCOND): New.
20294         (VCONQ): New.
20296 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20298         PR target/54222
20299         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20300         Add NULL LIBNAME argument to existing definitions.
20301         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20302         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20303         * config/avr/avr.c (DEF_BUILTIN): Same.
20304         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20305         (avr_expand_builtin): Expand to a vanilla call if a libgcc
20306         implementation is available (DECL_ASSEMBLER_NAME is set).
20307         (avr_fold_absfx): New static function.
20308         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20309         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20310         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20311         AVR_BUILTIN_ABSLLK.
20312         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20313         (abshk, absk, abslk, absllk): Provide as static inline functions.
20315 2013-01-25  Marek Polacek  <polacek@redhat.com>
20317         PR tree-optimization/56035
20318         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20320 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
20322         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20323         (*movtf_internal_rex64): Add (!o,C) alternative
20324         (*movxf_internal_rex64): Ditto.
20325         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20327 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
20329         * doc/invoke.texi: fix typo.
20330         * doc/objc.texi: fix typo.
20332 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
20334         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20335         for the first two alternatives.
20337 2013-01-24  Diego Novillo  <dnovillo@google.com>
20339         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
20340         (ggc-zone.o): Remove.
20341         * configure.ac: Remove option --with-gc.
20342         * configure: Re-generate.
20343         * doc/install.texi: Remove documentation for --with-gc.
20344         * gengtype.c (write_enum_defn): Remove.  Update all users.
20345         (write_Types_process_field): Remove generation of gt_e_* argument.
20346         (output_type_enum): Remove.  Update all users.
20347         (write_enum_defn): Remove.  Update all users.
20348         (enum alloc_zone): Remove.  Update all users.
20349         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20350         * ggc-common.c (ggc_splay_alloc): Remove first argument.
20351         Update all callers.
20352         (struct ptr_data): Remove field TYPE.  Update all users.
20353         (gt_pch_note_object): Remove argument TYPE.  Update all users.
20354         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20355         Update all users.
20356         * ggc-none.c (ggc_alloc_typed_stat): Remove.
20357         (struct alloc_zone): Remove.
20358         (ggc_internal_alloc_zone_stat): Remove.
20359         (ggc_internal_cleared_alloc_zone_stat): Remove.
20360         * ggc-page.c (ggc_alloc_typed_stat): Remove.
20361         (ggc_pch_count_object): Remove last argument.  Update all users.
20362         (ggc_pch_alloc_object): Remove last argument.  Update all users.
20363         (struct alloc_zone): Remove.
20364         * ggc-zone.c: Remove.
20365         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
20366         (struct alloc_zone): Remove.
20367         (ggc_alloc_typed_stat): Remove.
20368         (ggc_alloc_typed): Remove.
20369         (ggc_splay_alloc): Remove first argument.
20370         (rtl_zone): Remove.  Update all users.
20371         (tree_zone): Remove.  Update all users.
20372         (tree_id_zone): Remove.  Update all users.
20373         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
20374         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
20375         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
20376         * tree-ssanames.c: Remove references to zone allocator in comments.
20378 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
20380         * config/avr/avr.c (avr_out_fract): Make register numbers that
20381         might be outside of source operand signed.
20383 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
20385         * config/i386/constraints.md (Yf): New constraint.
20386         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20387         of f constraint to conditionaly disable x87 register preferences.
20388         (*movdf_internal): Ditto.
20389         (*movsf_internal): Ditto.
20391 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
20393         PR inline-asm/55934
20394         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20395         that have operands with impossible constraints.
20396         Add a FIXME for a speed-up opportunity.
20397         * lra-constraints.c (process_alt_operands): Verify that a class
20398         selected from constraints on asms is valid for the operand mode.
20399         (curr_insn_transform): Remove incorrect comment.
20401 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
20403         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20404         TOC operand is a valid symbol ref in the constant pool.
20406 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20408         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20410 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
20412         PR target/54222
20413         * config/avr/stdfix.h: New file.
20414         * t-avr (stdfix-gcc.h): New rule to build it.
20415         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20417 2013-01-23  Kostya Serebryany  <kcc@google.com>
20419         * config/darwin.h: remove dependency on
20420         CoreFoundation (asan on Mac OS).
20422 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
20424         PR target/49069
20425         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20426         instead of cmpdi_operand for first comparison operand.
20427         Don't assert that comparison operands aren't both constants.
20429 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20431         * doc/install.texi (Downloading the Source): Update references to
20432         downloading separate components.
20434 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20436         * doc/extend.texi (__int128): Improve grammar.
20438 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
20440         PR target/56028
20441         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20442         alternative to (o,r).
20443         (*movdi_internal_rex64): Remove (!o,n) alternative.
20444         (DImode immediate->memory splitter): Remove.
20445         (DImode immediate->memory peephole2): Remove.
20446         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20447         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20448         alternative to (!o,*r).
20449         (*movtf_internal_sse): New pattern.
20450         (*movxf_internal_rex64): New pattern.
20451         (*movxf_internal): Disable for TARGET_64BIT.
20452         (*movdf_internal_rex64): Remove (!o,F) alternative.
