* i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New tuning flag.
[official-gcc.git] / gcc / ChangeLog
blob020231a067de89e9a6b8f92495ccddf75f2027be
1 2013-10-22  Jan Hubicka  <jh@suse.cz>
3         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New tuning flag.
4         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
5         * i386.c (expand_small_movmem_or_setmem): New function.
6         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New function
7         (alg_usable_p): Add support for value ranges; cleanup.
8         (ix86_expand_set_or_movmem): Add support for misaligned moves.
10 2013-10-22  Sterling Augustine  <saugustine@google.com>
12         * doc/invoke.texi: Document -ggnu-pubnames.
13         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
14         logic.
15         * dwarf2out.c: Include gdb/gdb-index.h.
16         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
17         debug_generate_pub_sections.
18         (is_java, output_pubtables, output_pubname): New functions.
19         (include_pubname_in_output): Handle debug_generate_pub_sections at
20         level 2.
21         (size_of_pubnames): Use new local space_for_flags based on
22         debug_generate_pub_sections.
23         (output_pubnames): Unify pubnames and pubtypes output logic.
24         Genericize comments.  Call output_pubname.
25         (dwarf2out_finish): Move logic to output_pubnames and call it.
27 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
29         PR target/58779
30         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
31         Remove CCCmode handling.
32         <case LTU>: Return 'c' suffix for CCCmode.
33         <case GEU>: Return 'nc' suffix for CCCmode.
34         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
35         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
36         (*sub<mode>3_cc_overflow): Ditto.
37         (*subsi3_zext_cc_overflow): Ditto.
39 2013-10-22  Steve Ellcey  <sellcey@mips.com>
41         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
42         (AND (NOT OP1) (NOT OP2)).
44 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
46         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
47         meaning of merge-high and merge-low masks for little endian; avoid
48         use of vector-pack masks for little endian for mismatched modes.
50 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
52         * config/tilepro/tilepro.c: Include "tree.h".
54 2013-10-22  Andreas Schwab  <schwab@suse.de>
56         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
57         with PRE_DEC.
59 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
61         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
62         [Ed Smith-Rowland]): New entries.
63         ([Stephen M. Webb]): Update.
65 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
67         * tree-ssa-ter.h: Remove duplicate copy of file contents.
69 2013-10-21  Marek Polacek  <polacek@redhat.com>
71         PR middle-end/58809
72         * fold-const.c (fold_range_test): Return 0 if the type is not
73         an integral type.
75 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
77         * system.h: Move hwint.h include further down.
78         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
79         gcc_checking_asserts.
80         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
82 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
84         Fix volatile issues in optimize_bit_field_compare.
85         * fold-const.c (optimize_bit_field_compare): Bail out if
86         lvolatilep or rvolatilep.
88 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
90         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
91         and get_inner_reference returning different pmode for non-volatile
92         bit-field members dependent on flag_strict_volatile_bitfields.
93         * stor-layout.c (layout_decl): Remove special handling of
94         flag_strict_volatile_bitfields.
95         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
96         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
98 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
100         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
101         calculations.
103 2013-10-21  Jeff Law  <law@redhat.com>
105         * tree-ssa-threadedge.c (thread_through_normal_block): New
106         argument VISITED.  Remove VISISTED as a local variable.  When we
107         have a threadable jump, verify the destination of the jump has not
108         been visised.
109         (thread_across_edge): Allocate VISITED bitmap once at function
110         scope and use it throughout.  Make sure to set appropriate bits in
111         VISITED for E (start of jump thread path).
112         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
113         through a joiner if any edge on the path has a recorded jump thread.
115 2013-10-21  Ian Lance Taylor  <iant@google.com>
117         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
118         don't imply that attributes can solve all problems.
119         (Directory Options): Fix typo.
121 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
123         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
124         extend and extend_add.
126 2013-10-21  Richard Biener  <rguenther@suse.de>
128         PR tree-optimization/58794
129         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
130         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
132 2013-10-21  Richard Biener  <rguenther@suse.de>
134         PR middle-end/58742
135         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
136         to (T) X for sign-changing conversions (or no conversion).
138 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
140         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
142 2013-10-20  Jan Hubicka  <jh@suse.cz>
144         * config/i386/i386-tune.def: Add comment; organize into categories
146 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
148         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
149         argument.  Update function comment.
150         (expand_set_or_movmem_via_rep): New function combining
151         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
152         (expand_movmem_via_rep_mov): Remove.
153         expand_setmem_via_rep_stos): Remove.
154         (expand_movmem_epilogue): Update calls correspondingly.
155         (expand_setmem_epilogue_via_loop): Likewise.
156         (emit_memset): New.
157         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
158         (expand_set_or_movmem_prologue): New function combining
159         expand_movmem_prologue and expand_setmem_prologue.
160         (expand_movmem_prologue): Remove.
161         (expand_setmem_prologue): Remove.
162         (expand_set_or_movmem_constant_prologue): New function combining
163         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
164         (expand_constant_movmem_prologue): Remove.
165         (expand_constant_setmem_prologue): Remove.
166         (promote_duplicated_reg): Allow vector-const0 value.
167         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
168         and ix86_expand_setmem.
169         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
170         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
172 2013-10-21  Diego Novillo  <dnovillo@google.com>
174         * asan.c: Include tree.h
175         * bb-reorder.c: Likewise.
176         * cfgcleanup.c: Likewise.
177         * cfgloopmanip.c: Likewise.
178         * data-streamer-in.c: Likewise.
179         * data-streamer-out.c: Likewise.
180         * data-streamer.c: Likewise.
181         * dwarf2cfi.c: Likewise.
182         * graphite-blocking.c: Likewise.
183         * graphite-clast-to-gimple.c: Likewise.
184         * graphite-dependences.c: Likewise.
185         * graphite-interchange.c: Likewise.
186         * graphite-optimize-isl.c: Likewise.
187         * graphite-poly.c: Likewise.
188         * graphite-scop-detection.c: Likewise.
189         * graphite-sese-to-poly.c: Likewise.
190         * graphite.c: Likewise.
191         * ipa-devirt.c: Likewise.
192         * ipa-profile.c: Likewise.
193         * ipa.c: Likewise.
194         * ira.c: Likewise.
195         * loop-init.c: Likewise.
196         * loop-unroll.c: Likewise.
197         * lower-subreg.c: Likewise.
198         * lto/lto-object.c: Likewise.
199         * recog.c: Likewise.
200         * reginfo.c: Likewise.
201         * tree-loop-distribution.c: Likewise.
202         * tree-parloops.c: Likewise.
203         * tree-ssa-strlen.c: Likewise.
204         * tree-streamer.c: Likewise.
205         * value-prof.c: Likewise.
206         * target-globals.c: Likewise.
207         * expr.h: Include tree-core.h instead of tree.h.
208         * gimple.h: Likewise.
209         * ipa-prop.h: Likewise.
210         * ipa-utils.h: Likewise.
211         * lto-streamer.h: Likewise.
212         * streamer-hooks.h: Likewise.
213         * ipa-reference.h: Include cgraph.h instead of tree.h.
214         * cgraph.h: Include basic-block.h instead of tree.h.
215         * tree-streamer.h: Do not include tree.h.
216         * genattrtab.c (write_header): Generate inclusion of
217         tree.h.
218         * genautomata.c (main): Likewise.
219         * genemit.c: Likewise.
220         * genopinit.c: Likewise.
221         * genoutput.c (output_prologue): Likewise.
222         * genpeep.c: Likewise.
224 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
226         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
227         little endian.
228         (vec_unpacku_hi_v8hi): Likewise.
229         (vec_unpacku_lo_v16qi): Likewise.
230         (vec_unpacku_lo_v8hi): Likewise.
232 2013-10-20  Jan Hubicka  <jh@suse.cz>
234         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
235         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
236         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
238 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
240         * config/mips/mips.h (ISA_HAS_WSBH): Define.
241         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
242         constants.
243         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
245 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
247         PR target/58603
248         * system.h: Undef m_slot.
250 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
252         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
253         all elements for both endian flavors.
255 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
257         PR target/58792
258         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
259         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
260         and SI_REG for 64bit SYSV ABI targets.
262 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
264         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
265         to multi_reg_return.  Clarify that we are skipping USEs of multiple
266         return registers.  Use bool type where appropriate.
268 2013-10-18  Jan Hubicka  <jh@suse.cz>
270         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
271         for cold functions.
272         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
273         (X86_TUNE_PUSH_MEMORY): Likewise.
274         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
275         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
276         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
277         New.
278         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
279         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
280         Remove.
281         (ix86_option_override_internal): Update to use tune features instead
282         of variables.
284 2013-10-18  Cong Hou  <congh@google.com>
286         PR tree-optimization/58508
287         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
288         statement that contains data refs with zero-step.
290 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
292         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
293         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
294         * gimple-low.c (gimple_check_call_arg,
295         gimple_check_call_matching_types): Move to cgraph.c.
296         * gimple-low.h: Remove prototype.
297         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
298         Relocate from gimple-low.c.
299         * cgraph.h: Add prototype.  Don't include basic-block.h.
300         * gimplify.c: Add gimple-low to include list.
301         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
302         * tree-eh.c: Add gimple-low to include list.
303         * tree-nested.c: Likewise.
304         * cfgexpand.c: Add tree-ssa-address.h to include list.
305         * expr.c: Likewise.
306         * gimple-fold.c: Likewise.
307         * gimple-ssa-strength-reduction.c: Likewise.
308         * trans-mem.c: Likewise.
309         * tree-mudflap.c: Likewise.
310         * tree-ssa-loop-ivopts.c: Likewise.
311         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
312         (degenerate_phi_result): Move to tree-phinodes.c.
313         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
314         * tree-ssa-threadedge.c: Likewise.
315         * tree-vrp.c: Likewise.
316         * tree-phinodes.c (degenerate_phi_result): Relocate here.
317         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
318         * tree-phinodes.h (degenerate_phi_result): Add prototype.
319         * tree-ssa-copy.c: Include tree-ssa-dom.h.
320         * tree-ssa-forwprop.c: Likewise.
321         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
322         pass_data_cleanup_cfg_post_optimizing,
323         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
324         * tree-optimize.c: Delete File.
325         * graphite.c: Include tree-cfgcleanup.h.
326         * passes.c: Likewise.
327         * tree-cfg.c: Likewise.
328         * tree-profile.c: Likewise.
329         * tree-ssa-dse.c: Likewise.
330         * tree-ssa-loop-ivcanon.c: Likewise.
331         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
332         * tree-outof-ssa.c: Include sbitmap.h.
333         * tree-ssa-live.c: Likewise.
334         * tree-ssa-propagate.c: Likewise.
335         * tree-ssa-structalias.c: Likewise.
336         * tree-stdarg.c: Likewise.
337         * Makefile.in (OBJS): Delete tree-optimize.o.
338         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
339         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
340         * varasm.c: Include basic-block.h.
341         * cfgloop.h: Include function.h instead of basic-block.h
342         (bb_loop_depth): Move to cfgloop.c.
343         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
345 2013-10-18  Teresa Johnson  <tejohnson@google.com>
347         * predict.c (probably_never_executed): Compare frequency-based
348         count to number of training runs.
349         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
351 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
353         * config/arm/arm.c (cortexa9_extra_costs): New table.
354         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
356 2013-10-18  Jeff Law  <law@redhat.com>
358         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
360         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
361         (register_jump_thread): Add "registered_jump_thread" debug
362         counter support.
363         * dbgcnt.def (registered_jump_thread): New debug counter.
365 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
367         * config/rs6000/rs6000.c: Include cgraph.h.
369 2013-10-18  Teresa Johnson  <tejohnson@google.com>
371         * tree-ssa-tail-merge.c (replace_block_by): Update edge
372         weights during merging.
374 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
376         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
377         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
378         * tree-ssa-operands.h: Remove prototype.
379         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
380         * gimple.c (virtual_operand_p): Relocate from gimple.c.
381         * gimple.h: Add prototype.
382         * gimple-ssa.h: Include tree-ssa-operands.h.
383         * tree-dump.c: Add tree-cfg.h to include list.
384         * tree-ssa-alias.c: Add ipa-reference.h to include list.
385         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
386         * config/i386/i386.c: Don't include tree-flow.h.
387         * config/rs6000/rs6000.c: Likewise.
389 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
391         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
393 2013-10-18  Richard Biener  <rguenther@suse.de>
395         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
396         or TYPE_UNSIGNED of integral types.
397         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
398         NULL_TREE for zero-precision integral types.
400 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
402         * config/aarch64/arm_neon.h
403         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
405 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
407         * opth-gen.awk: Define target_flags_explicit.
409 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
411         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
412         fields to the reg_addr array that describes the valid addressing
413         mode for any register, general purpose registers, floating point
414         registers, and Altivec registers.
415         (FIRST_RELOAD_REG_CLASS): Likewise.
416         (LAST_RELOAD_REG_CLASS): Likewise.
417         (struct reload_reg_map_type): Likewise.
418         (reload_reg_map_type): Likewise.
419         (RELOAD_REG_VALID): Likewise.
420         (RELOAD_REG_MULTIPLE): Likewise.
421         (RELOAD_REG_INDEXED): Likewise.
422         (RELOAD_REG_OFFSET): Likewise.
423         (RELOAD_REG_PRE_INCDEC): Likewise.
424         (RELOAD_REG_PRE_MODIFY): Likewise.
425         (reg_addr): Likewise.
426         (mode_supports_pre_incdec_p): New helper functions to say whether
427         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
428         (mode_supports_pre_modify_p): Likewise.
429         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
430         print the valid address mode bits for each mode.
431         (rs6000_debug_print_mode): Likewise.
432         (rs6000_debug_reg_global): Likewise.
433         (rs6000_setup_reg_addr_masks): New function to set up the address
434         mask bits for each type.
435         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
436         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
437         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
438         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
439         PRE_MODIFY are supported.
440         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
441         registers, instead of {src,dest}_av_p.
442         (rs6000_print_options_internal): Tweak the debug output slightly.
444 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
446         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
447         isa attribute.
449 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
451         * tree-flow.h (struct omp_region): Move to omp-low.c.
452         Remove omp_ prototypes and variables.
453         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
454         (copy_var_decl): Relocate prototype from tree-flow.h.
455         * gimple.c (copy_var_decl): Relocate from omp-low.c.
456         * tree.h: Move prototype to omp-low.h.
457         * omp-low.h: New File.  Relocate prototypes here.
458         * omp-low.c (struct omp_region): Make local here.
459         (root_omp_region): Make static.
460         (copy_var_decl) Move to gimple.c.
461         (new_omp_region): Make static.
462         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
463         * tree-cfg.c: Include omp-low.h.
464         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
465         * gimplify.c: Include omp-low.h.
466         * tree-parloops.c: Likewise.
468 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
470         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
471         operands are in memory, prefer to force non-matched operand 1 to
472         the register.
474 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
476         PR target/58673
477         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
478         restrict TImode addresses to single indirect registers if both
479         -mquad-memory and -mvsx-timode are used.
480         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
481         we should emit load/store quad.  Remove using %y for quad memory
482         addresses.
484         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
485         constraints to allow load/store quad on machines where TImode is
486         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
487         for TImode to load integer constants.
489 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
491         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
493 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
495         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
496         handling of STACK_REG.
498 2013-10-17  Richard Biener  <rguenther@suse.de>
500         PR tree-optimization/58143
501         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
502         New function.
503         (rewrite_to_defined_overflow): Likewise.
504         (move_computations_dom_walker::before_dom): Rewrite stmts
505         with undefined signed overflow that are not always executed
506         into unsigned arithmetic.
508 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
510         PR target/57756
511         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
512         explicit isa flag to be an options variable, instead of using
513         global_options_set.  Remove define from rs6000.h.
514         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
516         * config/rs6000/rs6000.c (rs6000_option_override_internal):
517         Initialize rs6000_isa_flags_explicit.
518         (rs6000_function_specific_save): Add gcc_options* parameter, so
519         that the powerpc builds after the 2013-10-15 changes.
520         (rs6000_function_specific_restore): Likewise.
522 2013-10-16  DJ Delorie  <dj@redhat.com>
524         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
525         op is a REG before checking REGNO.
526         (rl78_alloc_physical_registers): Verify pattern is a SET before
527         checking SET_SRC.
529 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
531         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
532         endianness.
533         (vec_unpacks_lo_v4sf): Likewise.
534         (vec_unpacks_float_hi_v4si): Likewise.
535         (vec_unpacks_float_lo_v4si): Likewise.
536         (vec_unpacku_float_hi_v4si): Likewise.
537         (vec_unpacku_float_lo_v4si): Likewise.
539 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
541         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
542         (vsx_concat_v2sf): Likewise.
544 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
546         * config/aarch64/aarch64.md
547         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
549 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
551         PR tree-optimization/58697
552         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
553         estimated_loop_iterations_int.
554         (max_stmt_executions_int): Call get_max_loop_iterations_int.
555         (get_max_loop_iterations_int): New.  HWINT version of
556         get_max_loop_iterations.
557         * cfgloop.h: Add prototypes.
558         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
559         * loop-unroll.c (decide_peel_once_rolling): Call
560         get_estimated_loop_iterations_int.
561         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
562         * tree-ssa-loop-niter.h: Tweak prototypes.
564 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
566         * gengtype-parse.c (struct_field_seq): Ignore access-control
567         keywords ("public:" etc).
569 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
571         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
572         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
574 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
576         * config/arm/arm.opt (mlra): New option.
577         * config/arm/arm.c (arm_lra_p): New function.
578         (TARGET_LRA_P): Define.
580 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
582         * tree-core.h (tree_code_name): Remove.
583         * tree.h (get_tree_code_name): New prototype.
584         * tree.c (tree_code_name): Make static.
585         (get_tree_code_name): New function.
586         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
587         tree_class_check_failed, tree_range_check_failed,
588         tree_not_class_check_failed, omp_clause_check_failed,
589         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
590         wrapper get_tree_code_name instead of calling tree_code_name directly.
591         * tree-vrp.c (dump_asserts_for): Likewise.
592         * tree-dump.c (dequeue_and_dump): Likewise.
593         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
594         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
595         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
596         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
597         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
598         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
599         dump_gimple_omp_for): Likewise.
600         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
601         * tree-ssa-pre.c (print_pre_expr): Likewise.
602         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
603         * print-tree.c (print_node_brief, print_node): Likewise.
604         * gimple.c (gimple_check_failed): Likewise.
605         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
606         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
607         * config/mep/mep.c (mep_validate_vliw): Likewise.
608         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
609         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
611 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
613         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
614         for AMD bdver3.
616 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
618         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
619         (MULTILIB_MATCHES): Add multilib for -march=v8.
621 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
623         PR target/57756
624         * optc-save-gen.awk: Add extra parameter to the save and restore
625         target calls.
626         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
627         * tree.c (build_optimization_node): New parameter.  Add extra parameter
628         to call to cl_optimization_save.
629         (build_target_option_node): New parameter. Add extra parameter
630         to call to cl_target_option_save.
631         * tree.h (build_optimization_node): New parameter.
632         (build_target_option_node): New parameter.
633         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
634         build_optimization_node and build_target_option_node.
635         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
636         (handle_pragma_push_options): Ditto.
637         * toplev.c (process_options): Ditto.
638         * opts.c (init_options_struct): Check for opts_set non-null.
639         * target.def (target_option.save): New parameter.
640         (target_option.restore): New parameter.
641         * tm.texi: Generate.
642         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
643         (ix86_pragma_target_parse): Ditto.
644         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
645         parameters.
646         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
647         to  build_optimization_node and build_target_option_node.
648         (rs6000_valid_attribute_p): Ditto.
649         (rs6000_pragma_target_parse): Ditto.
650         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
651         data.
652         * config/i386/i386.h:
653         TARGET_64BIT_P: New Macro
654         TARGET_MMX_P: New Macro.
655         TARGET_3DNOW_P: New Macro.
656         TARGET_3DNOW_A_P: New Macro.
657         TARGET_SSE_P: New Macro.
658         TARGET_SSE2_P: New Macro.
659         TARGET_SSE3_P: New Macro.
660         TARGET_SSSE3_P: New Macro.
661         TARGET_SSE4_1_P: New Macro.
662         TARGET_SSE4_2_P: New Macro.
663         TARGET_AVX_P: New Macro.
664         TARGET_AVX2_P: New Macro.
665         TARGET_AVX512F_P: New Macro.
666         TARGET_AVX512PF_P: New Macro.
667         TARGET_AVX512ER_P: New Macro.
668         TARGET_AVX512CD_P: New Macro.
669         TARGET_FMA_P: New Macro.
670         TARGET_SSE4A_P: New Macro.
671         TARGET_FMA4_P: New Macro.
672         TARGET_XOP_P: New Macro.
673         TARGET_LWP_P: New Macro.
674         TARGET_ABM_P: New Macro.
675         TARGET_BMI_P: New Macro.
676         TARGET_BMI2_P: New Macro.
677         TARGET_LZCNT_P: New Macro.
678         TARGET_TBM_P: New Macro.
679         TARGET_POPCNT_P: New Macro.
680         TARGET_SAHF_P: New Macro.
681         TARGET_MOVBE_P: New Macro.
682         TARGET_CRC32_P: New Macro.
683         TARGET_AES_P: New Macro.
684         TARGET_PCLMUL_P: New Macro.
685         TARGET_CMPXCHG16B_P: New Macro.
686         TARGET_FSGSBASE_P: New Macro.
687         TARGET_RDRND_P: New Macro.
688         TARGET_F16C_P: New Macro.
689         TARGET_RTM_P: New Macro.
690         TARGET_HLE_P: New Macro.
691         TARGET_RDSEED_P: New Macro.
692         TARGET_PRFCHW_P: New Macro.
693         TARGET_ADX_P: New Macro.
694         TARGET_FXSR_P: New Macro.
695         TARGET_XSAVE_P: New Macro.
696         TARGET_XSAVEOPT_P: New Macro.
697         TARGET_LP64_P: New Macro.
698         TARGET_X32_P: New Macro.
699         TARGET_FPMATH_DEFAULT_P: New Macro.
700         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
701         * config/i386/i386.c (ix86_option_override_internal): New parameters.
702         opts and opts_set.
703         Change ix86_tune_string to access opts->x_ix86_tune_string.
704         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
705         Change ix86_arch_string to access opts->x_ix86_arch_string.
706         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
707         Change ix86_pmode to access opts->x_ix86_pmode.
708         Change ix86_abi to access opts->x_ix86_abi.
709         Change ix86_cmodel to access opts->x_ix86_cmodel.
710         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
711         Change ix86_isa_flags_explicit to access
712         opts->x_ix86_isa_flags_explicit.
713         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
714         Change ix86_regparm to access opts->x_ix86_regparm.
715         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
716         Change ix86_preferred_stack_boundary_arg to access
717         opts->x_ix86_preferred_stack_boundary_arg.
718         Change ix86_force_align_arg_pointer to access
719         opts->x_ix86_force_align_arg_pointer.
720         Change ix86_incoming_stack_boundar_arg to access
721         opts->x_ix86_incoming_stack_boundar_arg.
722         Change ix86_fpmath to access opts->x_ix86_fpmath.
723         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
724         Change ix86_recip_name to access opts->x_ix86_recip_name.
725         Change ix86_stack_protector_guard to access
726         opts->x_ix86_stack_protector_guard.
727         Change ix86_tune_memcpy_strategy to access
728         opts->x_ix86_tune_memcpy_strategy.
729         Change ix86_tune_memset_strategy to access
730         opts->x_ix86_tune_memset_strategy.
731         Change global_options to access opts.
732         Change global_options_set to access opts_set.
733         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
734         Change TARGET_MMX to TARGET_MMX_P (opts->...).
735         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
736         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
737         Change TARGET_SSE to TARGET_SSE_P (opts->...).
738         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
739         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
740         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
741         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
742         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
743         Change TARGET_AVX to TARGET_AVX_P (opts->...).
744         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
745         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
746         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
747         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
748         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
749         Change TARGET_FMA to TARGET_FMA_P (opts->...).
750         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
751         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
752         Change TARGET_XOP to TARGET_XOP_P (opts->...).
753         Change TARGET_LWP to TARGET_LWP_P (opts->...).
754         Change TARGET_ABM to TARGET_ABM_P (opts->...).
755         Change TARGET_BMI to TARGET_BMI_P (opts->...).
756         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
757         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
758         Change TARGET_TBM to TARGET_TBM_P (opts->...).
759         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
760         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
761         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
762         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
763         Change TARGET_AES to TARGET_AES_P (opts->...).
764         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
765         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
766         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
767         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
768         Change TARGET_F16C to TARGET_F16C_P (opts->...).
769         Change TARGET_RTM to TARGET_RTM_P (opts->...).
770         Change TARGET_HLE to TARGET_HLE_P (opts->...).
771         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
772         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
773         Change TARGET_ADX to TARGET_ADX_P (opts->...).
774         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
775         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
776         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
777         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
778         Change TARGET_X32 to TARGET_X32_P (opts->...).
779         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
780         Change TARGET_FLOAT_RETURNS_IN_80387 to
781         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
782         (ix86_function_specific_save): New parameter. Use opts-> fields
783         to replace global fields.
784         (ix86_function_specific_restore): Ditto.
785         (ix86_valid_target_attribute_inner_p): New parameters.
786         Fix recursive call.
787         Fix call to ix86_handle_option and set_option.
788         (ix86_valid_target_attribute_tree): New parameters.
789         Change global_options to access opts.
790         Change global_options_set to access opts_set.
791         Fix call to ix86_valid_target_attribute_inner_p.
792         Change ix86_tune_string to access opts->x_ix86_tune_string.
793         Change ix86_arch_string to access opts->x_ix86_arch_string.
794         Change ix86_fpmath to access opts->x_ix86_fpmath
795         Fix call to ix86_option_override_internal.
796         Fix call to ix86_add_new_builtins.
797         Fix calls to build_optimization_node and build_target_option_node.
798         (ix86_valid_target_attribute_p): Remove access to global_options.
799         Use new gcc_options structure func_options.
800         Fix call to ix86_valid_target_attribute_tree.
801         Fix call to  build_optimization_node.
802         (get_builtin_code_for_version): Fix call to
803         ix86_valid_target_attribute_tree.
805 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
807         * Makefile.in (PICFLAG): New.
808         (enable_host_shared): New.
809         (INTERNAL_CFLAGS): Use PICFLAG.
810         (LIBIBERTY): Use pic build of libiberty.a if configured with
811         --enable-host-shared.
812         * configure.ac: Add --enable-host-shared, setting up new
813         PICFLAG variable.
814         * configure: Regenerate.
815         * doc/install.texi (--enable-shared): Add note contrasting it with ...
816         (--enable-host-shared): New option.
818 2013-10-15  Richard Biener  <rguenther@suse.de>
820         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
821         for built-in functions.
823 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
825         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
826         (optimize_range_tests_1): New function,
827         extracted from optimize_range_tests.
828         (optimize_range_tests_xor): Similarly.
829         (optimize_range_tests_diff): New function.
830         (optimize_range_tests): Use optimize_range_tests_1.
832 2013-10-15  Cong Hou  <congh@google.com>
834         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
835         requirement of the reduction pattern so that one operand of the
836         reduction operation can come from outside of the loop.
838 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
840         * config/arm/neon-schedgen.ml: Remove.
841         * config/arm/cortex-a9-neon.md: Remove comment regarding
842         neon-schedgen.ml.
844 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
846         * config/arm/types: Remove old neon types.
848 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
850         * config/arm/cortex-a7.md
851         (cortex_a7_neon_type): New.
852         (cortex_a7_neon_mul): Update for new types.
853         (cortex_a7_neon_mla): Likewise.
854         (cortex_a7_neon): Likewise.
856 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
858         * config/arm/cortex-a15-neon.md
859         (cortex_a15_neon_type): New,
861         (cortex_a15_neon_int_1): Remove.
862         (cortex_a15_neon_int_2): Likewise.
863         (cortex_a15_neon_int_3): Likewise.
864         (cortex_a15_neon_int_4): Likewise.
865         (cortex_a15_neon_int_5): Likewise.
866         (cortex_a15_neon_vqneg_vqabs): Likewise.
867         (cortex_a15_neon_vmov): Likewise.
868         (cortex_a15_neon_vaba): Likewise.
869         (cortex_a15_neon_vaba_qqq): Likewise.
870         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
871         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
872         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
873         Likewise.
874         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
875         (cortex_a15_neon_mla_qqq_8_16): Likewise.
876         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
877         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
878         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
879         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
880         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
881         (cortex_a15_neon_shift_1): Likewise.
882         (cortex_a15_neon_shift_2): Likewise.
883         (cortex_a15_neon_shift_3): Likewise.
884         (cortex_a15_neon_vshl_ddd): Likewise.
885         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
886         (cortex_a15_neon_vsra_vrsra): Likewise.
887         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
888         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
889         (cortex_a15_neon_bp_3cycle): Likewise.
890         (cortex_a15_neon_ldm_2): Likewise.
891         (cortex_a15_neon_stm_2): Likewise.
892         (cortex_a15_neon_mcr): Likewise.
893         (cortex_a15_neon_mrc): Likewise.
894         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
895         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
896         (cortex_a15_neon_fp_vmul_ddd): Likewise.
897         (cortex_a15_neon_fp_vmul_qqd): Likewise.
898         (cortex_a15_neon_fp_vmla_ddd): Likewise.
899         (cortex_a15_neon_fp_vmla_qqq): Likewise.
900         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
901         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
902         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
903         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
904         (cortex_a15_neon_bp_simple): Likewise.
905         (cortex_a15_neon_bp_2cycle): Likewise.
906         (cortex_a15_neon_bp_3cycle): Likewise.
907         (cortex_a15_neon_vld1_1_2_regs): Likewise.
908         (cortex_a15_neon_vld1_3_4_regs): Likewise.
909         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
910         (cortex_a15_neon_vld2_4_regs): Likewise.
911         (cortex_a15_neon_vld3_vld4): Likewise.
912         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
913         (cortex_a15_neon_vst1_3_4_regs): Likewise.
914         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
915         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
916         (cortex_a15_neon_vst3_vst4): Rename to...
917         (cortex_a15_neon_vst4): This, update for new attributes.
918         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
919         (cortex_a15_neon_vld3_vld4_lane): Likewise.
920         (cortex_a15_neon_vst1_vst2_lane): Likewise.
921         (cortex_a15_neon_vst3_vst4_lane): Likewise.
922         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
923         (cortex_a15_neon_ldm_2): Likewise.
924         (cortex_a15_neon_stm_2): Likewise.
925         (cortex_a15_neon_mcr): Likewise.
926         (cortex_a15_neon_mcr_2_mcrr): Likewise.
927         (cortex_a15_neon_mrc): Likewise.
928         (cortex_a15_neon_mrrc): Likewise.
930         (cortex_a15_neon_abd): New.
931         (cortex_a15_neon_abd_q): Likewise.
932         (cortex_a15_neon_aba): Likewise.
933         (cortex_a15_neon_aba_q): Likewise.
934         (cortex_a15_neon_acc): Likewise.
935         (cortex_a15_neon_acc_q): Likewise.
936         (cortex_a15_neon_arith_basic): Likewise.
937         (cortex_a15_neon_arith_complex): Likewise.
938         (cortex_a15_neon_multiply): Likewise.
939         (cortex_a15_neon_multiply_q): Likewise.
940         (cortex_a15_neon_mla): Likewise.
941         (cortex_a15_neon_mla_q): Likewise.
942         (cortex_a15_neon_sat_mla_long): Likewise.
943         (cortex_a15_neon_shift_acc): Likewise.
944         (cortex_a15_neon_shift_imm_basic): Likewise.
945         (cortex_a15_neon_shift_imm_complex): Likewise.
946         (cortex_a15_neon_shift_reg_basic): Likewise.
947         (cortex_a15_neon_shift_reg_basic_q): Likewise.
948         (cortex_a15_neon_shift_reg_complex): Likewise.
949         (cortex_a15_neon_shift_reg_complex_q): Likewise.
950         (cortex_a15_neon_fp_negabs): Likewise
951         (cortex_a15_neon_fp_arith): Likewise
952         (cortex_a15_neon_fp_arith_q): Likewise
953         (cortex_a15_neon_fp_cvt_int): Likewise
954         (cortex_a15_neon_fp_cvt_int_q): Likewise
955         (cortex_a15_neon_fp_cvt_16): Likewise
956         (cortex_a15_neon_fp_mul): Likewise
957         (cortex_a15_neon_fp_mul_q): Likewise
958         (cortex_a15_neon_fp_mla): Likewise
959         (cortex_a15_neon_fp_mla_q): Likewise
960         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
961         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
962         (cortex_a15_neon_bitops): Likewise.
963         (cortex_a15_neon_bitops_q): Likewise.
964         (cortex_a15_neon_from_gp): Likewise.
965         (cortex_a15_neon_from_gp_q): Likewise.
966         (cortex_a15_neon_tbl3_tbl4): Likewise.
967         (cortex_a15_neon_zip_q): Likewise.
968         (cortex_a15_neon_to_gp): Likewise.
969         (cortex_a15_neon_load_a): Likewise.
970         (cortex_a15_neon_load_b): Likewise.
971         (cortex_a15_neon_load_c): Likewise.
972         (cortex_a15_neon_load_d): Likewise.
973         (cortex_a15_neon_load_e): Likewise.
974         (cortex_a15_neon_load_f): Likewise.
975         (cortex_a15_neon_store_a): Likewise.
976         (cortex_a15_neon_store_b): Likewise.
977         (cortex_a15_neon_store_c): Likewise.
978         (cortex_a15_neon_store_d): Likewise.
979         (cortex_a15_neon_store_e): Likewise.
980         (cortex_a15_neon_store_f): Likewise.
981         (cortex_a15_neon_store_g): Likewise.
982         (cortex_a15_neon_store_h): Likewise.
983         (cortex_a15_vfp_to_from_gp): Likewise.
985 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
987         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
989         (cortex_a9_neon_vshl_ddd): Remove.
990         (cortex_a9_neon_vst3_vst4): Likewise.
991         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
993         (cortex_a9_neon_bit_ops_q): New.
995         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
996         (cortex_a9_neon_int_2): Likewise.
997         (cortex_a9_neon_int_3): Likewise.
998         (cortex_a9_neon_int_4): Likewise.
999         (cortex_a9_neon_int_5): Likewise.
1000         (cortex_a9_neon_vqneg_vqabs): Likewise.
1001         (cortex_a9_neon_vmov): Likewise.
1002         (cortex_a9_neon_vaba): Likewise.
1003         (cortex_a9_neon_vaba_qqq): Likewise.
1004         (cortex_a9_neon_shift_1): Likewise.
1005         (cortex_a9_neon_shift_2): Likewise.
1006         (cortex_a9_neon_shift_3): Likewise.
1007         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1008         (cortex_a9_neon_vsra_vrsra): Likewise.
1009         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1010         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1011         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1012         Likewise.
1013         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1014         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1015         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1016         Likewise.
1017         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1018         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1019         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1020         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1021         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1022         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1023         (cortex_a9_neon_fp_vsum): Likewise.
1024         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1025         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1026         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1027         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1028         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1029         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1030         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1031         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1032         (cortex_a9_neon_bp_simple): Likewise.
1033         (cortex_a9_neon_bp_2cycle): Likewise.
1034         (cortex_a9_neon_bp_3cycle): Likewise.
1035         (cortex_a9_neon_ldr): Likewise.
1036         (cortex_a9_neon_str): Likewise.
1037         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1038         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1039         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1040         (cortex_a9_neon_vld2_4_regs): Likewise.
1041         (cortex_a9_neon_vld3_vld4): Likewise.
1042         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1043         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1044         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1045         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1046         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1047         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1048         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1049         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1050         (cortex_a9_neon_mcr): Likewise.
1051         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1052         (cortex_a9_neon_mrc): Likewise.
1053         (cortex_a9_neon_mrrc): Likewise.
1055 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1057         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
1059         (cortex_a8_neon_vshl_ddd): Remove.
1060         (cortex_a8_neon_vst3_vst4): Likewise.
1061         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1063         (cortex_a8_neon_bit_ops_q): New.
1065         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
1066         (cortex_a8_neon_int_2): Likewise..
1067         (cortex_a8_neon_int_3): Likewise.
1068         (cortex_a8_neon_int_5): Likewise.
1069         (cortex_a8_neon_vqneg_vqabs): Likewise.
1070         (cortex_a8_neon_int_4): Likewise.
1071         (cortex_a8_neon_vaba): Likewise.
1072         (cortex_a8_neon_vaba_qqq): Likewise.
1073         (cortex_a8_neon_shift_1): Likewise.
1074         (cortex_a8_neon_shift_2): Likewise.
1075         (cortex_a8_neon_shift_3): Likewise.
1076         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1077         (cortex_a8_neon_vsra_vrsra): Likewise.
1078         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1079         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1080         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1081         Likewise.
1082         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1083         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1084         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1085         Likewise.
1086         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1087         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1088         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1089         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1090         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1091         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1092         (cortex_a8_neon_fp_vsum): Likewise.
1093         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1094         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1095         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1096         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1097         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1098         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1099         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1100         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1101         (cortex_a8_neon_bp_simple): Likewise.
1102         (cortex_a8_neon_bp_2cycle): Likewise.
1103         (cortex_a8_neon_bp_3cycle): Likewise.
1104         (cortex_a8_neon_ldr): Likewise.
1105         (cortex_a8_neon_str): Likewise.
1106         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1107         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1108         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1109         (cortex_a8_neon_vld2_4_regs): Likewise.
1110         (cortex_a8_neon_vld3_vld4): Likewise.
1111         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1112         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1113         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1114         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1115         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1116         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1117         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1118         (cortex_a8_neon_mcr): Likewise.
1119         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1120         (cortex_a8_neon_mrc): Likewise.
1121         (cortex_a8_neon_mrrc): Likewise.
1123 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1125         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1126         (fp): New.
1127         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1128         (aarch64_simd_dup<mode>): Add "type" attribute.
1129         (aarch64_dup_lane<mode>): Likewise.
1130         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1131         (*aarch64_simd_mov<mode>): Likewise.
1132         (aarch64_simd_mov_from_<mode>low): Likewise.
1133         (aarch64_simd_mov_from_<mode>high): Likewise.
1134         (orn<mode>3): Likewise.
1135         (bic<mode>3): Likewise.
1136         (add<mode>3): Likewise.
1137         (sub<mode>3): Likewise.
1138         (mul<mode>3): Likewise.
1139         (*aarch64_mul3_elt<mode>): Likewise.
1140         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1141         (*aarch64_mul3_elt_to_128df): Likewise.
1142         (*aarch64_mul3_elt_to_64v2df): Likewise.
1143         (neg<mode>2): Likewise.
1144         (abs<mode>2): Likewise.
1145         (abd<mode>_3): Likewise.
1146         (aba<mode>_3): Likewise.
1147         (fabd<mode>_3): Likewise.
1148         (*fabd_scalar<mode>3): Likewise.
1149         (and<mode>3): Likewise.
1150         (ior<mode>3): Likewise.
1151         (xor<mode>3): Likewise.
1152         (one_cmpl<mode>2): Likewise.
1153         (aarch64_simd_vec_set<mode>): Likewise.
1154         (aarch64_simd_lshr<mode>): Likewise.
1155         (aarch64_simd_ashr<mode>): Likewise.
1156         (aarch64_simd_imm_shl<mode>): Likewise.
1157         (aarch64_simd_reg_sshl<mode): Likewise.
1158         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1159         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1160         (aarch64_simd_vec_setv2di): Likewise.
1161         (aarch64_simd_vec_set<mode>): Likewise.
1162         (aarch64_mla<mode>): Likewise.
1163         (*aarch64_mla_elt<mode>): Likewise.
1164         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1165         (aarch64_mls<mode>): Likewise.
1166         (*aarch64_mls_elt<mode>): Likewise.
1167         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1168         (<su><maxmin><mode>3): Likewise.
1169         (move_lo_quad_<mode>): Likewise.
1170         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1171         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1172         (vec_pack_trunc_<mode>): Likewise.
1173         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1174         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1175         (*aarch64_<su>mlal_lo<mode>): Likewise.
1176         (*aarch64_<su>mlal_hi<mode>): Likewise.
1177         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1178         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1179         (*aarch64_<su>mlal<mode>): Likewise.
1180         (*aarch64_<su>mlsl<mode>): Likewise.
1181         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1182         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1183         (add<mode>3): Likewise.
1184         (sub<mode>3): Likewise.
1185         (mul<mode>3): Likewise.
1186         (div<mode>3): Likewise.
1187         (neg<mode>2): Likewise.
1188         (abs<mode>2): Likewise.
1189         (fma<mode>4): Likewise.
1190         (*aarch64_fma4_elt<mode>): Likewise.
1191         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1192         (*aarch64_fma4_elt_to_128df): Likewise.
1193         (*aarch64_fma4_elt_to_64v2df): Likewise.
1194         (fnma<mode>4): Likewise.
1195         (*aarch64_fnma4_elt<mode>): Likewise.
1196         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1197         (*aarch64_fnma4_elt_to_128df): Likewise.
1198         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1199         (<frint_pattern><mode>2): Likewise.
1200         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1201         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1202         (vec_unpacks_lo_v4sf): Likewise.
1203         (aarch64_float_extend_lo_v2df): Likewise.
1204         (vec_unpacks_hi_v4sf): Likewise.
1205         (aarch64_float_truncate_lo_v2sf): Likewise.
1206         (aarch64_float_truncate_hi_v4sf): Likewise.
1207         (aarch64_vmls<mode>): Likewise.
1208         (<su><maxmin><mode>3): Likewise.
1209         (<maxmin_uns><mode>3): Likewise.
1210         (reduc_<sur>plus_<mode>): Likewise.
1211         (reduc_<sur>plus_v2di): Likewise.
1212         (reduc_<sur>plus_v2si): Likewise.
1213         (reduc_<sur>plus_<mode>): Likewise.
1214         (aarch64_addpv4sf): Likewise.
1215         (clz<mode>2): Likewise.
1216         (reduc_<maxmin_uns>_<mode>): Likewise.
1217         (reduc_<maxmin_uns>_v2di): Likewise.
1218         (reduc_<maxmin_uns>_v2si): Likewise.
1219         (reduc_<maxmin_uns>_<mode>): Likewise.
1220         (reduc_<maxmin_uns>_v4sf): Likewise.
1221         (aarch64_simd_bsl<mode>_internal): Likewise.
1222         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1223         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1224         (aarch64_get_lane<mode>): Likewise.
1225         (*aarch64_combinez<mode>): Likewise.
1226         (aarch64_combine<mode>): Likewise.
1227         (aarch64_simd_combine<mode>): Likewise.
1228         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1229         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1230         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1231         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1232         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1233         (aarch64_<sur>h<addsub><mode>): Likewise.
1234         (aarch64_<sur><addsub>hn<mode>): Likewise.
1235         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1236         (aarch64_pmul<mode>): Likewise.
1237         (aarch64_<su_optab><optab><mode>): Likewise.
1238         (aarch64_<sur>qadd<mode>): Likewise.
1239         (aarch64_sqmovun<mode>): Likewise.
1240         (aarch64_<sur>qmovn<mode>): Likewise.
1241         (aarch64_s<optab><mode>): Likewise.
1242         (aarch64_sq<r>dmulh<mode>): Likewise.
1243         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1244         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1245         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1246         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1247         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1248         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1249         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1250         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1251         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1252         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1253         (aarch64_sqdmull<mode>): Likewise.
1254         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1255         (aarch64_sqdmull_n<mode>): Likewise.
1256         (aarch64_sqdmull2<mode>_internal): Likewise.
1257         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1258         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1259         (aarch64_<sur>shl<mode>): Likewise.
1260         (aarch64_<sur>q<r>shl<mode>
1261         (aarch64_<sur>shll_n<mode>): Likewise.
1262         (aarch64_<sur>shll2_n<mode>): Likewise.
1263         (aarch64_<sur>shr_n<mode>): Likewise.
1264         (aarch64_<sur>sra_n<mode>): Likewise.
1265         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1266         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1267         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1268         (aarch64_cm<optab><mode>): Likewise.
1269         (aarch64_cm<optab>di): Likewise.
1270         (aarch64_cm<optab><mode>): Likewise.
1271         (aarch64_cm<optab>di): Likewise.
1272         (aarch64_cmtst<mode>): Likewise.
1273         (aarch64_cmtstdi): Likewise.
1274         (aarch64_cm<optab><mode>): Likewise.
1275         (*aarch64_fac<optab><mode>): Likewise.
1276         (aarch64_addp<mode>): Likewise.
1277         (aarch64_addpdi): Likewise.
1278         (sqrt<mode>2): Likewise.
1279         (vec_load_lanesoi<mode>): Likewise.
1280         (vec_store_lanesoi<mode>): Likewise.
1281         (vec_load_lanesci<mode>): Likewise.
1282         (vec_store_lanesci<mode>): Likewise.
1283         (vec_load_lanesxi<mode>): Likewise.
1284         (vec_store_lanesxi<mode>): Likewise.
1285         (*aarch64_mov<mode>): Likewise.
1286         (aarch64_ld2<mode>_dreg): Likewise.
1287         (aarch64_ld2<mode>_dreg): Likewise.
1288         (aarch64_ld3<mode>_dreg): Likewise.
1289         (aarch64_ld3<mode>_dreg): Likewise.
1290         (aarch64_ld4<mode>_dreg): Likewise.
1291         (aarch64_ld4<mode>_dreg): Likewise.
1292         (aarch64_tbl1<mode>): Likewise.
1293         (aarch64_tbl2v16qi): Likewise.
1294         (aarch64_combinev16qi): Likewise.
1295         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1296         (aarch64_st2<mode>_dreg): Likewise.
1297         (aarch64_st2<mode>_dreg): Likewise.
1298         (aarch64_st3<mode>_dreg): Likewise.
1299         (aarch64_st3<mode>_dreg): Likewise.
1300         (aarch64_st4<mode>_dreg): Likewise.
1301         (aarch64_st4<mode>_dreg): Likewise.
1302         (*aarch64_simd_ld1r<mode>): Likewise.
1303         (aarch64_frecpe<mode>): Likewise.
1304         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1305         (aarch64_frecps<mode>): Likewise.
1307 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1309         * config/arm/iterators.md (V_elem_ch): New.
1310         (q): Likewise.
1311         (VQH_type): Likewise.
1312         * config/arm/arm.md (is_neon_type): New.
1313         (conds): Use is_neon_type.
1314         (anddi3_insn): Update type attribute.
1315         (xordi3_insn): Likewise.
1316         (one_cmpldi2): Likewise.
1317         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1318         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1319         (*movmisalign<mode>_neon_store): Likewise.
1320         (*movmisalign<mode>_neon_load): Likewise.
1321         (vec_set<mode>_internal): Likewise.
1322         (vec_set<mode>_internal): Likewise.
1323         (vec_setv2di_internal): Likewise.
1324         (vec_extract<mode>): Likewise.
1325         (vec_extract<mode>): Likewise.
1326         (vec_extractv2di): Likewise.
1327         (*add<mode>3_neon): Likewise.
1328         (adddi3_neon): Likewise.
1329         (*sub<mode>3_neon): Likewise.
1330         (subdi3_neon): Likewise.
1331         (fma<VCVTF:mode>4): Likewise.
1332         (fma<VCVTF:mode>4_intrinsic): Likewise.
1333         (*fmsub<VCVTF:mode>4): Likewise.
1334         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1335         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1336         (ior<mode>3): Likewise.
1337         (and<mode>3): Likewise.
1338         (orn<mode>3_neon): Likewise.
1339         (orndi3_neon): Likewise.
1340         (bic<mode>3_neon): Likewise.
1341         (bicdi3_neon): Likewise.
1342         (xor<mode>3): Likewise.
1343         (one_cmpl<mode>2): Likewise.
1344         (abs<mode>2): Likewise.
1345         (neg<mode>2): Likewise.
1346         (negdi2_neon): Likewise.
1347         (*umin<mode>3_neon): Likewise.
1348         (*umax<mode>3_neon): Likewise.
1349         (*smin<mode>3_neon): Likewise.
1350         (*smax<mode>3_neon): Likewise.
1351         (vashl<mode>3): Likewise.
1352         (vashr<mode>3_imm): Likewise.
1353         (vlshr<mode>3_imm): Likewise.
1354         (ashl<mode>3_signed): Likewise.
1355         (ashl<mode>3_unsigned): Likewise.
1356         (neon_load_count): Likewise.
1357         (ashldi3_neon_noclobber): Likewise.
1358         (ashldi3_neon): Likewise.
1359         (signed_shift_di3_neon): Likewise.
1360         (unsigned_shift_di3_neon): Likewise.
1361         (ashrdi3_neon_imm_noclobber): Likewise.
1362         (lshrdi3_neon_imm_noclobber): Likewise.
1363         (<shift>di3_neon): Likewise.
1364         (widen_ssum<mode>3): Likewise.
1365         (widen_usum<mode>3): Likewise.
1366         (quad_halves_<code>v4si): Likewise.
1367         (quad_halves_<code>v4sf): Likewise.
1368         (quad_halves_<code>v8hi): Likewise.
1369         (quad_halves_<code>v16qi): Likewise.
1370         (reduc_splus_v2di): Likewise.
1371         (neon_vpadd_internal<mode>): Likewise.
1372         (neon_vpsmin<mode>): Likewise.
1373         (neon_vpsmax<mode>): Likewise.
1374         (neon_vpumin<mode>): Likewise.
1375         (neon_vpumax<mode>): Likewise.
1376         (*ss_add<mode>_neon): Likewise.
1377         (*us_add<mode>_neon): Likewise.
1378         (*ss_sub<mode>_neon): Likewise.
1379         (*us_sub<mode>_neon): Likewise.
1380         (neon_vadd<mode>_unspec): Likewise.
1381         (neon_vaddl<mode>): Likewise.
1382         (neon_vaddw<mode>): Likewise.
1383         (neon_vhadd<mode>): Likewise.
1384         (neon_vqadd<mode>): Likewise.
1385         (neon_vaddhn<mode>): Likewise.
1386         (neon_vmul<mode>): Likewise.
1387         (neon_vfms<VCVTF:mode>): Likewise.
1388         (neon_vmlal<mode>): Likewise.
1389         (neon_vmls<mode>): Likewise.
1390         (neon_vmlsl<mode>): Likewise.
1391         (neon_vqdmulh<mode>): Likewise.
1392         (neon_vqdmlal<mode>): Likewise.
1393         (neon_vqdmlsl<mode>): Likewise.
1394         (neon_vmull<mode>): Likewise.
1395         (neon_vqdmull<mode>): Likewise.
1396         (neon_vsub<mode>_unspec): Likewise.
1397         (neon_vsubl<mode>): Likewise.
1398         (neon_vsubw<mode>): Likewise.
1399         (neon_vqsub<mode>): Likewise.
1400         (neon_vhsub<mode>): Likewise.
1401         (neon_vsubhn<mode>): Likewise.
1402         (neon_vceq<mode>): Likewise.
1403         (neon_vcge<mode>): Likewise.
1404         (neon_vcgeu<mode>): Likewise.
1405         (neon_vcgt<mode>): Likewise.
1406         (neon_vcgtu<mode>): Likewise.
1407         (neon_vcle<mode>): Likewise.
1408         (neon_vclt<mode>): Likewise.
1409         (neon_vcage<mode>): Likewise.
1410         (neon_vcagt<mode>): Likewise.
1411         (neon_vtst<mode>): Likewise.
1412         (neon_vabd<mode>): Likewise.
1413         (neon_vabdl<mode>): Likewise.
1414         (neon_vaba<mode>): Likewise.
1415         (neon_vabal<mode>): Likewise.
1416         (neon_vmax<mode>): Likewise.
1417         (neon_vmin<mode>): Likewise.
1418         (neon_vpaddl<mode>): Likewise.
1419         (neon_vpadal<mode>): Likewise.