20454 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
20456         PR middle-end/56074
20457         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20458         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20459         * tree-vect-loop-manip.c (find_loop_location): Also ignore
20460         stmt locations where LOCATION_LOCUS of the stmt location is
20461         UNKNOWN_LOCATION or BUILTINS_LOCATION.
20463         PR target/55686
20464         * config/i386/i386.md (UNSPEC_STOS): New.
20465         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20466         *strsetqi_1): Add UNSPEC_STOS.
20468 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
20470         PR c++/56067
20471         * doc/invoke.texi: Remove left over -Wsynth example.
20473 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20475         PR tree-optimization/56051
20476         * fold-const.c (fold_binary_loc): Don't fold
20477         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20478         a narrowing conversion, or widening conversion from signed
20479         to unsigned.
20481 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20483         PR rtl-optimization/56023
20484         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20485         dependent on debug instruction.
20487 2013-01-21  Martin Jambor  <mjambor@suse.cz>
20489         PR middle-end/56022
20490         * function.c (allocate_struct_function): Call
20491         invoke_set_current_function_hook earlier.
20493 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20495         * reload1.c (init_reload): Only initialize reload_obstack
20496         during the first call.
20498 2013-01-21  Marek Polacek  <polacek@redhat.com>
20500         * cfgloop.c (verify_loop_structure): Fix up grammar.
20502 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
20504         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20505         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20507 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20509         PR target/56058
20510         * config/arm/marvell-pj4.md: Update copyright year.
20511         Fix up use of alu to alu_reg and simple_alu_imm.
20513 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20515         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20517 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
20519         PR target/55433
20520         * lra-constraints.c (curr_insn_transform): Don't reuse original
20521         insn for secondary memory move when memory mode should be different.
20523 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20525         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20526         atomic_storedi_1): New patterns.
20528 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
20530         btver2 pipeline descriptions.
20531         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20532         descriptions.
20533         * config/i386/i386.md (btver2_decode): New type attributes.
20534         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20535         type attributes.
20536         * config/i386/btver2.md: New file describing btver2 pipelines.
20538 2013-01-19  Andrew Pinski  <apinski@cavium.com>
20540         PR tree-optimization/52631
20541         * tree-ssa-sccvn (visit_use): Before looking up the original
20542         statement, try looking up the simplified expression.
20544 2013-01-19  Anthony Green  <green@moxielogic.com>
20546         * config/moxie/moxie.c (moxie_expand_prologue): Set
20547         current_function_static_stack_size.
20549 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20551         PR tree-optimization/56029
20552         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20553         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20555 2013-01-18  Sharad Singhai  <singhai@google.com>
20557         PR tree-optimization/55995
20558         * dumpfile.c (dump_loc): Print location only if available.
20559         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
20561 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
20563         PR target/55433
20564         * lra-constraints.c (curr_insn_transform): Reuse original insn for
20565         secondary memory move.
20566         (inherit_reload_reg): Use rclass instead of cl for
20567         check_secondary_memory_needed_p.
20569 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20571         PR middle-end/56015
20572         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
20573         the case where writing real complex part of target modifies op1.
20575 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
20577         * config/aarch64/aarch64-simd.md
20578         (aarch64_vcond_internal<mode>): Handle unordered cases.
20579         * config/aarch64/iterators.md (v_cmp_result): New.
20581 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
20582             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20584         * config/arm/marvell-pj4.md: New file.
20585         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
20586         * config/arm/arm.md (generic_sched): Add marvell_pj4.
20587         (generic_vfp): Likewise.
20588         * config/arm/arm-cores.def: Add marvell-pj4.
20589         * config/arm/arm-tune.md: Regenerate.
20590         * config/arm/arm-tables.opt: Regenerate.
20591         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
20592         * doc/invoke.texi: Document marvell-pj4.
20594 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
20596         * config/aarch64/arm_neon.h: Map scalar types to standard types.
20598 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20600         PR debug/54114
20601         PR debug/54402
20602         PR debug/49888
20603         * var-tracking.c (negative_power_of_two_p): New.
20604         (global_get_addr_cache, local_get_addr_cache): New.
20605         (get_addr_from_global_cache, get_addr_from_local_cache): New.
20606         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
20607         heading comment.
20608         (vt_stack_offset_p): Remove.
20609         (vt_canon_true_dep): Always canonicalize loc's address.
20610         (clobber_overlapping_mems): Make sure we have a MEM.
20611         (local_get_addr_clear_given_value): New.
20612         (val_reset): Clear local cached entries.
20613         (compute_bb_dataflow): Create and release the local cache.
20614         Disable duplicate MEMs clobbering.
20615         (emit_notes_in_bb): Clobber MEMs likewise.
20616         (vt_emit_notes): Create and release the local cache.
20617         (vt_initialize, vt_finalize): Create and release the global
20618         cache, respectively.
20619         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
20621 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20623         PR libmudflap/53359
20624         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
20625         not found in the symtab.
20627 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20629         PR debug/56006
20630         PR rtl-optimization/55547
20631         PR rtl-optimization/53827
20632         PR debug/53671
20633         PR debug/49888
20634         * alias.c (offset_overlap_p): New, factored out of...
20635         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
20636         the conservative special case for symbolic constants.  Don't
20637         adjust zero sizes on alignment.