1420         (neon_vpmax<mode>): Likewise.
1421         (neon_vpmin<mode>): Likewise.
1422         (neon_vrecps<mode>): Likewise.
1423         (neon_vrsqrts<mode>): Likewise.
1424         (neon_vqabs<mode>): Likewise.
1425         (neon_vqneg<mode>): Likewise.
1426         (neon_vcls<mode>): Likewise.
1427         (clz<mode>2): Likewise.
1428         (popcount<mode>2): Likewise.
1429         (neon_vrecpe<mode>): Likewise.
1430         (neon_vrsqrte<mode>): Likewise.
1431         (neon_vget_lane<mode>_sext_internal): Likewise.
1432         (neon_vget_lane<mode>_zext_internal): Likewise.
1433         (neon_vdup_n<mode>): Likewise.
1434         (neon_vdup_n<mode>): Likewise.
1435         (neon_vdup_nv2di): Likewise.
1436         (neon_vdup_lane<mode>_interal): Likewise.
1437         (*neon_vswp<mode>): Likewise.
1438         (neon_vcombine<mode>): Likewise.
1439         (float<mode><V_cvtto>2): Likewise.
1440         (floatuns<mode><V_cvtto>2): Likewise.
1441         (fix_trunc<mode><V_cvtto>2): Likewise.
1442         (fixuns_trunc<mode><V_cvtto>2
1443         (neon_vcvt<mode>): Likewise.
1444         (neon_vcvt<mode>): Likewise.
1445         (neon_vcvtv4sfv4hf): Likewise.
1446         (neon_vcvtv4hfv4sf): Likewise.
1447         (neon_vcvt_n<mode>): Likewise.
1448         (neon_vcvt_n<mode>): Likewise.
1449         (neon_vmovn<mode>): Likewise.
1450         (neon_vqmovn<mode>): Likewise.
1451         (neon_vqmovun<mode>): Likewise.
1452         (neon_vmovl<mode>): Likewise.
1453         (neon_vmul_lane<mode>): Likewise.
1454         (neon_vmul_lane<mode>): Likewise.
1455         (neon_vmull_lane<mode>): Likewise.
1456         (neon_vqdmull_lane<mode>): Likewise.
1457         (neon_vqdmulh_lane<mode>): Likewise.
1458         (neon_vqdmulh_lane<mode>): Likewise.
1459         (neon_vmla_lane<mode>): Likewise.
1460         (neon_vmla_lane<mode>): Likewise.
1461         (neon_vmlal_lane<mode>): Likewise.
1462         (neon_vqdmlal_lane<mode>): Likewise.
1463         (neon_vmls_lane<mode>): Likewise.
1464         (neon_vmls_lane<mode>): Likewise.
1465         (neon_vmlsl_lane<mode>): Likewise.
1466         (neon_vqdmlsl_lane<mode>): Likewise.
1467         (neon_vext<mode>): Likewise.
1468         (neon_vrev64<mode>): Likewise.
1469         (neon_vrev32<mode>): Likewise.
1470         (neon_vrev16<mode>): Likewise.
1471         (neon_vbsl<mode>_internal): Likewise.
1472         (neon_vshl<mode>): Likewise.
1473         (neon_vqshl<mode>): Likewise.
1474         (neon_vshr_n<mode>): Likewise.
1475         (neon_vshrn_n<mode>): Likewise.
1476         (neon_vqshrn_n<mode>): Likewise.
1477         (neon_vqshrun_n<mode>): Likewise.
1478         (neon_vshl_n<mode>): Likewise.
1479         (neon_vqshl_n<mode>): Likewise.
1480         (neon_vqshlu_n<mode>): Likewise.
1481         (neon_vshll_n<mode>): Likewise.
1482         (neon_vsra_n<mode>): Likewise.
1483         (neon_vsri_n<mode>): Likewise.
1484         (neon_vsli_n<mode>): Likewise.
1485         (neon_vtbl1v8qi): Likewise.
1486         (neon_vtbl2v8qi): Likewise.
1487         (neon_vtbl3v8qi): Likewise.
1488         (neon_vtbl4v8qi): Likewise.
1489         (neon_vtbl1v16qi): Likewise.
1490         (neon_vtbl2v16qi): Likewise.
1491         (neon_vcombinev16qi): Likewise.
1492         (neon_vtbx1v8qi): Likewise.
1493         (neon_vtbx2v8qi): Likewise.
1494         (neon_vtbx3v8qi): Likewise.
1495         (neon_vtbx4v8qi): Likewise.
1496         (*neon_vtrn<mode>_insn): Likewise.
1497         (*neon_vzip<mode>_insn): Likewise.
1498         (*neon_vuzp<mode>_insn): Likewise.
1499         (neon_vld1<mode>): Likewise.
1500         (neon_vld1_lane<mode>): Likewise.
1501         (neon_vld1_lane<mode>): Likewise.
1502         (neon_vld1_dup<mode>): Likewise.
1503         (neon_vld1_dup<mode>): Likewise.
1504         (neon_vld1_dupv2di): Likewise.
1505         (neon_vst1<mode>): Likewise.
1506         (neon_vst1_lane<mode>): Likewise.
1507         (neon_vst1_lane<mode>): Likewise.
1508         (neon_vld2<mode>): Likewise.
1509         (neon_vld2<mode>): Likewise.
1510         (neon_vld2_lane<mode>): Likewise.
1511         (neon_vld2_lane<mode>): Likewise.
1512         (neon_vld2_dup<mode>): Likewise.
1513         (neon_vst2<mode>): Likewise.
1514         (neon_vst2<mode>): Likewise.
1515         (neon_vst2_lane<mode>): Likewise.
1516         (neon_vst2_lane<mode>): Likewise.
1517         (neon_vld3<mode>): Likewise.
1518         (neon_vld3qa<mode>): Likewise.
1519         (neon_vld3qb<mode>): Likewise.
1520         (neon_vld3_lane<mode>): Likewise.
1521         (neon_vld3_lane<mode>): Likewise.
1522         (neon_vld3_dup<mode>): Likewise.
1523         (neon_vst3<mode>): Likewise.
1524         (neon_vst3qa<mode>): Likewise.
1525         (neon_vst3qb<mode>): Likewise.
1526         (neon_vst3_lane<mode>): Likewise.
1527         (neon_vst3_lane<mode>): Likewise.
1528         (neon_vld4<mode>): Likewise.
1529         (neon_vld4qa<mode>): Likewise.
1530         (neon_vld4qb<mode>): Likewise.
1531         (neon_vld4_lane<mode>): Likewise.
1532         (neon_vld4_lane<mode>): Likewise.
1533         (neon_vld4_dup<mode>): Likewise.
1534         (neon_vst4<mode>): Likewise.
1535         (neon_vst4qa<mode>): Likewise.
1536         (neon_vst4qb<mode>): Likewise.
1537         (neon_vst4_lane<mode>): Likewise.
1538         (neon_vst4_lane<mode>): Likewise.
1539         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1540         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1541         (neon_vec_<US>mult_lo_<mode>): Likewise.
1542         (neon_vec_<US>mult_hi_<mode>): Likewise.
1543         (neon_vec_<US>shiftl_<mode>): Likewise.
1544         (neon_unpack<US>_<mode>): Likewise.
1545         (neon_vec_<US>mult_<mode>): Likewise.
1546         (vec_pack_trunc_<mode>): Likewise.
1547         (neon_vec_pack_trunc_<mode>): Likewise.
1548         (neon_vabd<mode>_2): Likewise.
1549         (neon_vabd<mode>_3): Likewise.
1551 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1553         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1554         (load_pair): Update type attribute.
1555         (store_pair): Update type attribute.
1556         * config/aarch64/iterators.md (q): New.
1558 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1560         * config/arm/types.md: Add new types for Neon insns.
1562 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1563             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1564             Sergey Lega  <sergey.s.lega@intel.com>
1565             Anna Tikhonova  <anna.tikhonova@intel.com>
1566             Ilya Tocar  <ilya.tocar@intel.com>
1567             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1568             Ilya Verbin  <ilya.verbin@intel.com>
1569             Kirill Yukhin  <kirill.yukhin@intel.com>
1570             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1572         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1573         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1574         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1575         (rcp14<mode>): New.
1576         (srcp14<mode>): Ditto.
1577         (rsqrt14<mode>): Ditto.
1578         (rsqrt14<mode>): Ditto.
1579         (avx512f_vmscalef<mode>): Ditto.
1580         (avx512f_scalef<mode>): Ditto.
1581         (avx512f_getexp<mode>): Ditto.
1582         (avx512f_sgetexp<mode>): Ditto.
1583         (avx512f_fixupimm<mode>): Ditto.
1584         (avx512f_sfixupimm<mode>): Ditto.
1585         (avx512f_rndscale<mode>): Ditto.
1586         (*avx512er_exp2<mode>): Ditto.
1587         (*avx512er_rcp28<mode>): Ditto.
1588         (avx512er_rsqrt28<mode>): Ditto.
1589         (avx512f_getmant<mode>): Ditto.
1590         (avx512f_getmant<mode>): Ditto.
1591         (avx512f_rndscale<mode>): Fix formatting.
1593 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1595         * ipa-utils.h (ipa_edge_within_scc): Declare.
1596         * ipa-cp.c (edge_within_scc): Moved...
1597         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1599 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1600             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1601             Sergey Lega  <sergey.s.lega@intel.com>
1602             Anna Tikhonova  <anna.tikhonova@intel.com>
1603             Ilya Tocar  <ilya.tocar@intel.com>
1604             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1605             Ilya Verbin  <ilya.verbin@intel.com>
1606             Kirill Yukhin  <kirill.yukhin@intel.com>
1607             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1609         * config/i386/predicates.md (const_8_to_15_operand): New.
1610         (const_16_to_31_operand): Ditto.
1611         * config/i386/sse.md (V8FI): New.
1612         (V16FI): Ditto.
1613         (reduc_splus_v8df): Ditto.
1614         (reduc_splus_v16sf): Ditto.
1615         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1616         (vec_extract_hi_<mode>): Ditto.
1617         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1618         (vec_set_lo_<mode>): Ditto.
1619         (vec_set_hi_<mode>): Ditto.
1620         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1621         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1622         (avx512f_pshufd_1): Ditto.
1623         (avx512f_broadcast<mode>): Ditto.
1624         (avx512f_broadcast<mode>): Ditto.
1625         (define_split): Split vec_extract_lo into move.
1626         (ssequartermode): Ditto.
1627         (ssedoublemode): Extened with wider modes.
1628         (vec_extract_lo_<mode>): Ditto.
1630 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1631             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1632             Sergey Lega  <sergey.s.lega@intel.com>
1633             Anna Tikhonova  <anna.tikhonova@intel.com>
1634             Ilya Tocar  <ilya.tocar@intel.com>
1635             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1636             Ilya Verbin  <ilya.verbin@intel.com>
1637             Kirill Yukhin  <kirill.yukhin@intel.com>
1638             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1640         * config/i386/predicates.md (register_or_constm1_operand): New.
1641         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1642         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1643         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1644         UNSPEC_SCATTER_PREFETCH
1645         (VI48_512): New.
1646         (avx512f_ucmp<mode>3): Ditto.
1647         (avx512f_vternlog<mode>): Ditto.
1648         (avx512f_align<mode>): Ditto.
1649         (<shift_insn><mode>3): Ditto.
1650         (avx512f_<rotate>v<mode>): Ditto.
1651         (avx512f_<rotate><mode>): Ditto.
1652         (avx512f_eq<mode>3): Ditto.
1653         (avx512f_eq<mode>3_1): Ditto.
1654         (avx512f_gt<mode>3): Ditto.
1655         (avx512f_testm<mode>3): Ditto.
1656         (avx512f_testnm<mode>3): Ditto.
1657         (avx512pf_gatherpf<mode>): Ditto.
1658         (*avx512pf_gatherpf<mode>_mask): Ditto.
1659         (*avx512pf_gatherpf<mode>): Ditto.
1660         (avx512pf_scatterpf<mode>): Ditto.
1661         (*avx512pf_scatterpf<mode>_mask): Ditto.
1662         (*avx512pf_scatterpf<mode>): Ditto.
1663         (avx512f_vec_dup_gpr<mode>): Ditto.
1664         (clz<mode>2): Ditto.
1665         (conflict<mode>): Ditto.
1666         (REDUC_SMINMAX_MODE): Extened with wider modes.
1667         (reduc_<code>_<mode>): Ditto.
1668         (vlshr<mode>3): Ditto.
1669         (vashl<mode>3): Ditto.
1671 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1672             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1673             Sergey Lega  <sergey.s.lega@intel.com>
1674             Anna Tikhonova  <anna.tikhonova@intel.com>
1675             Ilya Tocar  <ilya.tocar@intel.com>
1676             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1677             Ilya Verbin  <ilya.verbin@intel.com>
1678             Kirill Yukhin  <kirill.yukhin@intel.com>
1679             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1681         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1682         UNSPEC_SCATTER.
1683         (VI48F_512): New.
1684         (avx512fmaskmode): Ditto.
1685         (bcstscalarsuff): Ditto.
1686         (avx512f_blendm<mode>): Ditto.
1687         (cmp_imm_predicate): Ditto.
1688         (avx512f_cmp<mode>3): Ditto.
1689         (avx512f_vec_dup<mode>): Ditto.
1690         (avx512f_vec_dup_mem<mode>): Ditto.
1691         (avx512f_vpermi2var<mode>3): Ditto.
1692         (avx512f_vpermt2var<mode>3): Ditto.
1693         (vec_init<mode>): Ditto.
1694         (avx512f_gathersi<mode>): Ditto.
1695         (*avx512f_gathersi<mode>): Ditto.
1696         (*avx512f_gathersi<mode>_2): Ditto.
1697         (avx512f_gatherdi<mode>): Ditto.
1698         (*avx512f_gatherdi<mode>): Ditto.
1699         (*avx512f_gatherdi<mode>_2): Ditto.
1700         (avx512f_scattersi<mode>): Ditto.
1701         (*avx512f_scattersi<mode>): Ditto.
1702         (avx512f_scatterdi<mode>): Ditto.
1703         (*avx512f_scatterdi<mode>): Ditto.
1704         (sseintprefix): Extened with wider modes.
1705         (VEC_GATHER_IDXSI): Ditto.
1706         (VEC_GATHER_IDXDI): Ditto.
1707         (VEC_GATHER_SRCDI): Ditto.
1709 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1710             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1712         * config/arm/t-aprofile: New file.
1713         * config.gcc: Handle --with-multilib-list option.
1715 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1717         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1718         the input for r2.
1720 2013-10-15  Richard Biener  <rguenther@suse.de>
1722         * tree-loop-distribution.c (build_empty_rdg): Inline into
1723         single user.
1724         (rdg_flag_vertex): Inline into single user.
1725         (rdg_flag_vertex_and_dependent): Likewise.
1726         (build_rdg_partition_for_vertex): Remove processed bitmap.
1727         (rdg_build_partitions): Simplify.
1729 2013-10-15  Richard Biener  <rguenther@suse.de>
1731         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1732         Restructure forwarding through conversions and copies to
1733         avoid performing copy-propagation the wrong way.  Adjust
1734         recursion invocations.
1735         (forward_propagate_addr_expr): Add argument stating if we
1736         are recursing from a single-use.
1737         (ssa_forward_propagate_and_combine): Adjust.
1739 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1741         * dumpfile.h (gcc::dump_manager): New class, to hold state
1742         relating to dumpfile management.
1743         (get_dump_file_name): Remove in favor of method of dump_manager.
1744         (dump_initialized_p): Likewise.
1745         (dump_start): Likewise.
1746         (dump_finish): Likewise.
1747         (dump_switch_p): Likewise.
1748         (dump_register): Likewise.
1749         (get_dump_file_info): Likewise.
1750         * context.c (gcc::context::context): Construct the dump_manager
1751         instance.
1752         * context.h (gcc::context::get_dumps): New.
1753         (gcc::context::m_dumps): New.
1754         * coverage.c (coverage_init): Port to dump_manager API.
1755         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1756         (extra_dump_files_in_use): Likewise.
1757         (extra_dump_files_alloced): Likewise.
1758         (gcc::dump_manager::dump_manager): New.
1759         (dump_register): Convert to...
1760         (gcc::dump_manager::dump_register): ...method, replacing
1761         function-static next_dump with m_next_dump field.
1762         (get_dump_file_info): Convert to...
1763         (gcc::dump_manager::get_dump_file_info): ...method.
1764         (get_dump_file_name): Convert to...
1765         (gcc::dump_manager::get_dump_file_name): ...method.
1766         (dump_start): Convert to...
1767         (gcc::dump_manager::dump_start): ...method.
1768         (dump_finish): Convert to...
1769         (gcc::dump_manager::dump_finish): ...method.
1770         (dump_begin): Replace body with...
1771         (gcc::dump_manager::dump_begin): ...new method.
1772         (dump_phase_enabled_p): Convert to...
1773         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1774         (dump_phase_enabled_p): Convert to...
1775         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1776         (dump_initialized_p):  Convert to...
1777         (gcc::dump_manager::dump_initialized_p): ...method.
1778         (dump_flag_name): Replace body with...
1779         (gcc::dump_manager::dump_flag_name): ...new method.
1780         (dump_enable_all): Convert to...
1781         (gcc::dump_manager::dump_enable_all): ...new method.
1782         (opt_info_enable_passes): Convert to...
1783         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1784         (dump_switch_p_1): Convert to...
1785         (gcc::dump_manager::dump_switch_p_1): ...new method.
1786         (dump_switch_p):  Convert to...
1787         (gcc::dump_manager::dump_switch_p): ...new method.
1788         (opt_info_switch_p): Port to dump_manager API.
1789         (enable_rtl_dump_file): Likewise.
1790         * opts-global.c (handle_common_deferred_options): Port to new
1791         dump_manager API.
1792         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1793         (pass_manager::register_one_dump_file): Likewise.
1794         (pass_manager::register_pass): Likewise.
1795         (pass_init_dump_file): Likewise.
1796         (pass_fini_dump_file): Likewise.
1797         * statistics.c (statistics_early_init): Likewise.
1799 2013-10-14  Richard Biener  <rguenther@suse.de>
1801         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1802         iterative_hash_canonical_type, gimple_canonical_type_hash,
1803         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1804         gimple_register_canonical_type, print_gimple_types_stats,
1805         free_gimple_type_tables): Move to lto/lto.c
1806         (gt-gimple.h): Do not include.
1807         * gimple.h (gimple_register_canonical_type,
1808         print_gimple_types_stats, free_gimple_type_tables): Remove.
1809         * Makefile.in (GTFILES): Remove gimple.c.
1811 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1813         PR target/58716
1814         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1815         scanning for msp430x targets.
1817 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1819         PR bootstrap/58509
1820         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1821         (registers_ok_for_ldd_peep): Move around.
1822         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1823         * config/sparc/sparc.md (widening peepholes): Use it.
1825 2013-10-14  Richard Biener  <rguenther@suse.de>
1827         PR middle-end/58712
1828         PR middle-end/55358
1829         * gimple.c (iterative_hash_canonical_type): Make sure to
1830         record the hash into the correct hashtable slot.
1832 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1834         PR rtl-optimization/58662
1835         * combine.c (try_combine): Take into account death nodes on I2 when
1836         splitting a PARALLEL of two independent SETs.  Fix dump message.
1838 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1840         PR target/51244
1841         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1842         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1843         * config/sh/t-sh (sh_treg_combine.o): New entry.
1844         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1845         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1846         (register_sh_passes): New function.  Register sh_treg_combine pass.
1847         (sh_option_override): Invoke it.
1848         (sh_canonicalize_comparison): Handle op0_preserve_value.
1849         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1850         opportunities.  Canonicalize branch condition.
1851         (nott): Allow only if pseudos can be created for non-SH2A.
1853 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1855         PR target/58690
1856         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1857         (ix86_expand_movmem): Replace copy_addr_to_reg with
1858         ix86_copy_addr_to_reg.
1859         (ix86_expand_setmem): Likewise.
1861 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1863         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1864         provided by ix86_fp_compare_mode instead of CCFPUmode.
1866 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1868         * config/aarch64/arm_neon.h
1869         (vtbx<1,3>_<psu>8): Fix register constriants.
1871 2013-10-11  Jeff Law  <law@redhat.com>
1873         PR tree-optimization/58640
1874         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1875         threading paths that cross over two loop entry points.
1877 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1879         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1880         handle vector float as well.
1881         (*vsx_le_perm_load_v4si): Likewise.
1882         (*vsx_le_perm_store_v2di): Likewise.
1883         (*vsx_le_perm_store_v4si): Likewise.
1885 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1887         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1888         directly to circumvent subtract from splat{31} workaround.
1889         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1890         prototype.
1891         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1892         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1893         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1894         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1895         separate big and little endian logic.
1896         (*altivec_vperm_<mode>_internal): New define_insn.
1897         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1898         separate big and little endian logic.
1899         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1900         (vec_permv16qi): Add little endian logic.
1902 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1904         * doc/extend.texi (returns_nonnull): Remove arguments.
1906 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1907             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1908             Sergey Lega  <sergey.s.lega@intel.com>
1909             Anna Tikhonova  <anna.tikhonova@intel.com>
1910             Ilya Tocar  <ilya.tocar@intel.com>
1911             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1912             Ilya Verbin  <ilya.verbin@intel.com>
1913             Kirill Yukhin  <kirill.yukhin@intel.com>
1914             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1916         * config/i386/sse.md (VI48F_256_512): New.
1917         (avx2_permvar<mode>): Change to ...
1918         (<avx2_avx512f>_permvar<mode>): This.
1920 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1921             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1922             Sergey Lega  <sergey.s.lega@intel.com>
1923             Anna Tikhonova  <anna.tikhonova@intel.com>
1924             Ilya Tocar  <ilya.tocar@intel.com>
1925             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1926             Ilya Verbin  <ilya.verbin@intel.com>
1927             Kirill Yukhin  <kirill.yukhin@intel.com>
1928             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1930         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1931         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1932         * config/i386/sse.md (VI4_AVX): New.
1933         (sf2simodelower): Ditto.
1934         (sse2_cvtps2dq): Change to ...
1935         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1937 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1938             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1939             Sergey Lega  <sergey.s.lega@intel.com>
1940             Anna Tikhonova  <anna.tikhonova@intel.com>
1941             Ilya Tocar  <ilya.tocar@intel.com>
1942             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1943             Ilya Verbin  <ilya.verbin@intel.com>
1944             Kirill Yukhin  <kirill.yukhin@intel.com>
1945             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1947         * config/i386/sse.md (V_512): New.
1948         (VI_512): Ditto.
1949         (vcond<V_512:mode><VF_512:mode>): Ditto.
1950         (vcond<V_512:mode><VI_512:mode>): Ditto.
1951         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1953 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1954             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1955             Sergey Lega  <sergey.s.lega@intel.com>
1956             Anna Tikhonova  <anna.tikhonova@intel.com>
1957             Ilya Tocar  <ilya.tocar@intel.com>
1958             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1959             Ilya Verbin  <ilya.verbin@intel.com>
1960             Kirill Yukhin  <kirill.yukhin@intel.com>
1961             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1963         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1964         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1965         (FMAMODE): Ditto.
1966         (fma<mode>4): Removed condition.
1967         (fms<mode>4): Ditto.
1968         (fnma<mode>4): Ditto.
1969         (fnms<mode>4): Ditto.
1970         (fma4i_fmadd_<mode>): Ditto.
1971         (*fma_fmadd_<mode>): Ditto.
1972         (*fma_fmsub_<mode>): Ditto.
1973         (*fma_fnmadd_<mode>): Ditto.
1974         (*fma_fnmsub_<mode>): Ditto.
1975         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1976         (*fma_fmaddsub_<mode>): Ditto.
1977         (*fma_fmsubadd_<mode>): Ditto.
1978         (*fmai_fmadd_<mode>): Ditto.
1979         (*fmai_fmsub_<mode>): Ditto.
1980         (*fmai_fnmadd_<mode>): Ditto.
1981         (*fmai_fnmsub_<mode>): Ditto.
1983 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1984             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1985             Sergey Lega  <sergey.s.lega@intel.com>
1986             Anna Tikhonova  <anna.tikhonova@intel.com>
1987             Ilya Tocar  <ilya.tocar@intel.com>
1988             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1989             Ilya Verbin  <ilya.verbin@intel.com>
1990             Kirill Yukhin  <kirill.yukhin@intel.com>
1991             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1993         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1994         (VI124_256): Changed to ...
1995         (VI124_256_48_512): This.
1996         (ssepackmode): Extended with wider modes.
1997         (<code><mode>3): Changed iterator.
1998         (*avx2_<code><mode>3): Ditto.
1999         (vec_pack_trunc_<mode>): Ditto.
2001 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2002             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2003             Sergey Lega  <sergey.s.lega@intel.com>
2004             Anna Tikhonova  <anna.tikhonova@intel.com>
2005             Ilya Tocar  <ilya.tocar@intel.com>
2006             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2007             Ilya Verbin  <ilya.verbin@intel.com>
2008             Kirill Yukhin  <kirill.yukhin@intel.com>
2009             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2011         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
2012         (VI8F_256_512): Ditto.
2013         (abs<mode>2): Changed iterator.
2014         (avx2_perm<mode>): Changed to ...
2015         (<avx2_avx512f>_perm<mode>): This.
2016         (avx2_perm<mode>_1): Changed to ...
2017         (<avx2_avx512f>_perm<mode>_1): This.
2019 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2020             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2021             Sergey Lega  <sergey.s.lega@intel.com>
2022             Anna Tikhonova  <anna.tikhonova@intel.com>
2023             Ilya Tocar  <ilya.tocar@intel.com>
2024             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2025             Ilya Verbin  <ilya.verbin@intel.com>
2026             Kirill Yukhin  <kirill.yukhin@intel.com>
2027             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2029         * config/i386/sse.md (VI48_AVX512F): New.
2030         (VI48_AVX2): Changed to ...
2031         (VI48_AVX2_48_AVX512F): This.
2032         (avx2_ashrv<mode>): Changed to ...
2033         (<avx2_avx512f>_ashrv<mode>): This.
2034         (avx2_<shift_insn>v<mode>): Changed to ...
2035         (<avx2_avx512f>_<shift_insn>v<mode>): This.
2037 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2038             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2039             Sergey Lega  <sergey.s.lega@intel.com>
2040             Anna Tikhonova  <anna.tikhonova@intel.com>
2041             Ilya Tocar  <ilya.tocar@intel.com>
2042             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2043             Ilya Verbin  <ilya.verbin@intel.com>
2044             Kirill Yukhin  <kirill.yukhin@intel.com>
2045             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2047         * config/i386/sse.md (VI4_AVX512F): New.
2048         (VI8_AVX2_AVX512F): Ditto.
2049         (mul<mode>3): Extended with wider modes.
2050         (*<sse4_1_avx2>_mul<mode>3): Ditto.
2051         (mul<mode>3): Ditto.
2052         (vec_widen_<s>mult_odd_<mode>): Ditto.
2054 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2055             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2056             Sergey Lega  <sergey.s.lega@intel.com>
2057             Anna Tikhonova  <anna.tikhonova@intel.com>
2058             Ilya Tocar  <ilya.tocar@intel.com>
2059             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2060             Ilya Verbin  <ilya.verbin@intel.com>
2061             Kirill Yukhin  <kirill.yukhin@intel.com>
2062             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2064         * config/i386/sse.md (VI2_AVX512F): New.
2065         (VI124_AVX512F): Ditto.
2066         (sseunpackmode): Extended with wider modes.
2067         (sseunpackfltmode): Ditto.
2068         (vec_unpacks_float_hi_<mode>): Ditto.
2069         (vec_unpacks_float_lo_<mode>): Ditto.
2070         (vec_unpacku_float_hi_<mode>): Ditto.
2071         (vec_unpacku_float_lo_<mode>): Ditto.
2072         (vec_unpacks_lo_<mode>): Ditto.
2073         (vec_unpacks_hi_<mode>): Ditto.
2074         (vec_unpacku_lo_<mode>): Ditto.
2075         (vec_unpacku_hi_<mode>): Ditto.
2077 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2078             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2079             Sergey Lega  <sergey.s.lega@intel.com>
2080             Anna Tikhonova  <anna.tikhonova@intel.com>
2081             Ilya Tocar  <ilya.tocar@intel.com>
2082             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2083             Ilya Verbin  <ilya.verbin@intel.com>
2084             Kirill Yukhin  <kirill.yukhin@intel.com>
2085             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2087         * config/i386/i386.md (multdiv): New.
2088         (multdiv_mnemonic): Ditto.
2089         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2090         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2091         (<sse>_vmdiv<mode>3): Removed.
2093 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2094             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2095             Sergey Lega  <sergey.s.lega@intel.com>
2096             Anna Tikhonova  <anna.tikhonova@intel.com>
2097             Ilya Tocar  <ilya.tocar@intel.com>
2098             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2099             Ilya Verbin  <ilya.verbin@intel.com>
2100             Kirill Yukhin  <kirill.yukhin@intel.com>
2101             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2103         * config/i386/sse.md (V): Extended with wider modes.
2104         (VF2): Ditto.
2105         (ssehalfvecmode): Ditto.
2106         (i128): Ditto.
2107         (ssepackfltmode): Ditto.
2108         (avx_vec_concat<mode>): Ditto.
2109         (V_256_512): New iterator.
2110         (VF2_512_256): Ditto.
2111         (si2dfmode): New attribute.
2112         (si2dfmodelower): Ditto.
2113         (sf2dfmode): Ditto.
2114         (concat_tg_mode): Ditto.
2115         (floatv4siv4df2): Changed to ...
2116         (float<si2dfmodelower><mode>2): This.
2117         (avx_cvtps2pd256): Changed to ...
2118         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2119         (vec_pack_trunc_v4df): Changed to ...
2120         (vec_pack_trunc_<mode>): This.
2121         (avx_vpermil<mode>): Changed to ...
2122         (<sse2_avx_avx512f>_vpermil<mode>): This.
2123         (<fixsuffix>fix_truncv8dfv8si2): New.
2124         (vec_pack_sfix_trunc_v8df): Ditto.
2125         (avx512f_rndscale<mode>): Ditto.
2126         (avx512f_roundpd512): Ditto.
2127         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2129 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2130             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2131             Sergey Lega  <sergey.s.lega@intel.com>
2132             Anna Tikhonova  <anna.tikhonova@intel.com>
2133             Ilya Tocar  <ilya.tocar@intel.com>
2134             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2135             Ilya Verbin  <ilya.verbin@intel.com>
2136             Kirill Yukhin  <kirill.yukhin@intel.com>
2137             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2139         * config/i386/i386.md (any_fix): New iterator.
2140         (fixsuffix): New attribute.
2141         * config/i386/sse.md (VF1): Extened with wider modes.
2142         (VI): Ditto.
2143         (VI_AVX2): Ditto.
2144         (VI8): Ditto.
2145         (sseintvecmodelower): Ditto.
2146         (ssescalarmode): Ditto.
2147         (ssescalarnum): Ditto.
2148         (VF1_128_256): New.
2149         (ssexmmmode): Ditto.
2150         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2151         (<sse>_rcp<mode>2): Change iterator.
2152         (rsqrt<mode>2): Ditto.
2153         (<sse>_rsqrt<mode>2): Ditto.
2154         (avx2_vec_dup<mode>): Ditto.
2155         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2156         (round<mode>2_sfix): Ditto.
2157         (avx2_pbroadcast<mode>): Ditto.
2158         (*andnot<mode>3): Handle XI mode.
2159         (*<code><mode>3): Ditto.
2160         (AVXTOSSEMODE): Removed.
2161         (avx_vpermil<mode>): Changed to ...
2162         (<sse2_avx_avx512f>_vpermil<mode>): This.
2164 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2165             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2166             Sergey Lega  <sergey.s.lega@intel.com>
2167             Anna Tikhonova  <anna.tikhonova@intel.com>
2168             Ilya Tocar  <ilya.tocar@intel.com>
2169             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2170             Ilya Verbin  <ilya.verbin@intel.com>
2171             Kirill Yukhin  <kirill.yukhin@intel.com>
2172             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2174         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2175         (<sse>_comi): Ditto.
2176         (<sse>_ucomi): Ditto.
2177         (sse_cvtss2siq_2): Ditto.
2178         (sse2_cvtsd2si): Ditto.
2179         (sse2_cvtsd2siq): Ditto.
2180         (sse2_cvttsd2si): Ditto.
2181         (sse2_cvttsd2siq): Ditto.
2182         (<shift_insn><mode>3): Ditto.
2183         (sse2_cvtsi2sdq): Update constraint and prefix.
2184         (sse_cvtsi2ss): Update prefix.
2185         (sse_cvtsi2ssq): Ditto.
2187 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2189         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2190         ignore internal calls.
2192 2013-10-11  Richard Biener  <rguenther@suse.de>
2194         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2195         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2196         TREE_OVERFLOW is set.
2198 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2200         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2202         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2203         * gimple.h: Likewise.
2205         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2207         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2208         macros.
2210         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2212         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2213         * configure: Regenerate.
2215 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2217         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2218         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2219         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2220         similarly pp_right_paren.
2221         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2222         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2223         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2224         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2225         clauses.
2226         (omp_declare_simd_clauses_equal,
2227         omp_remove_redundant_declare_simd_attrs): New functions.
2228         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2229         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2230         * tree.h (OMP_LOOP_CHECK): Define.
2231         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2232         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2233         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2234         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2235         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2236         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2237         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2238         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2239         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2240         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2241         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2242         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2243         (omp_remove_redundant_declare_simd_attrs): New prototype.
2244         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2245         GIMPLE_OMP_TEAMS): New codes.
2246         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2247         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2248         fields.
2249         (target_nesting_level): New variable.
2250         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2251         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2252         for collapse > 1 static schedule unless ordered.
2253         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2254         (determine_parallel_type): Adjust get_ws_args_for caller.
2255         (install_var_field): Handle mask & 4 for double indirection.
2256         (scan_sharing_clauses): Ignore shared clause on teams construct.
2257         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2258         (create_omp_child_function): If inside target or declare target
2259         constructs, set "omp declare target" attribute on the child function.
2260         (find_combined_for): New function.
2261         (scan_omp_parallel): Handle combined loops.
2262         (scan_omp_target, scan_omp_teams): New functions.
2263         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2264         restrictions and set ctx->cancellable for cancellable constructs.
2265         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2266         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2267         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2268         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2269         (omp_clause_aligned_alignment): New function.
2270         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2271         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2272         on teams constructs.  Handle user defined reductions and new
2273         OpenMP 4.0 clauses.
2274         (lower_reduction_clauses): Don't set placeholder to address of ref
2275         if it has already the right type.
2276         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2277         (expand_parallel_call): Use the new non-_start suffixed builtins,
2278         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2279         and GOMP_parallel_end after the call.
2280         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2281         (expand_omp_for_init_counts): Handle combined loops.
2282         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2283         loops.
2284         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2285         end of cancellable loops.
2286         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2287         Likewise.  Handle collapse > 1 loops.
2288         (expand_omp_simd): Handle combined loops.
2289         (expand_omp_for): Add inner_stmt argument, adjust callers of
2290         expand_omp_for* functions, use expand_omp_for_static*chunk even
2291         for collapse > 1 unless ordered.
2292         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2293         of cancellable sections.
2294         (expand_omp_single): Remove need_barrier variable, just rely on
2295         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2296         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2297         (expand_omp_atomic_load, expand_omp_atomic_store,
2298         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2299         (expand_omp_target): New function.
2300         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2301         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2302         (build_omp_regions_1): Immediately close region for
2303         GF_OMP_TARGET_KIND_UPDATE.
2304         (maybe_add_implicit_barrier_cancel): New function.
2305         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2306         cancellation.
2307         (lower_omp_single): Likewise.  Add clobber after the barrier.
2308         (lower_omp_taskgroup): New function.
2309         (lower_omp_for): Handle combined loops.  Adjust
2310         lower_rec_input_clauses caller.  Handle cancellation.
2311         (lower_depend_clauses): New function.
2312         (lower_omp_taskreg): Lower depend clauses.  Adjust
2313         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2314         cancellation.
2315         (lower_omp_target, lower_omp_teams): New functions.
2316         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2317         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2318         and GOMP_cancellation_point calls.
2319         (lower_omp): Fold stmts inside of target region.
2320         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2321         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2322         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2323         (BT_FN_VOID_OMPFN_PTR_UINT,
2324         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2325         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2326         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2327         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2328         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2329         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2330         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2331         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2332         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2333         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2334         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2335         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2336         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2337         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2338         BUILT_IN_GOMP_PARALLEL_END.
2339         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2340         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2341         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2342         GF_OMP_FOR_KIND_DISTRIBUTE.
2343         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2344         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2345         (dump_gimple_omp_return): Print lhs if it has any.
2346         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2347         gimple_omp_atomic_seq_cst_p.
2348         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2349         and GIMPLE_OMP_TEAMS.
2350         * langhooks.c (lhd_omp_mappable_type): New function.
2351         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2352         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2353         hook.
2354         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2355         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2356         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2357         (struct gimplify_omp_ctx): Add combined_loop field.
2358         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2359         on stmts inside of target region.
2360         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2361         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2362         ORT_TARGET and ORT_TARGET_DATA.
2363         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2364         Handle GOVD_MAP.
2365         (omp_notice_threadprivate_variable): Complain about threadprivate
2366         variables in target region.
2367         (omp_notice_variable): Complain about vars with non-mappable type
2368         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2369         (omp_check_private): Ignore ORT_TARGET* regions.
2370         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2371         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2372         (find_combined_omp_for): New function.
2373         (gimplify_omp_for): Handle gimplification of combined loops.
2374         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2375         OMP_TEAMS.
2376         (gimplify_omp_target_update): New function.
2377         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2378         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2379         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2380         (gimplify_body): If fndecl has "omp declare target" attribute, add
2381         implicit ORT_TARGET context around it.
2382         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2383         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2384         * tree-nested.c (convert_nonlocal_reference_stmt,
2385         convert_local_reference_stmt, convert_gimple_call): Handle
2386         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2387         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2388         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2389         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2390         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2391         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2392         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2393         BUILT_IN_GOMP_LOOP_END_CANCEL,
2394         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2395         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2396         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2397         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2398         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2399         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2400         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2401         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2402         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2403         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2404         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2405         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2406         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2407         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2408         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2409         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2410         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2411         gimple_build_omp_teams): New functions.
2412         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2413         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2414         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2415         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2416         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2417         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2418         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2419         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2420         GF_OMP_ATOMIC_SEQ_CST): New.
2421         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2422         gimple_build_omp_teams): New prototypes.
2423         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2424         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2425         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2426         GIMPLE_OMP_SINGLE as end of range.
2427         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2428         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2429         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2430         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2431         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2432         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2433         gimple_omp_target_kind, gimple_omp_target_set_kind,
2434         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2435         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2436         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2437         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2438         gimple_omp_teams_set_clauses): New inlines.
2439         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2440         and GIMPLE_OMP_TASKGROUP.
2441         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2442         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2443         enum omp_clause_proc_bind_kind): New.
2444         (union omp_clause_subcode): Add depend_kind, map_kind and
2445         proc_bind_kind fields.
2446         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2447         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2448         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2449         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2450         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2452 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2454         * predict.c (tree_estimate_probability): Add new parameter
2455         for estimate_bb_frequencies.
2456         (estimate_bb_frequencies): Add new parameter to force estimation.
2457         (rebuild_frequencies): When max frequency in function is small,
2458         recompute counts from frequencies.
2459         * predict.h (estimate_bb_frequencies): New parameter.
2461 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2463         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2464         optimizations are disabled.
2466 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2468         * coverage.c (coverage_finish): Fix leak of da_file_name.
2470 2013-10-10  Jan Hubicka  <jh@suse.cz>
2472         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2473         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2474         Bobcat and generic.
2476 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2478         PR middle-end/58670
2479         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2480         if any labels are in FALLTHRU_BB, use a special label emitted
2481         immediately after the asm goto insn rather than label_rtx
2482         of the LABEL_DECL.
2483         (expand_asm_stmt): Adjust caller.
2484         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2485         edge if the last insn in predecessor is a jump with single successor,
2486         but it isn't simplejump_p.
2488 2013-10-10  Richard Biener  <rguenther@suse.de>
2490         PR tree-optimization/58656
2491         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2493 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2495         * gimplify.c: Include expr.h and tm_p.h for targets with special
2496         va-arg padding requirements.
2498 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2500         * tree-flow.h: Move some prototypes to gimple.h.
2501         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2502         * gimple.h: Relocate some prototypes from tree-flow.h
2503         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2504         Move to gimplify.c.
2505         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2506         (build_va_arg_indirect_ref): Relocate and make static.
2507         (std_gimplify_va_arg_expr): Relocate here.
2508         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2509         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2511 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2513         * doc/md.texi: Document the mnemonic attribute.
2515 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2517         PR target/57377
2518         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2519         (set_attr_alternative x ...) when searching for user defined
2520         mnemonic attribute.
2522 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2524         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2526 2013-10-09  Easwaran Raman <eraman@google.com>
2528         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2529         * cfgexpand.c (defer_stack_allocation): ...use here
2530         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2532 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2534         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2535         (operand_equal_for_value_replacement): New function, extracted from
2536         value_replacement and enhanced to catch more cases.
2537         (value_replacement): Use operand_equal_for_value_replacement.
2539 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2541         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2542         get_max_loop_iterations.
2544 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2546         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2547         Place comment above function.
2549 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2551         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2552         are not related to tree-cfg.c.
2553         * tree-ssa-address.h: New file.  Relocate prototypes.
2554         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2555         (addr_for_mem_ref): New.  Combine call to get_address_description and
2556         return addr_for_mem_ref.
2557         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2558         * tree-ssa-live.h: Adjust prototypes.
2559         * passes.c: Include tree-ssa-live.h.
2560         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2561         * graphite.c (graphite_transform_loops): Make static.
2562         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2563         make_pass_graphite, pass_data_graphite_transforms,
2564         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2565         * ipa-pure-const.c (warn_function_noreturn): Make static.
2566         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2567         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2568         Relocate from tree-cfg.c
2569         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2570         static.
2571         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2572         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2573         Move to ipa-pure-const.c.
2574         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2575         Relocate from tree-optimize.c.
2576         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2577         make_pass_fixup_cfg): Move to tree-cfg.c.
2578         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2579         Relocate some prototypes.
2580         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2581         * tree-dump.c (dump_function_to_file): Remove prototype.
2582         Add tree-flow.h to the include file.
2583         * tree-dump.h: Remove prototype.
2584         * tree-parloops.h: New File.  Add prototypes.
2585         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2586         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2587         from tree-ssa-loop.c.
2588         * tree-predcom.c (run_tree_predictive_commoning,
2589         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2590         Relocate here from tree-ssa-loop.c.
2591         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2592         ssa_name_values.release ().
2593         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2594         (ssa_name_values): Relocate from tree-flow.h.
2595         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2596         * tree-ssa-loop.c (run_tree_predictive_commoning,
2597         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2598         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2599         make_pass_graphite, pass_data_graphite_transforms,
2600         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2601         tree_parallelize_loops, pass_data_parallelize_loops,
2602         make_pass_parallelize_loops): Move to other files.
2603         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2604         moved here.
2605         * tree.h: Remove prototypes from tree-address.c.
2607 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2609         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2610         (struct int_tree_map): Move to tree-hasher.h
2611         (SCALE, LABEL, PERCENT): Move to gimple.h
2612         * tree-flow-inline.h: Delete.  Move functions to other files.
2613         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2614         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2615         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2616         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2617         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2618         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2619         * tree-hasher.h: Don't include tree-flow.h.
2620         (struct int_tree_map): Relocate from tree-flow.h.
2621         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2622         tree-flow-inline.h and make static.
2623         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2624         tree-flow-inline.h.
2625         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2626         tree-flow-inline.h and make static.
2627         * tree.h (is_global_var, may_be_aliased): Relocate from
2628         tree-flow-inline.h.
2629         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2630         * value-prof.c: No longer include tree-flow-inline.h.
2631         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2633 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2635         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2636         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2637         (enum move_pos): Move to tree-ssa-loop-im.h
2638         * cfgloop.h: Move some prototypes.
2639         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2640         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2641         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2642         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2643         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2644         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2645         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2646         (enum move_pos): Relocate here.
2647         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2648         tree-ssa-loop.c.
2649         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2650         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2651         make_pass_lim): Relocate here from tree-ssa-loop.c.
2652         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2653         tree-ssa-loop.c.
2654         (loop_edge_to_cancel, unloop_loops): Make static.
2655         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2656         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2657         (tree_complete_unroll, gate_tree_complete_unroll,
2658         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2659         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2660         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2661         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2662         (stmt_invariant_in_loop_p): Remove unused function.
2663         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2664         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2665         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2666         (gcov_type_to_double_int): Move to cfgloop.h.
2667         (double_int_cmp, bound_index,
2668         estimate_numbers_of_iterations_loop): Make static.
2669         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2670         (max_loop_iterations): Factor out get_max_loop_iterations.
2671         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2672         cfgloop.c.
2673         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2674         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2675         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2676         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2677         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2678         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2679         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2680         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2681         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2682         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2683         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2684         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2685         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2686         pass_data_complete_unroll, make_pass_complete_unroll,
2687         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2688         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2689         tree-ssa-loop-ivcanon.c.
2690         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2691         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2692         tree-ssa-loop-prefetch.c.
2693         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2694         tree-ssa-loop-im.c.
2695         (get_lsm_tmp_name): Relocate and add suffix parameter.
2696         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2697         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2698         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2699         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2700         (get_estimated_loop_iterations): Factor out accessor from
2701         estimated_loop_iterations in tree-ssa-loop-niter.c.
2702         (get_max_loop_iterations): Factor out accessor from
2703         _max_loop_iterations in tree-ssa-niter.c.
2704         * loop-unroll.c (decide_unroll_constant_iterations,
2705         decide_unroll_runtime_iterations, decide_peel_simple,
2706         decide_unroll_stupid): Use new get_* accessors.
2708 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2710         PR tree-optimization/20318
2711         * doc/extend.texi (returns_nonnull): New function attribute.
2712         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2713         attribute.
2714         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2715         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2717 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2719         PR middle-end/58570
2720         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2721         false if both components are bitfields.
2723 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2725         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2726         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2727         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2728         * config/aarch64/aarch64.h
2729         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2730         * config/aarch64/aarch64-simd-builtins.def
2731         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2733 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2735         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2737 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2739         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2741 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2743         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2744         (vneg_f64): New intrinsic.
2745         (vneg_s8): Asm replaced with C.
2746         (vneg_s16): Likewise.
2747         (vneg_s32): Likewise.
2748         (vneg_s64): New intrinsic.
2749         (vnegq_f32): Asm replaced with C.
2750         (vnegq_f64): Likewise.
2751         (vnegq_s8): Likewise.
2752         (vnegq_s16): Likewise.
2753         (vnegq_s32): Likewise.
2754         (vnegq_s64): Likewise.
2756 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2758         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2760 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2762         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2763         packed stack special handling.
2764         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2765         back to fixed stack slots for FPRs saved due to stdarg.
2767 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2769         * config/s390/s390.c (s390_frame_info): Restructure function.
2771 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2773         * config/s390/s390.c (struct s390_frame_layout): New field
2774         gpr_save_slots.
2775         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2776         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2777         regs_ever_clobbered to char*.
2778         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2779         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2780         (s390_register_info): Enable FPR save slots.  Move/Copy some
2781         functionality into ...
2782         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2783         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2784         function.
2785         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2786         not imply a stack frame.
2787         (s390_init_frame_layout): Remove variable clobbered_regs.
2788         (s390_update_register_info): Remove function.
2789         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2790         cannot be used for register renaming.
2791         (s390_hard_regno_scratch_ok): New function.
2792         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2793         (s390_initial_elimination_offset): Change offset calculation of
2794         the return address pointer.
2795         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2796         (restore_gprs): Set frame related flag.
2797         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2798         (s390_emit_prologue): Call s390_register_info instead of
2799         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2800         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2801         (s390_optimize_prologue): Call s390_optimize_register_info.
2802         Try to remove also FPR slot save/restore INSNs.  Remove frame
2803         related flags from restore INSNs.
2805 2013-10-08  DJ Delorie  <dj@redhat.com>
2807         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2808         (movhi): Likewise.
2810         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2811         avoid conflict with the MI use of %c.
2812         * config/rl78/rl78-real.md: change %c to %C throughout.
2813         * config/rl78/rl78-virt.md: Likewise.
2815 2013-10-08  Jan Hubicka  <jh@suse.cz>
2817         * config/i386/i386.c (ix86_option_override_internal): Switch
2818         to SSE math for -ffast-math when target ISA supports SSE2.
2820 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2822         * tree-flow.h: Remove some prototypes.
2823         * tree.h: Remove some protypes, add a couple.
2824         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2825         using_eh_for_cleanups_p): Add interface routines for front ends.
2826         * tree-eh.h: New file.  Add protoptyes.
2827         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2828         (add_stmt_to_eh_lp_fn): Make static.
2829         (lower_try_finally): Use new using_eh_for_cleanups_p.
2830         * emit-rtl.c: Include tree-eh.h.
2831         * gimple.h: Include tree-eh.h.
2833 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2835         PR tree-optimization/58480
2836         * tree-vrp.c (infer_nonnull_range): New function.
2837         (infer_value_range): Call infer_nonnull_range.
2839 2013-10-08  Dehao Chen  <dehao@google.com>
2841         PR tree-optimization/58619
2842         * tree-inline.c (copy_phis_for_bb): Combine location data
2843         only if non-null.
2845 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2847         PR target/58423
2848         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2849         RTX_FRAME_RELATED_P on INSN.
2851 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2853         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2854         (altivec_expand_vec_perm_const): Call it.
2856 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2858         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2859         sequences for LE VSX loads and stores at expand time.
2860         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2861         prototype.
2862         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2863         (rs6000_gen_le_vsx_permute): New.
2864         (rs6000_gen_le_vsx_load): New.
2865         (rs6000_gen_le_vsx_store): New.
2866         (rs6000_gen_le_vsx_move): New.
2867         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2868         (*vsx_le_perm_load_v4si): New.
2869         (*vsx_le_perm_load_v8hi): New.
2870         (*vsx_le_perm_load_v16qi): New.
2871         (*vsx_le_perm_store_v2di): New.
2872         (*vsx_le_perm_store_v4si): New.
2873         (*vsx_le_perm_store_v8hi): New.
2874         (*vsx_le_perm_store_v16qi): New.
2875         (*vsx_xxpermdi2_le_<mode>): New.
2876         (*vsx_xxpermdi4_le_<mode>): New.
2877         (*vsx_xxpermdi8_le_V8HI): New.
2878         (*vsx_xxpermdi16_le_V16QI): New.
2879         (*vsx_lxvd2x2_le_<mode>): New.
2880         (*vsx_lxvd2x4_le_<mode>): New.
2881         (*vsx_lxvd2x8_le_V8HI): New.
2882         (*vsx_lxvd2x16_le_V16QI): New.
2883         (*vsx_stxvd2x2_le_<mode>): New.
2884         (*vsx_stxvd2x4_le_<mode>): New.
2885         (*vsx_stxvd2x8_le_V8HI): New.
2886         (*vsx_stxvd2x16_le_V16QI): New.
2888 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2890         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2892 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2894         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2895         loop to work also for 31bit ABI.
2896         Save the stack pointer for frame_size > 0.
2898 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2900         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2901         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2902         constraint letters from expanders.
2903         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2904         retry count to general_operand.
2905         ("tabort"): Give operand 0 a mode.
2906         ("tabort_1"): Add mode and constraint letter for operand 0.
2907         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2909 2013-10-04  Jeff Law  <law@redhat.com>
2911         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2913         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2914         out of ...
2915         (thread_across_edge): Here.  Call it.
2917 2013-10-04  Cary Coutant  <ccoutant@google.com>
2919         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2920         discarding a location list expression (or a piece of one).
2922 2013-10-03  Jan Hubicka  <jh@suse.cz>
2924         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2925         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2926         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2928 2013-10-03  Jan Hubicka  <jh@suse.cz>
2930         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2931         accumulate-outgoing-args when producing unwind info.
2933 2013-10-03  Wei Mi  <wmi@google.com>
2935         * lra-constraints.c (insert_move_for_subreg): New function
2936         extracted from simplify_operand_subreg.
2937         (simplify_operand_subreg): Add reload for paradoxical subreg.
2939 2013-10-03  Rong Xu  <xur@google.com>
2941         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2942         the candidate of builtin_expect such that we should fix the
2943         size/time estimation.
2944         (estimate_function_body_sizes): Do the acutally size/time fix-up
2945         for builtin_expect.
2947 2013-10-03  Rong Xu  <xur@google.com>
2949         * predict.c (tree_predict_by_opcode): Get the probability
2950         for builtin_expect from param builtin_expect_probability.
2951         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2952         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2953         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2955 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2957         PR c++/19476
2958         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2959         * calls.c (alloca_call_p): Use get_callee_fndecl.
2960         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2961         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2962         Likewise.
2963         (vrp_visit_stmt): Remove duplicated code.
2965 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2967         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2968         ceildf2, btruncdf2, instead of vsx_* name.
2970         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2971         iterators to only do V2DF and V4SF here.  Move the DF code to
2972         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2973         just 'v' since only vector operations are handled with these insns
2974         after moving the DF support to rs6000.md.
2975         (vsx_sub<mode>3): Likewise.
2976         (vsx_mul<mode>3): Likewise.
2977         (vsx_div<mode>3): Likewise.
2978         (vsx_fre<mode>2): Likewise.
2979         (vsx_neg<mode>2): Likewise.
2980         (vsx_abs<mode>2): Likewise.
2981         (vsx_nabs<mode>2): Likewise.
2982         (vsx_smax<mode>3): Likewise.
2983         (vsx_smin<mode>3): Likewise.
2984         (vsx_sqrt<mode>2): Likewise.
2985         (vsx_rsqrte<mode>2): Likewise.
2986         (vsx_fms<mode>4): Likewise.
2987         (vsx_nfma<mode>4): Likewise.
2988         (vsx_copysign<mode>3): Likewise.
2989         (vsx_btrunc<mode>2): Likewise.
2990         (vsx_floor<mode>2): Likewise.
2991         (vsx_ceil<mode>2): Likewise.
2992         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2993         (vsx_sminsf3): Likewise.
2994         (vsx_fmadf4): Likewise.
2995         (vsx_fmsdf4): Likewise.
2996         (vsx_nfmadf4): Likewise.
2997         (vsx_nfmsdf4): Likewise.
2998         (vsx_cmpdf_internal1): Likewise.
3000         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
3001         simpler to select whether a target has SPE or traditional floating
3002         point support in iterators.
3003         (TARGET_DF_SPE): Likewise.
3004         (TARGET_SF_FPR): Likewise.
3005         (TARGET_DF_FPR): Likewise.
3006         (TARGET_SF_INSN): Macros to say whether floating point support
3007         exists for a given operation for expanders.
3008         (TARGET_DF_INSN): Likewise.
3010         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
3011         combining of SF/DF mode operations, using both traditional and VSX
3012         registers.
3013         (Fvsx): Likewise.
3014         (Ff): Likewise.
3015         (Fv): Likewise.
3016         (Fs): Likewise.
3017         (Ffre): Likewise.
3018         (FFRE): Likewise.
3019         (abs<mode>2): Combine SF/DF modes using traditional floating point
3020         instructions.  Add support for using the upper DF registers with
3021         VSX support, and SF registers with power8-vector support.  Update
3022         expanders for operations supported by both the SPE and traditional
3023         floating point units.
3024         (abs<mode>2_fpr): Likewise.
3025         (nabs<mode>2): Likewise.
3026         (nabs<mode>2_fpr): Likewise.
3027         (neg<mode>2): Likewise.
3028         (neg<mode>2_fpr): Likewise.
3029         (add<mode>3): Likewise.
3030         (add<mode>3_fpr): Likewise.
3031         (sub<mode>3): Likewise.
3032         (sub<mode>3_fpr): Likewise.
3033         (mul<mode>3): Likewise.
3034         (mul<mode>3_fpr): Likewise.
3035         (div<mode>3): Likewise.
3036         (div<mode>3_fpr): Likewise.
3037         (sqrt<mode>3): Likewise.
3038         (sqrt<mode>3_fpr): Likewise.
3039         (fre<Fs>): Likewise.
3040         (rsqrt<mode>2): Likewise.
3041         (cmp<mode>_fpr): Likewise.
3042         (smax<mode>3): Likewise.
3043         (smin<mode>3): Likewise.
3044         (smax<mode>3_vsx): Likewise.
3045         (smin<mode>3_vsx): Likewise.
3046         (negsf2): Delete SF operations that are merged with DF.
3047         (abssf2): Likewise.
3048         (addsf3): Likewise.
3049         (subsf3): Likewise.
3050         (mulsf3): Likewise.
3051         (divsf3): Likewise.
3052         (fres): Likewise.
3053         (fmasf4_fpr): Likewise.
3054         (fmssf4_fpr): Likewise.
3055         (nfmasf4_fpr): Likewise.
3056         (nfmssf4_fpr): Likewise.
3057         (sqrtsf2): Likewise.
3058         (rsqrtsf_internal1): Likewise.
3059         (smaxsf3): Likewise.
3060         (sminsf3): Likewise.
3061         (cmpsf_internal1): Likewise.
3062         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
3063         (negdf2): Delete DF operations that are merged with SF.
3064         (absdf2): Likewise.
3065         (nabsdf2): Likewise.
3066         (adddf3): Likewise.
3067         (subdf3): Likewise.
3068         (muldf3): Likewise.
3069         (divdf3): Likewise.
3070         (fred): Likewise.
3071         (rsqrtdf_internal1): Likewise.
3072         (fmadf4_fpr): Likewise.
3073         (fmsdf4_fpr): Likewise.
3074         (nfmadf4_fpr): Likewise.
3075         (nfmsdf4_fpr): Likewise.
3076         (sqrtdf2): Likewise.
3077         (smaxdf3): Likewise.
3078         (smindf3): Likewise.
3079         (cmpdf_internal1): Likewise.
3080         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3081         (btrunc<mode>2): Delete separate expander, and combine with the
3082         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3083         (btrunc<mode>2_fpr): Likewise.
3084         (ceil<mode>2): Likewise.
3085         (ceil<mode>2_fpr): Likewise.
3086         (floor<mode>2): Likewise.
3087         (floor<mode>2_fpr): Likewise.
3088         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3089         Add support for using the upper registers with VSX and
3090         power8-vector.  Move insns to be closer to the define_expands. On
3091         VSX systems, prefer the traditional form of FMA over the VSX
3092         version, since the traditional form allows the target not to
3093         overlap with the inputs.
3094         (fms<mode>4_fpr): Likewise.
3095         (nfma<mode>4_fpr): Likewise.
3096         (nfms<mode>4_fpr): Likewise.
3098 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3099             Richard Earnshaw  <richard.earnshaw@arm.com>
3101         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3102         (struct mult_cost_table): Likewise.
3103         (struct mem_cost_table): Likewise.
3104         (struct fp_cost_table): Likewise.
3105         (struct vector_cost_table): Likewise.
3106         (cpu_cost_table): Likewise.
3107         * config/arm/arm.opt (mold-rts-costs): New option.
3108         (mnew-generic-costs): Likewise.
3109         * config/arm/arm.c (generic_extra_costs): New table.
3110         (cortexa15_extra_costs): Likewise.
3111         (arm_slowmul_tune): Use NULL as new costs.
3112         (arm_fastmul_tune): Likewise.
3113         (arm_strongarm_tune): Likewise.
3114         (arm_xscale_tune): Likewise.
3115         (arm_9e_tune): Likewise.
3116         (arm_v6t2_tune): Likewise.
3117         (arm_cortex_a5_tune): Likewise.
3118         (arm_cortex_a9_tune): Likewise.
3119         (arm_v6m_tune): Likewise.
3120         (arm_fa726te_tune): Likewise.
3121         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3122         (arm_cortex_tune): Use generict_extra_costs.
3123         (shifter_op_p): New function.
3124         (arm_unspec_cost): Likewise.
3125         (LIBCALL_COST): Define.
3126         (arm_new_rtx_costs): New function.
3127         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3128         table is available. Use old costs otherwise unless mnew-generic-costs
3129         is specified.
3130         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3131         (cpu_cost_table): Declare.
3133 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3135         PR target/58460
3136         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3137         (*subs_mul_imm_<mode>)
3138         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3139         (*sub_<shift>_<mode>)
3140         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3141         Remove k constraint.
3143 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3145         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3146         code.
3147         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3149 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3151         * predict.c (probably_never_executed): New function.
3152         (probably_never_executed_bb_p): Invoke probably_never_executed.
3153         (probably_never_executed_edge_p): Ditto.
3154         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3155         Treat profile insanities conservatively.
3157 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3159         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3161 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3163         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3164         use it.  Use smaller increase for scratch.  Don't increase reject
3165         for early clobber scratch.
3166         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3167         setting eliminated regs except setting fp from hfp.
3168         (lra_eliminate): Check lra_insn_recog_data on NULL.
3170 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3172         PR target/58587
3173         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3174         setting -mvsx-timode by default until the underlying problem is fixed.
3175         (RS6000_CPU, power7 defaults): Likewise.
3177 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3179         * config/x-linux (host-linux.o): Remove header dependencies.
3180         Use $(COMPILE) and $(POSTCOMPILE).
3181         * config/t-linux-android (linux-android.o): Ditto.
3183 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3185         * Makefile.in (expmed.o-warn): Remove.
3187 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3189         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3190         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3192 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3194         * dojump.c (do_jump_1): Divide probability between
3195         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3197 2013-10-02  Tom Tromey  <tromey@redhat.com>
3199         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3201 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3203         * tree-flow.h: Remove some prototypes.
3204         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3205         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3206         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3207         mark_virtual_phi_result_for_renaming): Relocate here.
3208         * tree-into-ssa.h: Add prototypes.
3209         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3210         single_pred_before_succ_order.
3211         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3212         (nonfreeing_call_p) Move to gimple.c.
3213         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3214         tree-ssa-phiopt.c.
3215         * basic-block.h (single_pred_before_succ_order): Add prototype.
3216         * gimple.c (nonfreeing_call_p): Relocate here.
3217         * gimple.h: Add prototype.
3218         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3219         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3220         * tree-ssa.h: Include tree-ssa-dom.h.
3222 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3224         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3225         Use $(COMPILE) and $(POSTCOMPILE).
3227         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3229 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3231         * tree-flow.h: Remove some prototypes.
3232         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3233         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3234         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3235         propagate_tree_value*): Move from here to...
3236         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3237         propagate_tree_value*): Relocate here.
3238         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3239         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3240         * gimple-fold.c: Remove gimple-fold.h from include list.
3241         * tree-vrp.c: Remove gimple-fold.h from include list.
3242         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3243         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3244         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3245         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3246         * sese.c: Add tree-ssa-propagate.h to include list.
3248 2013-10-02  Richard Biener  <rguenther@suse.de>
3250         * tree-loop-distribution.c: Include tree-vectorizer.h for
3251         find_loop_location.
3252         (enum partition_kind): Remove PKIND_REDUCTION.
3253         (struct partition_s): Remove has_writes member, add reduction_p member.
3254         (partition_alloc): Adjust.
3255         (partition_builtin_p): Likewise.
3256         (partition_has_writes): Remove.
3257         (partition_reduction_p): New function.
3258         (partition_merge_into): Likewise.
3259         (generate_code_for_partition): Commonize builtin partition
3260         handling tail.
3261         (rdg_cannot_recompute_vertex_p): Remove.
3262         (already_processed_vertex_p): Likewise.
3263         (rdg_flag_vertex): Do not set has_writes.
3264         (classify_partition): Adjust.
3265         (rdg_build_partitions): Do not set has_writes, treat all
3266         partitions as useful.
3267         (distribute_loop): Record number of library calls generated.  Adjust.
3268         (tree_loop_distribution): Report number of loops and library
3269         calls generated as opt-info.
3271 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3273         * tree-flow.h: Include new .h files.  Move prototypes.
3274         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3275         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3276         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3277         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3278         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3279         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3280         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3281         * tree.h (get_ref_base_and_extent): Move prototype out.
3282         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3283         tree-dfa.h.
3284         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3285         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3286         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3287         * tree-scalar-evolution.c: Include tree.h.
3288         * sese.c: Include tree.h.
3289         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3290         * dwarf2out.c: Include tree-dfa.h.
3291         * tree-chrec.c: Include tree.h.
3292         * tree-data-ref.c: Include tree.h.
3294 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3296         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3297         Fix whitespace.
3299 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3301         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3302         Use $(COMPILE) and $(POSTCOMPILE).
3303         (sol2-cxx.o): Likewise.
3304         (sol2-stubs.o): Likewise.
3305         (sol2.o): Likewise.
3306         * config/x-solaris (host-solaris.o): Likewise.
3308         * config/sparc/t-sparc (sparc.o): Remove.
3309         (sparc-c.o): Remove header dependencies.
3310         Use $(COMPILE) and $(POSTCOMPILE).
3311         * config/sparc/x-sparc: Likewise.
3313 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3315         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3316         * config/arc/arc700.md: Likewise.
3317         * config/arc/arc-modes.def: Likewise.
3318         * config/arc/arc-simd.h: Likewise.
3319         * config/arc/t-arc-uClibc: Likewise.
3320         * config/arc/t-arc-newlib: Likewise.
3322 2013-10-02  Renlin Li  <renlin.li@arm.com>
3324         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3325         plus_constant.
3326         (aarch64_expand_epilogue): Likewise.
3328 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3329             Yufeng Zhang  <yufeng.zhang@arm.com>
3331         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3332         declaration.
3333         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3334         'base_in' represent a conversion and legal_cast_p_1 holds; set
3335         'base_in' with the returned value from get_unwidened.
3337 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3339         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3340         plus_constant is not used.
3342 2013-10-01  Wei Mi  <wmi@google.com>
3344         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3345         * config/i386/i386.md: Add define_peephole2 to
3346         break partial reg stall for cvtss2sd/cvtsd2ss.
3348 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3350         * config/arc/arc.c (pass_arc_ifcvt::clone):
3351         Update for ctxt_ -> m_ctxt change.
3353 2013-10-01  Jeff Law  <law@redhat.com>
3355         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3356         outgoing_edge and intermediate_edge fields.  Instead store the path.
3357         (redirection_data::hash): Hash on the last edge's destination index.
3358         (redirection_data::equal): Check the entire thread path.
3359         (lookup_redirectio_data): Corresponding changes.
3360         (create_edge_and_update_destination_phis): Likewise.
3361         (thread_single_edge): Likewise.
3363 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3364             Diego Novillo <dnovillo@google.com>
3366         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3367         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3368         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3369         versions of these patterns.
3371         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3372         (__builtin_arc_aligned): Likewise.
3374         * config/arc/arc.md: Expand adc_0 comment stating the intended
3375         purpose and why it isn't ready.
3376         Replace commented out call_value_via_label_mixed with a
3377         plain comment about bl_s.
3379         * config/arc/arc.c (stdio.h): Don't include directly.
3380         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3381         Remove if (1) condition.
3382         (arc_encode_section_info): Fix comment.
3384 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3386         * config/arc/arc.c (arc_conditional_register_usage):
3387         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3388         Also set reg_alloc_order for DMA config regs.
3390 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3391             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3393         * doc/install.texi (--with-cpu): Mention ARC.
3394         (arc-*-elf32): New paragraph.
3395         (arc-linux-uclibc): Likewise.
3396         * doc/md.texi (Machine Constraints): Add ARC part.
3397         * doc/invoke.texi: (menu): Add ARC Options.
3398         (Machine Dependent Options) <ARC Options>: Add synopsis.
3399         (node ARC Options): Add.
3400         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3401         (ARC Built-in Functions): New section defining
3402         generic ARC built-in functions.
3403         (ARC SIMD Built-in Functions): New section defining SIMD specific
3404         built-in functions.
3405         (Declaring Attributes of Functions): Extended
3406         description of short_call and long_call attributes for ARC and
3407         added index entries.
3409 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3410             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3411             Joern Rennecke  <joern.rennecke@embecosm.com>
3412             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3413             Brendan Kehoe  <brendan@zen.org>
3414             Michael Eager  <eager@eagercon.com>
3415             Simon Cook  <simon.cook@embecosm.com>
3416             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3418         * config/arc, common/config/arc: New directories.
3420 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3421             Brendan Kehoe  <brendan@zen.org>
3422             Simon Cook  <simon.cook@embecosm.com>
3424         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3426 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3428         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3429         * tree-ssa-coalesce.h: New. Move prototype to here.
3430         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3431         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3432         (gimple_can_coalesce_p): Move to...
3433         * gimple.c (gimple_can_coalesce_p): Here.
3435 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3437         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3438         (dump_decl_set): Move to gimple.c.
3439         * gimple.h: Don't include tree-ssa-operands.h.
3440         (dump_decl_set): Add prototype.
3441         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3442         Move to gimple-ssa.h.
3443         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3444         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3445         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3446         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3447         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3448         than PHI_ARG_DEF.
3449         (dump_decl_set): Relocate here.
3450         * gimple-ssa.h: New file.
3451         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3452         Relocate from gimple.h.
3453         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3454         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3455         swap_tree_operands and remove non-ssa path.
3456         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3457         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3458         swap_ssa_operands.
3459         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3460         vect_is_simple_reduction_1): Use swap_ssa_operands.
3461         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3462         (enum need_phi_state): Move to tree-into-ssa.c.
3463         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3464         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3465         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3466         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3467         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3468         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3469         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3470         num_imm_uses): Move to ssa-iterators.h.
3471         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3472         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3473         tree-phinodes.h.
3474         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3475         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3476         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3477         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3478         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3479         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3480         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3481         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3482         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3483         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3484         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3485         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3486         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3487         (set_phi_nodes): Move to tree-phinodes.h.
3488         * tree-ssa-operands.h (enum ssa_op_iter_type,
3489         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3490         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3491         (dump_decl_set): Remove prototype.
3492         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3493         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3494         (set_phi_nodes): Relocate from tree-flow-inline.h.
3495         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3496         tree-flow-inline.h
3497         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3498         include list.  Temporarily add gimple.h to include list.
3499         * ssa-iterators.h: New file.
3500         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3501         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3502         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3503         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3504         Relocate from tree-ssa-operands.h.
3505         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3506         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3507         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3508         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3509         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3510         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3511         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3512         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3513         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3514         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3515         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3516         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3517         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3518         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3519         Relocate from tree-flow-inline.h.
3520         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3522 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3524         * aarch64-simd.md
3525         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3526         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3527         Insert '\t' to output template.
3528         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3529         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3530         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3531         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3533 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3535         * doc/install.texi (Host/target specific installation notes for GCC):
3536         Put @anchor before @heading.
3537         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3538         Use @email for email addresses.
3540 2013-10-01  Jeff Law  <law@redhat.com>
3542         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3543         a vec.  Only delete the path if we create one without successfully
3544         registering a jump thread.
3545         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3546         as a pointer.
3547         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3548         (paths): New vector of jump threading paths.
3549         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3550         (THREAD_PATH): New accessor macro for the entire thread path.
3551         (lookup_redirection_data): Get intermediate and final outgoing edge
3552         from the thread path.
3553         (create_edge_and_update_destination_phis): Copy the threading path.
3554         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3555         jump threading path.
3556         (ssa_redirect_edges): Get edges and block types from the jump threading
3557         path.  Free the path vector.
3558         (thread_block): Get edges from the jump threading path.  Look at the
3559         entire path to see if we thread to a loop exit.  If we cancel a jump
3560         thread request, then free the path vector.
3561         (thread_single_edge): Get edges and block types from the jump threading
3562         path.  Free the path vector.
3563         (thread_through_loop_header): Get edges and block types from the jump
3564         threading path.  Free the path vector.
3565         (mark_threaded_blocks): Iterate over the vector of paths and store
3566         the path on the appropriate edge.  Get edges and block types from the
3567         jump threading path.
3568         (mark_threaded_blocks): Get edges and block types from the jump
3569         threading path.  Free the path vector.
3570         (thread_through_all_blocks): Use the vector of paths rather than
3571         a vector of 3-edge sets.
3572         (register_jump_thread): Accept pointer to a path vector rather
3573         than the path vector itself.  Store the path vector for later use.
3574         Simplify.
3576 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3577             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3579         PR target/58574
3580         * config/s390/s390.c (s390_split_branches): Modify check for table
3581         jump insns.
3582         (s390_chunkify_start): Rearrange table jump insn check in order to
3583         deal with compare and branch insns correctly.
3585 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3587         PR target/58578
3588         Revert
3589         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3590         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3591         define_insn_and_split.
3592         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3593         (shiftsi3_compare): New pattern.
3594         (rrx): New pattern.
3595         * config/arm/unspecs.md (UNSPEC_RRX): New.
3597 2013-10-01  Alan Modra  <amodra@gmail.com>
3599         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3600         casing inout operands.
3602 2013-10-01  Richard Biener  <rguenther@suse.de>
3604         PR tree-optimization/58553
3605         * tree-loop-distribution.c (struct partition_s): Add niter member.
3606         (classify_partition): Populate niter member for the partition
3607         and properly identify whether the relevant store happens before
3608         or after the loop exit.
3609         (generate_memset_builtin): Use niter member from the partition.
3610         (generate_memcpy_builtin): Likewise.
3612 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3614         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3615         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3617 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3619         * basic-block.h (edge_list): Prefix member names with "m_".
3620         * context.h (context): Likewise.
3621         * domwalk.h (dom_walker): Likewise.
3622         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3623         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3624         * hash-table.h (hash_table): Likewise.
3625         * machmode.h (bit_field_mode_iterator): Likewise.
3626         * pass_manager.h (pass_list): Likewise.
3627         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3628         * tree-pass.h (pass_data): Likewise.
3629         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3630         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3631         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3632         * asan.c (pass_data_asan): Update accordingly.
3633         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3634         (control_dependences::control_dependences): Likewise.
3635         (control_dependences::~control_dependences): Likewise.
3636         (control_dependences::~control_dependences): Likewise.
3637         (control_dependences::get_edges_dependent_on): Likewise.
3638         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3639         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3640         * context.c (gcc::context::context): Likewise.
3641         * cprop.c (pass_rtl_cprop::clone): Likewise.
3642         * domwalk.c (dom_walker::walk): Likewise.
3643         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3644         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3645         * mode-switching.c (pass_mode_switching::clone): Likewise.
3646         * passes.c (opt_pass::opt_pass): Likewise.
3647         (pass_manager::pass_manager): Likewise.
3648         * predict.c (pass_strip_predict_hints::clone): Likewise.
3649         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3650         (pass_split_all_insns::clone): Likewise.
3651         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3652         Likewise.
3653         (bit_field_mode_iterator::next_mode): Likewise.
3654         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3655         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3656         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3657         * tree-complex.c (pass_lower_complex::clone): Likewise.
3658         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3659         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3660         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3661         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3662         (pass_fold_builtins::clone): Likewise.
3663         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3664         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3665         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3666         (pass_cd_dce::clone): Likewise.
3667         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3668         (pass_phi_only_cprop::clone): Likewise.
3669         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3670         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3671         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3672         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3673         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3674         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3675         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3676         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3677         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3678         * tree-vrp.c (pass_vrp::clone): Likewise.
3679         * tsan.c (pass_tsan::clone): Likewise.
3681 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3683         PR middle-end/58564
3684         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3685         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3687         PR middle-end/58564
3688         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3689         optimization, punt if sign_bit_p looked through any zero extension.
3691 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3693         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3694         Update redirected out edge count in joiner case.
3695         (ssa_redirect_edges): Common the joiner and non-joiner cases
3696         so that joiner case gets profile updates.
3698 2013-09-30  Richard Biener  <rguenther@suse.de>
3700         PR tree-optimization/58554
3701         * tree-loop-distribution.c (classify_partition): Require
3702         unconditionally executed stores for memcpy and memset recognition.
3703         (tree_loop_distribution): Calculate dominance info.
3705 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3707         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3708         (NO_PROFILE_COUNTERS): Likewise.
3709         (PROFILE_HOOK): Likewise.
3710         (FUNCTION_PROFILER): Likewise.
3711         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3713 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3715         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3716         calls and defines in TARGET_MACHO conditional.
3717         (load_macho_picbase_di): Likewise.
3718         (reload_macho_picbase): Likewise.
3719         (reload_macho_picbase_si): Likewise.
3720         (reload_macho_picbase_di): Likewise.
3721         (nonlocal_goto_receiver): Likewise.
3723 2013-09-30  Nick Clifton  <nickc@redhat.com>
3725         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3726         that use the MSP430X ISA.
3727         (msp430_option_override): Scan -mmcu command line option for any
3728         MCU name that supports the MSP430X ISA.
3729         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3730         -mmcu options which enable the MSP430X ISA.
3732 2013-09-30  Richard Biener  <rguenther@suse.de>
3734         PR middle-end/58532
3735         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3736         before looking for setjmp-like calls.
3738 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3740         PR target/10901
3741         * config/darwin-protos.h (machopic_get_function_picbase): New.
3742         * config/darwin.c (machopic_get_function_picbase): New.
3743         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3744         label for a new func.  (load_macho_picbase_di): Likewise.
3745         (reload_macho_picbase): New expand.
3746         (reload_macho_picbase_si): New insn.
3747         (reload_macho_picbase_di): New insn.
3748         (nonlocal_goto_receiver): New define and split.
3749         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3750         (unspecv enum): Add UNSPECV_NLGR.
3752 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3754         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3755         that altivec registers are correctly sized on Darwin.
3757 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3759         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3760         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3761         * config/x-darwin (host-darwin.o): Likewise.
3762         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3763         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3764         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3766 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3768         * doc/invoke.texi: Fix usage of @tie{} command.
3770 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3772         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3774 2013-09-28  Jan Hubicka  <jh@suse.cz>
3776         * config/i386/x86-tune.def: Add documentation for each of the options;
3777         add whitespace.
3779 2013-09-28  Jan Hubicka  <jh@suse.cz>
3781         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3782         generic.
3783         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3784         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3785         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3786         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3787         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3789 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3791         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3792         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3793         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3794         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3795         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3796         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3797         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3798         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3799         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3800         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3801         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3802         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3803         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3804         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3805         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3806         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3807         gimple.h, godump.c, graphite-clast-to-gimple.c,
3808         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3809         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3810         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3811         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3812         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3813         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3814         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3815         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3816         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3817         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3818         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3819         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3820         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3821         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3822         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3823         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3824         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3825         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3826         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3827         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3828         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3829         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3830         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3831         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3832         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3833         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3834         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3835         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3836         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3837         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3838         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3839         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3840         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3841         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3842         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3843         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3844         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3845         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3846         whitespace before "(".
3848 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3850         * expr.h (extract_bit_field): Remove packedp parameter.
3851         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3852         from forward declaration.
3853         (store_split_bit_field): Remove packedp arg from calls to
3854         extract_fixed_bit_field.
3855         (extract_bit_field_1): Remove packedp parameter and packedp
3856         argument from recursive calls and calls to extract_fixed_bit_field.
3857         (extract_bit_field): Remove packedp parameter and corresponding
3858         arg to extract_bit_field_1.
3859         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3860         to issue warnings.
3861         (extract_split_bit_field): Remove packedp arg from call to
3862         extract_fixed_bit_field.
3863         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3864         (copy_blkmode_from_reg): Likewise.
3865         (copy_blkmode_to_reg): Likewise.
3866         (read_complex_part): Likewise.
3867         (store_field): Likewise.
3868         (expand_expr_real_1): Likewise.
3869         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3870         to extract_bit_field.
3871         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3872         call to extract_bit_field.
3873         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3874         call to extract_bit_field.
3875         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3876         and special packedp behavior from -fstrict-volatile-bitfields
3877         documentation.
3879 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3881         * lra-eliminations.c (init_elim_table): Guard value_p.
3883 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3885         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3886         DFmode, DImode, and SFmode in the upper VSX registers based on the
3887         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3888         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3889         of the VSX register handling.
3891         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3892         (f32_sv): Likewise.
3893         (zero_extendsidi2_lfiwzx): Add support for loading into the
3894         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3895         only do VSX memory options on Altivec registers.
3896         (extendsidi2_lfiwax): Likewise.
3897         (extendsfdf2_fpr): Likewise.
3898         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3899         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3900         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3901         (movdi_internal64): Likewise.
3903 2013-09-27  Xinliang David Li  <davidxl@google.com>
3905         * opts.c (finish_options): Adjust parameters
3906         according to vect cost model.
3907         (common_handle_option): Set dynamic vect cost
3908         model for FDO.
3909         targhooks.c (default_add_stmt_cost): Compute stmt cost
3910         unconditionally.
3911         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3912         Use helper function.
3913         * tree-vectorizer.h (unlimited_cost_model): New function.
3914         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3915         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3916         function.
3917         (vect_enhance_data_refs_alignment): Ditto.
3918         * flag-types.h: New enum.
3919         * common/config/i386/i386-common.c (ix86_option_init_struct):
3920         No need to initialize vect_cost_model flag.
3921         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3922         unconditionally.
3924 2013-09-27  Diego Novillo  <dnovillo@google.com>
3926         * gimple.h (enum ssa_mode): Remove.
3928 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3930         * cfgloop.h (number_of_loops): Fix typo in check for null.
3932 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3934         PR middle-end/58551
3935         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3936         are children of outermost saved_cfun's loop, and set it up to
3937         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3938         if loop != loop0 and SESE region contains bbs that belong to loop0.
3940 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3942         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3943         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3944         (set_address_segment, set_address_base, set_address_index)
3945         (set_address_disp): Accept the argument unconditionally.
3946         (baseness): Remove must_be_base_p and must_be_index_p checks.
3947         (decompose_normal_address): Classify as much as possible in the
3948         main loop.
3950 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3952         * cse.c (count_reg_usage): Handle INT_LIST.
3953         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3954         * reginfo.c (reg_scan_mark_refs): Likewise.
3955         * reload1.c (eliminate_regs_1): Likewise.
3957 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3959         PR middle-end/58547
3960         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3961         signed.
3963 2013-09-27  Richard Biener  <rguenther@suse.de>
3965         PR tree-optimization/58459
3966         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3967         restriction not propagating into loops.
3969 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3971         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3972         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3973         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3975 2013-09-26  Richard Biener  <rguenther@suse.de>
3977         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3979 2013-09-26  Richard Biener  <rguenther@suse.de>
3981         * alias.h (component_uses_parent_alias_set): Rename to ...
3982         (component_uses_parent_alias_set_from): ... this.
3983         * alias.c (component_uses_parent_alias_set): Rename to ...
3984         (component_uses_parent_alias_set_from): ... this and return
3985         the desired parent.
3986         (reference_alias_ptr_type_1): Use the result from
3987         component_uses_parent_alias_set_from instead of stripping
3988         components one at a time.
3989         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3991 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3993         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3994         Move prototypes to...
3995         * tree-ssa-ter.h: New File.  Move prototypes here.
3996         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3997         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3998         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3999         from is_replaceable_p.
4000         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
4001         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
4002         (process_replaceable): Use ter_is_replaceable_p.
4003         (find_replaceable_in_bb): Use ter_is_replaceable_p.
4004         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
4005         newly refactored ssa_is_replaceable_p.
4006         * cfgexpand.c: Include tree-outof-ssa.h.
4007         * ssaexpand.h: Delete.
4009 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
4011         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
4012         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
4013         tree-ssa.c
4014         (create_gimple_tmp): Delete.
4015         (get_expr_type, build_assign, build_type_cast): Move to...
4016         * gimple-builder.c: New File.
4017         (get_expr_type): Relocate from gimple.c.
4018         (build_assign, build_type_cast): Change to only create ssanames.
4019         * gimple.h: Move prototypes to...
4020         * gimple-builder.h: New File. Here.
4021         * tree-ssa.h: And here.
4022         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
4023         count_uses_and_derefs): Relocate from gimple.c.
4024         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
4025         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
4026         * tree-ssa-math-opts (execute_cse_reciprocals): Use
4027         gimple_replace_ssa_lhs.
4028         * asan.c: Include gimple-builder.h.
4029         * Makefile.in: Add gimple-builder.o.
4031 2013-09-26  Richard Biener  <rguenther@suse.de>
4033         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
4034         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
4035         (loe_visit_block): Use gcc_checking_assert.
4036         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
4037         gimple_assign_ssa_name_copy_p.
4038         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
4039         change.
4041 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
4043         * config/rs6000/t-rs6000 (rs6000.o): Remove.
4044         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
4046 2013-09-26  Richard Biener  <rguenther@suse.de>
4048         PR tree-optimization/58539
4049         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
4050         the fact that debug statements are not taking part in loop-closed
4051         SSA construction.
4053 2013-09-26  Nick Clifton  <nickc@redhat.com>
4055         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4056         time warning message.
4057         (msp430_print_operand_raw): Delete unused letter parameter.
4058         (TARGET_PRINT_OPERAND_ADDRESS): Define.
4059         (msp430_print_operand_address): New function.
4060         (msp430_print_operand): Move address printing code from here to
4061         new function.
4062         * config/msp430/msp430.md (movsipsi2): Add comment in generated
4063         assembler.
4064         (zero_extendpsisi2): Likewise.
4065         (extendpsisi2): New pattern.
4066         (andneghi3): New pattern.
4068 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4070         * config/aarch64/aarch64.opt (mlra): New option.
4071         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4072         (TARGET_LRA_P): Define.
4074 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4076         * expr.c (expand_assignment): Remove obsolete comment.
4078 2013-09-25  Jeff Law  <law@redhat.com>
4080         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4081         tree-ssa-threadupdate.h.
4082         (register_jump_thread): Similarly.
4083         * tree-ssa-threadupdate.h: New header file.
4084         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4085         * tree-vrp.c: Likewise.
4086         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4087         (thread_around_empty_blocks): Change type of path vector argument to
4088         an edge,type pair from just an edge.  Initialize both elements when
4089         appending to a jump threading path.  Tweak references to elements
4090         appropriately.
4091         (thread_across_edge): Similarly.  Release memory for the elements
4092         as needed.
4093         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4094         (dump_jump_thread_path): New function broken out from
4095         register_jump_thread.
4096         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4097         path vector entries.  Search the path for NULL edges and dump
4098         the path if one is found.  Tweak the conversion of path to 3-edge
4099         form to use the block copy type information embedded in the path.
4101 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4103         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4105 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4106             Vladimir Makarov  <vmakarov@redhat.com>
4108         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4109         from the least significant bit.
4110         (strip_address_mutations): Add bitfield operations handling.
4111         (must_be_index_p): Add shifting and rotate operations handling.
4112         (set_address_base): Use must_be_base_p predicate.
4113         (set_address_index): Use must_be_index_p predicate.
4115 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4116             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4117             Sergey Lega  <sergey.s.lega@intel.com>
4118             Anna Tikhonova  <anna.tikhonova@intel.com>
4119             Ilya Tocar  <ilya.tocar@intel.com>
4120             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4121             Ilya Verbin  <ilya.verbin@intel.com>
4122             Kirill Yukhin  <kirill.yukhin@intel.com>
4123             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4125         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4126         Use new names.
4127         (ix86_expand_vector_move_misalign): Support new unaligned load and
4128         stores and use new names.
4129         (CODE_FOR_sse2_storedqu): Rename to ...
4130         (CODE_FOR_sse2_storedquv16qi): ... this.
4131         (CODE_FOR_sse2_loaddqu): Rename to ...
4132         (CODE_FOR_sse2_loaddquv16qi): ... this.
4133         (CODE_FOR_avx_loaddqu256): Rename to ...
4134         (CODE_FOR_avx_loaddquv32qi): ... this.
4135         (CODE_FOR_avx_storedqu256): Rename to ...
4136         (CODE_FOR_avx_storedquv32qi): ... this.
4137         * config/i386/i386.md (fpint_logic): New.
4138         * config/i386/sse.md (VMOVE): Extend for AVX512.
4139         (VF): Ditto.
4140         (VF_128_256): New.
4141         (VF_512): Ditto.
4142         (VI_UNALIGNED_LOADSTORE): Ditto.
4143         (sse2_avx_avx512f): Ditto.
4144         (sse2_avx2): Extend for AVX512.
4145         (sse4_1_avx2): Ditto.
4146         (avx2_avx512f): New.
4147         (sse): Extend for AVX512.
4148         (sse2): Ditto.
4149         (sse4_1): Ditto.
4150         (avxsizesuffix): Ditto.
4151         (sseintvecmode): Ditto.
4152         (ssePSmode): Ditto.
4153         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4154         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4155         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4156         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4157         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4158         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4159         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4160         (STORENT_MODE): Extend for AVX512.
4161         (*absneg<mode>2): Replace constraint "x" with "v".
4162         (*mul<mode>3): Ditto.
4163         (*ieee_smin<mode>3): Ditto.
4164         (*ieee_smax<mode>3): Ditto.
4165         (avx_cmp<mode>3): Replace VF with VF_128_256.
4166         (*<sse>_maskcmp<mode>3_comm): Ditto.
4167         (<sse>_maskcmp<mode>3): Ditto.
4168         (<sse>_andnot<mode>3): Extend for AVX512.
4169         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4170         (<code><mode>3, fpint_logic): New.
4171         (*<code><mode>3): Extend for AVX512.
4172         (avx512flogicsuff): New.
4173         (avx512f_<logic><mode>): Ditto.
4174         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4175         VF_128_256.
4176         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4177         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4178         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4179         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4180         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4181         (xop_vpermil2<mode>3): Ditto.
4182         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4183         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4184         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4185         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4187 2013-09-25  Tom Tromey  <tromey@redhat.com>
4189         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4190         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4191         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4192         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4193         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4194         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4195         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4196         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4197         (GRAPHITE_HTAB_H): Remove.
4199 2013-09-25  Tom Tromey  <tromey@redhat.com>
4201         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4203 2013-09-25  Tom Tromey  <tromey@redhat.com>
4205         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4207 2013-09-25  Tom Tromey  <tromey@redhat.com>
4209         * config/i386/t-i386 (i386.o): Remove.
4210         (i386-c.o): Use COMPILE and POSTCOMPILE.
4212 2013-09-25  Tom Tromey  <tromey@redhat.com>
4214         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4216 2013-09-25  Tom Tromey  <tromey@redhat.com>
4218         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4219         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4220         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4221         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4222         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4223         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4224         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4225         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4226         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4227         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4228         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4229         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4230         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4231         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4232         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4233         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4234         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4235         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4236         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4237         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4238         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4239         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4240         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4241         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4242         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4243         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4244         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4245         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4246         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4247         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4248         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4249         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4250         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4251         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4252         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4253         (graphite-clast-to-gimple.o, graphite-dependences.o)
4254         (graphite-interchange.o, graphite-poly.o)
4255         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4256         (graphite-optimize-isl.o, tree-vect-loop.o)
4257         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4258         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4259         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4260         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4261         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4262         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4263         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4264         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4265         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4266         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4267         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4268         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4269         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4270         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4271         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4272         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4273         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4274         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4275         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4276         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4277         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4278         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4279         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4280         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4281         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4282         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4283         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4284         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4285         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4286         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4287         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4288         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4289         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4290         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4291         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4292         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4293         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4294         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4295         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4296         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4297         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4298         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4299         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4300         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4301         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4302         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4303         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4304         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4305         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4306         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4307         (gcov-dump.o): Remove.
4308         (default-c.o): Use COMPILE and POSTCOMPILE.
4309         (CFLAGS-gcc.o): New variable.
4310         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4312 2013-09-25  Tom Tromey  <tromey@redhat.com>
4314         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4315         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4316         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4317         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4318         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4319         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4320         (c-family/array-notation-common.o, c-family/stub-objc.o)
4321         (c-family/c-ubsan.o): Remove.
4323 2013-09-25  Tom Tromey  <tromey@redhat.com>
4325         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4327 2013-09-25  Tom Tromey  <tromey@redhat.com>
4329         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4330         to add -DENABLE_SHARED_LIBGCC.
4331         (gcc.o): Don't use subshell.
4333 2013-09-25  Tom Tromey  <tromey@redhat.com>
4335         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4336         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4337         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4338         * configure, config.in: Rebuild.
4340 2013-09-25  Tom Tromey  <tromey@redhat.com>
4342         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4343         (COMPILE, POSTCOMPILE): New variables.
4344         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4345         (DEPFILES): New variable.
4346         Include ".Po" files.
4347         * configure.ac: Add checks for dependency checking.
4348         * configure, aclocal.m4: Regenerate.
4350 2013-09-25  Tom Tromey  <tromey@redhat.com>
4352         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4353         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4355 2013-09-25  Tom Tromey  <tromey@redhat.com>
4357         * Makefile.in (generated_files): Add options.h,
4358         target-hooks-def.h, insn-opinit.h,
4359         common/common-target-hooks-def.h, pass-instances.def,
4360         c-family/c-target-hooks-def.h.
4362 2013-09-25  Jeff Law  <law@redhat.com>
4364         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4365         than foo[foo.length () - 1] to access last member in a vec.
4366         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4368 2013-09-25  Richard Biener  <rguenther@suse.de>
4370         PR middle-end/58521
4371         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4373 2013-09-25  Jan Hubicka  <jh@suse.cz>
4375         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4376         test.
4378 2013-09-25  Marek Polacek  <polacek@redhat.com>
4380         PR sanitizer/58420
4381         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4382         when determining the type name.
4384 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4386         * config/sh/sh.md: Fix formatting.
4388 2013-09-24  Xinliang David Li  <davidxl@google.com>
4390         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4391         max peel iterations parameter.
4392         * param.def: New parameter.
4393         * doc/invoke.texi: Document New parameter.
4395 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4397         * gimple-pretty-print.c: Various whitespace tweaks.
4398         * tree-core.h: Likewise.
4399         * tree-pretty-print.c: Likewise.
4400         * tree-ssa-alias.c: Likewise.
4401         * tree-ssa-copy.c: Likewise.
4402         * tree-ssanames.c: Likewise.
4403         * tree-ssanames.h: Likewise.
4404         * tree-vrp.c: Likewise.
4406 2013-09-24  Alan Modra  <amodra@gmail.com>
4408         PR middle-end/57134
4409         PR middle-end/57586
4410         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4411         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4412         inout operands.
4414 2013-09-24  Richard Biener  <rguenther@suse.de>
4416         PR middle-end/58513
4417         * tree.c (reference_alias_ptr_type): Move ...
4418         * alias.c (reference_alias_ptr_type): ... here and implement
4419         in terms of the new reference_alias_ptr_type_1.
4420         (ref_all_alias_ptr_type_p): New helper.
4421         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4422         use ref_all_alias_ptr_type_p.
4423         (get_deref_alias_set): Add flag_strict_aliasing check here.
4424         (reference_alias_ptr_type_1): New function, split out from ...
4425         (get_alias_set): ... here.
4426         (alias_ptr_types_compatible_p): New function.
4427         * alias.h (reference_alias_ptr_type): Declare.
4428         (alias_ptr_types_compatible_p): Likewise.
4429         * tree.h (reference_alias_ptr_type): Remove.
4430         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4431         to compare MEM_REF alias types.
4433 2013-09-24  Richard Biener  <rguenther@suse.de>
4435         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4437 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4439         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4440         reload helper function arrays into a single array reg_addr.
4441         (reload_fpr_gpr): Likewise.
4442         (reload_gpr_vsx): Likewise.
4443         (reload_vsx_gpr): Likewise.
4444         (struct rs6000_reg_addr): Likewise.
4445         (reg_addr): Likewise.
4446         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4447         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4448         (rs6000_init_hard_regno_mode_ok): Likewise.
4449         (rs6000_secondary_reload_direct_move): Likewise.
4450         (rs6000_secondary_reload): Likewise.
4452         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4453         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4454         power8 changes.  Put wg constraint in alphabetical order.
4456         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4457         for future work to add ISA 2.07 VSX single precision support.
4458         (-mvsx-scalar-double): Change default from -1 to 1, update
4459         documentation comment.
4460         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4461         (-mupper-regs-df): New debug switch to control whether DF values
4462         can go in the traditional Altivec registers.
4463         (-mupper-regs-sf): New debug switch to control whether SF values
4464         can go in the traditional Altivec registers.
4466         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4467         and wy constraints.
4468         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4469         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4470         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4471         (rs6000_debug_legitimate_address_p): Print if we are running
4472         before, during, or after reload.
4473         (rs6000_secondary_reload): Add a comment.
4474         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4476         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4477         constraints.  Update documentation string.
4478         (wd constraint): Likewise.
4479         (wf constraint): Likewise.
4480         (wg constraint): Likewise.
4481         (wn constraint): Likewise.
4482         (ws constraint): Likewise.
4483         (wt constraint): Likewise.
4484         (wx constraint): Likewise.
4485         (wz constraint): Likewise.
4486         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4487         instructions.
4488         (ww constraint): Likewise.
4489         (wy constraint): Likewise.
4490         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4491         in the previous submissions.
4492         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4494 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4496         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4497         (REG_BR_PROB): Say that the probability is stored in an int_list.
4498         * reg-notes.def: Update commentary to mention INT_LIST.
4499         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4500         (INT_LIST): New rtx.
4501         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4502         * rtlanal.c (int_reg_note_p): New function.
4503         (alloc_reg_note): Assert that the note does not have an int argument.
4504         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4505         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4506         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4507         rather than an INSN_LIST.  Handle INT_LIST.
4508         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4509         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4510         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4511         Manipulate them as ints rather than rtxes.
4512         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4513         * regmove.c (copy_src_to_dest): Likewise.
4514         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4516         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4517         into the cases that need it.
4518         * config/arm/arm.c (arm_unwind_emit): Likewise.
4520         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4521         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4522         * loop-doloop.c (add_test, doloop_modify): Likewise.
4523         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4524         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4525         * predict.c (combine_predictions_for_insn): Likewise.
4526         * print-rtl.c (print_rtx): Handle INT_LIST.
4527         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4528         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4529         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4530         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4531         (ix86_split_fp_branch, predict_jump): Likewise.
4532         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4533         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4534         * config/spu/spu.c (ea_load_store_inline): Likewise.
4536         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4537         value of a REG_BR_PROB note.
4538         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4539         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4540         * emit-rtl.c (try_split): Likewise.
4541         * predict.c (br_prob_note_reliable_p): Likewise.
4542         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4543         * reorg.c (mostly_true_jump): Likewise.
4544         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4545         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4546         * config/i386/i386.c (ix86_print_operand): Likewise.
4547         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4548         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4549         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4550         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4551         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4552         * config/sparc/sparc.c (output_cbranch): Likewise.
4553         * config/spu/spu.c (get_branch_target): Likewise.
4554         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4555         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4557 2013-09-23  Jan Hubicka  <jh@suse.cz>
4559         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4560         for ipa-devirt.
4561         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4562         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4563         of external calls
4564         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4565         (gimple_fold_call): Dump inconsistent devirtualizations; add
4566         sanity check for type based devirtualizations.
4567         * ipa-prop.c: Include ipa-utils.h
4568         (ipa_intraprocedural_devirtualization): Add sanity check.
4569         (try_make_edge_direct_virtual_call): Likewise.
4571 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4573         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4574         assignment statements.
4576 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4578         * gimple-pretty-print.c (dump_ssaname_info): New function.
4579         (dump_gimple_phi): Call it.
4580         (pp_gimple_stmt_1): Likewise.
4581         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4582         (range_info_def): Declare.
4583         * tree-pretty-print.c (pp_double_int): New function.
4584         (dump_generic_node): Call it.
4585         * tree-pretty-print.h (pp_double_int): Declare.
4586         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4587         * tree-ssanames.h (range_info_def): New structure.
4588         (value_range_type): Move definition here.
4589         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4590         Declare.
4591         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4592         initialization.
4593         (set_range_info): New function.
4594         (get_range_info): Likewise.
4595         (duplicate_ssa_name_range_info): Likewise.
4596         (duplicate_ssa_name_fn): Check pointer type and call
4597         duplicate_ssa_name_range_info.
4598         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4599         * tree-vrp.c (value_range_type): Remove definition, now in
4600         tree-ssanames.h.
4601         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4602         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4603         (SSA_NAME_RANGE_INFO): New macro.
4605 2013-09-23  Richard Biener  <rguenther@suse.de>
4607         PR tree-optimization/58464
4608         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4609         (phi_trans_add): Change to add conditionally on being not present.
4610         (phi_translate_1): Remove recursion detection here.
4611         (phi_translate): Pre-seed the cache with NULL to catch
4612         recursion here in a more generic way.
4613         (bitmap_find_leader): Adjust comment.
4614         (get_representative_for): Dump value-numbers.
4615         (create_expression_by_pieces): Likewise.
4616         (insert_into_preds_of_block): Likewise.
4618 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4620         PR target/58475
4621         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4622         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4624 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4626         Revert r202780:
4627         2013-09-20  Renlin Li  <renlin.li@arm.com>
4629         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4630         plus_constant.
4631         (aarch64_expand_epilogue): Likewise.
4632         (aarch64_legitimize_reload_address): Likewise.
4634 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4636         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4637         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4639 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4641         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4642         comparison operators when -fno-trapping-math is in effect.
4643         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4644         comparison operators in TFmode and assert that unsupported operators
4645         cannot reach here.
4646         (ia64_print_operand): Likewise.
4648 2013-09-21  Jan Hubicka  <jh@suse.cz>
4650         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4651         (sse_typeless_stores): Enable for core
4652         (sse_load0_by_pxor): Likewise.
4653         (four_jump_limit): Disable for core.
4654         (pad_returns): Likewise.
4655         (avoid_vector_decode): Likewise.
4656         (fuse_cmp_and_branch): Enable for cores.
4657         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4659 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4661         PR middle-end/56791
4662         * config/pa/pa.c (pa_option_override): Disable auto increment and
4663         decrement instructions until reload is completed.
4665         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4666         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4667         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4669 2013-09-20  DJ Delorie  <dj@redhat.com>
4670             Nick Clifton  <nickc@redhat.com>
4672         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4673         (need_to_save): Save bank 0 on interrupts.
4674         (characterize_address): Strip far address wrappers.
4675         (rl78_as_legitimate_address): Likewise.
4676         (transcode_memory_rtx): Likewise.
4677         (rl78_peep_movhi_p): Disable this peephole after devirt.
4678         (rl78_propogate_register_origins): Forget all origins when a
4679         CLOBBER is seen.
4680         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4681         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4682         peephole2's.
4683         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4684         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4685         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4686         relocating.
4687         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4689 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4691         * config/pa/pa.md: In "scc" insn patterns, change output template to
4692         handle const0_rtx in reg_or_0_operand operands.
4694 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4696         PR target/56875
4697         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4698         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4700 2013-09-20  Richard Biener  <rguenther@suse.de>
4702         PR middle-end/58484
4703         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4704         remembering SSA name version and block index.
4705         (new_scev_info_str): Adjust.
4706         (hash_scev_info): Likewise.  Also hash the block index.
4707         (eq_scev_info): Adjust.
4708         (find_var_scev_info): Likewise.
4709         (struct instantiate_cache_entry): Remove.
4710         (struct instantiate_cache_type): Use a htab to map name, block
4711         to chrec.
4712         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4713         (get_instantiated_value_entry): Likewise.
4714         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4715         (instantiate_scev_name): Adjust.
4717 2013-09-20  Jeff Law  <law@redhat.com>
4719         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4721 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4723         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4724         Call aarch64_simd_expand_args to update op[argc].
4726 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4728         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4729         plugin argument.
4731 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4733         * gengtype.c (file_rules): Added rule for *.cc files.
4734         (get_output_file_with_visibility): Give fatal message when no
4735         rules found.