20639 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
20641         PR rtl-optimization/52573
20642         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
20643         REG_UNUSED for the same register.
20645 2013-01-17  Richard Biener  <rguenther@suse.de>
20646             Marek Polacek  <polacek@redhat.com>
20648         PR rtl-optimization/55833
20649         * loop-unswitch.c (unswitch_loops): Move loop verification...
20650         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
20651         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
20652         Set it to true when we're removing a loop from hierarchy tree in
20653         an irreducible region.
20654         (fix_bb_placements): Adjust caller.
20655         (fix_loop_placements): Likewise.
20657 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
20659         * config/avr/builtins.def (DEF_BUILTIN): Factor out
20660         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
20661         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
20662         Remove ID.  Adjust comments.
20663         * config/avr/avr-c.c (avr_builtin_name): Remove.
20664         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
20665         * config/avr/avr.c (avr_tolower): New static function.
20666         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
20667         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
20668         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
20669         default expansion.
20671 2013-01-17  Jan Hubicka  <jh@suse.cz>
20673         PR tree-optimization/55273
20674         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
20676 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
20678         PR target/55981
20679         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
20680         store through atomic_store<mode>_1.
20681         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
20683 2013-01-17  Martin Jambor  <mjambor@suse.cz>
20685         PR tree-optimizations/55264
20686         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
20687         for virtual methods.
20688         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
20689         virtual methods before inlining is over.
20690         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
20691         virtual functions.
20692         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
20693         non-virtual.
20695 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
20697         PR rtl-optimization/56005
20698         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
20699         pending reads for prefetch.
20701 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
20703         * config/aarch64/aarch64.md
20704         (*cstoresi_neg_uxtw): New pattern.
20705         (*cmovsi_insn_uxtw): New pattern.
20706         (*<optab>si3_uxtw): New pattern.
20707         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
20708         (*<optab>si3_insn_uxtw): New pattern.
20709         (*bswapsi2_uxtw): New pattern.
20711 2013-01-16  Richard Biener  <rguenther@suse.de>
20713         * tree-inline.c (tree_function_versioning): Remove set but
20714         never used variable.
20716 2013-01-16  Richard Biener  <rguenther@suse.de>
20718         PR tree-optimization/55964
20719         * tree-flow.h (rename_variables_in_loop): Remove.
20720         (rename_variables_in_bb): Likewise.
20721         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
20722         (copy_loop_before): Adjust and delete update-ssa status.
20723         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
20724         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
20725         (rename_variables_in_loop): Remove.
20726         (slpeel_update_phis_for_duplicate_loop): Likewise.
20727         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
20728         use available cfg machinery instead of duplicating it.
20729         Update PHI nodes and perform poor-mans SSA update here.
20730         (slpeel_tree_peel_loop_to_edge): Adjust.
20732 2013-01-16  Richard Biener  <rguenther@suse.de>
20734         PR tree-optimization/54767
20735         PR tree-optimization/53465
20736         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
20737         (vrp_visit_phi_node): For PHI arguments coming via backedges
20738         drop all symbolical range information.
20739         (execute_vrp): Compute backedges.
20741 2013-01-16  Richard Biener  <rguenther@suse.de>
20743         * doc/install.texi: Update CLooG and ISL requirements to
20744         0.18.0 and 0.11.1.
20746 2013-01-16  Christian Bruel  <christian.bruel@st.com>
20748         PR target/55301
20749         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
20750         (broken_move): Handle UNSPECV_SP_SWITCH_B.
20751         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
20753 2013-01-16  DJ Delorie  <dj@redhat.com>
20755         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
20756         (UNSPECV_SP_SWITCH_E): New.
20757         (sp_switch_1): Change to an unspec.
20758         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
20759         replace $r15.
20761 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
20763         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
20764         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
20765         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
20766         (expand_mem_thread_fence): Ditto.
20767         (expand_mem_signal_fence): Ditto.
20768         (expand_atomic_load): Ditto.
20769         (expand_atomic_store): Ditto.
20771 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
20773         PR rtl-optimization/55547
20774         PR rtl-optimization/53827
20775         PR debug/53671
20776         PR debug/49888
20777         * alias.c (memrefs_conflict_p): Set sizes to negative after
20778         AND adjustments.
20780 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20782         PR target/55940
20783         * function.c (thread_prologue_and_epilogue_insns): Always
20784         add crtl->drap_reg to set_up_by_prologue.set, even if
20785         stack_realign_drap is false.
20787 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
20789         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
20790         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
20791         *call): Fix indention.
20793 2013-01-15  Tom de Vries  <tom@codesourcery.com>
20795         PR target/55876
20796         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
20797         Update comment.
20799 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
20801         PR rtl-optimization/55153
20802         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
20804 2013-01-15  Martin Jambor  <mjambor@suse.cz>
20806         PR tree-optimization/55920
20807         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
20808         accesses as grp_to_be_debug_replaced.
20810 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20812         PR tree-optimization/55920
20813         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
20814         there is non-useless type conversion needed from debug rhs to lhs,
20815         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
20817 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
20818             Mikael Pettersson  <mikpe@it.uu.se>
20820         PR target/43961
20821         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
20822         Thumb.
20823         (ASM_OUTPUT_CASE_LABEL): Remove.