4737 2013-09-20  Renlin Li  <renlin.li@arm.com>
4739         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4740         (aarch64_expand_epilogue): Likewise.
4741         (aarch64_legitimize_reload_address): Likewise.
4743 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4745         PR middle-end/57748
4746         * expr.c (expand_assignment): Remove misalignp code path.
4748 2013-09-20  Marek Polacek  <polacek@redhat.com>
4750         PR sanitizer/58413
4751         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4752         TYPE_PRECISION.  Add asserts.
4754 2013-09-20  Richard Biener  <rguenther@suse.de>
4756         PR tree-optimization/58453
4757         * tree-loop-distribution.c (distribute_loop): Apply the cost
4758         model for -ftree-loop-distribute-patterns, too.
4760 2013-09-20  Richard Biener  <rguenther@suse.de>
4762         PR middle-end/58473
4763         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4764         make type comparison less strict.
4766 2013-09-20  Alan Modra  <amodra@gmail.com>
4768         * configure: Regenerate.
4769         * aclocal.m4: Regenerate.
4771 2013-09-20  Marek Polacek  <polacek@redhat.com>
4773         PR other/58467
4774         * doc/extend.texi: Document that attribute used is meant to be used
4775         on variables with static storage duration.
4777 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4779         PR tree-optimization/58472
4780         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4781         simd_lane_access set inv_p = false.
4782         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4783         the simduid magic VAR_DECL.
4785 2013-09-19  Jan Hubicka  <jh@suse.cz>
4787         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4789 2013-09-17  Jeff Law  <law@redhat.com>
4791         * tree-ssa-dom.c (record_temporary_equivalences): New function
4792         split out of dom_opt_dom_walker::after_dom_children.
4793         (dom_opt_dom_walker::thread_across_edge): Move common code
4794         in here from dom_opt_dom_walker::after_dom_children.
4795         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4797 2013-09-19  Jan Hubicka  <jh@suse.cz>
4799         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4800         (TARGET_GENERIC): Use PROCESOR_GENERIC
4801         (enum processor_type): Unify generic32 and 64.
4802         * i386.md (cpu): Likewise.
4803         * x86-tune.def (use_leave): Enable for generic32.
4804         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4805         * athlon.md: Change generic64 to generic in all occurences.
4806         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4807         (ix86_target_macros_internal): Likewise.
4808         * driver-i386.c (host_detect_local_cpu): Likewise.
4809         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4810         to ..
4811         (generic_memcpy, generic_memset, generic_cost): This one.
4812         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4813         (m_GENERIC32, m_GENERIC64): Remove.
4814         (m_GENERIC): Turn into one flag.
4815         (processor_target): Unify generic tunnings.
4816         (ix86_option_override_internal): Replace generic32/64 by generic.
4817         (ix86_issue_rate): Likewise.
4818         (ix86_adjust_cost): Likewise.
4820 2013-09-19  Jan Hubicka  <jh@suse.cz>
4822         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4823         of speculative flag.
4825 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4827         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4828         GOMP_sections_start, even if !exit_reachable.
4830 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4832         * lra-constraints.c (need_for_all_save_p): Use macro
4833         HARD_REGNO_CALL_PART_CLOBBERED.
4834         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4835         set up pseudo conflict hard regs.
4837 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4839         PR target/58452
4840         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4841         operands.
4843 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4845         PR rtl-optimization/58438
4846         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4847         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4848         from previous subpasses.
4850 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4852         * arm.c (arm_get_frame_offsets): Validate architecture supports
4853         LDRD/STRD before accepting the tuning preference.
4854         (arm_expand_prologue): Likewise.
4855         (arm_expand_epilogue): Likewise.
4857 2013-09-18  Richard Biener  <rguenther@suse.de>
4859         PR tree-optimization/58417
4860         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4861         have chrecs with symbols defined in the loop as operands.
4862         (chrec_fold_multiply): Likewise.
4863         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4864         parameters before folding binary operations.
4865         (struct instantiate_cache_entry_hasher): Remove.
4866         (struct instantiate_cache_type): Use a pointer-map.
4867         (instantiate_cache_type::instantiate_cache_type): New function.
4868         (instantiate_cache_type::get): Likewise.
4869         (instantiate_cache_type::set): Likewise.
4870         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4871         (get_instantiated_value_entry): Likewise.
4872         (global_cache): New global.
4873         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4874         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4875         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4876         (instantiate_scev_name): Adjust.
4877         (instantiate_scev): Construct global instead of local cache.
4878         (resolve_mixers): Likewise.
4880 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4881             Paolo Carlini  <paolo.carlini@oracle.com>
4883         PR c++/58458
4884         * doc/implement-cxx.texi: Fix references to the C++ standards.
4886 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4888         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4889         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4890         simd_lane_access drs, update also DR_ALIGNED_TO.
4892 2013-09-18  Marek Polacek  <polacek@redhat.com>
4894         PR sanitizer/58411
4895         * doc/extend.texi: Document no_sanitize_undefined attribute.
4896         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4897         no_sanitize_undefined attribute.
4899 2013-09-18  Nick Clifton  <nickc@redhat.com>
4901         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4902         (ASM_DECLARE_FUNCTION_NAME): Define.
4904 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4906         * compare-elim.c (find_comparison_dom_walker): New class
4907         (find_comparisons_in_bb): Rename to
4908         find_comparison_dom_walker::before_dom_children
4909         (find_comparisons): Adjust
4910         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4911         adjust.
4912         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4913         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4914         (init_walk_dominator_tree): Remove declaration.
4915         (fini_walk_dominator_tree): Remove declaration.
4916         * fwprop.c (single_def_use_dom_walker): New class
4917         (single_def_use_enter_block): Convert to
4918         single_def_use_dom_walker::before_dom_children.
4919         (single_def_use_leave_block): Convert to
4920         single_def_use_dom_walker::after_dom_children.
4921         (build_single_def_use_links): Adjust.
4922         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4923         class.
4924         (find_candidates_in_block): Convert to
4925         find_candidates_dom_walker::before_dom_children.
4926         (execute_strength_reduction): Adjust.
4927         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4928         (sese_dom_walker): New class.
4929         (sese_dom_walker::sese_dom_walker): New constructor.
4930         (sese_dom_walker::~sese_dom_walker): New destructor.
4931         (build_sese_conditions_before): Convert to
4932         sese_dom_walker::before_dom_children.
4933         (build_sese_conditions_after): Convert to
4934         sese_dom_walker::after_dom_children.
4935         (build_poly_scop): Adjust
4936         * tree-into-ssa.c (rewrite_dom_walker): New class
4937         (rewrite_enter_block): Convert to
4938         rewrite_dom_walker::before_dom_children.
4939         (rewrite_leave_block): Convert to
4940         rewrite_dom_walker::after_dom_children.
4941         (rewrite_update_dom_walker): New class.
4942         (rewrite_update_enter_block): Convert to
4943         rewrite_update_dom_walker::before_dom_children.
4944         (rewrite_update_leave_block): Convert to
4945         rewrite_update_dom_walker::after_dom_children.
4946         (rewrite_blocks, rewrite_into_ssa): Adjust.
4947         (mark_def_dom_walker): New class.
4948         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4949         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4950         (mark_def_sites_blocks): Convert to
4951         mark_def_dom_walker::before_dom_children.
4952         (mark_def_site_blocks): Remove.
4953         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4954         (tree_ssa_dominator_optimize): Adjust.
4955         (dom_thread_across_edge): Convert to method
4956         dom_opt_dom_walker::thread_across_edge.
4957         (dom_opt_enter_block): Convert to member function
4958         dom_opt_dom_walker::before_dom_children.
4959         (dom_opt_leave_block): Convert to member function
4960         dom_opt_dom_walker::after_dom_children.
4961         * tree-ssa-dse.c (dse_dom_walker): New class.
4962         (dse_enter_block): Convert to member function
4963         dse_dom_walker::before_dom_children.
4964         (tree_ssa_dse): Adjust.
4965         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4966         (determine_invariantness_stmt): Convert to method
4967         invariantness_dom_walker::before_dom_children.
4968         (determine_invariantness): Remove
4969         (move_computations_dom_walker): New class.
4970         (move_computations_stmt): Convert to method
4971         move_computations_dom_walker::before_dom_children.
4972         (move_computations, tree_ssa_lim): Adjust.
4973         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4974         (nt_init_block): Make method
4975         notrappping_dom_walker::before_dom_children.
4976         (nt_fini_block): Make
4977         method nontrapping_dom_walker::after_dom_children.
4978         (get_non_trapping): Adjust.
4979         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4980         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4981         (eliminate_leave_block): Make method.
4982         eliminate_dom_walker::after_dom_children.
4983         (eliminate): Adjust
4984         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4985         (strlen_enter_block): Make method
4986         strlen_dom_walker::before_dom_children.
4987         (strlen_leave_block): Make
4988         method strlen_dom_walker::after_dom_children.
4989         (tree_ssa_strlen): Adjust.
4990         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4991         (tree_ssa_uncprop): Adjust.
4992         (uncprop_leave_block): Make method
4993         uncprop_dom_walker::after_dom_children.
4994         (uncprop_leave_block): Make method
4995         uncprop_dom_walker::before_dom_children.
4997 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4999         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
5000         branch in same basic block.  Check both src and dest of the move insn.
5002 2013-09-17  Nick Clifton  <nickc@redhat.com>
5004         * config/rl78/rl78-real.md (bf): New pattern.
5005         (bt): New pattern.
5006         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
5007         (rl78_print_operand): Do not put a # before a %B.
5008         * config/rl78/rl78.opt: Tweak doc strings.
5010 2013-09-17  DJ Delorie  <dj@redhat.com>
5012         * config/rl78/constraints.md (Wcv): Allow up to $r31.
5013         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
5014         (rl78_option_override): Likewise, if -mallregs.
5015         (is_virtual_register): Likewise.
5016         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
5017         (REGNO_OK_FOR_BASE_P): Likewise.
5018         * config/rl78/rl78.opt (-mallregs): New.
5020 2013-09-17  Nick Clifton  <nickc@redhat.com>
5022         * config/rl78/rl78.c (need_to_save): Change return type to bool.
5023         For interrupt functions: save all call clobbered registers if the
5024         interrupt handler is not a leaf function.
5025         (rl78_expand_prologue): Always recompute the frame information.
5026         For interrupt functions: only select bank 0 if one of the bank 0
5027         registers is going to be psuhed.
5029 2013-09-17  DJ Delorie  <dj@redhat.com>
5031         * config/rl78/constraints.md: For each W* constraint, rename to C*
5032         and create a W* constraint that checks for an optional ES: prefix
5033         pattern also.
5034         * config/rl78/rl78.md (UNS_ES_ADDR): New.
5035         (es_addr): New.  Used to wrap far addresses.
5036         * config/rl78/rl78-protos.h (rl78_es_addr): New.
5037         (rl78_es_base): New.
5038         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
5039         wrapped far addresses.
5040         (rl78_print_operand_1): Unwrap far addresses before processing.
5041         (rl78_lo16): Wrap far addresses in unspecs.
5042         (rl78_es_addr): New.
5043         (rl78_es_base): New.
5044         (insn_ok_now): Check for not-yet-wrapped far addresses.
5045         (transcode_memory_rtx): Properly re-wrap far addresses.
5047 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5049         * config/sparc/t-rtems: Add leon3 multilibs.
5051 2013-09-17  Cong Hou  <congh@google.com>
5053         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
5054         when checking the dot production pattern. The type of rhs operand
5055         of multiply is now checked correctly.
5057 2013-09-17  Jeff Law  <law@redhat.com>
5059         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5060         edge implied equivalences into successor phis.
5061         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
5062         here from tree-ssa-threadedge.c.
5063         (mark_threaded_blocks): When threading through a joiner, if both
5064         successors of the joiner's clone reach the same block, verify the
5065         PHI arguments are equal.  If not, cancel the jump threading request.
5066         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
5067         tree-ssa-threadupdate.c
5068         (thread_across_edge): Don't check PHI argument equality when
5069         threading through joiner block here.
5071 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5073         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5074         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5075         (warn_uninit, warn_uninitialized_vars,
5076         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5077         Move to tree-ssa-uninit.c.
5078         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5079         (has_undefined_value_p): New.  Pass dependant parts of
5080         ssa_undefined_value_p.
5081         (uninit_undefined_value_p): Use has_undefined_value_p.
5082         (warn_uninit, warn_uninitialized_vars,
5083         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5084         Move from tree-ssa.c.
5085         * tree-ssa.h: Adjust prototypes.
5087 2013-09-17  Jan Hubicka  <jh@suse.cz>
5089         PR middle-end/58332
5090         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5091         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5092         FUNCTION_NOT_OPTIMIZED.
5093         * ipa-inline-analysis.c (compute_inline_parameters): Function
5094         not optimized is not inlinable unless it is alwaysinline.
5095         (inline_analyze_function): Force calls in not optimized
5096         function not inlinable.
5098 2013-09-17  Jan Hubicka  <jh@suse.cz>
5100         PR middle-end/58329
5101         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5102         to return NULL.
5103         * ipa.c (function_and_variable_visibility): Likewise.
5104         * ipa-profile.c (ipa_profile): Likewise.
5106 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5108         PR ipa/58398
5109         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5110         attribute, and don't inline the resolver in this case.
5112 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5114         * coverage.c (get_coverage_counts): Add missing newline.
5116 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5118         PR tree-optimization/58088
5119         * fold-const.c (mask_with_trailing_zeros): New function.
5120         (fold_binary_loc): Make sure we don't recurse infinitely
5121         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5122         Use mask_with_trailing_zeros where appropriate.
5124 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5126         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5127         of 'prev' var to get better distance estimation.
5129 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5131         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5132         * tree-inline.c (is_parameter_of): New predicate.
5133         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5134         a parameter has been remapped.
5135         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5136         (optimize_inline_calls): Initialize transform_parameter.
5137         (copy_gimple_seq_and_replace_locals): Likewise.
5138         (tree_function_versioning): Likewise.
5139         (maybe_inline_call_in_expr): Likewise.
5141 2013-09-17  Nick Clifton  <nickc@redhat.com>
5143         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5144         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5145         interrupt handlers.
5146         (is_attr_func): New function.
5147         (msp430_is_interrupt_func): New function.
5148         (is_naked_func): New function.
5149         (is_reentrant_func): New function.
5150         (is_critical_func): New function.
5151         (msp430_start_function): Add annotations for function attributes.
5152         (msp430_attr): New function.
5153         (msp430_attribute_table): New.
5154         (msp430_function_section): New function.
5155         (TARGET_ASM_FUNCTION_SECTION): Define.
5156         (msp430_builtin): New enum.
5157         (msp430_init_builtins): New function.
5158         (msp430_builtin_devl): New function.
5159         (msp430_expand_builtin): New function.
5160         (TARGET_INIT_BUILTINS): Define.
5161         (TARGET_EXPAND_BUILTINS): Define.
5162         (TARGET_BUILTIN_DECL): Define.
5163         (msp430_expand_prologue): Add support for naked, interrupt,
5164         critical and reentrant functions.
5165         (msp430_expand_epilogue): Likewise.
5166         (msp430_print_operand): Handle 'O' character.
5167         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5168         NO_TRAMPOLINES.
5169         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5170         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5171         (pushm): Use a 'n' rather than an 'i' constraint.
5172         (msp_return): Add generation of the interrupt return instruction.
5173         (disable_interrupts): New pattern.
5174         (enable_interrupts): New pattern.
5175         (push_intr_state): New pattern.
5176         (pop_intr_state): New pattern.
5177         (bic_SR): New pattern.
5178         (bis_SR): New pattern.
5179         * doc/extend.texi: Document MSP430 function attributes and builtin
5180         functions.
5182 2013-09-17  Richard Biener  <rguenther@suse.de>
5184         PR tree-optimization/58432
5185         * tree-loop-distribution.c (tree_loop_distribution): Also
5186         scan PHIs for outside loop uses and seed a partition from them.
5188 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5190         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5191         (restructure_reference): Call backtrace_base_for_ref.
5193 2013-09-17  Alan Modra  <amodra@gmail.com>
5195         PR target/57589
5196         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5197         patch.
5199 2013-09-16  DJ Delorie  <dj@redhat.com>
5201         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5202         vregs location for RL78/G10.
5203         (rl78_expand_prologue): Avoid SEL on G10.
5204         (rl78_expand_epilogue): Likewise.
5205         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5206         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5207         __RL78_G10__ when appropriate.
5208         (ASM_SPEC): Pass -mg10 along to the assembler.
5209         * config/rl78/rl78.md (sel_rb): Disable for G10.
5210         * config/rl78/rl78.opt: Add -mg10 option.
5211         * config/rl78/t-rl78: Add -mg10 multilib.
5213 2013-09-16  Xinliang David Li  <davidxl@google.com>
5215         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5216         * omp-low.c (omp_max_vf): Ditto.
5217         (expand_omp_simd): Ditto.
5218         * tree-vectorizer.c (vectorize_loops): Ditto.
5219         (gate_vect_slp): Ditto.
5220         (gate_increase_alignment): Ditto.
5221         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5222         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5223         (gate_tree_vectorize): Name change.
5224         (tree_vectorize): Ditto.
5225         (pass_vectorize::gate): Call new function.
5226         (pass_vectorize::execute): Ditto.
5227         * opts.c: O3 default setting change.
5228         (finish_options): Check new flag.
5229         * doc/invoke.texi: Document new flags.
5230         * common.opt: New flags.
5232 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5234         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5235         (Condition Code): Don't reference it.
5236         * doc/tm.texi: Regenerate.
5238 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5240         PR middle-end/58418
5241         * lra-constraints.c (undo_optional_reloads): Consider all optional
5242         reload even if it did not get a hard reg.
5244 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5246         * dumpfile.c (dump_loc): Remove newline emission.
5247         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5248         emission to dump_printf_loc calls where missing.
5249         (vect_mark_for_runtime_alias_test): Ditto.
5250         (vect_analyze_data_ref_dependence): Ditto.
5251         (vect_analyze_data_ref_dependences): Ditto.
5252         (vect_slp_analyze_data_ref_dependence): Ditto.
5253         (vect_slp_analyze_data_ref_dependences): Ditto.
5254         (vect_compute_data_ref_alignment): Ditto.
5255         (vect_update_misalignment_for_peel): Ditto.
5256         (vect_verify_datarefs_alignment): Ditto.
5257         (vector_alignment_reachable_p): Ditto.
5258         (vect_get_data_access_cost): Ditto.
5259         (vect_enhance_data_refs_alignment): Ditto.
5260         (vect_find_same_alignment_drs): Ditto.
5261         (vect_analyze_data_refs_alignment): Ditto.
5262         (vect_analyze_group_access): Ditto.
5263         (vect_analyze_data_ref_access): Ditto.
5264         (vect_analyze_data_ref_accesses): Ditto.
5265         (vect_prune_runtime_alias_test_list): Ditto.
5266         (vect_analyze_data_refs): Ditto.
5267         (vect_create_addr_base_for_vector_ref): Ditto.
5268         (vect_create_data_ref_ptr): Ditto.
5269         (vect_grouped_store_supported): Ditto.
5270         (vect_grouped_load_supported): Ditto.
5271         * value-prof.c (check_counter): Ditto.
5272         (check_ic_target): Ditto.
5273         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5274         (vect_recog_widen_mult_pattern): Ditto.
5275         (vect_recog_widen_sum_pattern): Ditto.
5276         (vect_recog_over_widening_pattern): Ditto.
5277         (vect_recog_widen_shift_pattern): Ditto.
5278         (vect_recog_rotate_pattern): Ditto.
5279         (vect_recog_vector_vector_shift_pattern): Ditto.
5280         (vect_recog_divmod_pattern): Ditto.
5281         (vect_recog_mixed_size_cond_pattern): Ditto.
5282         (vect_recog_bool_pattern): Ditto.
5283         (vect_pattern_recog_1): Ditto.
5284         (vect_pattern_recog): Ditto.
5285         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5286         (vect_is_simple_iv_evolution): Ditto.
5287         (vect_analyze_scalar_cycles_1): Ditto.
5288         (vect_get_loop_niters): Ditto.
5289         (vect_analyze_loop_1): Ditto.
5290         (vect_analyze_loop_form): Ditto.
5291         (vect_analyze_loop_operations): Ditto.
5292         (vect_analyze_loop_2): Ditto.
5293         (vect_analyze_loop): Ditto.
5294         (report_vect_op): Ditto.
5295         (vect_is_slp_reduction): Ditto.
5296         (vect_is_simple_reduction_1): Ditto.
5297         (vect_get_known_peeling_cost): Ditto.
5298         (vect_estimate_min_profitable_iters): Ditto.
5299         (vect_model_reduction_cost): Ditto.
5300         (vect_model_induction_cost): Ditto.
5301         (get_initial_def_for_induction): Ditto.
5302         (vect_create_epilog_for_reduction): Ditto.
5303         (vectorizable_reduction): Ditto.
5304         (vectorizable_induction): Ditto.
5305         (vectorizable_live_operation): Ditto.
5306         (vect_loop_kill_debug_uses): Ditto.
5307         (vect_transform_loop): Ditto.
5308         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5309         (vect_stmt_relevant_p): Ditto.
5310         (process_use): Ditto.
5311         (vect_mark_stmts_to_be_vectorized): Ditto.
5312         (vect_model_simple_cost): Ditto.
5313         (vect_model_promotion_demotion_cost): Ditto.
5314         (vect_model_store_cost): Ditto.
5315         (vect_get_store_cost): Ditto.
5316         (vect_model_load_cost): Ditto.
5317         (vect_get_load_cost): Ditto.
5318         (vect_init_vector_1): Ditto.
5319         (vect_get_vec_def_for_operand): Ditto.
5320         (vect_finish_stmt_generation): Ditto.
5321         (vectorizable_call): Ditto.
5322         (vectorizable_conversion): Ditto.
5323         (vectorizable_assignment): Ditto.
5324         (vectorizable_shift): Ditto.
5325         (vectorizable_operation): Ditto.
5326         (vectorizable_store): Ditto.
5327         (vectorizable_load): Ditto.
5328         (vectorizable_condition): Ditto.
5329         (vect_analyze_stmt): Ditto.
5330         (vect_transform_stmt): Ditto.
5331         (vect_is_simple_use): Ditto.
5332         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5333         (vect_can_advance_ivs_p): Ditto.
5334         (vect_update_ivs_after_vectorizer): Ditto.
5335         (vect_do_peeling_for_loop_bound): Ditto.
5336         (vect_gen_niters_for_prolog_loop): Ditto.
5337         (vect_update_inits_of_drs): Ditto.
5338         (vect_create_cond_for_alias_checks): Ditto.
5339         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5340         (vect_build_slp_tree_1): Ditto.
5341         (vect_supported_load_permutation_p): Ditto.
5342         (vect_analyze_slp_instance): Ditto.
5343         (vect_analyze_slp): Ditto.
5344         (vect_make_slp_decision): Ditto.
5345         (vect_detect_hybrid_slp): Ditto.
5346         (vect_bb_vectorization_profitable_p): Ditto.
5347         (vect_slp_analyze_bb_1): Ditto.
5348         (vect_update_slp_costs_according_to_vf): Ditto.
5349         (vect_get_mask_element): Ditto.
5350         (vect_transform_slp_perm_load): Ditto.
5351         (vect_schedule_slp_instance): Ditto.
5352         (vect_schedule_slp): Ditto.
5353         (vect_slp_transform_bb): Ditto.
5354         * profile.c (read_profile_edge_counts): Ditto.
5355         (compute_branch_probabilities): Ditto.
5356         * coverage.c (get_coverage_counts): Ditto.
5358 2013-09-16  Diego Novillo  <dnovillo@google.com>
5360         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5362 2013-09-16  Jan Hubicka  <jh@suse.cz>
5364         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5365         abstract functions; for static functions check the presence of body.
5367 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5369         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5370         * config/aarch64/aarch64-simd.md
5371         (aarch64_mla_elt<mode>): New.
5372         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5373         (aarch64_mls_elt<mode>): Likewise.
5374         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5375         (aarch64_fma4_elt<mode>): Likewise.
5376         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5377         (aarch64_fma4_elt_to_128v2df): Likewise.
5378         (aarch64_fma4_elt_to_64df): Likewise.
5379         (fnma<mode>4): Likewise.
5380         (aarch64_fnma4_elt<mode>): Likewise.
5381         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5382         (aarch64_fnma4_elt_to_128v2df): Likewise.
5383         (aarch64_fnma4_elt_to_64df): Likewise.
5384         * config/aarch64/iterators.md (VDQSF): New.
5385         * config/aarch64/arm_neon.h
5386         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5387         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5389 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5391         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5392         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5393         (aarch64_mul3_elt_to_128df): Likewise.
5394         (aarch64_mul3_elt_to_64v2df): Likewise.
5395         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5396         (VMUL): New.
5397         (VMUL_CHANGE_NLANES) Likewise.
5398         (h_con): Likewise.
5399         (f): Likewise.
5400         * config/aarch64/arm_neon.h
5401         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5403 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5405         * config/aarch64/arm_neon.h
5406         (vcvtx_high_f32_f64): Fix parameters.
5408 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5409             Uros Bizjak  <ubizjak@gmail.com>
5411         * config/alpha.c: Include tree-ssanames.h.
5413 2013-09-16  Richard Biener  <rguenther@suse.de>
5415         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5416         (dot_rdg_1): Handle control_dd.
5417         (create_edge_for_control_dependence): New function.
5418         (create_rdg_edges): Add control dependences if asked for.
5419         (build_rdg): Likewise.
5420         (generate_loops_for_partition): If there are not necessary
5421         control stmts remove all their dependencies.
5422         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5423         (distribute_loop): Pass on control dependences.
5424         (tree_loop_distribution): Compute control dependences and remove
5425         restriction on number of loop nodes.
5427 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5429         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5430         for internal calls.
5432 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5434         * cse.c (try_const_anchors): Punt on CC modes.
5436 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5438         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5440 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5442         PR target/58382
5443         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5444         calls to word_mode.
5446 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5448         PR target/48094
5449         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5450         seen.
5451         (darwin_objc1_section): Likewise.
5452         (darwin_file_end): Emit Image Info section when required.
5454 2013-09-14  Jan Hubicka  <jh@suse.cz>
5456         * tree-into-ssa.c (gate_into_ssa): New.
5457         (pass_data_build_ssa): Use it.
5458         * cgraph.h (expand_thunk): Update prototype.
5459         * cgraphunit.c (analyze_function): Expand thunks early.
5460         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5461         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5462         set lowered flag; do not add new function.
5463         (assemble_thunks_and_aliases): Update.
5464         * tree-ssa.c (gate_init_datastructures): New gate.
5465         (pass_data_init_datastructures): Use it.
5467 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5469         PR target/58269
5470         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5471         xmm register set described in the psABI.
5473 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5475         * dwarf2out.c (should_emit_struct_debug): Add check
5476         for type_decl variable is not NULL.
5478 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5480         * config.gcc: Use new winnt-c.c target hooks
5481         * config/t-winnt: New file
5482         * config/winnt-c.c: New file
5483         * doc/tm.texi.in: Document new hook
5484         * doc/tm.texi: Regenerated
5486 2013-09-13  Jan Hubicka  <jh@suse.cz>
5488         PR middle-end/58094
5489         * ipa-inline.c (check_callers): New function.
5490         (check_caller_edge): Remove.
5491         (want_inline_function_to_all_callers_p): Also permit alises that are
5492         called dirrectly.
5493         (inline_to_all_callers): Terminate the walk when devirtualization
5494         introduce new calls.
5496 2013-09-13  Jan Hubicka  <jh@suse.cz>
5498         * ipa-inline-analysis.c (struct growth_data): Add node.
5499         (do_estimate_growth_1): Fix detection of recursion.
5501 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5503         PR tree-optimization/58392
5504         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5505         to avoid shadowing of outer loop variable.  If
5506         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5507         replace_by_duplicate_decl simduid of loops that have it set and
5508         set dest_cfun->has_simduid_loops and/or
5509         dest_cfun->has_force_vect_loops.
5510         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5511         instead of maybe_lookup_decl.
5512         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5513         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5514         call.  Copy over force_vect and copy and remap simduid.  Set
5515         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5516         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5517         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5518         caller.  Don't set cfun->has_simduid_loops and/or
5519         cfun->has_force_vect_loops here.
5520         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5521         caller.
5522         (expand_call_inline, tree_function_versioning): Adjust copy_body
5523         callers.
5525 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5527         PR bootstrap/58388
5528         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5529         the assert if the edge was a speculative one.
5531 2013-09-13  Richard Biener  <rguenther@suse.de>
5533         * tree-data-ref.h (known_dependences_p): Move here ...
5534         * tree-loop-distribution.c (known_dependences_p): ... from here.
5535         (dump_rdg_component, debug_rdg_component): Remove.
5536         (dump_rdg): Adjust.
5537         (generate_loops_for_partition): Use gimple_uid instead of
5538         relying on matching stmt visit order.
5539         (rdg_build_partitions): Take starting stmt vector.
5540         (ldist_gen): Merge into ...
5541         (distribute_loop): ... this function.  Do not compute starting
5542         vertices vector.
5543         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5545 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5547         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5548         Set type attribute correctly. Set predicable_short_it attribute.
5549         (cmpsi_shiftsi): Remove %? from output template.
5551 2013-09-13  Richard Biener  <rguenther@suse.de>
5553         * tree-loop-distribution.c (struct rdg_component,
5554         rdg_defs_used_in_other_loops_p, free_rdg_components,
5555         rdg_build_components): Remove.
5556         (stmts_from_loop): Do not record virtual PHIs.
5557         (generate_loops_for_partition): Skip virtual PHIs.
5558         (build_rdg_partition_for_component): Rename to ...
5559         (build_rdg_partition_for_vertex): ... this and adjust.
5560         (rdg_build_partitions): Take a vector of starting vertices
5561         instead of components.  Remove unnecessary leftover handling.
5562         (ldist_gen): Do not build components or record other stores.
5563         (distribute_loop): Do not distribute loops containing stmts
5564         with side-effects.
5566 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5568         PR target/58314
5569         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5571 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5573         * config.gcc: Separate cases for mingw and cygwin targets,
5574         and add 64-bit cygwin target case.
5576         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5577         dll-export inline-functions.
5578         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5581 2013-09-13  Jeff Law  <law@redhat.com>
5583         PR middle-end/58387
5584         Revert:
5585         2013-09-06  Jeff Law  <law@redhat.com>
5587         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5588         edge implied equivalences into successor phis.
5590 2013-09-12  DJ Delorie  <dj@redhat.com>
5592         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5593         separators.
5595 2013-09-12  Brooks Moses  <bmoses@google.com>
5597         PR driver/42955
5598         * Makefile.in: Do not install driver binaries in $(target)/bin.
5600 2013-09-12  DJ Delorie  <dj@redhat.com>
5602         * config/rl78/rl78.opt (mrelax): New.
5603         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5604         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5606         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5607         between SP and FP.
5608         (rl78_expand_epilogue): Likewise.
5610 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5612         PR middle-end/58335
5613         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5614         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5615         pointer elimination with using remove_reg_equal_offset_note.
5617 2013-09-12  DJ Delorie  <dj@redhat.com>
5619         * config/msp430/: New port.
5620         * config.gcc (msp430): Added.
5621         * doc/invoke.texi: Document MSP430 options.
5622         * doc/install.texi: Document msp430-elf
5623         * doc/md.texi: Document msp430-elf
5624         * doc/contrib.texi: Document msp430-elf
5626         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5627         PSImode.
5629 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5631         PR ipa/58389
5632         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5633         reference descriptor is NULL.
5634         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5635         edge to NULL.
5637 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5639         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5640         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5641         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5642         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5643         make_temp_ssa_name): move to tree-ssanames.h
5644         * tree-ssa-alias.h: Move prototype.
5645         * tree-ssa.h: Include tree-ssanames.h.
5646         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5647         * tree-ssanames.h: New.  Move items from tree-flow*.h
5648         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5650 2013-09-12  Richard Biener  <rguenther@suse.de>
5652         PR tree-optimization/58404
5653         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5654         propagate non-invariant addresses into dereferences wrapped
5655         in component references.
5657 2013-09-12  Richard Biener  <rguenther@suse.de>
5659         PR tree-optimization/58402
5660         * passes.def: Move pass_late_warn_uninitialized later.
5662 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5664         * tree-ssa.h: New.  Move content from tree-flow.h and
5665         tree-flow-inline.h.
5666         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5667         Move prototypes belonging to tree-ssa.c.
5668         * tree-flow-inline.h (redirect_edge_var_map_def,
5669         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5670         tree-ssa.h.
5671         * gimple.h: Adjust prototypes.
5672         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5673         to...
5674         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5675         * tree.h: Move prototype to tree-ssa.h.
5676         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5677         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5678         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5679         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5680         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5681         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5682         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5683         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5684         graphite-blocking.c, graphite-clast-to-gimple.c,
5685         graphite-dependences.c, graphite-interchange.c,
5686         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5687         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5688         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5689         ipa-reference.c, ipa-split.c, ipa-utils.c,
5690         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5691         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5692         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5693         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5694         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5695         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5696         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5697         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5698         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5699         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5700         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5701         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5702         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5703         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5704         value-prof.c, var-tracking.c,
5705         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5707 2013-09-12  Richard Biener  <rguenther@suse.de>
5709         PR tree-optimization/58396
5710         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5711         (build_rdg): Take a loop-nest parameter, fix memleaks.
5712         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5714 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5716         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5717         for SLM.
5719 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5721         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5722         The FMA instruction names should have a 'v' prefix.
5724 2013-09-12  Richard Biener  <rguenther@suse.de>
5726         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5727         (dot_rdg): Use popen instead of system in optional code.
5728         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5729         (already_processed_vertex_p): Adjust.
5730         (has_anti_or_output_dependence, predecessor_has_mem_write,
5731         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5732         rdg_flag_uses): Remove.
5733         (rdg_flag_vertex): Simplify.
5734         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5735         remove recursion.
5736         (build_rdg_partition_for_component): Process the first vertex
5737         of a component only.
5738         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5740 2013-09-12  Alan Modra  <amodra@gmail.com>
5742         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5744 2013-09-11  DJ Delorie  <dj@redhat.com>
5745             Nick Clifton  <nickc@redhat.com>
5747         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5748         (rl78_stack_based_mem): New.
5749         * config/rl78/constraints.md (Iv08): New.
5750         (Iv16): New.
5751         (Iv24): New.
5752         (Is09): New.
5753         (Is17): New.
5754         (Is25): New.
5755         (ISsi): New.
5756         (IShi): New.
5757         (ISqi): New.
5758         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5759         (movhi): Likewise.
5760         (movsi): Change from expand to insn-and-split.
5761         (ashrsi3): Clobber AX.
5762         (lshrsi3): New.
5763         (ashlsi3): New.
5764         (cbranchsi4): New.
5765         * config/rl78/rl78.md (CC_REG): Fix.
5766         (addsi3): Allow memory and immediate operands.
5767         (addsi3_internal): Split into...
5768         (addsi3_internal_virt): ...new, and ...
5769         (addsi3_internal_real): ...new.
5770         (subsi): New.
5771         (subsi3_internal_virt): New.
5772         (subsi3_internal_real): New.
5773         (mulsi3): Add memory operand.
5774         (mulsi3_rl78): Likewise.
5775         (mulsi3_g13): Likewise.
5776         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5777         (cbranchqi4_real): Add more constraint options.
5778         (cbranchhi4_real): Expand pattern.
5779         (cbranchhi4_real_signed): New.
5780         (cbranchhi4_real_inverted): New.
5781         (cbranchsi4_real_lt): New.
5782         (cbranchsi4_real_ge): New.
5783         (cbranchsi4_real_signed): New.
5784         (cbranchsi4_real): New.
5785         (peephole2): New.
5786         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5787         constant shifts.
5788         (lshrsi3_virt): Likewise.
5789         (ashlsi3_virt): Likewise.
5790         (cbranchqi4_virt_signed): New.
5791         (cbranchhi4_virt_signed): New.
5792         (cbranchsi4_virt): New.
5793         * config/rl78/rl78.c: Whitespace fixes throughout.
5794         (move_elim_pass): New.
5795         (pass_data_rl78_move_elim): New.
5796         (pass_rl78_move_elim): New.
5797         (make_pass_rl78_move_elim): New.
5798         (rl78_devirt_info): Run devirt earlier.
5799         (rl78_move_elim_info): New.
5800         (rl78_asm_file_start): Register it.
5801         (rl78_split_movsi): New.
5802         (rl78_as_legitimate_address): Allow virtual base registers when
5803         appropriate.
5804         (rl78_addr_space_convert): Remove spurious debug stuff.
5805         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5806         (rl78_print_operand): More cases for not printing '#'.
5807         (rl78_expand_compare): Remove most of the logic.
5808         (content_memory): New.
5809         (clear_content_memory): New.
5810         (get_content_index): New.
5811         (get_content_name): New.
5812         (display_content_memory): New.
5813         (update_content): New.
5814         (record_content): New.
5815         (already_contains): New.
5816         (insn_ok_now): Re-recog insns with virtual registers.
5817         (add_postponed_content_update): New.
5818         (process_postponed_content_update): New.
5819         (gen_and_emit_move): New.
5820         (transcode_memory_rtx): Record new location content.
5821         Use gen_and_emit_move.
5822         (force_into_acc): New.
5823         (move_to_acc): Use gen_and_emit_move.
5824         (move_from_acc): Likewise.
5825         (move_acc_to_reg): Likewise.
5826         (move_to_x): Likewise.
5827         (move_to_hl): Likewise.
5828         (move_to_de): Likewise.
5829         (rl78_alloc_physical_registers_op1): Record location content.
5830         (has_constraint): New.
5831         (rl78_alloc_physical_registers_op2): Record location content.
5832         Optimize use of HL.
5833         (rl78_alloc_physical_registers_ro1): Likewise.
5834         (rl78_alloc_physical_registers_cmp): Likewise.
5835         (rl78_alloc_physical_registers_umul): Likewise.
5836         (rl78_alloc_address_registers_macax): New.
5837         (rl78_alloc_physical_registers): Initialize and set location
5838         content memory as needed.
5839         (rl78_reorg): Make sure split2 is called.
5840         (rl78_rtx_costs): New.
5842 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5844         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5845         for (not (neg ...)) and (neg (not ...)) cases.
5847 2013-09-11  Richard Biener  <rguenther@suse.de>
5849         PR middle-end/58377
5850         * passes.def: Split critical edges before late uninit warning passes.
5851         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5853 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5855         PR tree-optimization/58385
5856         * fold-const.c (build_range_check): If both low and high are NULL,
5857         use omit_one_operand_loc to preserve exp side-effects.
5859 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5861         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5863 2013-09-11  Richard Biener  <rguenther@suse.de>
5865         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5866         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5867         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5868         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5869         stmts_from_loop, known_dependences_p, build_empty_rdg,
5870         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5871         * tree-loop-distribution.c: ... here.
5872         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5873         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5874         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5875         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5876         * tree-loop-distribution.c: ... here.
5877         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5878         (struct partition_s): Add loops member.
5879         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5880         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5881         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5883 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5884             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5885             Sergey Lega  <sergey.s.lega@intel.com>
5886             Anna Tikhonova  <anna.tikhonova@intel.com>
5887             Ilya Tocar  <ilya.tocar@intel.com>
5888             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5889             Ilya Verbin  <ilya.verbin@intel.com>
5890             Kirill Yukhin  <kirill.yukhin@intel.com>
5891             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5893         * config/i386/constraints.md (k): New.
5894         (Yk): Ditto.
5895         * config/i386/i386.c (const regclass_map): Add new mask registers.
5896         (dbx_register_map): Ditto.
5897         (dbx64_register_map): Ditto.
5898         (svr4_dbx_register_map): Ditto.
5899         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5900         disabled.
5901         (ix86_preferred_reload_class): Disable constants for mask registers.
5902         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5903         (ix86_hard_regno_mode_ok): Support new mask registers.
5904         (x86_order_regs_for_local_alloc): Ditto.
5905         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5906         (FIXED_REGISTERS): Add new mask registers.
5907         (CALL_USED_REGISTERS): Ditto.
5908         (REG_ALLOC_ORDER): Ditto.
5909         (VALID_MASK_REG_MODE): New.
5910         (FIRST_MASK_REG): Ditto.
5911         (LAST_MASK_REG): Ditto.
5912         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5913         (MAYBE_MASK_CLASS_P): New.
5914         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5915         (REG_CLASS_CONTENTS): Ditto.
5916         (MASK_REGNO_P): New.
5917         (ANY_MASK_REG_P): Ditto.
5918         (HI_REGISTER_NAMES): Add new mask registers.
5919         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5920         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5921         mask registers.
5922         (attribute "type"): Add mskmov, msklog.
5923         (attribute "length_immediate"): Support them.
5924         (attribute "memory"): Ditto.
5925         (attribute "prefix_0f"): Ditto.
5926         (*movhi_internal): Support new mask registers.
5927         (*movqi_internal): Ditto.
5928         (define_split): Split out clobber pattern is a logic
5929         insn on mask registers.
5930         (*k<logic><mode>): New.
5931         (*andhi_1): Extend to support mask regs.
5932         (*andqi_1): Extend to support mask regs.
5933         (kandn<mode>): New.
5934         (define_split): Split and-not to and and not if operands
5935         are not mask regs.
5936         (*<code><mode>_1): Separate HI mode to new pattern...
5937         (*<code>hi_1): This.
5938         (*<code>qi_1): Extend to support mask regs.
5939         (kxnor<mode>): New.
5940         (kortestzhi): Ditto.
5941         (kortestchi): Ditto.
5942         (kunpckhi): Ditto.
5943         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5944         (*one_cmplhi2_1): ...Here, now with mask registers support.
5945         (*one_cmplqi2_1): Support new mask registers.
5946         (HI/QImode arithmetics splitter): Don't split if mask registers
5947         are used.
5948         (HI/QImode not splitter): Ditto.
5949         * config/i386/predicated.md (mask_reg_operand): New.
5950         (general_reg_operand): Ditto.
5952 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5954         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5955         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5957 2013-09-10  Jeff Law  <law@redhat.com>
5959         PR tree-optimization/58380
5960         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5961         of threading through a buried loop header.
5963         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5964         return value for single successor case.
5966 2013-09-10  Jan Hubicka  <jh@suse.cz>
5968         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5970 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5972         PR target/58361
5973         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5974         support conditional execution.
5975         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5977 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5979         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5980         constraint pass.
5981         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5982         Check destination too to check move insn.
5983         (undo_optional_reloads): Add check that the original peudo did not
5984         changed its allocation and the optional reload was inherited on last
5985         inheritance pass.  Break loop after deciding to keep optional reload.
5986         (lra_undo_inheritance): Add check that inherited pseudo still in
5987         memory.
5989 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5991         * config/aarch64/aarch64.md (generic_sched): New.
5992         * config/aarch64/aarch64-generic.md (load): Make conditional
5993         on generic_sched attribute.
5994         (nonload): Likewise.
5996 2013-09-10  Jan Hubicka  <jh@suse.cz>
5998         * lto-cgraph.c: Include ipa-utils.h.
5999         (compute_ltrans_boundary): Also add possible targets into the boundary.
6001 2013-09-10  Jan Hubicka  <jh@suse.cz>
6003         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
6004         VAR_DECL of vtable rather than full expression.
6006 2013-09-10  Jan Hubicka  <jh@suse.cz>
6007             Paolo Carlini  <paolo.carlini@oracle.com>
6009         * cgraphunit.c (analyze_functions): Save input_location, set it
6010         to UNKNOWN_LOCATION and restore it at the end.
6012 2013-09-10  Martin Jambor  <mjambor@suse.cz>
6014         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
6015         represented by a thunk.
6017 2013-09-10  Jeff Law  <law@redhat.com>
6019         PR tree-optimization/58343
6020         * tree-ssa-threadupdate.c (thread_block): Identify and disable
6021         jump threading requests through loop headers buried in the middle
6022         of a jump threading path.
6024         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
6025         in return value/type.
6027 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
6029         PR rtl-optimization/58365
6030         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
6031         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
6032         it differs.
6034 2013-09-10  Richard Biener  <rguenther@suse.de>
6036         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
6037         * tree-data-ref.c (create_rdg_vertices): Collect all data
6038         references, signal failure to the caller, use data-ref API.
6039         (build_rdg): Compute data references only once.  Maintain lifetime
6040         of data references and data dependences from within RDG.
6041         (free_rdg): Free dependence relations.
6042         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
6043         inventing extra dependences.
6044         (distribute_loop): Update for RDG API changes.
6046 2013-09-10  Kai Tietz  <ktietz@redhat.com>
6048         * doc/invoke.texi (fms-extensions): Document changed
6049         behavior for ms-abi targets.
6050         * config/i386/i386.c (ix86_option_override_internal):
6051         Set default value of option -fms-extension for ms-abi targets.
6053 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6055         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
6057 2013-09-10  Alan Modra  <amodra@gmail.com>
6059         PR target/58330
6060         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
6062 2013-09-10  Alan Modra  <amodra@gmail.com>
6064         * config/rs6000/predicates.md (add_cint_operand): New.
6065         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
6066         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
6067         using add_cint_operand.
6068         (largetoc_high_plus_aix): Likewise.
6070 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6072         PR tree-optimization/58364
6073         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6074         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6075         the current range can't be an unconditional true or false.
6077 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6079         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6081 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6083         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6085 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6087         PR c++/43452
6088         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6090 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6092         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6093         NO_REGS for immediate that can't be moved directly into FP_REGS.
6095 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6097         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6098         comparison with negated operand.
6099         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6100         RTL form.
6102 2013-09-09  Richard Biener  <rguenther@suse.de>
6104         PR middle-end/58326
6105         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6106         of a subloop record all its block as affecting loop-closed SSA form.
6108 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6110         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6111         of an rtx/bitpos pair.
6112         (store_fixed_bit_field): Update accordingly.
6114 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6116         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6117         GEN_INT.
6118         * builtins.c (expand_errno_check): Likewise.
6119         * dwarf2cfi.c (init_return_column_size): Likewise.
6120         * except.c (sjlj_mark_call_sites): Likewise.
6121         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6122         * lra-constraints.c (emit_inc): Likewise.
6123         * ree.c (combine_set_extension): Likewise.
6124         * regmove.c (fixup_match_2): Likewise.
6125         * reload1.c (inc_for_reload): Likewise.
6127 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6129         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6130         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6131         Use gen_int_mode with the mode of the associated simplify_* call.
6132         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6133         * expmed.c (expand_shift_1): Likewise.
6134         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6135         * loop-iv.c (iv_number_of_iterations): Likewise.
6136         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6137         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6139 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6141         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6142         of the associated expand_* call.
6143         (asan_emit_stack_protection): Likewise.
6144         * builtins.c (round_trampoline_addr): Likewise.
6145         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6146         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6147         (emit_store_flag): Likewise.
6148         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6149         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6150         Likewise.
6151         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6152         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6153         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6154         * modulo-sched.c (generate_prolog_epilog): Likewise.
6155         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6156         (expand_ctz, expand_ffs, expand_unop): Likewise.
6158 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6160         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6161         of the associated gen_rtx_* call.
6162         * caller-save.c (init_caller_save): Likewise.
6163         * combine.c (find_split_point, make_extraction): Likewise.
6164         (make_compound_operation): Likewise.
6165         * dwarf2out.c (mem_loc_descriptor): Likewise.
6166         * explow.c (plus_constant, probe_stack_range): Likewise.
6167         * expmed.c (expand_mult_const): Likewise.
6168         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6169         * reload1.c (init_reload): Likewise.
6170         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6171         * var-tracking.c (adjust_mems): Likewise.
6172         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6173         rather than gen_rtx_fmt_ee.
6175 2013-09-09  Jan Hubicka  <jh@suse.cz>
6177         PR middle-end/58294
6178         * value-prof.c (gimple_ic): Copy also abnormal edges.
6180 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6182         * asan.c (asan_shadow_cst): Use gen_int_mode.
6184 2013-09-08  Jan Hubicka  <jh@suse.cz>
6186         * ipa-profile.c: Add toplevel comment.
6187         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6188         (contains_hot_call_p): New function.
6189         (ipa_propagate_frequency): Set frequencies based on counts when
6190         profile is read.
6191         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6192         profile; do not tamper with profile after inlining if it is read.
6194 2013-09-08  Jan Hubicka  <jh@suse.cz>
6196         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6197         speculative edges.
6199 2013-09-08  Jan Hubicka  <jh@suse.cz>
6201         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6202         summary generation.
6204 2013-09-08  Jeff Law  <law@redhat.com>
6206         PR bootstrap/58340
6207         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6208         of 'found'.
6210 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6212         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6214 2013-09-08  Jan Hubicka  <jh@suse.cz>
6216         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6218 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6220         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6221         for non-debug insns.
6222         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6223         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6224         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6225         accordingly.
6227 2013-09-08  Jan Hubicka  <jh@suse.cz>
6229         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6230         targets and devirtualize to BUILT_IN_UNREACHABLE.
6231         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6232         * ipa.c (walk_polymorphic_call_targets): New function.
6233         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6234         functions; use the new timevar.
6235         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6236         was removed from the program.
6237         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6238         not consider it in the walk when its vtable is dead.
6239         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6240         record_binfo.
6241         (devirt_variable_node_removal_hook): New function.
6242         (possible_polymorphic_call_targets): Also register
6243         devirt_variable_node_removal_hook.
6244         (ipa_devirt): Do not do non-speculative devirtualization.
6245         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6247 2013-09-08  Jan Hubicka  <jh@suse.cz>
6249         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6250         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6251         varpool_remove_variable_insertion_hook): Declare.
6252         * varpool.c (varpool_node_hook_list): New structure.
6253         (first_varpool_node_removal_hook,
6254         first_varpool_variable_insertion_hook): New variables.
6255         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6256         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6257         varpool_remove_variable_insertion_hook,
6258         varpool_call_variable_insertion_hooks): New functions.
6259         (varpool_remove_node): Use it.
6261 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6263         PR c++/54941
6264         * diagnostic.c (diagnostic_build_prefix): When s.file is
6265         "<built-in>" don't output line and column numbers.
6267 2013-09-06  Jan Hubicka  <jh@suse.cz>
6269         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6270         * lto-streamer-out.c: Stream thunks, too.
6271         * lto-streamer-in.c (input_function): Pop cfun here
6272         (lto_read_body): Instead of here.
6274 2013-09-06  Caroline Tice  <cmtice@google.com>
6276         * doc/install.texi: Add documentation for the --enable-vtable-verify
6277         and the --disable-libvtv configure options.
6279 2013-09-06  Jeff Law  <law@redhat.com>
6281         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6282         edge implied equivalences into successor phis.
6284 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6286         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6288 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6290         * resource.c (mark_target_live_regs): Compute resources taking
6291         into account if a call is predicated or not.
6293 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6295         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6296         the compiler in the function names.
6298 2013-09-06  Jan Hubicka  <jh@suse.cz>
6300         PR middle-end/58094
6301         * ipa-inline.c (has_caller_p): New function.
6302         (want_inline_function_to_all_callers_p): Use it.
6303         (sum_callers, inline_to_all_callers): Break out from ...
6304         (ipa_inline): ... here.
6306 2013-09-06  Jan Hubicka  <jh@suse.cz>
6308         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6309         only when AVX is enabled.
6311 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6313         * config/aarch64/aarch64.md
6314         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6315         is fpsimd_<load/store>2.
6316         (load_pair<mode>): Likewise.
6317         (store_pair<mode>): Likewise.
6319 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6321         * config/arm/types.md (type): Add "mrs" type.
6322         * config/aarch64/aarch64.md
6323         (aarch64_load_tp_hard): Make type "mrs".
6324         * config/arm/arm.md
6325         (load_tp_hard): Make type "mrs".
6326         * config/arm/cortex-a15.md: Update with new attributes.