20824         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
20825         * final.c (shorten_branches): Update alignment of labels before
20826         jump tables if CASE_VECTOR_SHORTEN_MODE.
20828 2013-01-15  Richard Biener  <rguenther@suse.de>
20830         PR bootstrap/55961
20831         * system.h: Do not include gmp.h for building host tools.
20833 2013-01-15  Richard Biener  <rguenther@suse.de>
20835         PR middle-end/55882
20836         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
20837         account for bitpos when computing alignment.
20839 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
20841         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
20842         (ix86_target_macros_internal): Likewise.
20844         * config/i386/i386.c (m_CORE2I7): Removed.
20845         (m_CORE_HASWELL): New macro.
20846         (m_CORE_ALL): Likewise.
20847         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
20848         (initial_ix86_arch_features): Likewise.
20849         (processor_target_table): Initializations for Core avx2.
20850         (cpu_names): New names "core-avx2".
20851         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
20852         PROCESSOR_CORE_HASWELL.
20853         (ix86_issue_rate): New case.
20854         (ia32_multipass_dfa_lookahead): Likewise.
20855         (ix86_sched_init_global): Likewise.
20857         * config/i386/i386.h (TARGET_HASWELL): New macro.
20858         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
20859         (processor_type): New PROCESSOR_HASWELL.
20861 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20863         PR tree-optimization/55955
20864         * tree-vect-loop.c (vectorizable_reduction): Give up early on
20865         *SHIFT_EXPR and *ROTATE_EXPR codes.
20867         PR tree-optimization/48766
20868         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
20869         -ftrapv disable -fwrapv.
20871 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20873         PR target/55974
20874         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
20875         etc. to 1 and not to __flash.
20876         Use LL suffix for __INT24_MAX__ with -mint8.
20877         Use ULL suffix for __UINT24_MAX__ with -mint8.
20879 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20881         * config/avr/avr-arch.h
20882         (struct base_arch_s): Use typedef avr_arch_t instead.
20883         (struct arch_info_s): Use typedef avr_arch_info_t instead.
20884         (struct mcu_type_s): Use typedef avr_mcu_t instead.
20885         * config/avr/avr.c: Same.
20886         * config/avr/avr-devices.c: Same.
20887         * config/avr/driver-avr.c: Same.
20888         * config/avr/gen-avr-mmcu-texi.c: Same.
20889         * config/avr/avr-mcus.def: Adjust comment.
20891 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
20893         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
20894         * config/aarch64/iterators.md (VALLDI): New.
20896 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
20897             Andi Kleen  <ak@linux.intel.com>
20899         PR target/55948
20900         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
20901         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
20902         memmodel flag.
20904 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20906         * config/avr/avr-stdint.h: Remove trailing blanks.
20907         * config/avr/avr-log.h: Same.
20908         * config/avr/avr-arch.h: Same.
20909         * config/avr/avr-devices.c: Same.
20910         * config/avr/avr-dimode.md: Same.
20911         * config/avr/predicates.md: Same.
20912         * config/avr/avr-c.c: Same.  And fix typo.
20914         * config/avr/avr-protos.h: Same.  And:
20915         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
20916         (init_cumulative_args): Rename to avr_init_cumulative_args.
20917         (expand_prologue): Rename to avr_expand_prologue.
20918         (expand_epilogue): Rename to avr_expand_epilogue.
20919         (adjust_insn_length): Rename to avr_adjust_insn_length.
20920         (notice_update_cc): Rename to avr_notice_update_cc.
20921         (final_prescan_insn): Rename to avr_final_prescan_insn.
20922         * config/avr/avr.c: Same.
20923         * config/avr/avr.h: Same.
20924         * config/avr/avr.md: Remove trailing blanks.
20925         (prologue): Use avr_expand_prologue.
20926         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
20928 2013-01-14  Richard Biener  <rguenther@suse.de>
20930         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
20931         verify_location, collect_subblocks): New functions.
20932         (verify_gimple_in_cfg): Verify that locations only reference
20933         BLOCKs in the functions BLOCK tree.
20935 2013-01-14  Richard Biener  <rguenther@suse.de>
20937         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
20938         PHI argument.
20939         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
20940         unshare reference.
20941         (insert_out_of_ssa_copy_on_edge): Likewise.
20942         (rewrite_close_phi_out_of_ssa): Likewise.
20943         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
20944         debug expressions.
20945         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
20946         propagated constants.
20947         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
20948         can not be shared.
20950 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20952         * config/avr/avr-modes.def: Add GPL copyright notice.
20954 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
20956         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
20957         MEMMODEL_MASK to determine memory model.
20958         (atomic_store<mode>): Ditto from operands[2].
20959         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
20961 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
20963         PR fortran/55935
20964         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
20965         (fold_gimple_assign): Don't call unshare_expr here.
20966         (fold_ctor_reference): Call unshare_expr.
20968 2013-01-13  Terry Guo  <terry.guo@arm.com>
20970         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
20971         * doc/fragments.texi: Document MULTILIB_REUSE.
20972         * gcc.c (multilib_reuse): New internal spec.
20973         (set_multilib_dir): Also search multilib from multilib_reuse.
20974         * genmultilib (tmpmultilib3): Refactor code.
20975         (tmpmultilib4): Ditto.
20976         (multilib_reuse): New multilib argument.
20978 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
20980         * Makefile.in: Update copyright.
20982 2013-01-12  Tom de Vries  <tom@codesourcery.com>
20984         PR middle-end/55890
20985         * calls.c (expand_call): Check if arg_nr is valid.
20987 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
20989         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
20990         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
20991         documentation.  Add missing '__' in front of
20992         __builtin_ia32_packssdw256.
20994 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20996         PR target/55719
20997         * config/s390/s390.c (s390_preferred_reload_class): Do not return
20998         NO_REGS for larl operands.
20999         (s390_reload_larl_operand): Use s390_load_address instead of
21000         emit_move_insn.
21002 2013-01-11  Richard Biener  <rguenther@suse.de>
21004         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21005         (verify_node_sharing): ... here.
21006         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21008 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21010         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21011         Substitute TREECHECKING.
21012         * configure: Regenerate.
21013         * Makefile.in (TREECHECKING): New.
21015 2013-01-11  Richard Guenther  <rguenther@suse.de>
21017         PR tree-optimization/44061
21018         * tree-vrp.c (extract_range_basic): Compute zero as
21019         value-range for __builtin_constant_p of function parameters.
21021 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21023         Update copyright years.
21025 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21027         PR rtl-optimization/55672
21028         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21029         const to be eliminable.
21031 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21033         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21034         * configure: Regenerate.
21036 2013-01-10  Richard Biener  <rguenther@suse.de>
21038         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21040 2013-01-10  Richard Biener  <rguenther@suse.de>
21042         PR bootstrap/55792
21043         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21044         locations for virtual PHI arguments.
21045         (rewrite_update_phi_arguments): Likewise.
21047 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21049         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21050         on to assembler.
21052 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21054         PR tree-optimization/55921
21055         * tree-complex.c (expand_complex_asm): New function.
21056         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21058 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21060         PR target/55718
21061         * config/s390/s390.c (s390_symref_operand_p)
21062         (s390_loadrelative_operand_p): Merge the two functions.
21063         (s390_check_qrst_address, print_operand_address): Add parameters
21064         to s390_loadrelative_operand_p invokation.
21065         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21066         (s390_reload_larl_operand, s390_secondary_reload): Use
21067         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21068         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21070 2013-01-09  Mike Stump  <mikestump@comcast.net>
21072         * dse.c (record_store): Remove unnecessary assert.
21074 2013-01-09  Jan Hubicka  <jh@suse.cz>
21076         PR tree-optimization/55569
21077         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21078         * cfgloop.h (scale_loop_profile): Likewise.
21080 2013-01-09  Jan Hubicka  <jh@suse.cz>
21082         PR lto/45375
21083         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21084         functions.
21085         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21087 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21089         PR middle-end/55114
21090         * expr.h (maybe_emit_group_store): Declare.
21091         * expr.c (maybe_emit_group_store): New function.
21092         * builtins.c (expand_builtin_int_roundingfn): Call it.
21093         (expand_builtin_int_roundingfn_2): Likewise.
21095 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21097         PR rtl-optimization/55829
21098         * lra-constraints.c (match_reload): Add code for absent output.
21099         (curr_insn_transform): Add code for reloads of matched inputs
21100         without output.
21102 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21104         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21105         attribute of movddup insn to DF.
21106         (*vec_interleave_lowv2df): Ditto.
21107         (vec_dupv2df): Ditto.
21109 2013-01-09  Jan Hubicka  <jh@suse.cz>
21111         PR tree-optimiation/55875
21112         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21113         EVERY_ITERATION parameter.
21114         (number_of_iterations_exit): Check if exit is executed every iteration.
21115         (idx_infer_loop_bounds): Similarly here.
21116         (n_of_executions_at_most): Simplify
21117         to only test for cases where statement is dominated by the
21118         particular bound; handle correctly the "postdominance" test.
21119         (scev_probably_wraps_p): Use max loop iterations info
21120         as a global bound first.
21122 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21123             Nick Clifton  <nickc@redhat.com>
21125         * config/v850/v850.md (cbranchsf4): New pattern.
21126         (cstoresf4): New pattern.
21127         (cbranchdf4): New pattern.
21128         (cstoredf4): New pattern.
21129         (movsicc): Disallow floating point comparisons.
21130         (cmpsf_le_insn): Fix order of operators.
21131         (cmpsf_lt_insn): Likewise.
21132         (cmpsf_eq_insn): Likewise.
21133         (cmpdf_le_insn): Likewise.
21134         (cmpdf_lt_insn): Likewise.
21135         (cmpdf_eq_insn): Likewise.
21136         (cmpsf_ge_insn): Use LE comparison.
21137         (cmpdf_ge_insn): Likewise.
21138         (cmpsf_gt_insn): Use LT comparison.
21139         (cmpdf_gt_insn): Likewise.
21140         (cmpsf_ne_insn): Delete pattern.
21141         (cmpdf_ne_insn): Delete pattern.