6327         * config/arm/cortex-a5.md: Update with new attributes.
6328         * config/arm/cortex-a53.md: Update with new attributes.
6329         * config/arm/cortex-a7.md: Update with new attributes.
6330         * config/arm/cortex-a8.md: Update with new attributes.
6331         * config/arm/cortex-a9.md: Update with new attributes.
6332         * config/arm/cortex-m4.md: Update with new attributes.
6333         * config/arm/cortex-r4.md: Update with new attributes.
6334         * config/arm/fa526.md: Update with new attributes.
6335         * config/arm/fa606te.md: Update with new attributes.
6336         * config/arm/fa626te.md: Update with new attributes.
6337         * config/arm/fa726te.md: Update with new attributes.
6339 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6341         * config/aarch64/aarch64.md
6342         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6343         (*movtf_aarch64): Likewise.
6344         * config/arm/arm.md
6345         (thumb1_movdi_insn): Use "multiple" for type where more than one
6346         instruction is used for a move.
6347         (*arm32_movhf): Likewise.
6348         (*thumb_movdf_insn): Likewise.
6350 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6352         * config/arm/types.md (type): Rename fcpys to fmov.
6353         * config/arm/vfp.md
6354         (*arm_movsi_vfp): Rename type fcpys as fmov.
6355         (*thumb2_movsi_vfp): Likewise
6356         (*movhf_vfp_neon): Likewise
6357         (*movhf_vfp): Likewise
6358         (*movsf_vfp): Likewise
6359         (*thumb2_movsf_vfp): Likewise
6360         (*movsfcc_vfp): Likewise
6361         (*thumb2_movsfcc_vfp): Likewise
6362         * config/aarch64/aarch64-simd.md
6363         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6364         * config/aarch64/aarch64.md
6365         (*movsi_aarch64): Replace type mov_reg with fmovs.
6366         (*movdi_aarch64): Likewise
6367         (*movsf_aarch64): Likewise
6368         (*movdf_aarch64): Likewise
6369         * config/arm/arm.c
6370         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6371         * config/arm/iwmmxt.md
6372         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6373         * config/arm/arm1020e.md: Update with new attributes.
6374         * config/arm/cortex-a15-neon.md: Update with new attributes.
6375         * config/arm/cortex-a5.md: Update with new attributes.
6376         * config/arm/cortex-a53.md: Update with new attributes.
6377         * config/arm/cortex-a7.md: Update with new attributes.
6378         * config/arm/cortex-a8-neon.md: Update with new attributes.
6379         * config/arm/cortex-a9.md: Update with new attributes.
6380         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6381         * config/arm/cortex-r4f.md: Update with new attributes.
6382         * config/arm/marvell-pj4.md: Update with new attributes.
6383         * config/arm/vfp11.md: Update with new attributes.
6385 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6387         * config/aarch64/aarch64.md
6388         (*madd<mode>): Fix type attribute.
6389         (*maddsi_uxtw): Likewise.
6390         (*msub<mode>): Likewise.
6391         (*msubsi_uxtw): Likewise.
6392         (<su_optab>maddsidi4): Likewise.
6393         (<su_optab>msubsidi4): Likewise.
6395 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6397         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6398         * config/arm/arm.md (core_cycles): Remove fdiv.
6399         * config/arm/vfp.md:
6400         (*sqrtsf2_vfp): Update for attribute changes.
6401         (*sqrtdf2_vfp): Likewise.
6402         * config/aarch64/aarch64.md:
6403         (sqrt<mode>2): Update for attribute changes.
6404         * config/arm/arm1020e.md: Update with new attributes.
6405         * config/arm/cortex-a15-neon.md: Update with new attributes.
6406         * config/arm/cortex-a5.md: Update with new attributes.
6407         * config/arm/cortex-a53.md: Update with new attributes.
6408         * config/arm/cortex-a7.md: Update with new attributes.
6409         * config/arm/cortex-a8-neon.md: Update with new attributes.
6410         * config/arm/cortex-a9.md: Update with new attributes.
6411         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6412         * config/arm/cortex-r4f.md: Update with new attributes.
6413         * config/arm/marvell-pj4.md: Update with new attributes.
6414         * config/arm/vfp11.md: Update with new attributes.
6416 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6418         * config/arm/types.md
6419         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6420         * config/aarch64/aarch64.md
6421         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6422         new attributes.
6423         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6424         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6425         (float<GPI:mode><GPF:mode>2): Likewise.
6426         * config/arm/vfp.md
6427         (*truncsisf2_vfp): Update with new attributes.
6428         (*truncsidf2_vfp): Likewise.
6429         (fixuns_truncsfsi2): Likewise.
6430         (fixuns_truncdfsi2): Likewise.
6431         (*floatsisf2_vfp): Likewise.
6432         (*floatsidf2_vfp): Likewise.
6433         (floatunssisf2): Likewise.
6434         (floatunssidf2): Likewise.
6435         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6436         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6437         * config/arm/arm1020e.md: Update with new attributes.
6438         * config/arm/cortex-a15-neon.md: Update with new attributes.
6439         * config/arm/cortex-a5.md: Update with new attributes.
6440         * config/arm/cortex-a53.md: Update with new attributes.
6441         * config/arm/cortex-a7.md: Update with new attributes.
6442         * config/arm/cortex-a8-neon.md: Update with new attributes.
6443         * config/arm/cortex-a9.md: Update with new attributes.
6444         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6445         * config/arm/cortex-r4f.md: Update with new attributes.
6446         * config/arm/marvell-pj4.md: Update with new attributes.
6447         * config/arm/vfp11.md: Update with new attributes.
6449 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6451         * config/aarch64/arm_neon.h
6452         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6453         (vqtbx<1,2,3,4><q>_s8): Likewise.
6455 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6457         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6458         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6459         (add<mode>3): Add type attribute.
6460         (add<mode>3): Likewise.
6461         (usadd<mode>3): Likewise.
6462         (ssadd<mode>3): Likewise.
6463         (sub<mode>3): Likewise.
6464         (sub<mode>3): Likewise.
6465         (ussub<mode>3): Likewise.
6466         (sssub<mode>3): Likewise.
6467         (ssmulsa3): Likewise.
6468         (usmulusa3): Likewise.
6469         (arm_usatsihi): Likewise.
6470         * config/arm/vfp.md
6471         (*movdi_vfp): Add types for all instructions.
6472         (*movdi_vfp_cortexa8): Likewise.
6473         (*movhf_vfp_neon): Likewise.
6474         (*movhf_vfp): Likewise.
6475         (*movdf_vfp): Likewise.
6476         (*thumb2_movdf_vfp): Likewise.
6477         (*thumb2_movdfcc_vfp): Likewise.
6478         * config/arm/arm.md: Add type attribute to all insn patterns.
6479         (*thumb1_adddi3): Add type attribute.
6480         (*arm_adddi3): Likewise.
6481         (*adddi_sesidi_di): Likewise.
6482         (*adddi_zesidi_di): Likewise.
6483         (*thumb1_addsi3): Likewise.
6484         (addsi3_compare0): Likewise.
6485         (*addsi3_compare0_scratch): Likewise.
6486         (*compare_negsi_si): Likewise.
6487         (cmpsi2_addneg): Likewise.
6488         (*addsi3_carryin_<optab>): Likewise.
6489         (*addsi3_carryin_alt2_<optab>): Likewise.
6490         (*addsi3_carryin_clobercc_<optab>): Likewise.
6491         (*subsi3_carryin): Likewise.
6492         (*subsi3_carryin_const): Likewise.
6493         (*subsi3_carryin_compare): Likewise.
6494         (*subsi3_carryin_compare_const): Likewise.
6495         (*arm_subdi3): Likewise.
6496         (*thumb_subdi3): Likewise.
6497         (*subdi_di_zesidi): Likewise.
6498         (*subdi_di_sesidi): Likewise.
6499         (*subdi_zesidi_di): Likewise.
6500         (*subdi_sesidi_di): Likewise.
6501         (*subdi_zesidi_ze): Likewise.
6502         (thumb1_subsi3_insn): Likewise.
6503         (*arm_subsi3_insn): Likewise.
6504         (*anddi3_insn): Likewise.
6505         (*anddi_zesidi_di): Likewise.
6506         (*anddi_sesdi_di): Likewise.
6507         (*ne_zeroextracts): Likewise.
6508         (*ne_zeroextracts): Likewise.
6509         (*ite_ne_zeroextr): Likewise.
6510         (*ite_ne_zeroextr): Likewise.
6511         (*anddi_notdi_di): Likewise.
6512         (*anddi_notzesidi): Likewise.
6513         (*anddi_notsesidi): Likewise.
6514         (andsi_notsi_si): Likewise.
6515         (thumb1_bicsi3): Likewise.
6516         (*iordi3_insn): Likewise.
6517         (*iordi_zesidi_di): Likewise.
6518         (*iordi_sesidi_di): Likewise.
6519         (*thumb1_iorsi3_insn): Likewise.
6520         (*xordi3_insn): Likewise.
6521         (*xordi_zesidi_di): Likewise.
6522         (*xordi_sesidi_di): Likewise.
6523         (*arm_xorsi3): Likewise.
6524         (*andsi_iorsi3_no): Likewise.
6525         (*smax_0): Likewise.
6526         (*smax_m1): Likewise.
6527         (*arm_smax_insn): Likewise.
6528         (*smin_0): Likewise.
6529         (*arm_smin_insn): Likewise.
6530         (*arm_umaxsi3): Likewise.
6531         (*arm_uminsi3): Likewise.
6532         (*minmax_arithsi): Likewise.
6533         (*minmax_arithsi_): Likewise.
6534         (*satsi_<SAT:code>): Likewise.
6535         (arm_ashldi3_1bit): Likewise.
6536         (arm_ashrdi3_1bit): Likewise.
6537         (arm_lshrdi3_1bit): Likewise.
6538         (*arm_negdi2): Likewise.
6539         (*thumb1_negdi2): Likewise.
6540         (*arm_negsi2): Likewise.
6541         (*thumb1_negsi2): Likewise.
6542         (*negdi_extendsid): Likewise.
6543         (*negdi_zero_extend): Likewise.
6544         (*arm_abssi2): Likewise.
6545         (*thumb1_abssi2): Likewise.
6546         (*arm_neg_abssi2): Likewise.
6547         (*thumb1_neg_abss): Likewise.
6548         (one_cmpldi2): Likewise.
6549         (extend<mode>di2): Likewise.
6550         (*compareqi_eq0): Likewise.
6551         (*arm_extendhisi2addsi): Likewise.
6552         (*arm_movdi): Likewise.
6553         (*thumb1_movdi_insn): Likewise.
6554         (*arm_movt): Likewise.
6555         (*thumb1_movsi_insn): Likewise.
6556         (pic_add_dot_plus_four): Likewise.
6557         (pic_add_dot_plus_eight): Likewise.
6558         (tls_load_dot_plus_eight): Likewise.
6559         (*thumb1_movhi_insn): Likewise.
6560         (*thumb1_movsf_insn): Likewise.
6561         (*movdf_soft_insn): Likewise.
6562         (*thumb_movdf_insn): Likewise.
6563         (cbranchsi4_insn): Likewise.
6564         (cbranchsi4_scratch): Likewise.
6565         (*negated_cbranchsi4): Likewise.
6566         (*tbit_cbranch): Likewise.
6567         (*tlobits_cbranch): Likewise.
6568         (*tstsi3_cbranch): Likewise.
6569         (*cbranchne_decr1): Likewise.
6570         (*addsi3_cbranch): Likewise.
6571         (*addsi3_cbranch_scratch): Likewise.
6572         (*arm_cmpdi_insn): Likewise.
6573         (*arm_cmpdi_unsig): Likewise.
6574         (*arm_cmpdi_zero): Likewise.
6575         (*thumb_cmpdi_zero): Likewise.
6576         (*deleted_compare): Likewise.
6577         (*mov_scc): Likewise.
6578         (*mov_negscc): Likewise.
6579         (*mov_notscc): Likewise.
6580         (*cstoresi_eq0_thumb1_insn): Likewise.
6581         (cstoresi_nltu_thumb1): Likewise.
6582         (cstoresi_ltu_thu): Likewise.
6583         (thumb1_addsi3_addgeu): Likewise.
6584         (*arm_jump): Likewise.
6585         (*thumb_jump): Likewise.
6586         (*check_arch2): Likewise.
6587         (arm_casesi_internal): Likewise.
6588         (thumb1_casesi_dispatch): Likewise.
6589         (*arm_indirect_jump): Likewise.
6590         (*thumb1_indirect_jump): Likewise.
6591         (nop): Likewise.
6592         (*and_scc): Likewise.
6593         (*ior_scc): Likewise.
6594         (*compare_scc): Likewise.
6595         (*cond_move): Likewise.
6596         (*cond_arith): Likewise.
6597         (*cond_sub): Likewise.
6598         (*cmp_ite0): Likewise.
6599         (*cmp_ite1): Likewise.
6600         (*cmp_and): Likewise.
6601         (*cmp_ior): Likewise.
6602         (*ior_scc_scc): Likewise.
6603         (*ior_scc_scc_cmp): Likewise.
6604         (*and_scc_scc): Likewise.
6605         (*and_scc_scc_cmp): Likewise.
6606         (*and_scc_scc_nod): Likewise.
6607         (*negscc): Likewise.
6608         (movcond_addsi): Likewise.
6609         (movcond): Likewise.
6610         (*ifcompare_plus_move): Likewise.
6611         (*if_plus_move): Likewise.
6612         (*ifcompare_move_plus): Likewise.
6613         (*if_move_plus): Likewise.
6614         (*ifcompare_arith_arith): Likewise.
6615         (*if_arith_arith): Likewise.
6616         (*ifcompare_arith_move): Likewise.
6617         (*if_arith_move): Likewise.
6618         (*ifcompare_move_arith): Likewise.
6619         (*if_move_arith): Likewise.
6620         (*ifcompare_move_not): Likewise.
6621         (*if_move_not): Likewise.
6622         (*ifcompare_not_move): Likewise.
6623         (*if_not_move): Likewise.
6624         (*ifcompare_shift_move): Likewise.
6625         (*if_shift_move): Likewise.
6626         (*ifcompare_move_shift): Likewise.
6627         (*if_move_shift): Likewise.
6628         (*ifcompare_shift_shift): Likewise.
6629         (*ifcompare_not_arith): Likewise.
6630         (*ifcompare_arith_not): Likewise.
6631         (*if_arith_not): Likewise.
6632         (*ifcompare_neg_move): Likewise.
6633         (*if_neg_move): Likewise.
6634         (*ifcompare_move_neg): Likewise.
6635         (*if_move_neg): Likewise.
6636         (prologue_thumb1_interwork): Likewise.
6637         (*cond_move_not): Likewise.
6638         (*sign_extract_onebit): Likewise.
6639         (*not_signextract_onebit): Likewise.
6640         (stack_tie): Likewise.
6641         (align_4): Likewise.
6642         (align_8): Likewise.
6643         (consttable_end): Likewise.
6644         (consttable_1): Likewise.
6645         (consttable_2): Likewise.
6646         (consttable_4): Likewise.
6647         (consttable_8): Likewise.
6648         (consttable_16): Likewise.
6649         (*thumb1_tablejump): Likewise.
6650         (prefetch): Likewise.
6651         (force_register_use): Likewise.
6652         (thumb_eh_return): Likewise.
6653         (load_tp_hard): Likewise.
6654         (load_tp_soft): Likewise.
6655         (tlscall): Likewise.
6656         (*arm_movtas_ze): Likewise.
6657         (*arm_rev): Likewise.
6658         (*arm_revsh): Likewise.
6659         (*arm_rev16): Likewise.
6660         * config/arm/thumb2.md
6661         (*thumb2_smaxsi3): Likewise.
6662         (*thumb2_sminsi3): Likewise.
6663         (*thumb32_umaxsi3): Likewise.
6664         (*thumb2_uminsi3): Likewise.
6665         (*thumb2_negdi2): Likewise.
6666         (*thumb2_abssi2): Likewise.
6667         (*thumb2_neg_abss): Likewise.
6668         (*thumb2_movsi_insn): Likewise.
6669         (tls_load_dot_plus_four): Likewise.
6670         (*thumb2_movhi_insn): Likewise.
6671         (*thumb2_mov_scc): Likewise.
6672         (*thumb2_mov_negs): Likewise.
6673         (*thumb2_mov_negs): Likewise.
6674         (*thumb2_mov_nots): Likewise.
6675         (*thumb2_mov_nots): Likewise.
6676         (*thumb2_movsicc_): Likewise.
6677         (*thumb2_movsfcc_soft_insn): Likewise.
6678         (*thumb2_indirect_jump): Likewise.
6679         (*thumb2_and_scc): Likewise.
6680         (*thumb2_ior_scc): Likewise.
6681         (*thumb2_ior_scc_strict_it): Likewise.
6682         (*thumb2_cond_move): Likewise.
6683         (*thumb2_cond_arith): Likewise.
6684         (*thumb2_cond_ari): Likewise.
6685         (*thumb2_cond_sub): Likewise.
6686         (*thumb2_negscc): Likewise.
6687         (*thumb2_movcond): Likewise.
6688         (thumb2_casesi_internal): Likewise.
6689         (thumb2_casesi_internal_pic): Likewise.
6690         (*thumb2_alusi3_short): Likewise.
6691         (*thumb2_mov<mode>_shortim): Likewise.
6692         (*thumb2_addsi_short): Likewise.
6693         (*thumb2_subsi_short): Likewise.
6694         (thumb2_addsi3_compare0): Likewise.
6695         (*thumb2_cbz): Likewise.
6696         (*thumb2_cbnz): Likewise.
6697         (*thumb2_one_cmplsi2_short): Likewise.
6698         (*thumb2_negsi2_short): Likewise.
6699         (*orsi_notsi_si): Likewise.
6700         * config/arm/arm1020e.md: Update with new attributes.
6701         * config/arm/arm1026ejs.md: Update with new attributes.
6702         * config/arm/arm1136jfs.md: Update with new attributes.
6703         * config/arm/arm926ejs.md: Update with new attributes.
6704         * config/arm/cortex-a15.md: Update with new attributes.
6705         * config/arm/cortex-a5.md: Update with new attributes.
6706         * config/arm/cortex-a53.md: Update with new attributes.
6707         * config/arm/cortex-a7.md: Update with new attributes.
6708         * config/arm/cortex-a8.md: Update with new attributes.
6709         * config/arm/cortex-a9.md: Update with new attributes.
6710         * config/arm/cortex-m4.md: Update with new attributes.
6711         * config/arm/cortex-r4.md: Update with new attributes.
6712         * config/arm/fa526.md: Update with new attributes.
6713         * config/arm/fa606te.md: Update with new attributes.
6714         * config/arm/fa626te.md: Update with new attributes.
6715         * config/arm/fa726te.md: Update with new attributes.
6717 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6719         * config/aarch64/aarch64-simd.md
6720         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6721         <vwx> iterator to ensure correct register choice.
6722         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6723         (aarch64_sqdmull_n<mode>): Likewise.
6724         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6725         * config/aarch64/arm_neon.h
6726         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6727         (vml<as><q>_n_<su>16): Likewise.
6728         (vml<as>l_high_lane<q>_<su>16): Likewise.
6729         (vml<as>l_high_n_<su>16): Likewise.
6730         (vml<as>l_lane<q>_<su>16): Likewise.
6731         (vml<as>l_n_<su>16): Likewise.
6732         (vmul<q>_lane<q>_<su>16): Likewise.
6733         (vmul<q>_n_<su>16): Likewise.
6734         (vmull_lane<q>_<su>16): Likewise.
6735         (vmull_n_<su>16): Likewise.
6736         (vmull_high_lane<q>_<su>16): Likewise.
6737         (vmull_high_n_<su>16): Likewise.
6738         (vqrdmulh<q>_n_s16): Likewise.
6740 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6742         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6743         have the correct lane parameter.
6745 2013-09-06  Richard Biener <rguenther@suse.de>
6747         * cfganal.c (control_dependences::~control_dependences):
6748         Properly free all of the vector.
6750 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6752         PR target/58269
6753         * config/i386/i386.c (ix86_conditional_register_usage):
6754         Proper initialize extended SSE registers.
6756 2013-09-06  Jan Hubicka  <jh@suse.cz>
6758         PR tree-optimization/58311
6759         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6761 2013-09-06  Jan Hubicka  <jh@suse.cz>
6763         * Makefile.in (tree-sra.o): Update dependencies.
6764         * tree-sra.c: Include ipa-utils.h
6765         (scan_function): Use recursive_call_p.
6766         (has_caller_p): New function.
6767         (cgraph_for_node_and_aliases): Count also callers of aliases.
6769 2013-09-06  Jan Hubicka  <jh@suse.cz>
6771         PR middle-end/58094
6772         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6773         * tree-tailcall.c: Include ipa-utils.h.
6774         (find_tail_calls): Use it.
6775         * ipa-pure-const.c (check_call): Likewise.
6776         * ipa-utils.c (recursive_call_p): New function.
6777         * ipa-utils.h (recursive_call_p): Dclare.
6778         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6779         (symtab_semantically_equivalent_p): New function.
6780         * Makefile.in (tree-tailcall.o): Update dependencies.
6782 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6784         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6785         non-inlinable part.
6787 2013-09-06  Richard Biener  <rguenther@suse.de>
6789         * lto-streamer.h (lto_global_var_decls): Remove.
6790         * Makefile.in (OBJS): Remove lto-symtab.o.
6791         (lto-symtab.o): Remove.
6792         (GTFILES): Remove lto-symtab.c
6793         * lto-symtab.c: Move to lto/
6795 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6797         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6798         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6799         (UNSPEC_FPINT_RINT): New constant definitions.
6800         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6801         definition with 2 attributes.
6802         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6803         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6804         definitions.
6806 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6808         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6809         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6810         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6811         attribute to "z196_alone".
6812         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6813         "zEC12_simple".
6815 2013-09-06  Richard Biener  <rguenther@suse.de>
6817         * basic-block.h (class control_dependences): New.
6818         * tree-ssa-dce.c (control_dependence_map): Remove.
6819         (cd): New global.
6820         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6821         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6822         find_pdom, find_control_dependence, find_all_control_dependences):
6823         Move to cfganal.c.
6824         (mark_control_dependent_edges_necessary,
6825         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6826         tree_dce_done, perform_tree_ssa_dce): Adjust.
6827         * cfganal.c (set_control_dependence_map_bit,
6828         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6829         find_all_control_dependences): Move from tree-ssa-dce.c and
6830         implement as methods of control_dependences class.
6831         (control_dependences::control_dependences): New.
6832         (control_dependences::~control_dependences): Likewise.
6833         (control_dependences::get_edges_dependent_on): Likewise.
6834         (control_dependences::get_edge): Likewise.
6836 2013-09-04  Jan Hubicka  <jh@suse.cz>
6838         * tree.c (types_same_for_odr): Drop overactive check.
6839         * ipa-devirt.c (hash_type_name): Likewise.
6841 2013-09-04  Jan Hubicka  <jh@suse.cz>
6843         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6844         (analyze_functions): ... here.
6846 2013-09-04  Jan Hubicka  <jh@suse.cz>
6848         PR middle-end/58201
6849         * cgraphunit.c (analyze_functions): Clear AUX fields
6850         after processing; initialize assembler name has.
6852 2013-09-05  Jeff Law  <law@redhat.com>
6854         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6855         from thread_around_empty_block.  Record threading path into PATH.
6856         Recurse if threading through the initial block is successful.
6857         (thread_across_edge): Corresponding changes to slightly simplify.
6859 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6861         * config/aarch64/aarch64.md
6862         (type): Remove frecpe, frecps, frecpx.
6863         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6864         fix to be a TARGET_SIMD instruction.
6865         (aarch64_frecps): Remove.
6866         * config/aarch64/aarch64-simd.md
6867         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6868         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6870 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6871             Sofiane Naci  <sofiane.naci@arm.com>
6873         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6874         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6875         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6876         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6877         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6878         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6879         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6880         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6881         into "clz, "rbit".  Rename "shift" to "shift_imm".
6882         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6883         changes.  Update for attribute changes all occurrences of arlo_* and
6884         shift* types.
6885         * config/arm/arm-fixed.md: Update for attribute changes
6886         all occurrences of arlo_* types.
6887         * config/arm/thumb2.md: Update for attribute changes all occurrences
6888         of arlo_* types.
6889         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6890         (cortexa7_older_only): Likewise.
6891         (cortexa7_younger):  Likewise.
6892         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6893         (1020alu_shift_op): Likewise.
6894         (1020alu_shift_reg_op): Likewise.
6895         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6896         (alu_shift_op): Likewise.
6897         (alu_shift_reg_op): Likewise.
6898         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6899         (11_alu_shift_op): Likewise.
6900         (11_alu_shift_reg_op): Likewise.
6901         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6902         (9_alu_shift_reg_op): Likewise.
6903         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6904         attribute changes.
6905         (cortex_a15_alu_shift): Likewise.
6906         (cortex_a15_alu_shift_reg): Likewise.
6907         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6908         attribute changes.
6909         (cortex_a5_alu_shift): Likewise.
6910         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6911         attribute changes.
6912         (cortex_a53_alu_shift): Likewise.
6913         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6914         attribute changes.
6915         (cortex_a7_alu_reg): Likewise.
6916         (cortex_a7_alu_shift): Likewise.
6917         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6918         attribute changes.
6919         (cortex_a8_alu_shift): Likewise.
6920         (cortex_a8_alu_shift_reg): Likewise.
6921         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6922         (cortex_a9_dp_shift): Likewise.
6923         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6924         attribute changes.
6925         * config/arm/cortex-r4.md
6926         (cortex_r4_alu): Update for attribute changes.
6927         (cortex_r4_mov): Likewise.
6928         (cortex_r4_alu_shift_reg): Likewise.
6929         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6930         (526_alu_shift_op): Likewise.
6931         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6932         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6933         (626te_alu_shift_op): Likewise.
6934         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6935         (726te_alu_shift_op): Likewise.
6936         (726te_alu_shift_reg_op): Likewise.
6937         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6938         (mp626_alu_shift_op): Likewise.
6939         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6940         (pj4_alu_conds): Likewise.
6941         (pj4_shift): Likewise.
6942         (pj4_shift_conds): Likewise.
6943         (pj4_alu_shift): Likewise.
6944         (pj4_alu_shift_conds): Likewise.
6945         * config/aarch64/aarch64.md: Update for attribute change
6946         all occurrences of arlo_* and shift* types.
6948 2013-09-05  Mike Stump  <mikestump@comcast.net>
6950         * tree.h: Move documentation for tree_function_decl to tree-core.h
6951         with the declaration.
6953 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6955         PR target/58139
6956         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6957         looking for widest mode.
6959 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6961         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6963 2013-09-05  Richard Biener  <rguenther@suse.de>
6965         PR tree-optimization/58137
6966         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6967         Do not create vectors of pointers.
6968         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6969         types for the components of the vector initializer.
6970         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6971         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6973 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6975         * ipa-prop.c (remove_described_reference): Accept missing references,
6976         return false if that hppens, otherwise return true.
6977         (cgraph_node_for_jfunc): New function.
6978         (try_decrement_rdesc_refcount): Likewise.
6979         (try_make_edge_direct_simple_call): Use them.
6980         (ipa_edge_removal_hook): Remove references from rdescs.
6981         (ipa_edge_duplication_hook): Clone rdescs and their references
6982         when the new edge has the same caller as the old one.
6983         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6984         reference before removing any edges.
6986 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6988         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6989         initial store.
6990         * thumb2.md (thumb2_storewb_parisi): New pattern.
6992 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6994         * config/aarch64/aarch64-option-extensions.def: Add
6995         AARCH64_OPT_EXTENSION of 'crc'.
6996         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6997         (AARCH64_ISA_CRC): Ditto.
6998         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6999         description of the CRC extension.
7001 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7003         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
7004         * config/rs6000/linux.h: Ditto.
7005         * alpha/linux.h: Ditto.
7006         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
7007         no_c99_libc_has_function.
7008         * config/c6x/uclinux-elf.h: Ditto.
7009         * config/lm32/uclinux-elf.h: Ditto.
7010         * config/m68k/uclinux.h: Ditto.
7011         * config/moxie/uclinux.h: Ditto.
7012         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
7013         (crisv32-*-linux*, cris-*-linux*): Ditto.
7014         * config/bfin/bfin.c: Include "tm_p.h".
7016 2013-09-05  Richard Biener  <rguenther@suse.de>
7018         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
7019         check for a definition without a basic-block.
7021 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7022             Sofiane Naci <sofiane.naci@arm.com>
7024         * config/aarch64/aarch64.md
7025         (*movti_aarch64): Rename r_2_f and f_2_r.
7026         (*movsf_aarch64): Likewise.
7027         (*movdf_aarch64): Likewise.
7028         (*movtf_aarch64): Likewise.
7029         (aarch64_movdi_<mode>low): Likewise.
7030         (aarch64_movdi_<mode>high): Likewise.
7031         (aarch64_mov<mode>high_di): Likewise.
7032         (aarch64_mov<mode>low_di): Likewise.
7033         (aarch64_movtilow_tilow): Likewise.
7034         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
7035         values to config/arm/types.md
7036         (attribute "conds"): Update for attribute change.
7037         (anddi3_insn): Likewise.
7038         (iordi3_insn): Likewise.
7039         (xordi3_insn): Likewise.
7040         (one_cmpldi2): Likewise.
7041         * config/arm/types.md (type): Add Neon types.
7042         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
7043         use "type" attribute.
7044         (movmisalign<mode>_neon_store): Likewise.
7045         (movmisalign<mode>_neon_load): Likewise.
7046         (vec_set<mode>_internal): Likewise.
7047         (vec_setv2di_internal): Likewise.
7048         (vec_extract<mode>): Likewise.
7049         (vec_extractv2di): Likewise.
7050         (add<mode>3_neon): Likewise.
7051         (adddi3_neon): Likewise.
7052         (sub<mode>3_neon): Likewise.
7053         (subdi3_neon): Likewise.
7054         (mul<mode>3_neon): Likewise.
7055         (mul<mode>3add<mode>_neon): Likewise.
7056         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
7057         (fma<VCVTF:mode>4)): Likewise.
7058         (fma<VCVTF:mode>4_intrinsic): Likewise.
7059         (fmsub<VCVTF:mode>4)): Likewise.
7060         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
7061         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
7062         (ior<mode>3): Likewise.
7063         (and<mode>3): Likewise.
7064         (anddi3_neon): Likewise.
7065         (orn<mode>3_neon): Likewise.
7066         (orndi3_neon): Likewise.
7067         (bic<mode>3_neon): Likewise.
7068         (bicdi3_neon): Likewise.
7069         (xor<mode>3): Likewise.
7070         (one_cmpl<mode>2): Likewise.
7071         (abs<mode>2): Likewise.
7072         (neg<mode>2): Likewise.
7073         (umin<mode>3_neon): Likewise.
7074         (umax<mode>3_neon): Likewise.
7075         (smin<mode>3_neon): Likewise.
7076         (smax<mode>3_neon): Likewise.
7077         (vashl<mode>3): Likewise.
7078         (vashr<mode>3_imm): Likewise.
7079         (vlshr<mode>3_imm): Likewise.
7080         (ashl<mode>3_signed): Likewise.
7081         (ashl<mode>3_unsigned): Likewise.
7082         (neon_load_count): Likewise.
7083         (ashldi3_neon_noclobber): Likewise.
7084         (signed_shift_di3_neon): Likewise.
7085         (unsigned_shift_di3_neon): Likewise.
7086         (ashrdi3_neon_imm_noclobber): Likewise.
7087         (lshrdi3_neon_imm_noclobber): Likewise.
7088         (widen_ssum<mode>3): Likewise.
7089         (widen_usum<mode>3): Likewise.
7090         (quad_halves_<code>v4si): Likewise.
7091         (quad_halves_<code>v4sf): Likewise.
7092         (quad_halves_<code>v8hi): Likewise.
7093         (quad_halves_<code>v16qi): Likewise.
7094         (reduc_splus_v2di): Likewise.
7095         (neon_vpadd_internal<mode>): Likewise.
7096         (neon_vpsmin<mode>): Likewise.
7097         (neon_vpsmax<mode>): Likewise.
7098         (neon_vpumin<mode>): Likewise.
7099         (neon_vpumax<mode>): Likewise.
7100         (ss_add<mode>_neon): Likewise.
7101         (us_add<mode>_neon): Likewise.
7102         (ss_sub<mode>_neon): Likewise.
7103         (us_sub<mode>_neon): Likewise.
7104         (neon_vadd<mode>_unspec): Likewise.
7105         (neon_vaddl<mode>): Likewise.
7106         (neon_vaddw<mode>): Likewise.
7107         (neon_vhadd<mode>): Likewise.
7108         (neon_vqadd<mode>): Likewise.
7109         (neon_vaddhn<mode>): Likewise.
7110         (neon_vmul<mode>): Likewise.
7111         (neon_vmla<mode>): Likewise.
7112         (neon_vmlal<mode>): Likewise.
7113         (neon_vmls<mode>): Likewise.
7114         (neon_vmlsl<mode>): Likewise.
7115         (neon_vqdmulh<mode>): Likewise.
7116         (neon_vqdmlal<mode>): Likewise.
7117         (neon_vqdmlsl<mode>): Likewise.
7118         (neon_vmull<mode>): Likewise.
7119         (neon_vqdmull<mode>): Likewise.
7120         (neon_vsub<mode>_unspec): Likewise.
7121         (neon_vsubl<mode>): Likewise.
7122         (neon_vsubw<mode>): Likewise.
7123         (neon_vqsub<mode>): Likewise.
7124         (neon_vhsub<mode>): Likewise.
7125         (neon_vsubhn<mode>): Likewise.
7126         (neon_vceq<mode>): Likewise.
7127         (neon_vcge<mode>): Likewise.
7128         (neon_vcgeu<mode>): Likewise.
7129         (neon_vcgt<mode>): Likewise.
7130         (neon_vcgtu<mode>): Likewise.
7131         (neon_vcle<mode>): Likewise.
7132         (neon_vclt<mode>): Likewise.
7133         (neon_vcage<mode>): Likewise.
7134         (neon_vcagt<mode>): Likewise.
7135         (neon_vtst<mode>): Likewise.
7136         (neon_vabd<mode>): Likewise.
7137         (neon_vabdl<mode>): Likewise.
7138         (neon_vaba<mode>): Likewise.
7139         (neon_vabal<mode>): Likewise.
7140         (neon_vmax<mode>): Likewise.
7141         (neon_vmin<mode>): Likewise.
7142         (neon_vpaddl<mode>): Likewise.
7143         (neon_vpadal<mode>): Likewise.
7144         (neon_vpmax<mode>): Likewise.
7145         (neon_vpmin<mode>): Likewise.
7146         (neon_vrecps<mode>): Likewise.
7147         (neon_vrsqrts<mode>): Likewise.
7148         (neon_vqabs<mode>): Likewise.
7149         (neon_vqneg<mode>): Likewise.
7150         (neon_vcls<mode>): Likewise.
7151         (clz<mode>2): Likewise.
7152         (popcount<mode>2): Likewise.
7153         (neon_vrecpe): Likewise.
7154         (neon_vrsqrte): Likewise.
7155         (neon_vget_lane<mode>_sext_internal): Likewise.
7156         (neon_vget_lane<mode>_zext_internal): Likewise.
7157         (neon_vdup_n<mode>): Likewise.
7158         (neon_vdup_nv2di): Likewise.
7159         (neon_vdpu_lane<mode>_internal): Likewise.
7160         (neon_vswp<mode>): Likewise.
7161         (float<mode><V_cvtto>2): Likewise.
7162         (floatuns<mode><V_cvtto>2): Likewise.
7163         (fix_trunc<mode><V_cvtto>)2): Likewise
7164         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7165         (neon_vcvt<mode>): Likewise.
7166         (neon_vcvtv4sfv4hf): Likewise.
7167         (neon_vcvtv4hfv4sf): Likewise.
7168         (neon_vcvt_n<mode>): Likewise.
7169         (neon_vmovn<mode>): Likewise.
7170         (neon_vqmovn<mode>): Likewise.
7171         (neon_vqmovun<mode>): Likewise.
7172         (neon_vmovl<mode>): Likewise.
7173         (neon_vmul_lane<mode>): Likewise.
7174         (neon_vmull_lane<mode>): Likewise.
7175         (neon_vqdmull_lane<mode>): Likewise.
7176         (neon_vqdmulh_lane<mode>): Likewise.
7177         (neon_vmla_lane<mode>): Likewise.
7178         (neon_vmlal_lane<mode>): Likewise.
7179         (neon_vqdmlal_lane<mode>): Likewise.
7180         (neon_vmls_lane<mode>): Likewise.
7181         (neon_vmlsl_lane<mode>): Likewise.
7182         (neon_vqdmlsl_lane<mode>): Likewise.
7183         (neon_vext<mode>): Likewise.
7184         (neon_vrev64<mode>): Likewise.
7185         (neon_vrev32<mode>): Likewise.
7186         (neon_vrev16<mode>): Likewise.
7187         (neon_vbsl<mode>_internal): Likewise.
7188         (neon_vshl<mode>): Likewise.
7189         (neon_vqshl<mode>): Likewise.
7190         (neon_vshr_n<mode>): Likewise.
7191         (neon_vshrn_n<mode>): Likewise.
7192         (neon_vqshrn_n<mode>): Likewise.
7193         (neon_vqshrun_n<mode>): Likewise.
7194         (neon_vshl_n<mode>): Likewise.
7195         (neon_vqshl_n<mode>): Likewise.
7196         (neon_vqshlu_n<mode>): Likewise.
7197         (neon_vshll_n<mode>): Likewise.
7198         (neon_vsra_n<mode>): Likewise.
7199         (neon_vsri_n<mode>): Likewise.
7200         (neon_vsli_n<mode>): Likewise.
7201         (neon_vtbl1v8qi): Likewise.
7202         (neon_vtbl2v8qi): Likewise.
7203         (neon_vtbl3v8qi): Likewise.
7204         (neon_vtbl4v8qi): Likewise.
7205         (neon_vtbx1v8qi): Likewise.
7206         (neon_vtbx2v8qi): Likewise.
7207         (neon_vtbx3v8qi): Likewise.
7208         (neon_vtbx4v8qi): Likewise.
7209         (neon_vtrn<mode>_internal): Likewise.
7210         (neon_vzip<mode>_internal): Likewise.
7211         (neon_vuzp<mode>_internal): Likewise.
7212         (neon_vld1<mode>): Likewise.
7213         (neon_vld1_lane<mode>): Likewise.
7214         (neon_vld1_dup<mode>): Likewise.
7215         (neon_vld1_dupv2di): Likewise.
7216         (neon_vst1<mode>): Likewise.
7217         (neon_vst1_lane<mode>): Likewise.
7218         (neon_vld2<mode>): Likewise.
7219         (neon_vld2_lane<mode>): Likewise.
7220         (neon_vld2_dup<mode>): Likewise.
7221         (neon_vst2<mode>): Likewise.
7222         (neon_vst2_lane<mode>): Likewise.
7223         (neon_vld3<mode>): Likewise.
7224         (neon_vld3qa<mode>): Likewise.
7225         (neon_vld3qb<mode>): Likewise.
7226         (neon_vld3_lane<mode>): Likewise.
7227         (neon_vld3_dup<mode>): Likewise.
7228         (neon_vst3<mode>): Likewise.
7229         (neon_vst3qa<mode>): Likewise.
7230         (neon_vst3qb<mode>): Likewise.
7231         (neon_vst3_lane<mode>): Likewise.
7232         (neon_vld4<mode>): Likewise.
7233         (neon_vld4qa<mode>): Likewise.
7234         (neon_vld4qb<mode>): Likewise.
7235         (neon_vld4_lane<mode>): Likewise.
7236         (neon_vld4_dup<mode>): Likewise.
7237         (neon_vst4<mode>): Likewise.
7238         (neon_vst4qa<mode>): Likewise.
7239         (neon_vst4qb<mode>): Likewise.
7240         (neon_vst4_lane<mode>): Likewise.
7241         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7242         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7243         (neon_vec_<US>mult_lo_<mode>): Likewise.
7244         (neon_vec_<US>mult_hi_<mode>): Likewise.
7245         (neon_vec_<US>shiftl_<mode>): Likewise.
7246         (neon_unpack<US>_<mode>): Likewise.
7247         (neon_vec_<US>mult_<mode>): Likewise.
7248         (vec_pack_trunc_<mode>): Likewise.
7249         (neon_vec_pack_trunk_<mode>): Likewise.
7250         (neon_vabd<mode>_2): Likewise.
7251         (neon_vabd<mode>_3): Likewise.
7252         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7253         (thumb2_movsi_vfp): Likewise.
7254         (movdi_vfp): Likewise.
7255         (movdi_vfp_cortexa8): Likewise.
7256         (movhf_vfp_neon): Likewise.
7257         (movhf_vfp): Likewiwse.
7258         (movsf_vfp): Likewiwse.
7259         (thumb2_movsf_vfp): Likewiwse.
7260         (movdf_vfp): Likewise.
7261         (thumb2_movdf_vfp): Likewise.
7262         (movsfcc_vfp): Likewise.
7263         (thumb2_movsfcc_vfp): Likewise.
7264         (movdfcc_vfp): Likewise.
7265         (thumb2_movdfcc_vfp): Likewise.
7266         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7267         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7268         (v10_v2c): Likewise.
7269         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7270         attribute change.
7271         (cortex_a15_neon_int_2): Likewise.
7272         (cortex_a15_neon_int_3): Likewise.
7273         (cortex_a15_neon_int_4): Likewise.
7274         (cortex_a15_neon_int_5): Likewise.
7275         (cortex_a15_neon_vqneg_vqabs): Likewise.
7276         (cortex_a15_neon_vmov): Likewise.
7277         (cortex_a15_neon_vaba): Likewise.
7278         (cortex_a15_neon_vaba_qqq): Likewise.
7279         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7280         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7281         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7282         scalar_64_32_long_scalar): Likewise.
7283         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7284         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7285         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7286         lotype_qdd_64_32_long): Likewise.
7287         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7288         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7289         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7290         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7291         (cortex_a15_neon_shift_1): Likewise.
7292         (cortex_a15_neon_shift_2): Likewise.
7293         (cortex_a15_neon_shift_3): Likewise.
7294         (cortex_a15_neon_vshl_ddd): Likewise.
7295         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7296         (cortex_a15_neon_vsra_vrsra): Likewise.
7297         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7298         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7299         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7300         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7301         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7302         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7303         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7304         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7305         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7306         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7307         (cortex_a15_neon_bp_simple): Likewise.
7308         (cortex_a15_neon_bp_2cycle): Likewise.
7309         (cortex_a15_neon_bp_3cycle): Likewise.
7310         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7311         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7312         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7313         (cortex_a15_neon_vld2_4_regs): Likewise.
7314         (cortex_a15_neon_vld3_vld4): Likewise.
7315         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7316         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7317         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7318         (cortex_a15_neon_vst3_vst4): Likewise.
7319         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7320         (cortex_a15_neon_vld3_vld4_lane" 10
7321         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7322         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7323         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7324         (cortex_a15_neon_ldm_2): Likewise.0
7325         (cortex_a15_neon_stm_2): Likewise.
7326         (cortex_a15_neon_mcr): Likewise.
7327         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7328         (cortex_a15_neon_mrc): Likewise.
7329         (cortex_a15_neon_mrrc): Likewise.
7330         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7331         change.
7332         (cortex_a15_alu_shift): Likewise.
7333         (cortex_a15_alu_shift_reg): Likewise.
7334         (cortex_a15_mult32): Likewise.
7335         (cortex_a15_mult64): Likewise.
7336         (cortex_a15_block): Likewise.
7337         (cortex_a15_branch): Likewise.
7338         (cortex_a15_load1): Likewise.
7339         (cortex_a15_load3): Likewise.
7340         (cortex_a15_store1): Likewise.
7341         (cortex_a15_store3): Likewise.
7342         (cortex_a15_call): Likewise.
7343         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7344         (cortex_a5_f2r): Likewise.
7345         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7346         change.
7347         (cortex_a53_f2r): Likewise.
7348         * config/arm/cortex-a7.md
7349         (cortex_a7_branch): Update for attribute change.
7350         (cortex_a7_call): Likewise.
7351         (cortex_a7_alu_imm): Likewise.
7352         (cortex_a7_alu_reg): Likewise.
7353         (cortex_a7_alu_shift): Likewise.
7354         (cortex_a7_mul): Likewise.
7355         (cortex_a7_load1): Likewise.
7356         (cortex_a7_store1): Likewise.
7357         (cortex_a7_load2): Likewise.
7358         (cortex_a7_store2): Likewise.
7359         (cortex_a7_load3): Likewise.
7360         (cortex_a7_store3): Likewise.
7361         (cortex_a7_load4): Likewise.
7362         (cortex_a7_store4): Likewise.
7363         (cortex_a7_fpalu): Likewise.
7364         (cortex_a7_fconst): Likewise.
7365         (cortex_a7_fpmuls): Likewise.
7366         (cortex_a7_neon_mul): Likewise.
7367         (cortex_a7_fpmacs): Likewise.
7368         (cortex_a7_neon_mla: Likewise.
7369         (cortex_a7_fpmuld: Likewise.
7370         (cortex_a7_fpmacd: Likewise.
7371         (cortex_a7_fpfmad: Likewise.
7372         (cortex_a7_fdivs: Likewise.
7373         (cortex_a7_fdivd: Likewise.
7374         (cortex_a7_r2f: Likewise.
7375         (cortex_a7_f2r: Likewise.
7376         (cortex_a7_f_flags: Likewise.
7377         (cortex_a7_f_loads: Likewise.
7378         (cortex_a7_f_loadd: Likewise.
7379         (cortex_a7_f_stores: Likewise.
7380         (cortex_a7_f_stored: Likewise.
7381         (cortex_a7_neon): Likewise.
7382         * config/arm/cortex-a8-neon.md
7383         (cortex_a8_neon_mrc): Update for attribute change.
7384         (cortex_a8_neon_mrrc): Likewise.
7385         (cortex_a8_neon_int_1): Likewise.
7386         (cortex_a8_neon_int_2): Likewise.
7387         (cortex_a8_neon_int_3): Likewise.
7388         (cortex_a8_neon_int_4): Likewise.
7389         (cortex_a8_neon_int_5): Likewise.
7390         (cortex_a8_neon_vqneg_vqabs): Likewise.
7391         (cortex_a8_neon_vmov): Likewise.
7392         (cortex_a8_neon_vaba): Likewise.
7393         (cortex_a8_neon_vaba_qqq): Likewise.
7394         (cortex_a8_neon_vsma): Likewise.
7395         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7396         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7397         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7398         long_scalar): Likewise.
7399         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7400         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7401         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7402         long_scalar_qdd_64_32_long): Likewise.
7403         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7404         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7405         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7406         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7407         (cortex_a8_neon_shift_1): Likewise.
7408         (cortex_a8_neon_shift_2): Likewise.
7409         (cortex_a8_neon_shift_3): Likewise.
7410         (cortex_a8_neon_vshl_ddd): Likewise.
7411         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7412         (cortex_a8_neon_vsra_vrsra): Likewise.
7413         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7414         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7415         (cortex_a8_neon_fp_vsum): Likewise.
7416         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7417         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7418         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7419         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7420         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7421         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7422         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7423         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7424         (cortex_a8_neon_bp_simple): Likewise.
7425         (cortex_a8_neon_bp_2cycle): Likewise.
7426         (cortex_a8_neon_bp_3cycle): Likewise.
7427         (cortex_a8_neon_ldr): Likewise.
7428         (cortex_a8_neon_str): Likewise.
7429         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7430         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7431         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7432         (cortex_a8_neon_vld2_4_regs): Likewise.
7433         (cortex_a8_neon_vld3_vld4): Likewise.
7434         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7435         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7436         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7437         (cortex_a8_neon_vst3_vst4): Likewise.
7438         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7439         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7440         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7441         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7442         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7443         (cortex_a8_neon_mcr): Likewise.
7444         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7445         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7446         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7447         change.
7448         (ca9_neon_mrrc): Likewise.
7449         (cortex_a9_neon_int_1): Likewise.
7450         (cortex_a9_neon_int_2): Likewise.
7451         (cortex_a9_neon_int_3): Likewise.
7452         (cortex_a9_neon_int_4): Likewise.
7453         (cortex_a9_neon_int_5): Likewise.
7454         (cortex_a9_neon_vqneg_vqabs): Likewise.
7455         (cortex_a9_neon_vmov): Likewise.
7456         (cortex_a9_neon_vaba): Likewise.
7457         (cortex_a9_neon_vaba_qqq): Likewise.
7458         (cortex_a9_neon_vsma): Likewise.
7459         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7460         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7461         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7462         long_scalar): Likewise.
7463         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7464         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7465         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7466         long_scalar_qdd_64_32_long): Likewise.
7467         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7468         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7469         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7470         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7471         (cortex_a9_neon_shift_1): Likewise.
7472         (cortex_a9_neon_shift_2): Likewise.
7473         (cortex_a9_neon_shift_3): Likewise.
7474         (cortex_a9_neon_vshl_ddd): Likewise.
7475         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7476         (cortex_a9_neon_vsra_vrsra): Likewise.
7477         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7478         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7479         (cortex_a9_neon_fp_vsum): Likewise.
7480         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7481         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7482         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7483         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7484         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7485         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7486         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7487         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7488         (cortex_a9_neon_bp_simple): Likewise.
7489         (cortex_a9_neon_bp_2cycle): Likewise.
7490         (cortex_a9_neon_bp_3cycle): Likewise.
7491         (cortex_a9_neon_ldr): Likewise.
7492         (cortex_a9_neon_str): Likewise.
7493         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7494         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7495         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7496         (cortex_a9_neon_vld2_4_regs): Likewise.
7497         (cortex_a9_neon_vld3_vld4): Likewise.
7498         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7499         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7500         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7501         (cortex_a9_neon_vst3_vst4): Likewise.
7502         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7503         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7504         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7505         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7506         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7507         (cortex_a9_neon_mcr): Likewise.
7508         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7509         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7510         (cortex_a9_fps): Likewise.
7511         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7512         change.
7513         (cortex_m4_fmuls): Likewise.
7514         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7515         change.
7516         (cortex_r4_mrc): Likewise.
7517         * config/arm/iterators.md: Update comment referring to neon_type.
7518         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7519         (iwmmxt_movsi_insn): Likewise.
7520         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7521         attribute change.
7522         (pj4_core_to_vfp): Likewise.
7523         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7524         attribute change.
7525         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7526         (vfp_fstore): Likewise.
7527         * doc/md.texi: Change references to neon_type to refer to type.
7529 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7531         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7533 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7535         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7536         lglobl if not weak.
7538 2013-09-04  Easwaran Raman  <eraman@google.com>
7540         PR middle-end/57370
7541         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7542         (build_and_add_sum): Use it.
7543         (appears_later_in_bb): Simplify code.
7545 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7547         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7549 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7551         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7553 2013-09-04  Jan Hubicka  <jh@suse.cz>
7555         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7556         * ipa-devirt.c: Include diganostic.h
7557         (odr_type_d): Add types and types_set.
7558         (hash_type_name): Work for types with vtables during LTO.
7559         (odr_hasher::remove): Fix comment; destroy types_set.
7560         (add_type_duplicate): New function,
7561         (get_odr_type): Use it.
7562         (dump_type_inheritance_graph): Dump type duplicates.
7563         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7564         graph.
7565         * tree.c (types_same_for_odr): Give exact answers on types with
7566         virtual tables.
7568 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7570         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7571         explaining their differences.
7573 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7575         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7577 2013-09-03  Jeff Law  <law@redhat.com>
7579         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7580         when not threading through a joiner block.  Pass joiner/no joiner
7581         state to register_jump_thread.
7582         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7583         state from argument rather than implying on path length.