21142         * config/v850/v850.c (v850_gen_float_compare): Use
21143         gen_cmpdf_eq_insn for NE comparison.
21144         (v850_float_z_comparison_operator)
21145         (v850_float_nz_comparison_operator): Move from here ...
21146         * config/v850/predicates.md: ... to here.  Move GT and GE
21147         comparisons into v850_float_z_comparison_operator.
21148         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21149         Delete prototype.
21150         (v850_float_nz_comparison_operator): Likewise.
21152 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21154         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21155         with calls to gen_insvsi/gen_insvdi.
21157 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21159         * config/i386/i386.c (initial_ix86_tune_features): Set up
21160         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21162 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21163             Jakub Jelinek  <jakub@redhat.com>
21165         PR tree-optimization/48189
21166         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21167         If nitercst is 0, don't predict the exit edge.
21169 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21171         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21172         in asm_fprintf with reg_names.
21173         (aarch64_print_operand_address): Likewise.
21174         (aarch64_return_addr): Likewise.
21175         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21177 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21179         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21180         (INT_U6_BITS): Likewise.
21181         * config/pa/predicates.md (uint6_operand): New predicate.
21182         (shift5_operand, shift6_operand): Likewise.
21183         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21184         arith32_operand.
21185         (lshrdi3): Use shift6_operand.
21186         (shrpsi4, shrpdi4): New insn patterns.
21187         (extzv): Delete expander.
21188         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21189         predicates in unamed zero extract patterns.  Tighten common constraint.
21190         (extv): Delete expander.
21191         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21192         predicates in unamed sign extract patterns.  Tighten common constraint.
21193         (insv): Delete expander.
21194         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21195         predicates in unamed insert patterns.  Tighten common constraint.
21196         Change uint32_operand predicate to uint6_operand predicate in unamed
21197         DImode pattern to insert constant values of type 1...1xxxx.
21199 2013-01-04  Jan Hubicka  <jh@suse.cz>
21201         PR tree-optimization/55823
21202         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21203         issue.
21205 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21206             Uros Bizjak  <ubizjak@gmail.com>
21208         PR rtl-optimization/55845
21209         * df-problems.c (can_move_insns_across): Stop scanning at
21210         volatile_insn_p source instruction or give up if
21211         across_from .. across_to range contains any volatile_insn_p
21212         instructions.
21214 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21216         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21217         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21218         Declare.
21219         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21220         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21222 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21224         PR fortran/55341
21225         * asan.c (asan_clear_shadow): New function.
21226         (asan_emit_stack_protection): Use it.
21228 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21230         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21231         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21232         with tab instead of space.
21234 2013-01-08  Nick Clifton  <nickc@redhat.com>
21236         * config/rl78/rl78.c (rl78_expand_prologue): Always select
21237         register bank 0 at the start of an interrupt handler.
21238         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21239         MDBH registers.
21241 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21243         * config/aarch64/aarch64-simd.md
21244         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21245         (aarch64_simd_bsl): Likewise.
21246         (aarch64_vcond_internal<mode>): Likewise.
21247         (vcond<mode><mode>): Likewise.
21248         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21249         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21251 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21253         * config/aarch64/aarch64-builtins.c
21254         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21256 2013-01-08  Martin Jambor  <mjambor@suse.cz>
21258         PR debug/55579
21259         * tree-sra.c (analyze_access_subtree): Return true also after
21260         potentially creating a debug-only replacement.
21262 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21264         PR middle-end/55890
21265         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21267         PR tree-optimization/54120
21268         * tree-vrp.c (range_fits_type_p): Don't allow
21269         src_precision < precision from signed vr to unsigned_p
21270         if vr->min or vr->max is negative.
21271         (simplify_float_conversion_using_ranges): Test can_float_p
21272         against CODE_FOR_nothing.
21274 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21275             Richard Biener  <rguenther@suse.de>
21277         PR middle-end/55851
21278         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21279         types instead of just INTEGER_TYPE types.
21281 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
21283         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21284         TF_SIZE): Define.
21286 2013-01-07  Steve Ellcey  <sellcey@mips.com>
21288         PR target/42661
21289         * config/mips/mips.opt: Change mad to mmad to match documentation.
21291 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21293         PR target/55897
21294         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21295         .progmemx.data now.
21297 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21299         PR target/55897
21300         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21301         (avr_addrspace_t): Add .section_name field.
21302         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21303         array size.
21304         (avr_addrspace): Same.  Initialize .section_name.  Remove last
21305         NULL entry.  Put __memx into .progmemx.data.
21306         (progmem_section_prefix): Remove.
21307         (avr_asm_init_sections): No need to initialize progmem_section.
21308         (avr_asm_named_section): Use avr_addrspace[].section_name to get
21309         section name prefix.
21310         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
21311         retrieve the progmem section.
21312         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21313         boundary to run over avr_addrspace[].
21314         (avr_register_target_pragmas): Ditto.
21316 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
21318         * varasm.c (output_constant_def_contents): For asan_protect_global
21319         protected strings, adjust DECL_ALIGN if needed, before testing for
21320         anchored symbols.
21321         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21322         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
21323         normal decls.
21324         (output_object_block): For asan protected decls, emit asan padding
21325         after their contents.