7584         Dump the entire jump thread path into debugging dump.
7585         * tree-flow.h (register_jump_thread): Update prototype.
7587 2013-08-29  Xinliang David Li  <davidxl@google.com>
7589         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7590         Remove a trivial gcc_assert.
7592 2013-08-29  Xinliang David Li  <davidxl@google.com>
7594         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7595         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7596         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7597         Delay base decl alignment adjustment.
7598         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7599         * tree-vectorizer.h: New data structure.
7600         (set_dr_misalignment): New function.
7601         (dr_misalignment): Ditto.
7602         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7603         (vectorizable_load): Ditto.
7604         (ensure_base_align): New function.
7605         (vectorize_loops): Add dbg_cnt support.
7606         (execute_vect_slp): Ditto.
7607         * dbgcnt.def: New debug counter.
7608         * Makefile: New dependency.
7610 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7612         Revert:
7614         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7616         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7618 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7620         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7621         function descriptor.
7623 2013-09-03  Richard Biener  <rguenther@suse.de>
7625         * tree-affine.c (add_elt_to_tree): Fix association issue,
7626         avoid useless converts and make sure to always return a
7627         properly typed result.
7629 2013-09-03  Richard Biener  <rguenther@suse.de>
7631         PR middle-end/57656
7632         * fold-const.c (negate_expr_p): Fix division case.
7633         (negate_expr): Likewise.
7635 2013-09-03  Richard Biener  <rguenther@suse.de>
7637         PR lto/58285
7638         * tree-streamer-out.c: Include tm.h.
7639         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7641 2013-09-03  Jan Hubicka  <jh@suse.cz>
7643         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7645 2013-09-03  Alan Modra  <amodra@gmail.com>
7647         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7648         multilibs to big-endian target and vice versa.
7649         * config/rs6000/t-linux64: Use := assignment on all vars.
7650         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7651         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7652         * config/rs6000/t-linux64le: New file.
7653         * config/rs6000/t-linux64bele: New file.
7654         * config/rs6000/t-linux64lebe: New file.
7656 2013-09-02  Jan Hubicka  <jh@suse.cz>
7658         * ipa-inline-transform.c (inline_transform): Do not
7659         optimize_inline_calls when not optimizing.
7661 2013-09-02  Jan Hubicka  <jh@suse.cz>
7663         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7664         duplicated nodes for assembler names.
7665         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7666         hard registers from assembler name hash.
7668 2013-09-02  Jan Hubicka  <jh@suse.cz>
7670         * ipa-split.c (execute_split_functions): Split externally visible
7671         functions called once.
7673 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7675         PR ipa/58106
7676         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7677         linked list.  When finding the correct duplicate, also consider also
7678         the caller in additon to its inlined_to node.
7680 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7682         * config/aarch64/aarch64-simd-builtins.def
7683         (dup_lane_scalar): Remove.
7684         * config/aarch64/aarch64-simd.md
7685         (aarch64_simd_dup): Add 'w->w' alternative.
7686         (aarch64_dup_lane<mode>): Allow for VALL.
7687         (aarch64_dup_lane_scalar<mode>): Remove.
7688         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7689         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7690         (aarch64_get_lane_unsigned<mode>): Likewise.
7691         (aarch64_get_lane<mode>): Likewise.
7692         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7693         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7694         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7695         (VCON): Change container of V2SF.
7696         (vswap_width_name): Likewise.
7697         * config/aarch64/arm_neon.h
7698         (__aarch64_vdup_lane_any): New.
7699         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7700         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7701         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7703 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7705         PR middle-end/56382
7706         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7707         the source or the destination is a single hard register.
7709 2013-09-02  Richard Biener  <rguenther@suse.de>
7711         PR middle-end/57511
7712         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7713         non-linear SCEVs.
7715 2013-09-02  Richard Biener  <rguenther@suse.de>
7717         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7718         arithmetic to sizetype.
7720 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7722         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7723         Find auto-increment use both before and after candidate.
7725 2013-09-02  Marek Polacek  <polacek@redhat.com>
7727         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7729 2013-09-01  Jan Hubicka  <jh@suse.cz>
7731         * Makefile.in: Add ipa-profile.o
7732         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7733         * cgraph.c (struct cgraph_propagate_frequency_data,
7734         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7735         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7736         * cgraph.h (cgraph_propagate_frequency): Remove.
7737         * ipa-inline-analysis.c: Include ipa-utils.h;
7738         drop duplicated cfgloop.h.
7739         (inline_update_callee_summaries): Update.
7740         * ipa-profile.c: New file.
7741         * ipa-utils.h (ipa_propagate_frequency): Declare.
7742         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7743         data-streamer.h, value-prof.h.
7744         (symtab_remove_unreachable_nodes): Update profile.
7745         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7746         account_time_size, cmp_counts, dump_histogram,
7747         ipa_profile_generate_summary, ipa_profile_write_summary,
7748         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7749         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7750         Move to ipa-profile.c.
7752 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7754         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7756 2013-09-01  Jan Hubicka  <jh@suse.cz>
7758         * common.opt (fdevirtualize-speculatively): New function.
7759         * invoke.texi (fdevirtualize-speculatively): Document.
7760         * ipa-devirt.c: Include ipa-inline.h
7761         (likely_target_p): New function.
7762         (ipa_devirt): New function.
7763         (gate_ipa_devirt): New function.
7764         (pass_data_ipa_devirt): New static var.
7765         (pass_ipa_devirt): Likewise.
7766         (make_pass_ipa_devirt): New function.
7767         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7768         (common_handle_option): Disable devirtualization when
7769         value range profiling is available.
7770         * passes.def (pass_ipa_devirt): Add.
7771         * timever.def (TV_IPA_DEVIRT): New timevar.
7772         * tree-pass.h (make_pass_ipa_devirt):
7774 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7776         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7777         include sanitize(undefined).
7779 2013-08-31  Diego Novillo  <dnovillo@google.com>
7781         * Makefile.in (TREE_CORE_H): Define.
7782         (TREE_H): Use.
7783         (GTFILES): Add tree-core.h.
7784         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7785         size the array.
7786         * tree-core.h: New file.
7787         Move all data structures, enum, typedefs, global
7788         declarations and constants from ...
7789         * tree.h: ... here.
7791 2013-08-31  Jan Hubicka  <jh@suse.cz>
7793         * bulitins.c (expand_builtin): Do not early exit for gcov
7794         instrumented functions.
7796 2013-08-31  Marek Polacek  <polacek@redhat.com>
7798         * ubsan.c: Include tm_p.h.
7800 2013-08-31  Jan Hubicka  <jh@suse.cz>
7802         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7803         warning.
7805         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7806         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7807         * tree-cfg.c (verify_gimple_label): ... here.
7808         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7809         (ipa_merge_profiles): New function.
7810         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7811         (lto_input_function_body): Likewise.
7812         * ipa-utils.h (ipa_merge_profiles): Declare.
7813         * lto-streamer.h (lto_input_function_body): Update prototype.
7814         (emit_label_in_global_context_p): Remove.
7815         * lto-symtab.c: Include ipa-utils.h
7816         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7818 2013-08-31  Jan Hubicka  <jh@suse.cz>
7820         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7822 2013-08-31  Jan Hubicka  <jh@suse.cz>
7824         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7826 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7828         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7829         "cmp" RTX before signed_comparison_operator check to account
7830         for "code" changes.
7832 2013-08-30  Jan Hubicka  <jh@suse.cz>
7834         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7835         (detect_type_change_1): Rename to ...
7836         (detect_type_change): ... this one; early return on non-polymorphic
7837         types.
7838         (detect_type_change_ssa): Add comp_type parameter; update
7839         use of detect_type_change.
7840         (compute_complex_assign_jump_func): Add param_type parameter;
7841         update use of detect_type_change_ssa.
7842         (compute_complex_ancestor_jump_func): Likewise.
7843         (ipa_get_callee_param_type): New function.
7844         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7845         update calls to the jump function computation functions.
7847 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7848             Steven Bosscher  <steven@gcc.gnu.org>
7850         * cfgrtl.c (fixup_new_cold_bb): New routine.
7851         (commit_edge_insertions): Invoke fixup_partitions.
7852         (find_partition_fixes): New routine.
7853         (fixup_partitions): Ditto.
7854         (verify_hot_cold_block_grouping): Update comments.
7855         (rtl_verify_edges): Invoke find_partition_fixes.
7856         (rtl_verify_bb_pointers): Update comments.
7857         (rtl_verify_bb_layout): Ditto.
7858         * basic-block.h (probably_never_executed_edge_p): Declare.
7859         (fixup_partitions): Ditto.
7860         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7861         * bb-reorder.c (sanitize_hot_paths): New function.
7862         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7863         sanitize_hot_paths.
7864         * predict.c (probably_never_executed_edge_p): New routine.
7865         * cfg.c (check_bb_profile): Add partition insanity warnings.
7867 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7869         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7871 2013-08-30  Marek Polacek  <polacek@redhat.com>
7873         * Makefile.in (ubsan.o): Add.
7874         (c-family/c-ubsan.o): Add.
7875         (builtins.o): Add ubsan.h dependency.
7876         * ubsan.h: New file.
7877         * ubsan.c: New file.
7878         * common.opt: Add -fsanitize=undefined option.
7879         (flag_sanitize): Add variable.
7880         (fsanitize=): Add option.  Add Driver.
7881         (fsanitize=thread): Remove option.
7882         (fsanitize=address): Likewise.
7883         (static-libubsan): New option.
7884         * doc/invoke.texi: Document the new flag and -static-libubsan.
7885         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7886         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7887         * builtin-attrs.def (ATTR_COLD): Define.
7888         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7889         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7890         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7891         * flag-types.h (sanitize_code): New enum.
7892         * opts.c (common_handle_option): Parse command line arguments
7893         of -fsanitize=.  Add -fsanitize=unreachable option.
7894         * varasm.c (get_variable_section): Adjust.
7895         (assemble_noswitch_variable): Likewise.
7896         (assemble_variable): Likewise.
7897         (output_constant_def_contents): Likewise.
7898         (categorize_decl_for_section): Likewise.
7899         (place_block_symbol): Likewise.
7900         (output_object_block): Likewise.
7901         * builtins.def: Likewise.
7902         * toplev.c (compile_file): Likewise.
7903         (process_options): Likewise.
7904         * cppbuiltin.c: Likewise.
7905         * tsan.c (tsan_pass): Likewise.
7906         (tsan_gate): Likewise.
7907         (tsan_gate_O0): Likewise.
7908         * cfgexpand.c (partition_stack_vars): Likewise.
7909         (expand_stack_vars): Likewise.
7910         (defer_stack_allocation): Likewise.
7911         (expand_used_vars): Likewise.
7912         * cfgcleanup.c (old_insns_match_p): Likewise.
7913         * asan.c (asan_finish_file): Likewise.
7914         (asan_instrument): Likewise.
7915         (gate_asan): Likewise.
7916         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7917         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7918         (asan_global_struct): Use pointer_sized_int_node instead
7919         calling build_nonstandard_integer_type.
7920         (initialize_sanitizer_builtins): Likewise.
7921         (asan_finish_file): Likewise.
7922         * gcc.c: Document %{%:function(args):X}.
7923         (static_spec_functions): Add sanitize.
7924         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7925         store funcval != NULL there.
7926         (do_spec_1): Adjust handle_spec_function caller.
7927         (handle_braces): Allow %:function(args) as condition.
7928         (sanitize_spec_function): New function.
7929         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7930         (LIBUBSAN_SPEC): Likewise.
7931         (LIBUBSAN_EARLY_SPEC): Likewise.
7932         (SANITIZER_SPEC): Handle libubsan.
7933         (SANITIZER_EARLY_SPEC): Likewise.
7934         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7935         instead of fsanitize=address.
7936         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7937         instead of fsanitize=address*.
7938         * builtins.c: Include ubsan.h.
7939         (fold_builtin_0): Instrument __builtin_unreachable.
7940         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7941         instead of flag_asan.
7942         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7943         (pointer_sized_int_node): Define.
7944         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7946 2013-08-30  Mike Stump  <mikestump@comcast.net>
7948         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7949         with RE patterns.
7951 2013-08-29  Jan Hubicka  <jh@suse.cz>
7953         * cgraph.c (cgraph_function_body_availability): Handle weakref
7954         correctly.
7955         * passes.def: Remove pass_fixup_cfg.
7956         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7957         track when we need to remove functions.
7958         (gate_ipa_inline): Execute inlining always; add comment why.
7959         (pass_data_ipa_inline): Remove TODO_remove_functions.
7960         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7961         do not produce summaries.
7962         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7963         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7964         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7965         constant pool and vtable.
7967 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7969         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7970         New arm_neon.h's internal macros to specify 64-bit constants.
7971         Avoid using stdint.h's macros.
7973 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7975         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7976         before applying REG_P.
7978 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7980         PR tree-optimization/58277
7981         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7982         after seeing too many stmts with vdef in between dombb and current
7983         bb, invalidate everything.
7985 2013-08-30  Richard Biener  <rguenther@suse.de>
7987         * fold-const.c (fold_single_bit_test): Fix overflow test.
7989 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7991         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7992         and which can live in a register, always retrieve the value on entry.
7993         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7994         passed by invisible reference specially.
7995         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7996         (vt_add_function_parameter): Correctly deal with a parameter passed by
7997         invisible reference.
7999 2013-08-30  Jan Hubicka  <jh@suse.cz>
8001         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
8003 2013-08-30  Richard Biener  <rguenther@suse.de>
8005         PR tree-optimization/58228
8006         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
8007         allow invariant loads in nested loop vectorization.
8009 2013-08-30  Richard Biener  <rguenther@suse.de>
8011         PR tree-optimization/58223
8012         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
8013         (has_anti_or_output_dependence): ... this and adjust to also
8014         look for output dependences.
8015         (mark_nodes_having_upstream_mem_writes): Adjust.
8016         (rdg_flag_uses): Likewise.
8018 2013-08-30  Richard Biener  <rguenther@suse.de>
8020         PR tree-optimization/58010
8021         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
8022         assert that we have a loop-closed PHI.
8024 2013-08-29  Jan Hubicka  <jh@suse.cz>
8026         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
8027         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
8028         * lto-section-in.c (lto_free_function_in_decl_state): New function.
8029         (lto_free_function_in_decl_state_for_node): New function.
8031 2013-08-29  Xinliang David Li  <davidxl@google.com>
8033         * loop-unroll.c (report_unroll_peel): Minor message change.
8034         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
8035         Emit alignment peeling message with default -fopt-info.
8036         (vect_loop_versioning): Emit loop version info message.
8037         * tree-vectorizer.c (vectorize_loops): Minor message change.
8038         (execute_vect_slp): Ditto.
8040 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
8042         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
8043         of the clone from the DECL_NAME of the original function.
8045 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
8047         * passes.c (register_pass): Add overload.
8048         * tree-pass.h (register_pass): Forward declare it.  Add comment.
8050 2013-08-29  Jan Hubicka  <jh@suse.cz>
8052         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
8053         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
8054         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
8055         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8056         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
8057         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
8058         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8059         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
8061 2013-08-29  Teresa Johnson  <tejohnson@google.com>
8063         * dumpfile.c (dump_loc): Output column number.
8064         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
8065         * doc/invoke.texi: Document optall -fopt-info flag.
8066         * profile.c (read_profile_edge_counts): Use new dump framework.
8067         (compute_branch_probabilities): Ditto.
8068         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8069         when pass not in any opt group.
8070         * pass_manager.h (pass_manager::get_pass_profile): New method.
8071         * value-prof.c (check_counter): Use new dump framework.
8072         (check_ic_target): Ditto.
8073         * coverage.c (get_coverage_counts): Ditto.
8074         (coverage_init): Setup new dump framework.
8076 2013-08-29  Richard Biener  <rguenther@suse.de>
8078         PR tree-optimization/58246
8079         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8080         handle the dominance check inside a basic-block.
8082 2013-08-29  Richard Biener  <rguenther@suse.de>
8084         PR middle-end/57287
8085         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8086         of default defs that appear in abnormal PHI nodes.
8088 2013-08-29  Richard Biener  <rguenther@suse.de>
8090         PR tree-optimization/57685
8091         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8092         single-use operands to avoid exponential complexity.
8094 2013-08-28  Dehao Chen  <dehao@google.com>
8096         * ipa-inline.c (edge_badness): Fix integer underflow.
8098 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8100         * gtm-builtins.def (_ITM_free): Declare leaf.
8102 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8104         PR target/58067
8105         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8106         (*tls_local_dynamic_base_64_largepic): Likewise.
8107         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8108         Remove predicate from call operand.
8109         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8110         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8112 2013-08-28  Jeff Law  <law@redhat.com>
8114         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8115         checks for the number of predecessors and successors allowed.
8116         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8117         which require copying a joiner block if there is a request which
8118         is a subpath that requires no joiner block copying.
8120 2013-08-28  Jan Hubicka  <jh@suse.cz>
8122         * lto-streamer-out.c (DFS_write_tree_body): Drop
8123         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8124         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8125         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8126         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8127         Do not read DECL_ERROR_ISSUED.
8128         (unpack_ts_decl_with_vis_value_fields): Do not read
8129         DECL_DEFER_OUTPUT.
8130         (lto_input_ts_binfo_tree_pointers): Do not read
8131         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8132         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8133         write DECL_ERROR_ISSUED..
8134         (pack_ts_decl_with_vis_value_fields): Do not write
8135         DECL_DEFER_OUTPUT.
8136         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8137         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8138         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8139         * tree.h (tree_decl_common): Update comment.
8140         (DECL_ERROR_ISSUED): Remove.
8142 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8144         PR middle-end/58257
8145         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8147 2013-08-28  Jan Hubicka  <jh@suse.cz>
8149         * builtins.def (free): Declare leaf.
8151 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8153         * gdbhooks.py: New.
8154         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8155         * configure: Regenerate.
8157 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8159         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8160         (ipa_ancestor_jf_data): Likewise.
8161         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8162         (ipa_get_jf_pass_through_type_preserved): New function.
8163         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8164         (ipa_get_jf_ancestor_type_preserved): New function.
8165         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8166         (ipa_get_jf_ancestor_result): Likewise.
8167         (propagate_vals_accross_pass_through): Use
8168         ipa_get_jf_pass_through_result to do all the value mappings.
8169         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8170         type_preserved flag.
8171         (ipa_set_jf_cst_copy): New function.
8172         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8173         (ipa_set_jf_arith_pass_through): Likewise.
8174         (ipa_set_ancestor_jf): Likewise.
8175         (compute_complex_assign_jump_func): Set type_preserved instead of
8176         punting.
8177         (ipa_compute_jump_functions_for_edge): Likewise.
8178         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8179         (update_jump_functions_after_inlining): Update type_preserved.
8180         Explicitely create jump functions when combining one with pass_through.
8181         (ipa_write_jump_function): Stream the type_preserved flags.
8182         (ipa_read_jump_function): Likewise.
8184 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8185             Aldy Hernandez  <aldyh@redhat.com>
8187         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8188         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8189         * function.h (struct function): Add has_force_vect_loops and
8190         has_simduid_loops.
8191         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8192         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8193         handle it.
8194         * gimple.def: Update CLAUSES comments.
8195         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8196         (gimple_build_omp_for): Add argument to prototype.
8197         (gimple_omp_for_kind): New.
8198         (gimple_omp_for_set_kind): New.
8199         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8200         GOVD_DATA_SHARE_CLASS.
8201         (enum omp_region_type): Add ORT_SIMD.
8202         (gimple_add_tmp_var): Handle ORT_SIMD.
8203         (gimplify_var_or_parm_decl): Same.
8204         (is_gimple_stmt): Same.
8205         (omp_firstprivatize_variable): Same.
8206         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8207         (omp_notice_variable): Handle ORT_SIMD.
8208         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8209         (omp_check_private): Handle ORT_SIMD.
8210         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8211         OMP_CLAUSE_SAFELEN.
8212         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8213         Handle OMP_CLAUSE_LASTPRIVATE.
8214         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8215         OMP_CLAUSE_SAFELEN.
8216         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8217         (gimplify_expr): Handle OMP_SIMD.
8218         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8219         (expand_GOMP_SIMD_VF): New.
8220         (expand_GOMP_SIMD_LAST_LANE): New.
8221         * internal-fn.def (GOMP_SIMD_LANE): New.
8222         (GOMP_SIMD_VF): New.
8223         (GOMP_SIMD_LAST_LANE): New.
8224         * omp-low.c: Include target.h.
8225         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8226         OMP_CLAUSE_SAFELEN.
8227         (check_omp_nesting_restrictions): Same.
8228         (omp_max_vf): New.
8229         (lower_rec_simd_input_clauses): New.
8230         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8231         OMP_CLAUSE_LINEAR.
8232         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8233         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8234         (expand_omp_build_assign): New.
8235         (expand_omp_for_init_counts): New.
8236         (expand_omp_for_init_vars): New.
8237         (extract_omp_for_update_vars): New.
8238         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8239         and rewrite accordingly.
8240         (expand_omp_simd): New.
8241         (expand_omp_for): Use expand_omp_simd.
8242         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8243         (lower_omp_for): Do not lower the body.
8244         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8245         in their own loops.
8246         * tree-flow.h (find_omp_clause): Remove prototype.
8247         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8248         forcing vectorization of the loop, or if flag_tree_vectorize.
8249         (gate_tree_if_conversion): Similarly.
8250         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8251         gimple_build_omp_for.
8252         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8253         * tree-parloops (create_parallel_loop): Pass kind argument to
8254         gimple_build_omp_for.
8255         * tree-pretty-print.c (dump_omp_clause): Add cases for
8256         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8257         OMP_CLAUSE__SIMDUID_.
8258         (dump_generic_node): Handle OMP_SIMD.
8259         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8260         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8261         unroll OMP_SIMD loops here.
8262         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8263         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8264         loop->safelen.
8265         (vect_analyze_data_refs): Handle simd loops.
8266         * tree-vect-loop.c (vectorizable_live_operation): Handle
8267         IFN_GOMP_SIMD*.
8268         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8269         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8270         (vectorizable_load): Same.
8271         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8272         (struct simduid_to_vf): New.
8273         (simduid_to_vf::hash): New.
8274         (simduid_to-vf::equal): New.
8275         (struct simd_array_to_simduid): New.
8276         (simd_array_to_simduid::hash): New.
8277         (simd_array_to_simduid::equal): New.
8278         (adjust_simduid_builtins): New.
8279         (struct note_simd_array_uses_struct): New.
8280         (note_simd_array_uses_cb): New.
8281         (note_simd_array_uses): New.
8282         (vectorize_loops): Handle simd hints and adjust simd builtins
8283         accordingly.
8284         * tree-vectorizer.h (struct _stmt_vec_info): Add
8285         simd_lane_access_p field.
8286         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8287         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8288         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8289         (omp_clause_code_name): Same.
8290         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8291         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8292         * tree.def (OMP_SIMD): New entry.
8293         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8294         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8295         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8296         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8297         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8298         (OMP_CLAUSE_LINEAR_STEP): New.
8299         (OMP_CLAUSE_SAFELEN_EXPR): New.
8300         (OMP_CLAUSE__SIMDUID__DECL): New.
8301         (find_omp_clause): New prototype.
8303 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8305         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8306         Haswell processor detection.
8308 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8310         PR target/57927
8311         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8312         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8313         AVX2 capable processors.
8315 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8317         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8318         of vget_low_* with implementations in terms of other intrinsics.
8320 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8322         PR middle-end/57219
8323         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8324         values to -1, 0 and 1.
8326 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8328         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8329         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8330         (<optab><mode>3_insn): Remove.
8331         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8332         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8333         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8334         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8335         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8336         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8337         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8338         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8339         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8340         (ror<mode>3_insn): Likewise.
8341         * config/aarch64/predicates.md (aarch64_simd_register): New.
8343 2013-08-27  Richard Biener  <rguenther@suse.de>
8345         PR tree-optimization/57521
8346         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8347         one edge is non-critical.
8348         (find_phi_replacement_condition): Make sure to use a non-critical
8349         edge.  Cleanup and remove old bug workarounds.
8350         (bb_postdominates_preds): Remove.
8351         (if_convertible_loop_p_1): Do not compute post-dominators.
8352         (combine_blocks): Do not free post-dominators.
8353         (main_tree_if_conversion): Likewise.
8354         (pass_data_if_conversion): Add TODO_verify_ssa.
8356 2013-08-27  DJ Delorie  <dj@redhat.com>
8358         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8360 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8362         * function.c (assign_parm_find_data_types): Set passed_mode and
8363         nominal_mode to the TYPE_MODE of nominal_type for the built
8364         pointer type in case of the struct-pass-by-reference.
8366 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8368         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8369         (UINT16_TYPE): Change default to "unsigned int".
8371         * config/avr/avr.opt (mfract-convert-truncate): New option.
8372         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8373         is set, round negative fractional integers according to n1169
8374         when converting to integer types.
8376 2013-08-26  Jan Hubicka  <jh@suse.cz>
8378         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8379         methods can not be called indirectly when their address is not taken.
8381 2013-08-26  Jan Hubicka  <jh@suse.cz>
8383         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8384         ctor_for_folding.
8386 2013-08-26  Jan Hubicka  <jh@suse.cz>
8388         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8389         can be unshared.
8391 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8393         * reload.c (find_valid_class): Allow classes that do not include
8394         FIRST_PSEUDO_REGISTER - 1.
8396 2013-08-26  Jan Hubicka  <jh@suse.cz>
8398         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8399         fix edge count/frequency when speculation failed; fix type check
8400         for the direct call.
8402 2013-08-26  Jan Hubicka  <jh@suse.cz>
8404         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8406 2013-08-26  Jan Hubicka  <jh@suse.cz>
8408         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8409         to be changed by edge redirection.
8411 2013-08-26  Jan Hubicka  <jh@suse.cz>
8413         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8414         formating; add sanity check.
8415         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8416         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8417         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8418         (resolve_noninline_speculation): Update callee keys, too.
8420 2013-08-26  Jan Hubicka  <jh@suse.cz>
8422         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8423         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8425 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8427         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8428         into proper place.
8430 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8432         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8433         (x86_64_elf_select_section): Ditto.
8434         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8435         arguments.
8436         (ix86_pass_by_reference): Ditto.
8437         (output_set_got): Ditto.
8438         (ix86_unary_operator_ok): Ditto.
8439         (ix86_expand_builtin): Ditto.
8441 2013-08-23  Jan Hubicka  <jh@suse.cz>
8443         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8445 2013-08-23  Jan Hubicka  <jh@suse.cz>
8447         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8448         (tree_decl_with_vis): Add FINAL field.
8450 2013-08-23  Jeff Law  <law@redhat.com>
8452         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8453         the debugging dump when the expression is fully redundant.
8455 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8457         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8458         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8459         * pretty-print.c (pp_formatted_text_data): Likewise.
8460         (pp_write_text_to_stream): Likewise.
8461         (pp_write_text_as_dot_label_to_stream): Likewise.
8462         (pp_append_r): Likewise.
8463         (pp_format): Likewise.
8464         (pp_flush): Likewise.
8465         (pp_clear_output_area): Likewise.
8466         (pp_append_text): Likewise.
8467         (pp_formatted_text): Likewise.
8468         (pp_remaining_character_count_for_line): Likewise.
8469         (pp_newline): Likewise.
8470         (pp_character): Likewise.
8471         (output_buffer::~output_buffer): Define.
8472         (pretty_printer::~pretty_printer): Destruct output buffer.
8473         * pretty-print.h (output_buffer::~output_buffer): Declare.
8474         (pretty_printer::~pretty_printer): Declare virtual.
8476 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8478         PR other/57324
8479         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8480         HOST_WIDE_INT_M1U): New macros.
8481         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8482         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8483         unsigned -1 for lshift.
8484         * cse.c (cse_insn): Likewise.
8485         * double-int.c (rshift_double, lshift_double): Likewise.
8486         * builtins.c (fold_builtin_bitop): Likewise.
8487         * combine.c (force_to_mode): Likewise.
8488         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8489         * simplify-rtx.c (simplify_const_unary_operation,
8490         simplify_const_binary_operation): Likewise.
8491         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8492         check_va_list_escapes): Likewise.
8493         * rtlanal.c (nonzero_bits1): Likewise.
8494         * expmed.c (expand_smod_pow2): Likewise.
8495         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8497 2013-08-23  Jan Hubicka  <jh@suse.cz>
8499         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8500         as having address taken.
8502 2013-08-23  Jan Hubicka  <jh@suse.cz>
8504         * ipa-utils.h (method_class_type): Declare.
8505         * ipa-devirt.c (method_class_type): Export.
8507         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8508         do not walk base classes of anonymous types.
8510 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8512         PR rtl-optimization/58220
8513         PR regression/58221
8514         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8515         handle SEQUENCE insns properly.
8517 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8519         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8520         definition.
8521         (pp_newline_and_indent): Likewise.
8522         (pp_separate_with): Likewise.
8523         * pretty-print.c (pp_newline_and_flush): Define.
8524         (pp_newline_and_indent): Likewise.
8525         (pp_separate_with): Likewise.
8527 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8529         PR target/58218
8530         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8531         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8533 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8535         * config/i386/predicates.md (ext_sse_reg_operand): New.
8536         * config/i386/i386.md (*movti_internal): Use
8537         predicate to determine if EVEX is needed.
8538         (*movsi_internal): Ditto.
8539         (*movdf_internal): Ditto.
8540         (*movsf_internal): Ditto.
8541         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8543 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8545         PR tree-optimization/58209
8546         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8547         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8548         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8549         emit POINTER_PLUS_EXPR.
8550         (create_tailcall_accumulator): For pointer result type accumulate in
8551         sizetype type.
8553 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8555         * configure.ac: Add backslashes missing from the last change.
8556         * configure: Regenerate.
8558 2013-08-22  Jan Hubicka  <jh@susue.cz>
8560         * ipa.c (function_and_variable_visibility): First remember function
8561         was global and then make it local.
8563 2013-08-22  Julian Brown  <julian@codesourcery.com>
8565         * configure.ac: Add aarch64 to list of arches which use "nop" in
8566         debug_line test.
8567         * configure: Regenerate.
8569 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8571         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8572         gnu_libc_has_function.
8573         * config/s390/tpf.h: Likewise.
8575 2013-08-22  Jan Hubicka  <jh@susue.cz>
8577         * timevar.c (validate_phases): Add cast.
8579 2013-08-22  Jan Hubicka  <jh@susue.cz>
8581         * timevar.c (validate_phases): Use size_t for memory.
8582         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8584 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8586         * pretty-print.h (output_buffer::output_buffer): Declare.
8587         (pretty_printer::pretty_printer): Likewise.
8588         (pp_construct): Remove.
8589         * pretty-print.c (output_buffer::output_buffer): Define.
8590         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8591         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8592         (print_gimple_expr): Likewise.
8593         (print_gimple_seq): Likewise.
8594         (gimple_dump_bb): Likewise.
8595         * sched-vis.c (dump_value_slim): Likewise.
8596         (dump_insn_slim): Likewise.
8597         (dump_rtl_slim): Likewise.
8598         (str_pattern_slim): Likewise.
8599         * tree-mudflap.c (mf_varname_tree): Likewise.
8600         * graph.c (print_graph_cfg): Likewise.
8601         (start_graph_dump): Likewise.
8602         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8603         placement-new.
8604         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8605         pretty printer initialization.
8606         * coretypes.h (diagnostic_context): Remove superflous type alias
8607         declaration.
8608         (pretty_printer): Likewise.  Declare directly as a class.
8609         (pretty_print_info): Remove declaration as class.
8610         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8611         and pp_clear_output_area.
8612         (asan_add_global): Likewise.
8614 2013-08-22  Jan Hubicka  <jh@suse.cz>
8616         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8617         * ipa-utils.h (update_type_inheritance_graph): Declare.
8618         (possible_polymorphic_call_target_p): Declare.
8619         (possible_polymorphic_call_target_p): New.
8620         * ipa-devirt.c: Update toplevel comments.
8621         (cached_polymorphic_call_targets): Move up.
8622         (odr_type_d): Move ID down.
8623         (polymorphic_type_binfo_p): Update comment.
8624         (odr_hasher::remove): Likewise;
8625         (get_odr_type): Set anonymous_namespace.
8626         (dump_odr_type): Dump it.
8627         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8628         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8629         (record_binfo): Add comment.
8630         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8631         built.
8632         (devirt_node_removal_hook): Do not iCE when cache is freed.
8633         (possible_polymorphic_call_target_p): New predicate.
8634         (update_type_inheritance_graph): New function.
8636 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8637             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8638             Sergey Lega  <sergey.s.lega@intel.com>
8639             Anna Tikhonova  <anna.tikhonova@intel.com>
8640             Ilya Tocar  <ilya.tocar@intel.com>
8641             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8642             Ilya Verbin  <ilya.verbin@intel.com>
8643             Kirill Yukhin  <kirill.yukhin@intel.com>
8644             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8646         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8647         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8648         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8649         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8650         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8651         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8652         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8653         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8654         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8655         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8656         OPT_mavx512pf, OPT_mavx512er cases.
8657         * config/i386/constraints.md (v): New constraint.
8658         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8659         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8660         (bit_AVX512CD): New.
8661         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8662         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8663         * config/i386/i386-c.c (ix86_target_macros_internal):
8664         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8665         __AVX512PF__.
8666         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8667         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8668         * config/i386/i386.c (regclass_map, dbx_register_map)
8669         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8670         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8671         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8672         -mavx512pf options.
8673         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8674         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8675         -mavx512pf options.  Fix formatting.
8676         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8677         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8678         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8679         -mavx512cd, -mavx512pf options.
8680         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8681         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8682         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8683         ALL_SSE_REGS.
8684         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8685         (ix86_set_reg_reg_cost): Ditto.
8686         (x86_order_regs_for_local_alloc): Ditto.
8687         (MAX_VECT_LEN): Extend to 64-byte.
8688         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8689         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8690         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8691         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8692         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8693         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8694         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8695         (SSE_REG_MODE_P): Support new modes.
8696         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8697         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8698         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8699         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8700         (REG_CLASS_CONTENTS): Add new registers.
8701         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8702         (EXT_REX_SSE_REGNO_P): New.
8703         (HI_REGISTER_NAMES): Add new registers.
8704         * config/i386/i386.md: Define constants for new registers.
8705         (mode): Add new 512-bit modes.
8706         (prefix): Support evex prefix.
8707         (isa): Support avx512f, noavx512f, fma_avx512f.
8708         (ssemodesuffix): Add new 512-bit modes.
8709         (movxi): New.
8710         (*movxi_internal_avx512f): Ditto.
8711         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8712         Support MODE_XI.
8713         (*movsi_internal): Likewise.
8714         (*movdf_internal): Likewise.
8715         (*movsf_internal): Likewise.
8716         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8717         (<code><mode>3): Likewise.
8718         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8719         New.
8720         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8721         with the new constraint "v".
8722         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8723         modes.
8724         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8725         with the new constraint "v".
8726         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8727         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8728         (*<plusminus_insn><mode>3): Likewise.
8729         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8730         (*mul<mode>3): Likewise.
8731         (<sse>_vmmul<mode>3): Likewise.
8732         (<sse>_div<mode>3): Likewise.
8733         (<sse>_vmdiv<mode>3): Likewise.
8734         (<sse>_sqrt<mode>2): Likewise.
8735         (<sse>_vmsqrt<mode>2): Likewise.
8736         (*<code><mode>3_finite): Likewise.
8737         (*<code><mode>3) <smaxmin>: Likewise.
8738         (<sse>_vm<code><mode>3): Likewise.
8739         (*<code><mode>3) <any_logic>: Likewise.
8740         (*fma_fmadd_<mode>): Likewise.
8741         (*fma_fmsub_<mode>): Likewise.
8742         (*fma_fnmadd_<mode>): Likewise.
8743         (*fma_fnmsub_<mode>): Likewise.
8744         (*fma_fmaddsub_<mode>): Likewise.
8745         (*fma_fmsubadd_<mode>): Likewise.
8746         (*fmai_fmadd_<mode>): Likewise.
8747         (*fmai_fmsub_<mode>): Likewise.
8748         (*fmai_fnmadd_<mode>): Likewise.
8749         (*fmai_fnmsub_<mode>): Likewise.
8750         (sse_cvtsi2ss): Likewise.
8751         (sse_cvtsi2ssq): Likewise.
8752         (sse_cvtss2si): Likewise.
8753         (sse_cvtss2si_2): Likewise.
8754         (sse_cvtss2siq): Likewise.
8755         (sse_cvtss2siq_2): Likewise.
8756         (sse_cvttss2si): Likewise.
8757         (sse_cvtss2siq_2): Likewise.
8758         (float<sseintvecmodelower><mode>2): Likewise.
8759         (sse2_cvtsd2si_2): Likewise.
8760         (sse2_cvtsd2siq_2): Likewise.
8761         (*<plusminus_insn><mode>3): Likewise.
8762         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8763         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8764         (ashr<mode>3): Likewise.
8765         (<shift_insn><mode>3): Likewise.
8766         (avx2_<code><mode>3): Likewise.
8767         (*avx2_<code><mode>3): Likewise.
8768         (*andnot<mode>3): Likewise.
8769         (*<code><mode>3) <any_logic>: Likewise.
8770         (abs<mode>2): Likewise.
8771         (avx2_permvar<mode>): Likewise.
8772         (avx2_perm<mode>_1): Likewise.
8773         (*avx_vpermilp<mode>): Likewise.
8774         (avx_vpermilvar<mode>3): Likewise.
8775         (avx2_ashrv<mode>): Likewise.
8776         (avx2_<shift_insn>v<mode>): Likewise.
8777         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8778         -mavx512cd.
8779         * doc/rtl.texi: Document XImode.
8781 2013-08-21  Jeff Law  <law@redhat.com>
8783         * tree-flow.h (register_jump_thread): Pass vector of edges
8784         instead of each important edge.
8785         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8786         thread path into a vector and pass that to register_jump_thread.
8787         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8788         passed in edge vector to the current 3-edge form.
8790         Revert:
8791         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8793         * dce.c (fini_dce): Call df_analyze again just in case
8794         delete_unmarked_insns removed anything.
8796 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8798         * reload.h (struct reg_equivs): Rename to ..
8799         (struct reg_equivs_s): .. this.
8801 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8803         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8805 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8807         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8809 2013-08-21  Jeff Law  <law@redhat.com>
8811         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8812         simplify assignments too.  If the RHS collapses to a singleton
8813         range, then return the value for the range.
8815 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8817         * config/i386/sse.md (V16): Rename to...
8818         (VMOVE): this.
8819         (mov<mode>): Update iterator name.
8820         (*mov<mode>_internal): Ditto.
8821         (push<mode>1): Ditto.
8822         (movmisalign<mode>): Ditto.
8824 2013-08-20  Jan Hubicka  <jh@suse.cz>
8826         PR bootstrap/58186
8827         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8828         entry for direct edges.
8829         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8831 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8833         Revert my last two changes, r201865 and r201864:
8835         Revert r201865:
8836         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8838         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8839         instances can own GC refs.
8841         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8842         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8843         (gcc::context::gt_pch_nx): Likewise.
8844         (gcc::context::gt_pch_nx):  Likewise.
8845         * ggc.h (gt_ggc_mx <T>): New.
8846         (gt_pch_nx_with_op <T>): New.
8847         (gt_pch_nx <T>): New.
8848         * passes.c (opt_pass::gt_ggc_mx): New.
8849         (opt_pass::gt_pch_nx): New.
8850         (opt_pass::gt_pch_nx_with_op): New.
8851         (pass_manager::gt_ggc_mx): New.
8852         (pass_manager::gt_pch_nx): New.
8853         (pass_manager::gt_pch_nx_with_op): New.
8854         (pass_manager::operator new): Use
8855         ggc_internal_cleared_alloc_stat rather than xcalloc.
8856         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8857         (pass_manager::gt_ggc_mx): New.
8858         (pass_manager::gt_pch_nx): New.
8859         (pass_manager::gt_pch_nx_with_op): New.
8860         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8861         (opt_pass::operator new): New.
8862         (opt_pass::gt_ggc_mx): New.
8863         (opt_pass::gt_pch_nx): New.
8864         (opt_pass::gt_pch_nx_with_op): New.
8866         Revert r201864:
8867         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8869         * Makefile.in (GTFILES): Add context.h.
8870         * context.c (gcc::context::operator new): New.
8871         (gcc::context::gt_ggc_mx): New.
8872         (gcc::context::gt_pch_nx): New.
8873         (gcc::context::gt_pch_nx): New.
8874         * context.h (gcc::context): Add GTY((user)) marking.
8875         (gcc::context::operator new): New.
8876         (gcc::context::gt_ggc_mx): New.
8877         (gcc::context::gt_pch_nx): New.
8878         (gcc::context::gt_pch_nx): New.
8879         (g): Add GTY marking.
8880         (gt_ggc_mx (gcc::context *)): New.
8881         (gt_pch_nx (gcc::context *)): New.
8882         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8883         void *cookie)): New.
8884         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8886 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8888         * dce.c (fini_dce): Call df_analyze again just in case
8889         delete_unmarked_insns removed anything.
8891 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8893         PR rtl-optimizations/57451
8894         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8895         from crossing split section boundaries.
8897 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8899         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8900         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8901         with MULTLIB_DEFAULTS.
8903 2013-08-20  Nick Clifton  <nickc@redhat.com>
8905         * target.def (narrow_volatile_bitfield): Note that the default
8906         value is false, not !TARGET_STRICT_ALIGN.
8907         * doc/tm.texi: Regenerate.
8909 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8911         Fix LIB_SPEC for systems without libpthread.
8913         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8914         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8915         for Android.
8916         * config/i386/linux-common.h: Likewise.
8917         * config/mips/linux-common.h: Likewise.
8919 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8921         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8922         checks.
8924 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8926         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8927         instances can own GC refs.
8929         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8930         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8931         (gcc::context::gt_pch_nx): Likewise.
8932         (gcc::context::gt_pch_nx):  Likewise.
8933         * ggc.h (gt_ggc_mx <T>): New.
8934         (gt_pch_nx_with_op <T>): New.
8935         (gt_pch_nx <T>): New.
8936         * passes.c (opt_pass::gt_ggc_mx): New.
8937         (opt_pass::gt_pch_nx): New.
8938         (opt_pass::gt_pch_nx_with_op): New.
8939         (pass_manager::gt_ggc_mx): New.
8940         (pass_manager::gt_pch_nx): New.
8941         (pass_manager::gt_pch_nx_with_op): New.
8942         (pass_manager::operator new): Use
8943         ggc_internal_cleared_alloc_stat rather than xcalloc.
8944         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8945         (pass_manager::gt_ggc_mx): New.
8946         (pass_manager::gt_pch_nx): New.
8947         (pass_manager::gt_pch_nx_with_op): New.
8948         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8949         (opt_pass::operator new): New.
8950         (opt_pass::gt_ggc_mx): New.
8951         (opt_pass::gt_pch_nx): New.
8952         (opt_pass::gt_pch_nx_with_op): New.
8954 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8956         * Makefile.in (GTFILES): Add context.h.
8957         * context.c (gcc::context::operator new): New.
8958         (gcc::context::gt_ggc_mx): New.
8959         (gcc::context::gt_pch_nx): New.
8960         (gcc::context::gt_pch_nx): New.
8961         * context.h (gcc::context): Add GTY((user)) marking.
8962         (gcc::context::operator new): New.
8963         (gcc::context::gt_ggc_mx): New.
8964         (gcc::context::gt_pch_nx): New.
8965         (gcc::context::gt_pch_nx): New.
8966         (g): Add GTY marking.
8967         (gt_ggc_mx (gcc::context *)): New.
8968         (gt_pch_nx (gcc::context *)): New.
8969         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8970         void *cookie)): New.
8971         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8973 2013-08-20  Alan Modra  <amodra@gmail.com>
8975         PR target/57865
8976         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8977         (rs6000_emit_epilogue): Likewise.
8979 2013-08-19  Dehao Chen  <dehao@google.com>
8981         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8983 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8984             Jakub Jelinek  <jakub@redhat.com>
8986         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8987         (BUILT_IN_FABSD64): Likewise.
8988         (BUILT_IN_FABSD128): Likewise.
8989         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8990         (fold_builtin_1): Likewise.
8991         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8992         destination and source operands.
8993         (*abstd2_fpr): Likewise.
8994         (*nabstd2_fpr): Likewise.
8996 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8998         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8999         JUMP_P and INSN_P.
9001 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
9003         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
9004         incomplete.
9006 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9008         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
9009         * builtins.c (default_libc_has_function): New.
9010         (gnu_libc_has_function): Ditto.
9011         (no_c99_libc_has_function): Ditto.
9012         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
9013         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
9014         (fold_builtin_sincos): Likewise.
9015         (fold_builtin_cexp): Likewise.
9016         * builtins.def (DEF_C94_BUILTIN): Likewise.
9017         (DEF_C99_BUILTIN): Likewise.
9018         (DEF_C99_C90RES_BUILTIN): Likewise.
9019         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
9020         definitions to using this define.
9021         * config/darwin-protos.h (darwin_libc_has_function): New.
9022         * config/darwin.c (darwin_libc_has_function): Ditto.
9023         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
9024         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
9025         * config/darwin.h: Ditto.
9026         * config/elfos.h: Ditto.
9027         * config/freebsd.h: Ditto.
9028         * config/i386/cygming.h: Ditto.
9029         * config/i386/djgpp.h: Ditto.
9030         * config/i386/i386-interix.h: Ditto.
9031         * config/microblaze/microblaze.h: Ditto.
9032         * config/mmix/mmix.h: Ditto.
9033         * config/gnu-user.h: Ditto.
9034         * config/ia64/hpux.h: Ditto.
9035         * config/pa/pa-hpux.h: Ditto.
9036         * config/pdp11/pdp11.h: Ditto.
9037         * config/picochip/picochip.h: Ditto.
9038         * config/linux.h: Ditto.
9039         * config/netbsd.h: Ditto.
9040         * config/openbsd.h: Ditto.
9041         * config/rs6000/aix43.h: Ditto.
9042         * config/rs6000/aix51.h: Ditto.
9043         * config/rs6000/aix52.h: Ditto.
9044         * config/rs6000/aix53.h: Ditto.
9045         * config/rs6000/aix61.h: Ditto.
9046         * config/rs6000/darwin.h: Ditto.
9047         * config/rs6000/linux.h: Ditto.
9048         * config/rs6000/linux64.h: Ditto.
9049         * config/s390/tpf.h: Ditto.
9050         * config/sol2-10.h: Ditto.
9051         * config/sol2.h: Ditto.
9052         * config/vms/vms.h: Ditto.
9053         * config/vxworks.h: Ditto.
9054         * config/linux-android.c (linux_android_libc_has_function):
9055         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
9056         * config/linux-protos.h (linux_android_libc_has_function):
9057         New declaration.
9058         * config/i386/i386.c (ix86_libc_has_function): New.
9059         * config/i386/i386-protos.h
9060         (ix86_libc_has_function): New declaration.
9061         * config/i386/i386.md
9062         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
9063         ("isinf<mode>2): Likewise.
9064         * convert.c (convert_to_integer): Using new target hook
9065         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
9066         TARGET_C99_FUNCTIONS.
9067         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9068         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9069         * coretypes.h (function_class): New enum for different
9070         classes of functions.
9071         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9072         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9073         (TARGET_HAS_SINCOS): Likewise.
9074         (TARGET_LIBC_HAS_FUNCTION): New.
9075         * doc/tm.texi: Regenerated.
9076         * targhooks.h (default_libc_has_function): New declaration.
9077         (no_c99_libc_has_function): Ditto.
9078         (gnu_libc_has_function): Ditto.
9079         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9080         and TARGET_HAS_SINCOS.
9082 2013-08-18  Jan Hubicka  <jh@suse.cz>
9084         * Makeifle-in (ipa-devirt.o): New.
9085         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9086         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9087         (analyze_functions): Look into possible targets of polymorphic call.
9088         * dumpfile.c (dump_files): Add type-inheritance dump.
9089         * dumpfile.h (TDI_inheritance): New.
9090         * ipa-devirt.c: New file.
9091         * ipa-utils.h (odr_type_d): Forward declare.
9092         (odr_type): New type.
9093         (build_type_inheritance_graph): Declare.
9094         (possible_polymorphic_call_targets): Declare and introduce inline
9095         variant when only edge is pased.
9096         (dump_possible_polymorphic_call_targets): Likewise.
9097         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9098         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9099         (types_same_for_odr): ... here.
9100         * tree.h (type_in_anonymous_namespace_p): Declare.
9102 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9104         PR tree-optimization/58006
9105         * tree-parloops.c (take_address_of): Don't ICE if get_name
9106         returns NULL.
9107         (eliminate_local_variables_stmt): Remove clobber stmts.
9109 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9111         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9112         error message is issued for an alias to undefined symbol.
9114 2013-08-18  Jan Hubicka  <jh@suse.cz>
9116         * cgraph.c (cgraph_create_indirect_edge): Discover
9117         polymorphic calls and record basic info into indirect_info.
9118         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9119         devirtualization, ignore objc function calls.
9120         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9121         call with no parm index info.
9122         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9123         * tree.c (virtual_method_call_p): New function.
9124         * tree.h (virtual_method_call_p): Declare.
9126 2013-08-16  Jan Hubicka  <jh@suse.cz>
9128         PR middle-end/58179
9129         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9131 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9133         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9134         attribute.
9136 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9138         * gengtype.c (type_for_name): Add special-case support for
9139         locating types within the "gcc::" namespace.
9140         (open_base_files): Emit a "using namespace gcc" directive.
9142 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9144         PR target/58160
9145         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9146         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9148         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9149         array instead of each individual operand as a separate argument.
9150         (emit_fusion_gpr_load): Likewise.
9151         (expand_fusion_gpr_load): Add new function declaration.
9153         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9154         signature to have the operands passed as an array, instead of as
9155         separate arguments.  Allow ZERO_EXTEND to be in the memory
9156         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9157         depend on the register live/dead flags when peepholes are run.
9158         (expand_fusion_gpr_load): New function to be called from the
9159         peephole2 pass, to change the register that addis sets to be the
9160         target register.
9161         (emit_fusion_gpr_load): Change the calling signature to have the
9162         operands passed as an array, instead of as separate arguments.
9163         Allow ZERO_EXTEND to be in the memory address, and also
9164         SIGN_EXTEND if -mpower8-fusion-sign.
9166         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9167         unspec enumeration.
9168         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9169         adjust the register addis loads up in the peephole2 pass.  Do not
9170         depend on the register live/dead state when the peephole pass is done.
9172 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9174         * gengtype.c (create_user_defined_type): Ensure that the kind
9175         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9176         declaration is seen before the GTY((user)) marking.
9178 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9180         PR target/58105
9181         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9183 2013-08-16  Jan Hubicka  <jh@suse.cz>
9185         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9186         arugment expected_type.
9187         (gimple_fold_call): Use it.
9188         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9189         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9190         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9191         (try_make_edge_direct_virtual_call): Likewise.
9192         * tree.c (obj_type_ref_class): New.
9193         * tree.h (obj_type_ref_class): Use it.
9195 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9197         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9198         (rtl_slim_pp): Likewise.
9199         (init_rtl_slim_pretty_print): Likewise.
9200         (dump_value_slim):  Don't call it.  Use local pretty printer.
9201         (dump_insn_slim): Likewise.
9202         (dump_rtl_slim): Likewise.
9203         (str_pattern_slim): Likewise.
9204         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9205         Simplify.
9207 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9209         PR tree-optimization/58164
9210         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9211         walk gimple_goto_dest of GIMPLE_GOTO.
9213         PR tree-optimization/58165
9214         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9215         bi_call must be the last stmt in a bb, don't split_block, instead
9216         use fallthru edge from it and give up if there is none.