21326         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21327         (asan_finish_file): Test it here instead.
21329 2013-01-07  Nick Clifton  <nickc@redhat.com>
21330             Matthias Klose  <doko@debian.org>
21331             Doug Kwan  <dougkwan@google.com>
21332             H.J. Lu  <hongjiu.lu@intel.com>
21334         PR driver/55470
21335         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21337         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21339         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21341         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21343         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21345 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21347         PR target/54461
21348         * doc/install.texi (Cross-Compiler-Specific Options): Document
21349         --with-avrlibc.
21351 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
21353         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21354         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21355         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21356         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21357         vqmovun_high_s64): Fix source operand number and update copyright.
21359 2013-01-07  Richard Biener  <rguenther@suse.de>
21361         PR middle-end/55890
21362         * gimple.h (gimple_call_builtin_p): New overload.
21363         * gimple.c (validate_call): New function.
21364         (gimple_call_builtin_p): Likewise.
21365         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21366         Use gimple_call_builtin_p.
21367         (find_func_clobbers): Likewise.
21368         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21369         (strlen_optimize_stmt): Likewise.
21371 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21373         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21374         (vld1q_dup_*): Likewise.
21375         (vld1_*): Likewise.
21376         (vld1q_*): Likewise.
21377         (vld1_lane_*): Likewise.
21378         (vld1q_lane_*): Likewise.
21380 2013-01-07  Richard Biener  <rguenther@suse.de>
21382         * lto-streamer.h (LTO_minor_version): Bump to 2.
21384 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21386         * config/aarch64/aarch64-protos.h
21387         (aarch64_const_double_zero_rtx_p): Rename to...
21388         (aarch64_float_const_zero_rtx_p): ...this.
21389         (aarch64_float_const_representable_p): New.
21390         (aarch64_output_simd_mov_immediate): Likewise.
21391         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21392         move immediate case.
21393         * config/aarch64/aarch64.c
21394         (aarch64_const_double_zero_rtx_p): Rename to...
21395         (aarch64_float_const_zero_rtx_p): ...this.
21396         (aarch64_print_operand): Allow printing of new constants.
21397         (aarch64_valid_floating_const): New.
21398         (aarch64_legitimate_constant_p): Check for valid floating-point
21399         constants.
21400         (aarch64_simd_valid_immediate): Likewise.
21401         (aarch64_vect_float_const_representable_p): New.
21402         (aarch64_float_const_representable_p): Likewise.
21403         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21404         (aarch64_output_simd_mov_immediate): New.
21405         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21406         (*movdf_aarch64): Likewise.
21407         * config/aarch64/constraints.md (Ufc): New.
21408         (Y): call aarch64_float_const_zero_rtx.
21409         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21411 2013-01-07  Richard Biener  <rguenther@suse.de>
21413         PR tree-optimization/55888
21414         PR tree-optimization/55862
21415         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21416         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21417         not if it is contained therein.
21419 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21421         * config/avr/t-avr: Typo.
21423 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21425         PR55243
21426         * config/avr/t-avr: Don't automatically rebuild
21427         $(srcdir)/config/avr/t-multilib
21428         $(srcdir)/config/avr/avr-tables.opt
21429         $(srcdir)/doc/avr-mmcu.texi
21430         (avr-mcus): New phony target to build them on request.
21431         (s-avr-mlib, s-avr-mmcu-texi): Remove.
21432         * avr/avr-mcus.def: Adjust comments.
21434 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
21436         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21438 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21440         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21442 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21444         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21446 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
21448         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21449         to generate profiling.
21450         * config/rs6000/aix64.h (LIB_SPEC): Same.
21452 2013-01-04  Andrew Pinski  <apinski@cavium.com>
21454         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21455         New function.
21456         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21458 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
21460         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21461         unconditionally.
21462         (ix86_expand_move): Ditto.
21463         (ix86_zero_extend_to_Pmode): Ditto.
21464         (ix86_expand_call): Ditto.
21465         (ix86_expand_special_args_builtin): Ditto.
21466         (ix86_expand_builtin): Ditto.
21468 2013-01-04  Richard Biener  <rguenther@suse.de>
21470         PR tree-optimization/55862
21471         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21472         translating them through PHI nodes.
21474 2013-01-04  Martin Jambor  <mjambor@suse.cz>
21476         PR tree-optimization/55755
21477         * tree-sra.c (sra_modify_assign): Do not check that an access has no
21478         children when trying to avoid producing a VIEW_CONVERT_EXPR.
21480 2013-01-04  Marek Polacek  <polacek@redhat.com>
21482         PR middle-end/55859
21483         * opts.c (default_options_optimization): Clarify error message.
21485 2013-01-04  Richard Biener  <rguenther@suse.de>
21487         PR middle-end/55863
21488         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21489         reassociation.
21491 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21493         PR target/53789
21494         * config/pa/pa.md (movsi): Revert previous change.
21495         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21496         references.
21498 2013-01-03  Richard Henderson  <rth@redhat.com>
21500         * config/i386/i386.c (ix86_expand_move): Always assign to op1
21501         after eliminating TLS symbols.