9217         Release conds vector when returning early.
9219 2013-08-14  Xinliang David Li  <davidxl@google.com>
9221         * config/i386/i386.c (ix86_option_override_internal):
9222         Remove unused variable and field.
9224 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9226         PR target/57949
9227         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9228         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9229         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9230         and Linux, correct BLKmode alignment when 128-bit alignment is
9231         required and compatibility flag is not set.
9232         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9233         for zero-size arguments when compatibility flag is not set.
9235 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9237         PR tree-optimization/58145
9238         * tree-sra.c (build_ref_for_offset): If prev_base has
9239         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9241 2013-08-14  Xinliang David Li  <davidxl@google.com>
9243         * config/i386/i386.c (ix86_option_override_internal):
9244         Fix uninitialized variable error.
9246 2013-08-14  Xinliang David Li  <davidxl@google.com>
9248         * config/i386/i386.opt: Define two new options.
9249         * config/i386/x86-tune.def: Add arch selector field in macros.
9250         * config/i386/i386.h: Adjust macro definition.
9251         * config/i386/i386.c (ix86_option_override_internal):
9252         Refactor the code.
9253         (parse_mtune_ctrl_str): New function.
9254         (set_ix86_tune_features): New function.
9255         (ix86_function_specific_restore): Call the new helper function.
9257 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9259         PR rtl-optimization/57662
9260         * sel-sched.c (code_motion_process_successors): When the current insn
9261         is removed after the recursive traversal, break from the loop.
9262         Add comments and debug printouts.
9264 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9265             Alexandre Oliva  <aoliva@redhat.com>
9267         PR target/58067
9268         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9269         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9270         there also UNSPEC_PLTOFF.
9272 2013-08-14  Marek Polacek  <polacek@redhat.com>
9274         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9275         NUM_CONDITIONS bit positions.
9277 2013-08-13  Cary Coutant  <ccoutant@google.com>
9279         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9280         (attr_checksum): Hash vector contents instead of pointer.
9281         (attr_checksum_ordered): Likewise.
9283 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9285         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9286         when Pmode != word_mode.  Add length_address attribute.
9287         (sse3_monitor_<mode>): Merge from sse3_monitor and
9288         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9289         Pmode != word_mode.  Update insn length attribute.
9290         * config/i386/i386.c (ix86_option_override_internal): Update
9291         ix86_gen_monitor selection for merged sse3_monitor insn.
9293 2013-08-13  Julian Brown  <julian@codesourcery.com>
9295         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9296         perform invalid legitimization on greater-than-word-size modes for
9297         TARGET_E500_DOUBLE.
9299 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9301         * ira.c (setup_class_translate_array): Use aclass instead of cl
9302         for classes not fully covered by allocno classes.
9304 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9306         PR tree-optimization/57661
9307         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9308         * tree-inline.c (tree_function_versioning): Initialize it.
9309         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9310         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9311         that is not being copied.
9313         PR sanitizer/56417
9314         * asan.c (instrument_strlen_call): Fix typo in comment.
9315         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9317 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9319         * config/mips/mips.md (prefetch): Use lw instead of ld on
9320         loongson in 32bit mode.
9322 2013-08-13  Nick Clifton  <nickc@redhat.com>
9324         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9326 2013-08-13  Jan Hubicka  <jh@suse.cz>
9328         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9329         introduced edge; fix typo in sanity check.
9330         (cgraph_resolve_speculation): Export; improve diagnostic.
9331         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9332         speculation at type mismatch.
9333         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9334         (cgraph_resolve_speculation): Declare.
9335         (symtab_can_be_discarded): New function.
9336         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9337         * ipa-inline-transform.c (speculation_removed): New global var.
9338         (clone_inlined_nodes): See if speculation can be removed.
9339         (inline_call): If speculations was removed, we growths may not match.
9340         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9341         (speculation_useful_p): New function.
9342         (resolve_noninline_speculation): New function.
9343         (inline_small_functions): Resolve useless speculations.
9344         * ipa-inline.h (speculation_useful_p): Declare
9345         * ipa.c (can_replace_by_local_alias): Simplify.
9346         (ipa_profile): Produce speculative calls in non-lto, too;
9347         add simple cost model; produce local aliases.
9349 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9351         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9352         PASS_MANAGER_H.
9354 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9356         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9358 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9360         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9361         iterator instead of X87MODEF.
9363 2013-08-12  Perez Read  <netfirewall@gmail.com>
9365         PR target/58132
9366         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9367         operand 0 for intel asm alternative.
9368         (*movabs<mode>_2): Ditto for operand 1.
9370 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9372         * config/aarch64/arm_none.h
9373         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9375 2013-08-12  Nick Clifton  <nickc@redhat.com>
9377         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9379 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9381         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9382         expand for QI/HImode operand to produce more effictive code for
9383         unsigned char(short) --> float(double) conversion.
9385 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9387         * doc/invoke.texi: Mention that -ftls-model does not force the final
9388         model.
9390 2013-08-12  Marek Polacek  <polacek@redhat.com>
9391             Marc Glisse  <marc.glisse@inria.fr>
9393         PR tree-optimization/57980
9394         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9395         when creating -1 constant.
9397 2013-08-10  Jan Hubicka  <jh@suse.cz>
9399         Workaround binutils PR14342.
9400         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9401         (gimple_init_edge_profiler): Likewise.
9402         (gimple_gen_ic_func_profiler): Likewise.
9404 2013-08-09  Jan Hubicka  <jh@suse.cz>
9406         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9408 2013-08-09  Xinliang David Li  <davidxl@google.com>
9410         * config/i386/stringop.def: New file.
9411         * config/i386/stringop.opt: New file.
9412         * config/i386/i386-opts.h: Include stringopt.def.
9413         * config/i386/i386.opt: Include stringopt.opt.
9414         * config/i386/i386.c (ix86_option_override_internal):
9415         Override default size based stringop inline strategies with options.
9416         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9417         New function.
9419 2013-08-09  Jan Hubicka  <jh@suse.cz>
9421         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9423 2013-08-09  Jan Hubicka  <jh@suse.cz>
9425         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9426         CGRAPH_FREQ_MAX.
9427         (dump_cgraph_node): Dump profile-id.
9428         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9429         and common_target_probability.
9430         * lto-cgraph.c (lto_output_edge): Stream common targets.
9431         (lto_output_node): Stream profile ids.
9432         (input_node): Stream profile ids.
9433         (input_edge): Stream common targets.
9434         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9435         * ipa.c: Include value-prof.h
9436         (ipa_profile_generate_summary): Turn indirect call statement histograms
9437         into common targets.
9438         (ipa_profile): Turn common targets into speculative edges.
9440 2013-08-09  Jan Hubicka  <jh@suse.cz>
9442         * cgraph.h (cgraph_node): Add profile_id.
9443         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9444         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9445         (del_node_map): Update.
9446         (find_func_by_funcdef_no): Replace by ...
9447         (find_func_by_profile_id): ... this one.
9448         (gimple_ic_transform): Do not remove useful histograms when
9449         speculation is not done; dump info when indirect call removal
9450         can happen at LTO.
9451         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9452         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9453         (__gcov_indirect_call_profiler_v2): .. this one.
9454         * profile.h (init_node_map): Update.
9455         * coverage.c (coverage_compute_profile_id): New function.
9456         * coverage.h (coverage_compute_profile_id): Declare.
9457         * tree-profile.c (init_ic_make_global_vars): Make
9458         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9459         (gimple_init_edge_profiler): Update prototype of
9460         __gcov_indirect_call_profiler.
9461         (gimple_gen_ic_func_profiler): Simplify.
9462         (tree_profiling): Use init_node_map
9464 2013-08-09  Jan Hubicka  <jh@suse.cz>
9466         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9467         non-speculative refs.
9468         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9469         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9470         (cgraph_set_call_stmt): Likewise.
9471         (cgraph_create_edge_1): Fix release checking compilatoin;
9472         clear lto_stmt_uid.
9473         (cgraph_free_edge): Free indirect info.
9474         (cgraph_turn_edge_to_speculative): New function.
9475         (cgraph_speculative_call_info): New function.
9476         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9477         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9478         (dump_cgraph_node): Dump speculation.
9479         (verify_edge_count_and_frequency): Accept speculative edges.
9480         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9481         (verify_cgraph_node): Handle speculation.
9482         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9483         (cgraph_set_call_stmt): Update prototype.
9484         (cgraph_make_edge_direct): Update prototype.
9485         (cgraph_speculative_call_info): Declare.
9486         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9487         to change; update call of ipa_find_references.
9488         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9489         lto_stmt_uid and speculative flags.
9490         (ipa_dump_references): Dump speculation.
9491         (ipa_clone_references): Clone speculative flag.
9492         (ipa_clone_referring): Likewise.
9493         (ipa_clone_ref): New function.
9494         (ipa_find_reference): Look into lto_stmt_uids
9495         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9496         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9497         (ipa_find_reference): Update declaration.
9498         (ipa_clone_ref): Declare.
9499         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9500         stream speculative flag.
9501         (lto_output_ref): Stream statements uids and speculation.
9502         (input_ref): Likewise.
9503         (input_edge): Stream speuclation.
9504         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9505         (cgraph_set_call_stmt_including_clones): Handle speculation.
9506         * ipa-inline.c (heap_edge_removal_hook): New function.
9507         (inline_small_functions): Register it.
9508         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9509         also initialize refs.
9510         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9511         edge to change.
9512         (try_make_edge_direct_simple_call): Likewise.
9513         (try_make_edge_direct_simple_call): Likewise.
9514         (update_indirect_edges_after_inlining): Likewise.
9515         (remove_described_reference): Look proper lto_stmt_uid.
9516         (propagate_controlled_uses): Likewise.
9517         (propagate_controlled_uses): Liekwise.
9518         * tree-inline.c (copy_bb): Copy speculative edges.
9519         (redirect_all_calls): New function.
9520         (copy_cfg_body): Do redirection after loop info is updated.
9521         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9523 2013-08-09  Jan Hubicka  <jh@suse.cz>
9525         * lto-streamer-out.c (output_function): Renumber PHIs.
9526         * lto-streamer-in.c (input_function): Likewise.
9528 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9530         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9531         (get_lane_unsigned): Likewise.
9532         (dup_lane_scalar): Likewise.
9533         (get_lane): enable for VALL.
9534         * config/aarch64/aarch64-simd.md
9535         (aarch64_dup_lane_scalar<mode>): Remove.
9536         (aarch64_get_lane_signed<mode>): Likewise.
9537         (aarch64_get_lane_unsigned<mode>): Likewise.
9538         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9539         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9540         (aarch64_get_lane<mode>): Enable for all vector modes.
9541         (aarch64_get_lanedi): Remove misleading constraints.
9542         * config/aarch64/arm_neon.h
9543         (__aarch64_vget_lane_any): Define.
9544         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9545         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9546         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9547         * config/aarch64/iterators.md (VDQQH): New.
9548         (VDQQHS): Likewise.
9549         (vwcore): Likewise.
9551 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9553         * configure.ac: Add GAS check for LEON instructions on SPARC.
9554         * configure: Regenerate.
9555         * config.in: Likewise.
9556         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9557         sparc*-*-* block.
9558         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9559         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9560         for LEON or LEON3.
9561         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9562         (AS_LEON_FLAG): New macro.
9563         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9564         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9565         Deal with LEON and LEON3 for the memory model.
9566         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9567         (atomic_compare_and_swap<mode>_1): Likewise.
9568         (*atomic_compare_and_swap<mode>_1): Likewise.
9570 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9572         * config/arm/neon.md (vcond): Fix floating-point vector
9573         comparisons against 0.
9575 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9577         * lra-constraints.c (emit_spill_move): Remove assert.
9578         (process_alt_operands): Add more debugging
9579         output.  Increase reject for spilling into memory.  Decrease
9580         reject for reloading scratch.
9581         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9583 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9585         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9586         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9587         micromips incompatible.  Add nan2008.
9588         (MULTILIB_DIRNAMES): Add nan2008.
9589         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9590         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9591         and micromips incompatible.  Add nan2008.
9592         (MULTILIB_DIRNAMES): Add nan2008.
9593         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9595 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9597         PR rtl-optimization/58079
9598         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9599         simplify_comparison has widened a comparison with an integer.
9601 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9603         * config/arm/neon.md (movmisalign<mode>): Disable when we
9604         don't allow unaligned accesses.
9605         (*movmisalign<mode>_neon_store): Likewise.
9606         (*movmisalign<mode>_neon_load): Likewise.
9607         (*movmisalign<mode>_neon_store): Likewise.
9608         (*movmisalign<mode>_neon_load): Likewise.
9610 2013-08-08  Jan Hubicka  <jh@suse.cz>
9612         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9613         (make_pass_rebuild_cgraph_edges): Also clear references.
9614         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9615         * ipa-inline-transform.c (inline_transform): Remove all references
9616         after inlining.
9617         * cgraphunit.c (expand_function): Remove all references after
9618         expansion.
9619         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9620         (ipa_find_reference): Rewrite to iterator.
9621         (remove_stmt_references): Likewise.
9622         (ipa_clear_stmts_in_references): New function.
9623         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9624         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9625         clear references.
9626         * ipa-split.c (split_function): Remove references in split function.
9628 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9630         PR target/57431
9631         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9632         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9634 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9636         PR target/56979
9637         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9638         suggested mode for the assignment isn't compatible with the
9639         registers required.
9641 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9643         PR target/58065
9644         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9646 2013-08-07  Xinliang David Li  <davidxl@google.com>
9648         * config/i386/i386.opt: New option -mtune-ctrl=.
9649         * config/i386/x86-tune.def: New file.
9650         * config/i386/i386.h: include x86-tune.def.
9651         * config/i386/i386.c (ix86_option_override_internal):
9652         Parsing -mtune-ctrl= option and set tune features.
9654 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9656         PR other/12081
9657         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9658         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9659         to gen_2arg_fn_t.
9661 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9663         * rtl.h (update_alignments): Declare.
9664         * final.c (grow_label_align): New function extracted from...
9665         (shorten_branches): ...here.  Call it.
9666         (update_alignments): New function.
9667         * reorg.c (sibling_labels): New variable.
9668         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9669         the new label along with it onto the sibling_labels vector.
9670         (fill_simple_delay_slots): Adjust call to get_label_before.
9671         (fill_slots_from_thread): Likewise.
9672         (relax_delay_slots): Likewise.
9673         (make_return_insns): Likewise.
9674         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9676 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9678         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9679         document its semantics.
9680         (diagnostic_report_diagnostic): Adjust accordingly.
9682 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9684         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9685         (sparc_option_override): ...and port to new C++ pass API.
9686         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9688 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9690         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9692 2013-08-06  Caroline Tice  <cmtice@google.com>
9694         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9695         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9696         * tree-pass.h: Add pass_vtable_verify.
9697         * varasm.c (assemble_variable): Add code to properly set the comdat
9698         section and name for the .vtable_map_vars section.
9699         (assemble_vtyv_preinit_initializer): New function.
9700         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9701         LINK_ONCE flag.
9702         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9703         * vtable-verify.c: New file.
9704         * vtable-verify.h: New file.
9705         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9706         initialiation levels.
9707         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9708         * passes.def: Insert pass_vtable_verify.
9709         * aclocal.m4: Reorder includes.
9710         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9711         -fvtv-counts options.
9712         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9713         as appropriate, if -fvtable-verify=... is used.
9714         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9715         -fvtable-verify=... is used.
9716         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9717         (vtable-verify.o): Add new build rule.
9718         (GTFILES): Add vtable-verify.c to list.
9719         * common.opt (fvtable-verify=): New flag.
9720         (vtv_priority): Values for fvtable-verify= flag.
9721         (fvtv-counts): New flag.
9722         (fvtv-debug): New flag.
9723         * tree.h (save_vtable_map_decl): New extern function decl.
9725 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9727         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9728         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9729         (pass_data_rl78_devirt): ...new pass_data instance and...
9730         (make_pass_rl78_devirt): ...new function.
9731         (rl78_asm_file_start): Port pass registration to new C++ API.
9733 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9735         * coretypes.h (rtl_opt_pass): Add.
9736         (gcc::context): Add.
9737         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9738         (epiphany_init): Port to new C++ pass API.
9739         (epiphany_optimize_mode_switching): Likewise.
9740         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9741         (pass_manager::get_pass_mode_switching): New.
9742         (pass_manager::get_pass_peephole2): New.
9743         * mode-switching.c (pass_mode_switching): Add clone method.
9744         * recog.c (pass_peephole2): Add clone method.
9745         (pass_split_all_insns): Add clone method.
9747 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9749         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9750         (mips_option_override): ...here, porting to new C++ API for passes.
9752 2013-08-06  Jan Hubicka  <jh@suse.cz>
9754         * cgraph.c (cgraph_get_body): New function based on lto.c
9755         implementation.
9756         * cgraph.h (cgraph_get_body): Declare.
9757         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9758         LTO paths.
9759         * cgraphunit.c (expand_function): Get body prior expanding.
9760         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9761         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9762         really need.
9763         * passes.c (do_per_function_toporder): Get body.
9764         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9765         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9767 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9769         PR fortran/57987
9770         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9771         is not re-finalized.  Rename second parameter to no_collect.
9773 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9775         PR middle-end/58041
9776         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9777         MEM_REF has proper alignment information.
9779 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9781         PR other/12081
9782         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9783         class insn_gen_fn.
9784         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9785         rtx (*) (rtx, ...) with insn_gen_fn.
9786         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9787         insn_gen_fn::stored_funcptr.  Add initializer braces.
9789 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9791         Rewrite how instances of passes are cloned to remove assumptions
9792         about their sizes (thus allowing pass subclasses to have
9793         additional data fields, albeit non-GC-managed ones at this point).
9795         * passes.c (make_pass_instance): Now that passes have clone
9796         methods, rewrite this function to eliminate XNEW and memcpy
9797         calls that used hardcoded sizes.  Since this function no longer
9798         creates pass instances, rename it to...
9799         (add_pass_instance): ...this.  Document the old way that passes were
9800         numbered and flagged, and rework this function to continue using it.
9801         (next_pass_1): Add an initial_pass argument for use by
9802         add_pass_instance.
9803         (position_pass): When adding multiple instances of a pass, use
9804         the pass's clone method, rather than relying on the XNEW/memcpy
9805         within the former make_pass_instance (now add_pass_instance).
9806         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9807         the initial instance of the current pass within the pass manager.
9809 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9811         This is the automated part of the conversion of passes from C
9812         structs to C++ classes.
9814         Patch autogenerated by refactor_passes.py from
9815         https://github.com/davidmalcolm/gcc-refactoring-scripts
9816         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9818         * asan.c (pass_asan): Convert from a global struct to a subclass of
9819         gimple_opt_pass along with...
9820         (pass_data_asan): ...new pass_data instance and...
9821         (make_pass_asan): ...new function.
9822         (pass_asan_O0): Convert from a global struct to a subclass of
9823         gimple_opt_pass along with...
9824         (pass_data_asan_O0): ...new pass_data instance and...
9825         (make_pass_asan_O0): ...new function.
9826         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9827         subclass of rtl_opt_pass along with...
9828         (pass_data_inc_dec): ...new pass_data instance and...
9829         (make_pass_inc_dec): ...new function.
9830         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9831         a subclass of rtl_opt_pass along with...
9832         (pass_data_reorder_blocks): ...new pass_data instance and...
9833         (make_pass_reorder_blocks): ...new function.
9834         (pass_duplicate_computed_gotos): Convert from a global struct to a
9835         subclass of rtl_opt_pass along with...
9836         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9837         (make_pass_duplicate_computed_gotos): ...new function.
9838         (pass_partition_blocks): Convert from a global struct to a subclass of
9839         rtl_opt_pass along with...
9840         (pass_data_partition_blocks): ...new pass_data instance and...
9841         (make_pass_partition_blocks): ...new function.
9842         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9843         struct to a subclass of rtl_opt_pass along with...
9844         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9845         and...
9846         (make_pass_branch_target_load_optimize1): ...new function.
9847         (pass_branch_target_load_optimize2): Convert from a global struct to a
9848         subclass of rtl_opt_pass along with...
9849         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9850         and...
9851         (make_pass_branch_target_load_optimize2): ...new function.
9852         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9853         of rtl_opt_pass along with...
9854         (pass_data_jump): ...new pass_data instance and...
9855         (make_pass_jump): ...new function.
9856         (pass_jump2): Convert from a global struct to a subclass of
9857         rtl_opt_pass along with...
9858         (pass_data_jump2): ...new pass_data instance and...
9859         (make_pass_jump2): ...new function.
9860         * cfgexpand.c (pass_expand): Convert from a global struct to a
9861         subclass of rtl_opt_pass along with...
9862         (pass_data_expand): ...new pass_data instance and...
9863         (make_pass_expand): ...new function.
9864         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9865         of rtl_opt_pass along with...
9866         (pass_data_free_cfg): ...new pass_data instance and...
9867         (make_pass_free_cfg): ...new function.
9868         (pass_into_cfg_layout_mode): Convert from a global struct to a
9869         subclass of rtl_opt_pass along with...
9870         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9871         (make_pass_into_cfg_layout_mode): ...new function.
9872         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9873         subclass of rtl_opt_pass along with...
9874         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9875         (make_pass_outof_cfg_layout_mode): ...new function.
9876         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9877         struct to a subclass of gimple_opt_pass along with...
9878         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9879         (make_pass_build_cgraph_edges): ...new function.
9880         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9881         subclass of gimple_opt_pass along with...
9882         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9883         (make_pass_rebuild_cgraph_edges): ...new function.
9884         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9885         subclass of gimple_opt_pass along with...
9886         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9887         and...
9888         (make_pass_remove_cgraph_callee_edges): ...new function.
9889         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9890         struct to a subclass of rtl_opt_pass along with...
9891         (pass_data_stack_adjustments): ...new pass_data instance and...
9892         (make_pass_stack_adjustments): ...new function.
9893         * combine.c (pass_combine): Convert from a global struct to a subclass
9894         of rtl_opt_pass along with...
9895         (pass_data_combine): ...new pass_data instance and...
9896         (make_pass_combine): ...new function.
9897         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9898         global struct to a subclass of rtl_opt_pass along with...
9899         (pass_data_compare_elim_after_reload): ...new pass_data instance
9900         and...
9901         (make_pass_compare_elim_after_reload): ...new function.
9902         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9903         of rtl_opt_pass along with...
9904         (pass_data_rtl_cprop): ...new pass_data instance and...
9905         (make_pass_rtl_cprop): ...new function.
9906         * cse.c (pass_cse): Convert from a global struct to a subclass of
9907         rtl_opt_pass along with...
9908         (pass_data_cse): ...new pass_data instance and...
9909         (make_pass_cse): ...new function.
9910         (pass_cse2): Convert from a global struct to a subclass of
9911         rtl_opt_pass along with...
9912         (pass_data_cse2): ...new pass_data instance and...
9913         (make_pass_cse2): ...new function.
9914         (pass_cse_after_global_opts): Convert from a global struct to a
9915         subclass of rtl_opt_pass along with...
9916         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9917         (make_pass_cse_after_global_opts): ...new function.
9918         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9919         of rtl_opt_pass along with...
9920         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9921         (make_pass_ud_rtl_dce): ...new function.
9922         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9923         rtl_opt_pass along with...
9924         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9925         (make_pass_fast_rtl_dce): ...new function.
9926         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9927         a subclass of rtl_opt_pass along with...
9928         (pass_data_df_initialize_opt): ...new pass_data instance and...
9929         (make_pass_df_initialize_opt): ...new function.
9930         (pass_df_initialize_no_opt): Convert from a global struct to a
9931         subclass of rtl_opt_pass along with...
9932         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9933         (make_pass_df_initialize_no_opt): ...new function.
9934         (pass_df_finish): Convert from a global struct to a subclass of
9935         rtl_opt_pass along with...
9936         (pass_data_df_finish): ...new pass_data instance and...
9937         (make_pass_df_finish): ...new function.
9938         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9939         rtl_opt_pass along with...
9940         (pass_data_rtl_dse1): ...new pass_data instance and...
9941         (make_pass_rtl_dse1): ...new function.
9942         (pass_rtl_dse2): Convert from a global struct to a subclass of
9943         rtl_opt_pass along with...
9944         (pass_data_rtl_dse2): ...new pass_data instance and...
9945         (make_pass_rtl_dse2): ...new function.
9946         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9947         subclass of rtl_opt_pass along with...
9948         (pass_data_dwarf2_frame): ...new pass_data instance and...
9949         (make_pass_dwarf2_frame): ...new function.
9950         * except.c (pass_set_nothrow_function_flags): Convert from a global
9951         struct to a subclass of rtl_opt_pass along with...
9952         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9953         and...
9954         (make_pass_set_nothrow_function_flags): ...new function.
9955         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9956         subclass of rtl_opt_pass along with...
9957         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9958         and...
9959         (make_pass_convert_to_eh_region_ranges): ...new function.
9960         * final.c (pass_compute_alignments): Convert from a global struct to a
9961         subclass of rtl_opt_pass along with...
9962         (pass_data_compute_alignments): ...new pass_data instance and...
9963         (make_pass_compute_alignments): ...new function.
9964         (pass_final): Convert from a global struct to a subclass of
9965         rtl_opt_pass along with...
9966         (pass_data_final): ...new pass_data instance and...
9967         (make_pass_final): ...new function.
9968         (pass_shorten_branches): Convert from a global struct to a subclass of
9969         rtl_opt_pass along with...
9970         (pass_data_shorten_branches): ...new pass_data instance and...
9971         (make_pass_shorten_branches): ...new function.
9972         (pass_clean_state): Convert from a global struct to a subclass of
9973         rtl_opt_pass along with...
9974         (pass_data_clean_state): ...new pass_data instance and...
9975         (make_pass_clean_state): ...new function.
9976         * function.c (pass_instantiate_virtual_regs): Convert from a global
9977         struct to a subclass of rtl_opt_pass along with...
9978         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9979         (make_pass_instantiate_virtual_regs): ...new function.
9980         (pass_leaf_regs): Convert from a global struct to a subclass of
9981         rtl_opt_pass along with...
9982         (pass_data_leaf_regs): ...new pass_data instance and...
9983         (make_pass_leaf_regs): ...new function.
9984         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9985         subclass of rtl_opt_pass along with...
9986         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9987         and...
9988         (make_pass_thread_prologue_and_epilogue): ...new function.
9989         (pass_match_asm_constraints): Convert from a global struct to a
9990         subclass of rtl_opt_pass along with...
9991         (pass_data_match_asm_constraints): ...new pass_data instance and...
9992         (make_pass_match_asm_constraints): ...new function.
9993         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9994         subclass of rtl_opt_pass along with...
9995         (pass_data_rtl_fwprop): ...new pass_data instance and...
9996         (make_pass_rtl_fwprop): ...new function.
9997         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9998         rtl_opt_pass along with...
9999         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
10000         (make_pass_rtl_fwprop_addr): ...new function.
10001         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
10002         rtl_opt_pass along with...
10003         (pass_data_rtl_pre): ...new pass_data instance and...
10004         (make_pass_rtl_pre): ...new function.
10005         (pass_rtl_hoist): Convert from a global struct to a subclass of
10006         rtl_opt_pass along with...
10007         (pass_data_rtl_hoist): ...new pass_data instance and...
10008         (make_pass_rtl_hoist): ...new function.
10009         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
10010         subclass of gimple_opt_pass along with...
10011         (pass_data_lower_cf): ...new pass_data instance and...
10012         (make_pass_lower_cf): ...new function.
10013         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
10014         from a global struct to a subclass of gimple_opt_pass along with...
10015         (pass_data_strength_reduction): ...new pass_data instance and...
10016         (make_pass_strength_reduction): ...new function.
10017         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
10018         of rtl_opt_pass along with...
10019         (pass_data_rtl_ifcvt): ...new pass_data instance and...
10020         (make_pass_rtl_ifcvt): ...new function.
10021         (pass_if_after_combine): Convert from a global struct to a subclass of
10022         rtl_opt_pass along with...
10023         (pass_data_if_after_combine): ...new pass_data instance and...
10024         (make_pass_if_after_combine): ...new function.
10025         (pass_if_after_reload): Convert from a global struct to a subclass of
10026         rtl_opt_pass along with...
10027         (pass_data_if_after_reload): ...new pass_data instance and...
10028         (make_pass_if_after_reload): ...new function.
10029         * init-regs.c (pass_initialize_regs): Convert from a global struct to
10030         a subclass of rtl_opt_pass along with...
10031         (pass_data_initialize_regs): ...new pass_data instance and...
10032         (make_pass_initialize_regs): ...new function.
10033         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
10034         of ipa_opt_pass_d along with...
10035         (pass_data_ipa_cp): ...new pass_data instance and...
10036         (make_pass_ipa_cp): ...new function.
10037         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
10038         global struct to a subclass of gimple_opt_pass along with...
10039         (pass_data_inline_parameters): ...new pass_data instance and...
10040         (make_pass_inline_parameters): ...new function.
10041         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
10042         subclass of gimple_opt_pass along with...
10043         (pass_data_early_inline): ...new pass_data instance and...
10044         (make_pass_early_inline): ...new function.
10045         (pass_ipa_inline): Convert from a global struct to a subclass of
10046         ipa_opt_pass_d along with...
10047         (pass_data_ipa_inline): ...new pass_data instance and...
10048         (make_pass_ipa_inline): ...new function.
10049         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
10050         struct to a subclass of gimple_opt_pass along with...
10051         (pass_data_local_pure_const): ...new pass_data instance and...
10052         (make_pass_local_pure_const): ...new function.
10053         (pass_ipa_pure_const): Convert from a global struct to a subclass of
10054         ipa_opt_pass_d along with...
10055         (pass_data_ipa_pure_const): ...new pass_data instance and...
10056         (make_pass_ipa_pure_const): ...new function.
10057         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
10058         to a subclass of ipa_opt_pass_d along with...
10059         (pass_data_ipa_reference): ...new pass_data instance and...
10060         (make_pass_ipa_reference): ...new function.
10061         * ipa-split.c (pass_split_functions): Convert from a global struct to
10062         a subclass of gimple_opt_pass along with...
10063         (pass_data_split_functions): ...new pass_data instance and...
10064         (make_pass_split_functions): ...new function.
10065         (pass_feedback_split_functions): Convert from a global struct to a
10066         subclass of gimple_opt_pass along with...
10067         (pass_data_feedback_split_functions): ...new pass_data instance and...
10068         (make_pass_feedback_split_functions): ...new function.
10069         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10070         global struct to a subclass of simple_ipa_opt_pass along with...
10071         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10072         instance and...
10073         (make_pass_ipa_function_and_variable_visibility): ...new function.
10074         (pass_ipa_free_inline_summary): Convert from a global struct to a
10075         subclass of simple_ipa_opt_pass along with...
10076         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10077         (make_pass_ipa_free_inline_summary): ...new function.
10078         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10079         subclass of ipa_opt_pass_d along with...
10080         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10081         and...
10082         (make_pass_ipa_whole_program_visibility): ...new function.
10083         (pass_ipa_profile): Convert from a global struct to a subclass of
10084         ipa_opt_pass_d along with...
10085         (pass_data_ipa_profile): ...new pass_data instance and...
10086         (make_pass_ipa_profile): ...new function.
10087         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10088         ipa_opt_pass_d along with...
10089         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10090         (make_pass_ipa_cdtor_merge): ...new function.
10091         * ira.c (pass_ira): Convert from a global struct to a subclass of
10092         rtl_opt_pass along with...
10093         (pass_data_ira): ...new pass_data instance and...
10094         (make_pass_ira): ...new function.
10095         (pass_reload): Convert from a global struct to a subclass of
10096         rtl_opt_pass along with...
10097         (pass_data_reload): ...new pass_data instance and...
10098         (make_pass_reload): ...new function.
10099         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10100         subclass of rtl_opt_pass along with...
10101         (pass_data_cleanup_barriers): ...new pass_data instance and...
10102         (make_pass_cleanup_barriers): ...new function.
10103         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10104         of rtl_opt_pass along with...
10105         (pass_data_loop2): ...new pass_data instance and...
10106         (make_pass_loop2): ...new function.
10107         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10108         rtl_opt_pass along with...
10109         (pass_data_rtl_loop_init): ...new pass_data instance and...
10110         (make_pass_rtl_loop_init): ...new function.
10111         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10112         rtl_opt_pass along with...
10113         (pass_data_rtl_loop_done): ...new pass_data instance and...
10114         (make_pass_rtl_loop_done): ...new function.
10115         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10116         subclass of rtl_opt_pass along with...
10117         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10118         (make_pass_rtl_move_loop_invariants): ...new function.
10119         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10120         rtl_opt_pass along with...
10121         (pass_data_rtl_unswitch): ...new pass_data instance and...
10122         (make_pass_rtl_unswitch): ...new function.
10123         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10124         subclass of rtl_opt_pass along with...
10125         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10126         and...
10127         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10128         (pass_rtl_doloop): Convert from a global struct to a subclass of
10129         rtl_opt_pass along with...
10130         (pass_data_rtl_doloop): ...new pass_data instance and...
10131         (make_pass_rtl_doloop): ...new function.
10132         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10133         a subclass of rtl_opt_pass along with...
10134         (pass_data_lower_subreg): ...new pass_data instance and...
10135         (make_pass_lower_subreg): ...new function.
10136         (pass_lower_subreg2): Convert from a global struct to a subclass of
10137         rtl_opt_pass along with...
10138         (pass_data_lower_subreg2): ...new pass_data instance and...
10139         (make_pass_lower_subreg2): ...new function.
10140         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10141         struct to a subclass of ipa_opt_pass_d along with...
10142         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10143         (make_pass_ipa_lto_gimple_out): ...new function.
10144         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10145         of ipa_opt_pass_d along with...
10146         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10147         (make_pass_ipa_lto_finish_out): ...new function.
10148         * mode-switching.c (pass_mode_switching): Convert from a global struct
10149         to a subclass of rtl_opt_pass along with...
10150         (pass_data_mode_switching): ...new pass_data instance and...
10151         (make_pass_mode_switching): ...new function.
10152         * modulo-sched.c (pass_sms): Convert from a global struct to a
10153         subclass of rtl_opt_pass along with...
10154         (pass_data_sms): ...new pass_data instance and...
10155         (make_pass_sms): ...new function.
10156         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10157         subclass of gimple_opt_pass along with...
10158         (pass_data_expand_omp): ...new pass_data instance and...
10159         (make_pass_expand_omp): ...new function.
10160         (pass_lower_omp): Convert from a global struct to a subclass of
10161         gimple_opt_pass along with...
10162         (pass_data_lower_omp): ...new pass_data instance and...
10163         (make_pass_lower_omp): ...new function.
10164         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10165         of gimple_opt_pass along with...
10166         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10167         (make_pass_diagnose_omp_blocks): ...new function.
10168         * passes.c (pass_early_local_passes): Convert from a global struct to
10169         a subclass of simple_ipa_opt_pass along with...
10170         (pass_data_early_local_passes): ...new pass_data instance and...
10171         (make_pass_early_local_passes): ...new function.
10172         (pass_all_early_optimizations): Convert from a global struct to a
10173         subclass of gimple_opt_pass along with...
10174         (pass_data_all_early_optimizations): ...new pass_data instance and...
10175         (make_pass_all_early_optimizations): ...new function.
10176         (pass_all_optimizations): Convert from a global struct to a subclass
10177         of gimple_opt_pass along with...
10178         (pass_data_all_optimizations): ...new pass_data instance and...
10179         (make_pass_all_optimizations): ...new function.
10180         (pass_all_optimizations_g): Convert from a global struct to a subclass
10181         of gimple_opt_pass along with...
10182         (pass_data_all_optimizations_g): ...new pass_data instance and...
10183         (make_pass_all_optimizations_g): ...new function.
10184         (pass_rest_of_compilation): Convert from a global struct to a subclass
10185         of rtl_opt_pass along with...
10186         (pass_data_rest_of_compilation): ...new pass_data instance and...
10187         (make_pass_rest_of_compilation): ...new function.
10188         (pass_postreload): Convert from a global struct to a subclass of
10189         rtl_opt_pass along with...
10190         (pass_data_postreload): ...new pass_data instance and...
10191         (make_pass_postreload): ...new function.
10192         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10193         subclass of rtl_opt_pass along with...
10194         (pass_data_gcse2): ...new pass_data instance and...
10195         (make_pass_gcse2): ...new function.
10196         * postreload.c (pass_postreload_cse): Convert from a global struct to
10197         a subclass of rtl_opt_pass along with...
10198         (pass_data_postreload_cse): ...new pass_data instance and...
10199         (make_pass_postreload_cse): ...new function.
10200         * predict.c (pass_profile): Convert from a global struct to a subclass
10201         of gimple_opt_pass along with...
10202         (pass_data_profile): ...new pass_data instance and...
10203         (make_pass_profile): ...new function.
10204         (pass_strip_predict_hints): Convert from a global struct to a subclass
10205         of gimple_opt_pass along with...
10206         (pass_data_strip_predict_hints): ...new pass_data instance and...
10207         (make_pass_strip_predict_hints): ...new function.
10208         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10209         of rtl_opt_pass along with...
10210         (pass_data_peephole2): ...new pass_data instance and...
10211         (make_pass_peephole2): ...new function.
10212         (pass_split_all_insns): Convert from a global struct to a subclass of
10213         rtl_opt_pass along with...
10214         (pass_data_split_all_insns): ...new pass_data instance and...
10215         (make_pass_split_all_insns): ...new function.
10216         (pass_split_after_reload): Convert from a global struct to a subclass
10217         of rtl_opt_pass along with...
10218         (pass_data_split_after_reload): ...new pass_data instance and...
10219         (make_pass_split_after_reload): ...new function.
10220         (pass_split_before_regstack): Convert from a global struct to a
10221         subclass of rtl_opt_pass along with...
10222         (pass_data_split_before_regstack): ...new pass_data instance and...
10223         (make_pass_split_before_regstack): ...new function.
10224         (pass_split_before_sched2): Convert from a global struct to a subclass
10225         of rtl_opt_pass along with...
10226         (pass_data_split_before_sched2): ...new pass_data instance and...
10227         (make_pass_split_before_sched2): ...new function.
10228         (pass_split_for_shorten_branches): Convert from a global struct to a
10229         subclass of rtl_opt_pass along with...
10230         (pass_data_split_for_shorten_branches): ...new pass_data instance
10231         and...
10232         (make_pass_split_for_shorten_branches): ...new function.
10233         * ree.c (pass_ree): Convert from a global struct to a subclass of
10234         rtl_opt_pass along with...
10235         (pass_data_ree): ...new pass_data instance and...
10236         (make_pass_ree): ...new function.
10237         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10238         subclass of rtl_opt_pass along with...
10239         (pass_data_stack_regs): ...new pass_data instance and...
10240         (make_pass_stack_regs): ...new function.
10241         (pass_stack_regs_run): Convert from a global struct to a subclass of
10242         rtl_opt_pass along with...
10243         (pass_data_stack_regs_run): ...new pass_data instance and...
10244         (make_pass_stack_regs_run): ...new function.
10245         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10246         subclass of rtl_opt_pass along with...
10247         (pass_data_cprop_hardreg): ...new pass_data instance and...
10248         (make_pass_cprop_hardreg): ...new function.
10249         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10250         subclass of rtl_opt_pass along with...
10251         (pass_data_reginfo_init): ...new pass_data instance and...
10252         (make_pass_reginfo_init): ...new function.
10253         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10254         of rtl_opt_pass along with...
10255         (pass_data_regmove): ...new pass_data instance and...
10256         (make_pass_regmove): ...new function.
10257         * regrename.c (pass_regrename): Convert from a global struct to a
10258         subclass of rtl_opt_pass along with...
10259         (pass_data_regrename): ...new pass_data instance and...
10260         (make_pass_regrename): ...new function.
10261         * reorg.c (pass_delay_slots): Convert from a global struct to a
10262         subclass of rtl_opt_pass along with...
10263         (pass_data_delay_slots): ...new pass_data instance and...
10264         (make_pass_delay_slots): ...new function.
10265         (pass_machine_reorg): Convert from a global struct to a subclass of
10266         rtl_opt_pass along with...
10267         (pass_data_machine_reorg): ...new pass_data instance and...
10268         (make_pass_machine_reorg): ...new function.
10269         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10270         of rtl_opt_pass along with...
10271         (pass_data_sched): ...new pass_data instance and...
10272         (make_pass_sched): ...new function.
10273         (pass_sched2): Convert from a global struct to a subclass of
10274         rtl_opt_pass along with...
10275         (pass_data_sched2): ...new pass_data instance and...
10276         (make_pass_sched2): ...new function.
10277         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10278         to a subclass of rtl_opt_pass along with...
10279         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10280         (make_pass_stack_ptr_mod): ...new function.
10281         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10282         to a subclass of rtl_opt_pass along with...
10283         (pass_data_rtl_store_motion): ...new pass_data instance and...
10284         (make_pass_rtl_store_motion): ...new function.
10285         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10286         of gimple_opt_pass along with...
10287         (pass_data_tracer): ...new pass_data instance and...
10288         (make_pass_tracer): ...new function.
10289         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10290         to a subclass of gimple_opt_pass along with...
10291         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10292         (make_pass_diagnose_tm_blocks): ...new function.
10293         (pass_lower_tm): Convert from a global struct to a subclass of
10294         gimple_opt_pass along with...
10295         (pass_data_lower_tm): ...new pass_data instance and...
10296         (make_pass_lower_tm): ...new function.
10297         (pass_tm_init): Convert from a global struct to a subclass of
10298         gimple_opt_pass along with...
10299         (pass_data_tm_init): ...new pass_data instance and...
10300         (make_pass_tm_init): ...new function.
10301         (pass_tm_mark): Convert from a global struct to a subclass of
10302         gimple_opt_pass along with...
10303         (pass_data_tm_mark): ...new pass_data instance and...
10304         (make_pass_tm_mark): ...new function.
10305         (pass_tm_edges): Convert from a global struct to a subclass of
10306         gimple_opt_pass along with...
10307         (pass_data_tm_edges): ...new pass_data instance and...
10308         (make_pass_tm_edges): ...new function.
10309         (pass_tm_memopt): Convert from a global struct to a subclass of
10310         gimple_opt_pass along with...
10311         (pass_data_tm_memopt): ...new pass_data instance and...
10312         (make_pass_tm_memopt): ...new function.
10313         (pass_ipa_tm): Convert from a global struct to a subclass of
10314         simple_ipa_opt_pass along with...
10315         (pass_data_ipa_tm): ...new pass_data instance and...
10316         (make_pass_ipa_tm): ...new function.
10317         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10318         subclass of gimple_opt_pass along with...
10319         (pass_data_call_cdce): ...new pass_data instance and...
10320         (make_pass_call_cdce): ...new function.
10321         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10322         subclass of gimple_opt_pass along with...
10323         (pass_data_build_cfg): ...new pass_data instance and...
10324         (make_pass_build_cfg): ...new function.
10325         (pass_split_crit_edges): Convert from a global struct to a subclass of
10326         gimple_opt_pass along with...
10327         (pass_data_split_crit_edges): ...new pass_data instance and...
10328         (make_pass_split_crit_edges): ...new function.
10329         (pass_warn_function_return): Convert from a global struct to a
10330         subclass of gimple_opt_pass along with...
10331         (pass_data_warn_function_return): ...new pass_data instance and...
10332         (make_pass_warn_function_return): ...new function.
10333         (pass_warn_function_noreturn): Convert from a global struct to a
10334         subclass of gimple_opt_pass along with...
10335         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10336         (make_pass_warn_function_noreturn): ...new function.
10337         (pass_warn_unused_result): Convert from a global struct to a subclass
10338         of gimple_opt_pass along with...
10339         (pass_data_warn_unused_result): ...new pass_data instance and...
10340         (make_pass_warn_unused_result): ...new function.
10341         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10342         a subclass of gimple_opt_pass along with...
10343         (pass_data_merge_phi): ...new pass_data instance and...
10344         (make_pass_merge_phi): ...new function.
10345         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10346         a subclass of gimple_opt_pass along with...
10347         (pass_data_lower_complex): ...new pass_data instance and...
10348         (make_pass_lower_complex): ...new function.
10349         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10350         gimple_opt_pass along with...
10351         (pass_data_lower_complex_O0): ...new pass_data instance and...
10352         (make_pass_lower_complex_O0): ...new function.
10353         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10354         subclass of gimple_opt_pass along with...
10355         (pass_data_lower_eh): ...new pass_data instance and...
10356         (make_pass_lower_eh): ...new function.
10357         (pass_refactor_eh): Convert from a global struct to a subclass of
10358         gimple_opt_pass along with...
10359         (pass_data_refactor_eh): ...new pass_data instance and...
10360         (make_pass_refactor_eh): ...new function.
10361         (pass_lower_resx): Convert from a global struct to a subclass of
10362         gimple_opt_pass along with...
10363         (pass_data_lower_resx): ...new pass_data instance and...
10364         (make_pass_lower_resx): ...new function.
10365         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10366         of gimple_opt_pass along with...
10367         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10368         (make_pass_lower_eh_dispatch): ...new function.
10369         (pass_cleanup_eh): Convert from a global struct to a subclass of
10370         gimple_opt_pass along with...
10371         (pass_data_cleanup_eh): ...new pass_data instance and...
10372         (make_pass_cleanup_eh): ...new function.
10373         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10374         to a subclass of simple_ipa_opt_pass along with...
10375         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10376         (make_pass_ipa_lower_emutls): ...new function.
10377         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10378         a subclass of gimple_opt_pass along with...
10379         (pass_data_if_conversion): ...new pass_data instance and...
10380         (make_pass_if_conversion): ...new function.
10381         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10382         subclass of gimple_opt_pass along with...
10383         (pass_data_build_ssa): ...new pass_data instance and...
10384         (make_pass_build_ssa): ...new function.
10385         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10386         global struct to a subclass of gimple_opt_pass along with...
10387         (pass_data_loop_distribution): ...new pass_data instance and...
10388         (make_pass_loop_distribution): ...new function.
10389         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10390         subclass of gimple_opt_pass along with...
10391         (pass_data_mudflap_1): ...new pass_data instance and...
10392         (make_pass_mudflap_1): ...new function.
10393         (pass_mudflap_2): Convert from a global struct to a subclass of
10394         gimple_opt_pass along with...
10395         (pass_data_mudflap_2): ...new pass_data instance and...
10396         (make_pass_mudflap_2): ...new function.
10397         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10398         subclass of gimple_opt_pass along with...
10399         (pass_data_mudflap_1): ...new pass_data instance and...
10400         (make_pass_mudflap_1): ...new function.
10401         (pass_mudflap_2): Convert from a global struct to a subclass of
10402         gimple_opt_pass along with...
10403         (pass_data_mudflap_2): ...new pass_data instance and...
10404         (make_pass_mudflap_2): ...new function.
10405         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10406         gimple_opt_pass along with...
10407         (pass_data_nrv): ...new pass_data instance and...
10408         (make_pass_nrv): ...new function.
10409         (pass_return_slot): Convert from a global struct to a subclass of
10410         gimple_opt_pass along with...
10411         (pass_data_return_slot): ...new pass_data instance and...
10412         (make_pass_return_slot): ...new function.
10413         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10414         to a subclass of gimple_opt_pass along with...
10415         (pass_data_object_sizes): ...new pass_data instance and...
10416         (make_pass_object_sizes): ...new function.
10417         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10418         global struct to a subclass of gimple_opt_pass along with...
10419         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10420         and...
10421         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10422         (pass_fixup_cfg): Convert from a global struct to a subclass of
10423         gimple_opt_pass along with...
10424         (pass_data_fixup_cfg): ...new pass_data instance and...
10425         (make_pass_fixup_cfg): ...new function.
10426         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10427         (make_pass_mudflap_1): ...new function.
10428         (pass_mudflap_2): Replace declaration with that of...
10429         (make_pass_mudflap_2): ...new function.
10430         (pass_asan): Replace declaration with that of...
10431         (make_pass_asan): ...new function.
10432         (pass_asan_O0): Replace declaration with that of...
10433         (make_pass_asan_O0): ...new function.
10434         (pass_tsan): Replace declaration with that of...
10435         (make_pass_tsan): ...new function.
10436         (pass_tsan_O0): Replace declaration with that of...
10437         (make_pass_tsan_O0): ...new function.
10438         (pass_lower_cf): Replace declaration with that of...
10439         (make_pass_lower_cf): ...new function.
10440         (pass_refactor_eh): Replace declaration with that of...
10441         (make_pass_refactor_eh): ...new function.
10442         (pass_lower_eh): Replace declaration with that of...
10443         (make_pass_lower_eh): ...new function.
10444         (pass_lower_eh_dispatch): Replace declaration with that of...
10445         (make_pass_lower_eh_dispatch): ...new function.
10446         (pass_lower_resx): Replace declaration with that of...
10447         (make_pass_lower_resx): ...new function.
10448         (pass_build_cfg): Replace declaration with that of...
10449         (make_pass_build_cfg): ...new function.
10450         (pass_early_tree_profile): Replace declaration with that of...
10451         (make_pass_early_tree_profile): ...new function.
10452         (pass_cleanup_eh): Replace declaration with that of...
10453         (make_pass_cleanup_eh): ...new function.
10454         (pass_sra): Replace declaration with that of...
10455         (make_pass_sra): ...new function.
10456         (pass_sra_early): Replace declaration with that of...
10457         (make_pass_sra_early): ...new function.
10458         (pass_early_ipa_sra): Replace declaration with that of...
10459         (make_pass_early_ipa_sra): ...new function.
10460         (pass_tail_recursion): Replace declaration with that of...
10461         (make_pass_tail_recursion): ...new function.
10462         (pass_tail_calls): Replace declaration with that of...
10463         (make_pass_tail_calls): ...new function.
10464         (pass_tree_loop): Replace declaration with that of...
10465         (make_pass_tree_loop): ...new function.
10466         (pass_tree_loop_init): Replace declaration with that of...
10467         (make_pass_tree_loop_init): ...new function.
10468         (pass_lim): Replace declaration with that of...
10469         (make_pass_lim): ...new function.
10470         (pass_tree_unswitch): Replace declaration with that of...
10471         (make_pass_tree_unswitch): ...new function.
10472         (pass_predcom): Replace declaration with that of...
10473         (make_pass_predcom): ...new function.
10474         (pass_iv_canon): Replace declaration with that of...
10475         (make_pass_iv_canon): ...new function.
10476         (pass_scev_cprop): Replace declaration with that of...
10477         (make_pass_scev_cprop): ...new function.
10478         (pass_empty_loop): Replace declaration with that of...
10479         (make_pass_empty_loop): ...new function.
10480         (pass_record_bounds): Replace declaration with that of...
10481         (make_pass_record_bounds): ...new function.
10482         (pass_graphite): Replace declaration with that of...
10483         (make_pass_graphite): ...new function.
10484         (pass_graphite_transforms): Replace declaration with that of...
10485         (make_pass_graphite_transforms): ...new function.
10486         (pass_if_conversion): Replace declaration with that of...
10487         (make_pass_if_conversion): ...new function.
10488         (pass_loop_distribution): Replace declaration with that of...
10489         (make_pass_loop_distribution): ...new function.
10490         (pass_vectorize): Replace declaration with that of...
10491         (make_pass_vectorize): ...new function.
10492         (pass_slp_vectorize): Replace declaration with that of...
10493         (make_pass_slp_vectorize): ...new function.
10494         (pass_complete_unroll): Replace declaration with that of...
10495         (make_pass_complete_unroll): ...new function.
10496         (pass_complete_unrolli): Replace declaration with that of...
10497         (make_pass_complete_unrolli): ...new function.
10498         (pass_parallelize_loops): Replace declaration with that of...
10499         (make_pass_parallelize_loops): ...new function.
10500         (pass_loop_prefetch): Replace declaration with that of...
10501         (make_pass_loop_prefetch): ...new function.
10502         (pass_iv_optimize): Replace declaration with that of...
10503         (make_pass_iv_optimize): ...new function.