21503 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
21505         PR bootstrap/50167
21506         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21507         * graphite-poly.c (debug_gmp_value): Likewise.
21509 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
21511         PR target/55712
21512         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21513         selected code model, define __code_mode_small__, __code_model_medium__,
21514         __code_model_large__, __code_model_32__ or __code_model_kernel__.
21515         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21516         xchg temporary register with %k.  Declare temporary register as
21517         early clobbered.
21518         [__x86_64__]: For medium and large code models, preserve %rbx register.
21520 2013-01-03  Richard Biener  <rguenther@suse.de>
21522         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21523         (dump_subscript): Adjust.
21524         (finalize_ddr_dependent): Do not dump redundant info.
21525         (analyze_siv_subscript): Adjust.
21526         (subscript_dependence_tester): Likewise.
21527         (compute_affine_dependence): Likewise.
21529 2013-01-03  Richard Biener  <rguenther@suse.de>
21531         Revert
21532         2013-01-03  Richard Biener  <rguenther@suse.de>
21534         PR tree-optimization/55857
21535         * tree-vect-stmts.c (vectorizable_load): Do not setup
21536         re-alignment for invariant loads.
21538         2013-01-02  Richard Biener  <rguenther@suse.de>
21540         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21541         invariant load do not generate a vector load from the scalar location.
21543 2013-01-03  Richard Biener  <rguenther@suse.de>
21545         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21546         for not vectorizing.
21547         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21548         not build INDIRECT_REFs, call get_name once only.
21549         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
21550         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21552 2013-01-03  Richard Biener  <rguenther@suse.de>
21554         PR tree-optimization/55857
21555         * tree-vect-stmts.c (vectorizable_load): Do not setup
21556         re-alignment for invariant loads.
21558 2013-01-03  Richard Biener  <rguenther@suse.de>
21560         PR lto/55848
21561         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
21562         prefer a built-in decl.
21564 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21566         * gcc.c (process_command): Update copyright notice dates.
21567         * gcov.c (print_version): Likewise.
21568         * gcov-dump.c (print_version): Likewise.
21570         PR rtl-optimization/55838
21571         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
21572         iv0.step, iv1.step and step.
21574 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21575             Marc Glisse  <marc.glisse@inria.fr>
21577         PR tree-optimization/55832
21578         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
21579         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
21580         integer_{one,zero}_node.
21582 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21584         PR debug/54402
21585         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
21586         * var-tracking.c (reverse_op): Don't add reverse ops to
21587         VALUEs that have already
21588         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
21590 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
21592         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
21594 2013-01-02  Teresa Johnson  <tejohnson@google.com>
21596         * dumpfile.c (dump_loc): Print filename with location.
21597         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
21598         new location_t parameter to emit complete unroll message with
21599         new dump framework.
21600         (canonicalize_loop_induction_variables): Compute loops location
21601         and pass to try_unroll_loop_completely.
21602         * loop-unroll.c (report_unroll_peel): New function.
21603         (peel_loops_completely): Use new dump format with location
21604         for main dumpfile message, and invoke report_unroll_peel on success.
21605         (decide_unrolling_and_peeling): Ditto.
21606         (decide_peel_once_rolling): Remove old dumpfile message subsumed
21607         by report_unroll_peel.
21608         (decide_peel_completely): Ditto.
21609         (decide_unroll_constant_iterations): Ditto.
21610         (decide_unroll_runtime_iterations): Ditto.
21611         (decide_peel_simple): Ditto.
21612         (decide_unroll_stupid): Ditto.
21613         * cfgloop.c (get_loop_location): New function.
21614         * cfgloop.h (get_loop_location): Declare.
21616 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21618         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
21619         NULL.
21621 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21623         PR middle-end/55198
21624         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
21625         BLKmode objects when EXPAND_MEMORY is specified.
21627 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21629         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
21630         in loop predicate.
21631         (fold_builtin_cpu): Do not share cpu model decls across statements.
21633 2013-01-02  Jason Merrill  <jason@redhat.com>
21635         PR c++/55804
21636         * tree.c (build_array_type_1): Revert earlier change.
21638 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
21640         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
21641         "cortex-a57".
21642         * config/aarch64/aarch64-tune.md: Re-generate.
21644 2013-01-02  Richard Biener  <rguenther@suse.de>
21646         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21647         invariant load do not generate a vector load from the scalar location.
21649 2013-01-02  Richard Biener  <rguenther@suse.de>
21651         PR bootstrap/55784
21652         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
21653         * configure: Regenerate.
21655 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21657         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
21658         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
21659         (expand_builtin_int_roundingfn_2): Keep the original target around
21660         for the fallback case.
21662 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21664         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
21665         to be clear for sign changes.
21667 2013-01-01  Jan Hubicka  <jh@suse.cz>
21669         * ipa-inline-analysis.c: Fix formatting.
21671 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
21673         PR tree-optimization/55831
21674         * tree-vect-loop.c (get_initial_def_for_induction): Use
21675         gsi_after_labels instead of gsi_start_bb.
21677 Copyright (C) 2013 Free Software Foundation, Inc.
21679 Copying and distribution of this file, with or without modification,
21680 are permitted in any medium without royalty provided the copyright
21681 notice and this notice are preserved.