10504         (pass_tree_loop_done): Replace declaration with that of...
10505         (make_pass_tree_loop_done): ...new function.
10506         (pass_ch): Replace declaration with that of...
10507         (make_pass_ch): ...new function.
10508         (pass_ccp): Replace declaration with that of...
10509         (make_pass_ccp): ...new function.
10510         (pass_phi_only_cprop): Replace declaration with that of...
10511         (make_pass_phi_only_cprop): ...new function.
10512         (pass_build_ssa): Replace declaration with that of...
10513         (make_pass_build_ssa): ...new function.
10514         (pass_build_alias): Replace declaration with that of...
10515         (make_pass_build_alias): ...new function.
10516         (pass_build_ealias): Replace declaration with that of...
10517         (make_pass_build_ealias): ...new function.
10518         (pass_dominator): Replace declaration with that of...
10519         (make_pass_dominator): ...new function.
10520         (pass_dce): Replace declaration with that of...
10521         (make_pass_dce): ...new function.
10522         (pass_dce_loop): Replace declaration with that of...
10523         (make_pass_dce_loop): ...new function.
10524         (pass_cd_dce): Replace declaration with that of...
10525         (make_pass_cd_dce): ...new function.
10526         (pass_call_cdce): Replace declaration with that of...
10527         (make_pass_call_cdce): ...new function.
10528         (pass_merge_phi): Replace declaration with that of...
10529         (make_pass_merge_phi): ...new function.
10530         (pass_split_crit_edges): Replace declaration with that of...
10531         (make_pass_split_crit_edges): ...new function.
10532         (pass_pre): Replace declaration with that of...
10533         (make_pass_pre): ...new function.
10534         (pass_profile): Replace declaration with that of...
10535         (make_pass_profile): ...new function.
10536         (pass_strip_predict_hints): Replace declaration with that of...
10537         (make_pass_strip_predict_hints): ...new function.
10538         (pass_lower_complex_O0): Replace declaration with that of...
10539         (make_pass_lower_complex_O0): ...new function.
10540         (pass_lower_complex): Replace declaration with that of...
10541         (make_pass_lower_complex): ...new function.
10542         (pass_lower_vector): Replace declaration with that of...
10543         (make_pass_lower_vector): ...new function.
10544         (pass_lower_vector_ssa): Replace declaration with that of...
10545         (make_pass_lower_vector_ssa): ...new function.
10546         (pass_lower_omp): Replace declaration with that of...
10547         (make_pass_lower_omp): ...new function.
10548         (pass_diagnose_omp_blocks): Replace declaration with that of...
10549         (make_pass_diagnose_omp_blocks): ...new function.
10550         (pass_expand_omp): Replace declaration with that of...
10551         (make_pass_expand_omp): ...new function.
10552         (pass_expand_omp_ssa): Replace declaration with that of...
10553         (make_pass_expand_omp_ssa): ...new function.
10554         (pass_object_sizes): Replace declaration with that of...
10555         (make_pass_object_sizes): ...new function.
10556         (pass_strlen): Replace declaration with that of...
10557         (make_pass_strlen): ...new function.
10558         (pass_fold_builtins): Replace declaration with that of...
10559         (make_pass_fold_builtins): ...new function.
10560         (pass_stdarg): Replace declaration with that of...
10561         (make_pass_stdarg): ...new function.
10562         (pass_early_warn_uninitialized): Replace declaration with that of...
10563         (make_pass_early_warn_uninitialized): ...new function.
10564         (pass_late_warn_uninitialized): Replace declaration with that of...
10565         (make_pass_late_warn_uninitialized): ...new function.
10566         (pass_cse_reciprocals): Replace declaration with that of...
10567         (make_pass_cse_reciprocals): ...new function.
10568         (pass_cse_sincos): Replace declaration with that of...
10569         (make_pass_cse_sincos): ...new function.
10570         (pass_optimize_bswap): Replace declaration with that of...
10571         (make_pass_optimize_bswap): ...new function.
10572         (pass_optimize_widening_mul): Replace declaration with that of...
10573         (make_pass_optimize_widening_mul): ...new function.
10574         (pass_warn_function_return): Replace declaration with that of...
10575         (make_pass_warn_function_return): ...new function.
10576         (pass_warn_function_noreturn): Replace declaration with that of...
10577         (make_pass_warn_function_noreturn): ...new function.
10578         (pass_cselim): Replace declaration with that of...
10579         (make_pass_cselim): ...new function.
10580         (pass_phiopt): Replace declaration with that of...
10581         (make_pass_phiopt): ...new function.
10582         (pass_forwprop): Replace declaration with that of...
10583         (make_pass_forwprop): ...new function.
10584         (pass_phiprop): Replace declaration with that of...
10585         (make_pass_phiprop): ...new function.
10586         (pass_tree_ifcombine): Replace declaration with that of...
10587         (make_pass_tree_ifcombine): ...new function.
10588         (pass_dse): Replace declaration with that of...
10589         (make_pass_dse): ...new function.
10590         (pass_nrv): Replace declaration with that of...
10591         (make_pass_nrv): ...new function.
10592         (pass_rename_ssa_copies): Replace declaration with that of...
10593         (make_pass_rename_ssa_copies): ...new function.
10594         (pass_sink_code): Replace declaration with that of...
10595         (make_pass_sink_code): ...new function.
10596         (pass_fre): Replace declaration with that of...
10597         (make_pass_fre): ...new function.
10598         (pass_check_data_deps): Replace declaration with that of...
10599         (make_pass_check_data_deps): ...new function.
10600         (pass_copy_prop): Replace declaration with that of...
10601         (make_pass_copy_prop): ...new function.
10602         (pass_vrp): Replace declaration with that of...
10603         (make_pass_vrp): ...new function.
10604         (pass_uncprop): Replace declaration with that of...
10605         (make_pass_uncprop): ...new function.
10606         (pass_return_slot): Replace declaration with that of...
10607         (make_pass_return_slot): ...new function.
10608         (pass_reassoc): Replace declaration with that of...
10609         (make_pass_reassoc): ...new function.
10610         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10611         (make_pass_rebuild_cgraph_edges): ...new function.
10612         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10613         (make_pass_remove_cgraph_callee_edges): ...new function.
10614         (pass_build_cgraph_edges): Replace declaration with that of...
10615         (make_pass_build_cgraph_edges): ...new function.
10616         (pass_local_pure_const): Replace declaration with that of...
10617         (make_pass_local_pure_const): ...new function.
10618         (pass_tracer): Replace declaration with that of...
10619         (make_pass_tracer): ...new function.
10620         (pass_warn_unused_result): Replace declaration with that of...
10621         (make_pass_warn_unused_result): ...new function.
10622         (pass_diagnose_tm_blocks): Replace declaration with that of...
10623         (make_pass_diagnose_tm_blocks): ...new function.
10624         (pass_lower_tm): Replace declaration with that of...
10625         (make_pass_lower_tm): ...new function.
10626         (pass_tm_init): Replace declaration with that of...
10627         (make_pass_tm_init): ...new function.
10628         (pass_tm_mark): Replace declaration with that of...
10629         (make_pass_tm_mark): ...new function.
10630         (pass_tm_memopt): Replace declaration with that of...
10631         (make_pass_tm_memopt): ...new function.
10632         (pass_tm_edges): Replace declaration with that of...
10633         (make_pass_tm_edges): ...new function.
10634         (pass_split_functions): Replace declaration with that of...
10635         (make_pass_split_functions): ...new function.
10636         (pass_feedback_split_functions): Replace declaration with that of...
10637         (make_pass_feedback_split_functions): ...new function.
10638         (pass_strength_reduction): Replace declaration with that of...
10639         (make_pass_strength_reduction): ...new function.
10640         (pass_ipa_lower_emutls): Replace declaration with that of...
10641         (make_pass_ipa_lower_emutls): ...new function.
10642         (pass_ipa_function_and_variable_visibility): Replace declaration with
10643         that of...
10644         (make_pass_ipa_function_and_variable_visibility): ...new function.
10645         (pass_ipa_tree_profile): Replace declaration with that of...
10646         (make_pass_ipa_tree_profile): ...new function.
10647         (pass_early_local_passes): Replace declaration with that of...
10648         (make_pass_early_local_passes): ...new function.
10649         (pass_ipa_whole_program_visibility): Replace declaration with that
10650         of...
10651         (make_pass_ipa_whole_program_visibility): ...new function.
10652         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10653         (make_pass_ipa_lto_gimple_out): ...new function.
10654         (pass_ipa_increase_alignment): Replace declaration with that of...
10655         (make_pass_ipa_increase_alignment): ...new function.
10656         (pass_ipa_inline): Replace declaration with that of...
10657         (make_pass_ipa_inline): ...new function.
10658         (pass_ipa_free_lang_data): Replace declaration with that of...
10659         (make_pass_ipa_free_lang_data): ...new function.
10660         (pass_ipa_free_inline_summary): Replace declaration with that of...
10661         (make_pass_ipa_free_inline_summary): ...new function.
10662         (pass_ipa_cp): Replace declaration with that of...
10663         (make_pass_ipa_cp): ...new function.
10664         (pass_ipa_reference): Replace declaration with that of...
10665         (make_pass_ipa_reference): ...new function.
10666         (pass_ipa_pure_const): Replace declaration with that of...
10667         (make_pass_ipa_pure_const): ...new function.
10668         (pass_ipa_pta): Replace declaration with that of...
10669         (make_pass_ipa_pta): ...new function.
10670         (pass_ipa_lto_finish_out): Replace declaration with that of...
10671         (make_pass_ipa_lto_finish_out): ...new function.
10672         (pass_ipa_tm): Replace declaration with that of...
10673         (make_pass_ipa_tm): ...new function.
10674         (pass_ipa_profile): Replace declaration with that of...
10675         (make_pass_ipa_profile): ...new function.
10676         (pass_ipa_cdtor_merge): Replace declaration with that of...
10677         (make_pass_ipa_cdtor_merge): ...new function.
10678         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10679         of...
10680         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10681         (pass_init_datastructures): Replace declaration with that of...
10682         (make_pass_init_datastructures): ...new function.
10683         (pass_fixup_cfg): Replace declaration with that of...
10684         (make_pass_fixup_cfg): ...new function.
10685         (pass_expand): Replace declaration with that of...
10686         (make_pass_expand): ...new function.
10687         (pass_instantiate_virtual_regs): Replace declaration with that of...
10688         (make_pass_instantiate_virtual_regs): ...new function.
10689         (pass_rtl_fwprop): Replace declaration with that of...
10690         (make_pass_rtl_fwprop): ...new function.
10691         (pass_rtl_fwprop_addr): Replace declaration with that of...
10692         (make_pass_rtl_fwprop_addr): ...new function.
10693         (pass_jump): Replace declaration with that of...
10694         (make_pass_jump): ...new function.
10695         (pass_jump2): Replace declaration with that of...
10696         (make_pass_jump2): ...new function.
10697         (pass_lower_subreg): Replace declaration with that of...
10698         (make_pass_lower_subreg): ...new function.
10699         (pass_cse): Replace declaration with that of...
10700         (make_pass_cse): ...new function.
10701         (pass_fast_rtl_dce): Replace declaration with that of...
10702         (make_pass_fast_rtl_dce): ...new function.
10703         (pass_ud_rtl_dce): Replace declaration with that of...
10704         (make_pass_ud_rtl_dce): ...new function.
10705         (pass_rtl_dce): Replace declaration with that of...
10706         (make_pass_rtl_dce): ...new function.
10707         (pass_rtl_dse1): Replace declaration with that of...
10708         (make_pass_rtl_dse1): ...new function.
10709         (pass_rtl_dse2): Replace declaration with that of...
10710         (make_pass_rtl_dse2): ...new function.
10711         (pass_rtl_dse3): Replace declaration with that of...
10712         (make_pass_rtl_dse3): ...new function.
10713         (pass_rtl_cprop): Replace declaration with that of...
10714         (make_pass_rtl_cprop): ...new function.
10715         (pass_rtl_pre): Replace declaration with that of...
10716         (make_pass_rtl_pre): ...new function.
10717         (pass_rtl_hoist): Replace declaration with that of...
10718         (make_pass_rtl_hoist): ...new function.
10719         (pass_rtl_store_motion): Replace declaration with that of...
10720         (make_pass_rtl_store_motion): ...new function.
10721         (pass_cse_after_global_opts): Replace declaration with that of...
10722         (make_pass_cse_after_global_opts): ...new function.
10723         (pass_rtl_ifcvt): Replace declaration with that of...
10724         (make_pass_rtl_ifcvt): ...new function.
10725         (pass_into_cfg_layout_mode): Replace declaration with that of...
10726         (make_pass_into_cfg_layout_mode): ...new function.
10727         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10728         (make_pass_outof_cfg_layout_mode): ...new function.
10729         (pass_loop2): Replace declaration with that of...
10730         (make_pass_loop2): ...new function.
10731         (pass_rtl_loop_init): Replace declaration with that of...
10732         (make_pass_rtl_loop_init): ...new function.
10733         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10734         (make_pass_rtl_move_loop_invariants): ...new function.
10735         (pass_rtl_unswitch): Replace declaration with that of...
10736         (make_pass_rtl_unswitch): ...new function.
10737         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10738         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10739         (pass_rtl_doloop): Replace declaration with that of...
10740         (make_pass_rtl_doloop): ...new function.
10741         (pass_rtl_loop_done): Replace declaration with that of...
10742         (make_pass_rtl_loop_done): ...new function.
10743         (pass_web): Replace declaration with that of...
10744         (make_pass_web): ...new function.
10745         (pass_cse2): Replace declaration with that of...
10746         (make_pass_cse2): ...new function.
10747         (pass_df_initialize_opt): Replace declaration with that of...
10748         (make_pass_df_initialize_opt): ...new function.
10749         (pass_df_initialize_no_opt): Replace declaration with that of...
10750         (make_pass_df_initialize_no_opt): ...new function.
10751         (pass_reginfo_init): Replace declaration with that of...
10752         (make_pass_reginfo_init): ...new function.
10753         (pass_inc_dec): Replace declaration with that of...
10754         (make_pass_inc_dec): ...new function.
10755         (pass_stack_ptr_mod): Replace declaration with that of...
10756         (make_pass_stack_ptr_mod): ...new function.
10757         (pass_initialize_regs): Replace declaration with that of...
10758         (make_pass_initialize_regs): ...new function.
10759         (pass_combine): Replace declaration with that of...
10760         (make_pass_combine): ...new function.
10761         (pass_if_after_combine): Replace declaration with that of...
10762         (make_pass_if_after_combine): ...new function.
10763         (pass_ree): Replace declaration with that of...
10764         (make_pass_ree): ...new function.
10765         (pass_partition_blocks): Replace declaration with that of...
10766         (make_pass_partition_blocks): ...new function.
10767         (pass_match_asm_constraints): Replace declaration with that of...
10768         (make_pass_match_asm_constraints): ...new function.
10769         (pass_regmove): Replace declaration with that of...
10770         (make_pass_regmove): ...new function.
10771         (pass_split_all_insns): Replace declaration with that of...
10772         (make_pass_split_all_insns): ...new function.
10773         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10774         (make_pass_fast_rtl_byte_dce): ...new function.
10775         (pass_lower_subreg2): Replace declaration with that of...
10776         (make_pass_lower_subreg2): ...new function.
10777         (pass_mode_switching): Replace declaration with that of...
10778         (make_pass_mode_switching): ...new function.
10779         (pass_sms): Replace declaration with that of...
10780         (make_pass_sms): ...new function.
10781         (pass_sched): Replace declaration with that of...
10782         (make_pass_sched): ...new function.
10783         (pass_ira): Replace declaration with that of...
10784         (make_pass_ira): ...new function.
10785         (pass_reload): Replace declaration with that of...
10786         (make_pass_reload): ...new function.
10787         (pass_clean_state): Replace declaration with that of...
10788         (make_pass_clean_state): ...new function.
10789         (pass_branch_prob): Replace declaration with that of...
10790         (make_pass_branch_prob): ...new function.
10791         (pass_value_profile_transformations): Replace declaration with that
10792         of...
10793         (make_pass_value_profile_transformations): ...new function.
10794         (pass_postreload_cse): Replace declaration with that of...
10795         (make_pass_postreload_cse): ...new function.
10796         (pass_gcse2): Replace declaration with that of...
10797         (make_pass_gcse2): ...new function.
10798         (pass_split_after_reload): Replace declaration with that of...
10799         (make_pass_split_after_reload): ...new function.
10800         (pass_branch_target_load_optimize1): Replace declaration with that
10801         of...
10802         (make_pass_branch_target_load_optimize1): ...new function.
10803         (pass_thread_prologue_and_epilogue): Replace declaration with that
10804         of...
10805         (make_pass_thread_prologue_and_epilogue): ...new function.
10806         (pass_stack_adjustments): Replace declaration with that of...
10807         (make_pass_stack_adjustments): ...new function.
10808         (pass_peephole2): Replace declaration with that of...
10809         (make_pass_peephole2): ...new function.
10810         (pass_if_after_reload): Replace declaration with that of...
10811         (make_pass_if_after_reload): ...new function.
10812         (pass_regrename): Replace declaration with that of...
10813         (make_pass_regrename): ...new function.
10814         (pass_cprop_hardreg): Replace declaration with that of...
10815         (make_pass_cprop_hardreg): ...new function.
10816         (pass_reorder_blocks): Replace declaration with that of...
10817         (make_pass_reorder_blocks): ...new function.
10818         (pass_branch_target_load_optimize2): Replace declaration with that
10819         of...
10820         (make_pass_branch_target_load_optimize2): ...new function.
10821         (pass_leaf_regs): Replace declaration with that of...
10822         (make_pass_leaf_regs): ...new function.
10823         (pass_split_before_sched2): Replace declaration with that of...
10824         (make_pass_split_before_sched2): ...new function.
10825         (pass_compare_elim_after_reload): Replace declaration with that of...
10826         (make_pass_compare_elim_after_reload): ...new function.
10827         (pass_sched2): Replace declaration with that of...
10828         (make_pass_sched2): ...new function.
10829         (pass_stack_regs): Replace declaration with that of...
10830         (make_pass_stack_regs): ...new function.
10831         (pass_stack_regs_run): Replace declaration with that of...
10832         (make_pass_stack_regs_run): ...new function.
10833         (pass_df_finish): Replace declaration with that of...
10834         (make_pass_df_finish): ...new function.
10835         (pass_compute_alignments): Replace declaration with that of...
10836         (make_pass_compute_alignments): ...new function.
10837         (pass_duplicate_computed_gotos): Replace declaration with that of...
10838         (make_pass_duplicate_computed_gotos): ...new function.
10839         (pass_variable_tracking): Replace declaration with that of...
10840         (make_pass_variable_tracking): ...new function.
10841         (pass_free_cfg): Replace declaration with that of...
10842         (make_pass_free_cfg): ...new function.
10843         (pass_machine_reorg): Replace declaration with that of...
10844         (make_pass_machine_reorg): ...new function.
10845         (pass_cleanup_barriers): Replace declaration with that of...
10846         (make_pass_cleanup_barriers): ...new function.
10847         (pass_delay_slots): Replace declaration with that of...
10848         (make_pass_delay_slots): ...new function.
10849         (pass_split_for_shorten_branches): Replace declaration with that of...
10850         (make_pass_split_for_shorten_branches): ...new function.
10851         (pass_split_before_regstack): Replace declaration with that of...
10852         (make_pass_split_before_regstack): ...new function.
10853         (pass_convert_to_eh_region_ranges): Replace declaration with that
10854         of...
10855         (make_pass_convert_to_eh_region_ranges): ...new function.
10856         (pass_shorten_branches): Replace declaration with that of...
10857         (make_pass_shorten_branches): ...new function.
10858         (pass_set_nothrow_function_flags): Replace declaration with that of...
10859         (make_pass_set_nothrow_function_flags): ...new function.
10860         (pass_dwarf2_frame): Replace declaration with that of...
10861         (make_pass_dwarf2_frame): ...new function.
10862         (pass_final): Replace declaration with that of...
10863         (make_pass_final): ...new function.
10864         (pass_rtl_seqabstr): Replace declaration with that of...
10865         (make_pass_rtl_seqabstr): ...new function.
10866         (pass_release_ssa_names): Replace declaration with that of...
10867         (make_pass_release_ssa_names): ...new function.
10868         (pass_early_inline): Replace declaration with that of...
10869         (make_pass_early_inline): ...new function.
10870         (pass_inline_parameters): Replace declaration with that of...
10871         (make_pass_inline_parameters): ...new function.
10872         (pass_update_address_taken): Replace declaration with that of...
10873         (make_pass_update_address_taken): ...new function.
10874         (pass_convert_switch): Replace declaration with that of...
10875         (make_pass_convert_switch): ...new function.
10876         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10877         to a subclass of simple_ipa_opt_pass along with...
10878         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10879         (make_pass_ipa_tree_profile): ...new function.
10880         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10881         subclass of gimple_opt_pass along with...
10882         (pass_data_sra_early): ...new pass_data instance and...
10883         (make_pass_sra_early): ...new function.
10884         (pass_sra): Convert from a global struct to a subclass of
10885         gimple_opt_pass along with...
10886         (pass_data_sra): ...new pass_data instance and...
10887         (make_pass_sra): ...new function.
10888         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10889         gimple_opt_pass along with...
10890         (pass_data_early_ipa_sra): ...new pass_data instance and...
10891         (make_pass_early_ipa_sra): ...new function.
10892         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10893         subclass of gimple_opt_pass along with...
10894         (pass_data_ccp): ...new pass_data instance and...
10895         (make_pass_ccp): ...new function.
10896         (pass_fold_builtins): Convert from a global struct to a subclass of
10897         gimple_opt_pass along with...
10898         (pass_data_fold_builtins): ...new pass_data instance and...
10899         (make_pass_fold_builtins): ...new function.
10900         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10901         subclass of gimple_opt_pass along with...
10902         (pass_data_copy_prop): ...new pass_data instance and...
10903         (make_pass_copy_prop): ...new function.
10904         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10905         global struct to a subclass of gimple_opt_pass along with...
10906         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10907         (make_pass_rename_ssa_copies): ...new function.
10908         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10909         subclass of gimple_opt_pass along with...
10910         (pass_data_dce): ...new pass_data instance and...
10911         (make_pass_dce): ...new function.
10912         (pass_dce_loop): Convert from a global struct to a subclass of
10913         gimple_opt_pass along with...
10914         (pass_data_dce_loop): ...new pass_data instance and...
10915         (make_pass_dce_loop): ...new function.
10916         (pass_cd_dce): Convert from a global struct to a subclass of
10917         gimple_opt_pass along with...
10918         (pass_data_cd_dce): ...new pass_data instance and...
10919         (make_pass_cd_dce): ...new function.
10920         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10921         subclass of gimple_opt_pass along with...
10922         (pass_data_dominator): ...new pass_data instance and...
10923         (make_pass_dominator): ...new function.
10924         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10925         gimple_opt_pass along with...
10926         (pass_data_phi_only_cprop): ...new pass_data instance and...
10927         (make_pass_phi_only_cprop): ...new function.
10928         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10929         subclass of gimple_opt_pass along with...
10930         (pass_data_dse): ...new pass_data instance and...
10931         (make_pass_dse): ...new function.
10932         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10933         a subclass of gimple_opt_pass along with...
10934         (pass_data_forwprop): ...new pass_data instance and...
10935         (make_pass_forwprop): ...new function.
10936         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10937         struct to a subclass of gimple_opt_pass along with...
10938         (pass_data_tree_ifcombine): ...new pass_data instance and...
10939         (make_pass_tree_ifcombine): ...new function.
10940         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10941         subclass of gimple_opt_pass along with...
10942         (pass_data_ch): ...new pass_data instance and...
10943         (make_pass_ch): ...new function.
10944         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10945         subclass of gimple_opt_pass along with...
10946         (pass_data_tree_loop): ...new pass_data instance and...
10947         (make_pass_tree_loop): ...new function.
10948         (pass_tree_loop_init): Convert from a global struct to a subclass of
10949         gimple_opt_pass along with...
10950         (pass_data_tree_loop_init): ...new pass_data instance and...
10951         (make_pass_tree_loop_init): ...new function.
10952         (pass_lim): Convert from a global struct to a subclass of
10953         gimple_opt_pass along with...
10954         (pass_data_lim): ...new pass_data instance and...
10955         (make_pass_lim): ...new function.
10956         (pass_tree_unswitch): Convert from a global struct to a subclass of
10957         gimple_opt_pass along with...
10958         (pass_data_tree_unswitch): ...new pass_data instance and...
10959         (make_pass_tree_unswitch): ...new function.
10960         (pass_predcom): Convert from a global struct to a subclass of
10961         gimple_opt_pass along with...
10962         (pass_data_predcom): ...new pass_data instance and...
10963         (make_pass_predcom): ...new function.
10964         (pass_vectorize): Convert from a global struct to a subclass of
10965         gimple_opt_pass along with...
10966         (pass_data_vectorize): ...new pass_data instance and...
10967         (make_pass_vectorize): ...new function.
10968         (pass_graphite): Convert from a global struct to a subclass of
10969         gimple_opt_pass along with...
10970         (pass_data_graphite): ...new pass_data instance and...
10971         (make_pass_graphite): ...new function.
10972         (pass_graphite_transforms): Convert from a global struct to a subclass
10973         of gimple_opt_pass along with...
10974         (pass_data_graphite_transforms): ...new pass_data instance and...
10975         (make_pass_graphite_transforms): ...new function.
10976         (pass_check_data_deps): Convert from a global struct to a subclass of
10977         gimple_opt_pass along with...
10978         (pass_data_check_data_deps): ...new pass_data instance and...
10979         (make_pass_check_data_deps): ...new function.
10980         (pass_iv_canon): Convert from a global struct to a subclass of
10981         gimple_opt_pass along with...
10982         (pass_data_iv_canon): ...new pass_data instance and...
10983         (make_pass_iv_canon): ...new function.
10984         (pass_scev_cprop): Convert from a global struct to a subclass of
10985         gimple_opt_pass along with...
10986         (pass_data_scev_cprop): ...new pass_data instance and...
10987         (make_pass_scev_cprop): ...new function.
10988         (pass_record_bounds): Convert from a global struct to a subclass of
10989         gimple_opt_pass along with...
10990         (pass_data_record_bounds): ...new pass_data instance and...
10991         (make_pass_record_bounds): ...new function.
10992         (pass_complete_unroll): Convert from a global struct to a subclass of
10993         gimple_opt_pass along with...
10994         (pass_data_complete_unroll): ...new pass_data instance and...
10995         (make_pass_complete_unroll): ...new function.
10996         (pass_complete_unrolli): Convert from a global struct to a subclass of
10997         gimple_opt_pass along with...
10998         (pass_data_complete_unrolli): ...new pass_data instance and...
10999         (make_pass_complete_unrolli): ...new function.
11000         (pass_parallelize_loops): Convert from a global struct to a subclass
11001         of gimple_opt_pass along with...
11002         (pass_data_parallelize_loops): ...new pass_data instance and...
11003         (make_pass_parallelize_loops): ...new function.
11004         (pass_loop_prefetch): Convert from a global struct to a subclass of
11005         gimple_opt_pass along with...
11006         (pass_data_loop_prefetch): ...new pass_data instance and...
11007         (make_pass_loop_prefetch): ...new function.
11008         (pass_iv_optimize): Convert from a global struct to a subclass of
11009         gimple_opt_pass along with...
11010         (pass_data_iv_optimize): ...new pass_data instance and...
11011         (make_pass_iv_optimize): ...new function.
11012         (pass_tree_loop_done): Convert from a global struct to a subclass of
11013         gimple_opt_pass along with...
11014         (pass_data_tree_loop_done): ...new pass_data instance and...
11015         (make_pass_tree_loop_done): ...new function.
11016         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
11017         struct to a subclass of gimple_opt_pass along with...
11018         (pass_data_cse_reciprocals): ...new pass_data instance and...
11019         (make_pass_cse_reciprocals): ...new function.
11020         (pass_cse_sincos): Convert from a global struct to a subclass of
11021         gimple_opt_pass along with...
11022         (pass_data_cse_sincos): ...new pass_data instance and...
11023         (make_pass_cse_sincos): ...new function.
11024         (pass_optimize_bswap): Convert from a global struct to a subclass of
11025         gimple_opt_pass along with...
11026         (pass_data_optimize_bswap): ...new pass_data instance and...
11027         (make_pass_optimize_bswap): ...new function.
11028         (pass_optimize_widening_mul): Convert from a global struct to a
11029         subclass of gimple_opt_pass along with...
11030         (pass_data_optimize_widening_mul): ...new pass_data instance and...
11031         (make_pass_optimize_widening_mul): ...new function.
11032         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
11033         subclass of gimple_opt_pass along with...
11034         (pass_data_phiopt): ...new pass_data instance and...
11035         (make_pass_phiopt): ...new function.
11036         (pass_cselim): Convert from a global struct to a subclass of
11037         gimple_opt_pass along with...
11038         (pass_data_cselim): ...new pass_data instance and...
11039         (make_pass_cselim): ...new function.
11040         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
11041         subclass of gimple_opt_pass along with...
11042         (pass_data_phiprop): ...new pass_data instance and...
11043         (make_pass_phiprop): ...new function.
11044         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
11045         subclass of gimple_opt_pass along with...
11046         (pass_data_pre): ...new pass_data instance and...
11047         (make_pass_pre): ...new function.
11048         (pass_fre): Convert from a global struct to a subclass of
11049         gimple_opt_pass along with...
11050         (pass_data_fre): ...new pass_data instance and...
11051         (make_pass_fre): ...new function.
11052         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
11053         subclass of gimple_opt_pass along with...
11054         (pass_data_reassoc): ...new pass_data instance and...
11055         (make_pass_reassoc): ...new function.
11056         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
11057         subclass of gimple_opt_pass along with...
11058         (pass_data_sink_code): ...new pass_data instance and...
11059         (make_pass_sink_code): ...new function.
11060         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
11061         subclass of gimple_opt_pass along with...
11062         (pass_data_strlen): ...new pass_data instance and...
11063         (make_pass_strlen): ...new function.
11064         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
11065         struct to a subclass of gimple_opt_pass along with...
11066         (pass_data_build_alias): ...new pass_data instance and...
11067         (make_pass_build_alias): ...new function.
11068         (pass_build_ealias): Convert from a global struct to a subclass of
11069         gimple_opt_pass along with...
11070         (pass_data_build_ealias): ...new pass_data instance and...
11071         (make_pass_build_ealias): ...new function.
11072         (pass_ipa_pta): Convert from a global struct to a subclass of
11073         simple_ipa_opt_pass along with...
11074         (pass_data_ipa_pta): ...new pass_data instance and...
11075         (make_pass_ipa_pta): ...new function.
11076         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11077         subclass of gimple_opt_pass along with...
11078         (pass_data_uncprop): ...new pass_data instance and...
11079         (make_pass_uncprop): ...new function.
11080         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11081         global struct to a subclass of gimple_opt_pass along with...
11082         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11083         (make_pass_late_warn_uninitialized): ...new function.
11084         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11085         to a subclass of gimple_opt_pass along with...
11086         (pass_data_init_datastructures): ...new pass_data instance and...
11087         (make_pass_init_datastructures): ...new function.
11088         (pass_early_warn_uninitialized): Convert from a global struct to a
11089         subclass of gimple_opt_pass along with...
11090         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11091         (make_pass_early_warn_uninitialized): ...new function.
11092         (pass_update_address_taken): Convert from a global struct to a
11093         subclass of gimple_opt_pass along with...
11094         (pass_data_update_address_taken): ...new pass_data instance and...
11095         (make_pass_update_address_taken): ...new function.
11096         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11097         struct to a subclass of gimple_opt_pass along with...
11098         (pass_data_release_ssa_names): ...new pass_data instance and...
11099         (make_pass_release_ssa_names): ...new function.
11100         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11101         subclass of gimple_opt_pass along with...
11102         (pass_data_stdarg): ...new pass_data instance and...
11103         (make_pass_stdarg): ...new function.
11104         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11105         global struct to a subclass of gimple_opt_pass along with...
11106         (pass_data_convert_switch): ...new pass_data instance and...
11107         (make_pass_convert_switch): ...new function.
11108         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11109         to a subclass of gimple_opt_pass along with...
11110         (pass_data_tail_recursion): ...new pass_data instance and...
11111         (make_pass_tail_recursion): ...new function.
11112         (pass_tail_calls): Convert from a global struct to a subclass of
11113         gimple_opt_pass along with...
11114         (pass_data_tail_calls): ...new pass_data instance and...
11115         (make_pass_tail_calls): ...new function.
11116         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11117         struct to a subclass of gimple_opt_pass along with...
11118         (pass_data_lower_vector): ...new pass_data instance and...
11119         (make_pass_lower_vector): ...new function.
11120         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11121         gimple_opt_pass along with...
11122         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11123         (make_pass_lower_vector_ssa): ...new function.
11124         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11125         to a subclass of gimple_opt_pass along with...
11126         (pass_data_slp_vectorize): ...new pass_data instance and...
11127         (make_pass_slp_vectorize): ...new function.
11128         (pass_ipa_increase_alignment): Convert from a global struct to a
11129         subclass of simple_ipa_opt_pass along with...
11130         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11131         (make_pass_ipa_increase_alignment): ...new function.
11132         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11133         gimple_opt_pass along with...
11134         (pass_data_vrp): ...new pass_data instance and...
11135         (make_pass_vrp): ...new function.
11136         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11137         subclass of simple_ipa_opt_pass along with...
11138         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11139         (make_pass_ipa_free_lang_data): ...new function.
11140         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11141         gimple_opt_pass along with...
11142         (pass_data_tsan): ...new pass_data instance and...
11143         (make_pass_tsan): ...new function.
11144         (pass_tsan_O0): Convert from a global struct to a subclass of
11145         gimple_opt_pass along with...
11146         (pass_data_tsan_O0): ...new pass_data instance and...
11147         (make_pass_tsan_O0): ...new function.
11148         * var-tracking.c (pass_variable_tracking): Convert from a global
11149         struct to a subclass of rtl_opt_pass along with...
11150         (pass_data_variable_tracking): ...new pass_data instance and...
11151         (make_pass_variable_tracking): ...new function.
11152         * web.c (pass_web): Convert from a global struct to a subclass of
11153         rtl_opt_pass along with...
11154         (pass_data_web): ...new pass_data instance and...
11155         (make_pass_web): ...new function.
11156         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11157         declaration with that of...
11158         (make_pass_mode_switch_use): ...new function.
11159         (pass_resolve_sw_modes): Replace declaration with that of...
11160         (make_pass_resolve_sw_modes): ...new function.
11161         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11162         from a global struct to a subclass of rtl_opt_pass along with...
11163         (pass_data_mode_switch_use): ...new pass_data instance and...
11164         (make_pass_mode_switch_use): ...new function.
11165         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11166         from a global struct to a subclass of rtl_opt_pass along with...
11167         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11168         (make_pass_resolve_sw_modes): ...new function.
11169         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11170         struct to a subclass of rtl_opt_pass along with...
11171         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11172         (make_pass_insert_vzeroupper): ...new function.
11173         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11174         global struct to a subclass of rtl_opt_pass along with...
11175         (pass_data_work_around_errata): ...new pass_data instance and...
11176         (make_pass_work_around_errata): ...new function.
11177         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11178         struct to a subclass of rtl_opt_pass along with...
11179         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11180         (make_pass_mips_machine_reorg2): ...new function.
11182 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11184         * passes.c (pass_manager::operator new): New.
11186 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11188         Handwritten part of conversion of passes to C++ classes.
11190         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11191         (toplev.o): Add dep on PASS_MANAGER_H.
11192         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11193         of early local pases to reflect this moving from a global to a
11194         member of gcc::pass_manager.
11195         (cgraph_add_new_function): Likewise.
11196         * lto-cgraph.c (lto_output_node): Update for conversion of
11197         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11198         * passes.c (opt_pass::clone): New.
11199         (opt_pass::gate): New.
11200         (opt_pass::execute): New.
11201         (opt_pass::opt_pass): New.
11202         (pass_manager::execute_early_local_passes): New.
11203         (pass_manager::execute_pass_mode_switching): new.
11204         (finish_optimization_passes): Convert to...
11205         (pass_manager::finish_optimization_passes): ...this.
11206         (finish_optimization_passes): Update for conversion of passes to
11207         C++ classes.
11208         (register_dump_files_1): Use has_gate since we cannot portably
11209         check a vtable entry against NULL.
11210         (dump_one_pass): Likewise.
11211         (ipa_write_summaries_2): Likewise.
11212         (ipa_write_optimization_summaries_1): Likewise.
11213         (ipa_read_summaries_1): Likewise.
11214         (ipa_read_optimization_summaries_1): Likewise.
11215         (execute_ipa_stmt_fixups): Likewise.
11216         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11217         pass-creation functions rather than wiring up globals, and
11218         storing the results in fields of pass_manager generated using
11219         pass-instances.def.
11220         (pass_manager::dump_profile_report): Update for conversion of
11221         passes to C++ classes.
11222         (pass_manager::execute_ipa_summary_passes): Likewise.
11223         (execute_one_ipa_transform_pass): Likewise.
11224         (execute_one_pass): Use has_gate and has_execute since we cannot
11225         portably check a vtable entry against NULL.
11226         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11227         (pass_manager): Use pass-instances.def to add fields for the
11228         various pass instances.
11229         * toplev.c (finalize): Update for move of
11230         finish_optimization_passes to a method of gcc::pass_manager.
11231         * toplev.h (finish_optimization_passes): Move to method of class
11232         pass_manager.
11233         * tree-pass.h (struct pass_data): New.
11234         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11235         (opt_pass::gate): Convert to virtual function.
11236         (opt_pass::~opt_pass): New.
11237         (opt_pass::clone): New.
11238         (opt_pass::execute): Convert to virtual function.
11239         (opt_pass::opt_pass): New.
11240         (opt_pass::ctxt_): new.
11241         (gimple_opt_pass): Convert to subclass of opt_pass.
11242         (gimple_opt_pass::gimple_opt_pass): New.
11243         (rtl_opt_pass): Convert to subclass of opt_pass.
11244         (rtl_opt_pass::rtl_opt_pass): New.
11245         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11246         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11247         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11248         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11249         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11250         invocation of pass_mode_switching to reflect this moving from a
11251         global to a member of gcc::pass_manager.
11252         (ix86_option_override): Rework how pass_insert_vzeroupper is
11253         added to the pass_manager to reflect autogenerated changes.
11254         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11256 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11258         PR rtl-optimization/57708
11259         * recog.c (peep2_find_free_register): Validate all regs in a
11260         multi-reg mode.
11262 2013-08-05  Jan Hubicka  <jh@suse.cz>
11264         PR lto/57602
11265         * cgraph.c (verify_cgraph_node): Accept local flags from other
11266         partitions.
11267         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11268         (function_and_variable_visibility): Likewise.
11269         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11271 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11273         * graph.c (init_graph_slim_pretty_print): Remove.
11274         (print_graph_cfg): Do not call it.  Use local pretty printer.
11275         (start_graph_dump): Likewise.
11277 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11279         * gimple-pretty-print.c (buffer): Remove.
11280         (initialized): Likewise.
11281         (maybe_init_pretty_print): Likewise.
11282         (print_gimple_stmt): Do not call it.  Use non-static local
11283         pretty_printer variable.
11284         (print_gimple_expr): Likewise.
11285         (print_gimple_seq): Likewise.
11286         (gimple_dump_bb): Likewise.
11288 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11290         * asan.c (asan_pp): Remove.
11291         (asan_pp_initialized): Likewise.
11292         (asan_pp_initialize): Likewise.
11293         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11294         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11295         (asan_add_global): Likewise.
11297 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11299         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11300         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11301         * pretty-print.c (pp_formatted_text_data): Likewise.
11302         (pp_indent): Rename from pp_base_indent.
11303         (pp_format): Rename from pp_base_format.
11304         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11305         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11306         (pp_flush): Rename from pp_base_flush.
11307         (pp_set_line_maximum_length): Rename from
11308         pp_base_set_line_maximum_length.
11309         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11310         (pp_set_prefix): Rename from pp_base_set_prefix.
11311         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11312         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11313         (pp_append_text): Rename from pp_base_append_text.
11314         (pp_formatted_text): Rename from pp_base_formatted_text.
11315         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11316         (pp_remaining_character_count_for_line): Rename from
11317         pp_base_remaining_character_count_for_line.
11318         (pp_newline): Rename from pp_base_newline.
11319         (pp_character): Rename from pp_base_character.
11320         (pp_string): Rename from pp_base_string.
11321         (pp_maybe_space): Rename from pp_base_maybe_space.
11322         * asan.c (asan_pp_string): Adjust.
11323         (asan_emit_stack_protection): Likewise.
11324         (asan_add_global): Likewise.
11325         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11326         * tree-mudflap.c (mf_varname_tree): Likewise.
11327         * tree-pretty-print.c (pp_tree_identifier): Rename from
11328         pp_base_tree_identifier.
11329         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11330         Declare as function.
11332 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11334         * pretty-print.h (pp_bar_bar): New.
11335         (pp_ampersand_ampersand): Likewise.
11336         (pp_less_equal): Likewise.
11337         (pp_greater_equal): Likewise.
11338         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11339         printer functions instead of pp_string or operators and punctuators.
11340         (dump_gimple_call): Likewise.
11341         (dump_gimple_omp_for): Likewise.
11342         (dump_gimple_transaction): Likewise.
11343         (dump_gimple_phi): Likewise.
11344         (pp_gimple_stmt_1): Likewise.
11345         * sched-vis.c (print_insn): Likewise.
11346         * tree-mudflap.c (mf_varname_tree): Likewise.
11347         * tree-pretty-print.c (dump_block_node): Likewise.
11348         (dump_generic_node): Likewise.
11350 2013-08-02  Jan Hubicka  <jh@suse.cz>
11352         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11353         boundaries.
11354         * lto-streamer-out.c (tree_is_indexable): Results decls and
11355         parm decls are not indexable.
11356         (DFS_write_tree_body): Do not follow args and results.
11357         (hash_tree): Likewise.
11358         (output_functions): Rearrange so struct function is needed
11359         only when real body is output; be able to also ouptut abstract
11360         functions; output DECL_ARGUMENTS and DECL_RESULT.
11361         (lto_output): When not in WPA, ale store abstract functions.
11362         (write_symbol): Do not care about RESULT_DECL.
11363         (output_symbol_p): Handle correctly sbtract decls.
11364         * lto-streamer-in.c (input_function): Rearrange so struct
11365         function can be NULL at entry; allow streaming of
11366         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11367         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11368         sanity check during LTO.
11369         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11370         RESULT_DECl and DECL_ARGUMENTS.
11371         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11372         Likewise.
11374 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11376         * pretty-print.h (pp_underscore): New.
11377         (pp_comma): Tidy.
11378         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11379         printer functions instead of pp_character.
11380         (dump_binary_rhs): Likewise.
11381         (dump_ternary_rhs): Likewise.
11382         (dump_gimple_call_args): Likewise.
11383         (pp_points_to_solution): Likewise.
11384         (dump_gimple_call): Likewise.
11385         (dump_gimple_switch): Likewise.
11386         (dump_gimple_cond): Likewise.
11387         (dump_gimple_bind): Likewise.
11388         (dump_gimple_try): Likewise.
11389         (dump_gimple_omp_for): Likewise.
11390         (dump_gimple_omp_continue): Likewise.
11391         (dump_gimple_omp_single): Likewise.
11392         (dump_gimple_omp_sections): Likewise.
11393         (dump_gimple_omp_block): Likewise.
11394         (dump_gimple_omp_critical): Likewise.
11395         (dump_gimple_transaction): Likewise.
11396         (dump_gimple_asm): Likewise.
11397         (dump_gimple_phi): Likewise.
11398         (dump_gimple_omp_parallel): Likewise.
11399         (dump_gimple_omp_task): Likewise.
11400         (dump_gimple_omp_atomic_load): Likewise.
11401         (dump_gimple_omp_atomic_store): Likewise.
11402         (dump_gimple_mem_ops): Likewise.
11403         (pp_gimple_stmt_1): Likewise.
11404         (pp_cfg_jump): Likewise.
11405         (dump_implicit_edges): Likewise.
11406         (gimple_dump_bb_for_graph): Likewise.
11407         * graph.c (draw_cfg_node): Likewise.
11408         * langhooks.c (lhd_print_error_function): Likewise.
11409         * sched-vis.c (print_exp): Likewise.
11410         (print_value): Likewise.
11411         (print_pattern): Likewise.
11412         (print_insn): Likewise.
11413         (rtl_dump_bb_for_graph): Likewise.
11414         * tree-pretty-print.c (dump_function_declaration): Likewise.
11415         (dump_array_domain): Likewise.
11416         (dump_omp_clause): Likewise.
11417         (dump_location): Likewise.
11418         (dump_generic_node): Likewise.
11419         (print_struct_decl): Likewise.
11420         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11422 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11424         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11425         candidate table when replacing a candidate statement.
11426         (replace_rhs_if_not_dup): Likewise.
11427         (replace_one_candidate): Likewise.
11429 2013-08-02  Jan Hubicka  <jh@suse.cz>
11430             Martin Liska  <marxin.liska@gmail.com>
11432         * cgraphunit.c (add_new_function): Fix logic when adding from
11433         late IPA pass.
11434         (assemble_thunk): Rename to ...
11435         (expand_thunk); .. this one; export; get it working with
11436         general functions; make produced gimple valid.
11437         * cgraph.h (expand_thunk): Declare.
11439 2013-08-02  Jan Hubicka  <jh@suse.cz>
11441         * ipa-cp.c (gather_context_independent_values): Use
11442         ipa_get_param_move_cost.
11443         (get_replacement_map): Remove PARAM; move parameter folding
11444         into tree-inline.c
11445         (create_specialized_node): Update.
11446         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11447         assert that we have gimple body; update move_cost.
11448         (count_formal_params): Assert that we have gimple body.
11449         (ipa_dump_param): New function.
11450         (ipa_alloc_node_params): Break out from ...
11451         (ipa_initialize_node_params): ... here.
11452         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11453         (ipa_write_node_info): Stream move costs.
11454         (ipa_read_node_info): Read move costs.
11455         (ipa_update_after_lto_read): Do not recompute node params.
11456         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11457         (ipa_get_param): Check we are not in WPA.
11458         (ipa_get_param_move_cost): New.
11459         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11460         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11461         parm numbers to be present.
11463 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11465         PR rtl-optimization/58048
11466         * lra-constraints.c (process_alt_operands): Don't check asm
11467         operand on register.
11469 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11471         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11472         the implied StoreLoad barrier for atomic operations if before.
11474 2013-08-02  Jan Hubicka  <jh@suse.cz>
11475             Martin Liska  <marxin.liska@gmail.com>
11477         * cgraph.c (cgraph_function_body_availability): Do not check
11478         cgraph flags.
11479         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11480         symtab_node_availability): Declare.
11481         * ipa.c (can_replace_by_local_alias): New.
11482         (function_and_variable_visibility): Use it.
11483         * symtab.c (symtab_for_node_and_aliases,
11484         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11486 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11488         PR rtl-optimization/57963
11489         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11490         (lra_constraints): Use them.
11492 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11494         * config/arm/types.md (define_attr "type"): Add "load_acq"
11495         and "store_rel".
11496         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11497         changes.
11498         (cortex_a53_store1): Likewise.
11500 2013-08-01  Jan Hubicka  <jh@suse.cz>
11502         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11503         partitions are not needed.
11505 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11507         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11508         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11509         MAYBE_NON_Q_CLASS_P where appropriate.
11511 2013-08-01  Jan Hubicka  <jh@suse.cz>
11513         * cgraph.h (release_function_body): Declare.
11514         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11515         of unused delcarations.
11517 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11519         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11520         RTL form when subtracting a constant.
11522 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11524         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11525         Generate canonical plus rtx with negated immediate instead of minus
11526         where appropriate.
11527         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11529 2013-08-01  Jan Hubicka  <jh@suse.cz>
11531         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11532         (cgraph_release_function_body): Likewise.
11533         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11534         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11535         to used_as_abstract_origin.
11536         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11537         symbols used as abstract origins.
11538         * cgraphunit.c (analyze_functions): Update.
11539         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11540         used_as_abstract_origin.
11541         * tree-inline.c (tree_function_versioning): Update
11542         used_as_abstract_origin; be ready for DECL_RESULT and
11543         DECL_ARGUMENTS to be NULL.
11545         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11546         for abstract functions.
11547         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11548         real symbols.
11550 2013-08-01  Jan Hubicka  <jh@suse.cz>
11552         * profile.c (compute_value_histograms): Fix thinko.
11554 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11556         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11557         aarch-common-protos.h to extra_headers.
11558         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11559         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11560         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11562 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11564         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11565         Include "../arm/types.md".  Define "type" attribute for all patterns.
11566         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11567         attribute changes.
11569 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11571         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11572         to support power8 load fusion.
11573         (fusion_gpr_mem_load): Likewise.
11575         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11577         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11578         declarations for power8 load fusion.
11579         (emit_fusion_gpr_load): Likewise.
11581         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11582         tuning for power8, turn on fusion mode by default.  Turn on sign
11583         extending fusion mode if normal fusion mode is on, and we are at
11584         -O2 or -O3.
11585         (fusion_gpr_load_p): New function, return true if we can fuse an
11586         addis instruction with a dependent load to a GPR.
11587         (emit_fusion_gpr_load): Emit the instructions for power8 load
11588         fusion to GPRs.
11590         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11591         (VSX load fusion peepholes): New peepholes to fuse together an
11592         addi instruction with a VSX load instruction.
11594         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11595         peepholes to fuse an addis instruction with a load to a GPR base
11596         register.  If we are supporting sign extending fusions, convert
11597         sign extending loads to zero extending loads and add an explicit
11598         sign extension.
11600 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11602         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11603         aarch-common-protos.h to extra_headers.
11604         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11605         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11606         (arm_early_store_addr_dep): Likewise.
11607         (arm_no_early_alu_shift_dep): Likewise.
11608         (arm_no_early_alu_shift_value_dep): Likewise.
11609         (arm_no_early_mul_dep): Likewise.
11610         (arm_no_early_store_addr_dep): Likewise.
11611         (arm_mac_accumulator_is_mul_result): Likewise.
11612         (arm_mac_accumulator_is_result): Likewise.
11613         * config/arm/aarch-common.c: ... here.  New file.
11614         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11615         here to ...
11616         (arm_early_store_addr_dep): Likewise.
11617         (arm_no_early_alu_shift_dep): Likewise.
11618         (arm_no_early_alu_shift_value_dep): Likewise.
11619         (arm_no_early_mul_dep): Likewise.
11620         (arm_no_early_store_addr_dep): Likewise.
11621         (arm_mac_accumulator_is_mul_result): Likewise.
11622         (arm_mac_accumulator_is_result): Likewise.
11623         * config/arm/aarch-common-protos.h: ... here.  New file.
11624         * config/arm/t-arm (aarch-common.o): Define.
11626 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11628         * config/arm/arm.md: Include new file "types.md".
11629         (define_attr "type"): Move from here to ...
11630         (define_attr "mul32"): Likewise.
11631         (define_attr "mul64"): Likewise.
11632         * config/arm/types.md: ... here.  New file.
11634 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11636         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11637         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11639 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11641         * gen-pass-instances.awk: Fix offset of substr().
11643 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11645         * Makefile.in (pass-instances.def): New.
11646         (passes.o): Replace dependency on passes.def with one on
11647         pass-instances.def
11649         * gen-pass-instances.awk: New.
11651         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11652         rather than passes.def, updating local definition of NEXT_PASS
11653         macro to add an extra NUM parameter (currently unused).
11655 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11657         * Makefile.in (PASS_MANAGER_H): New.
11658         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11659         (passes.o): Likewise.
11660         (statistics.o): Likewise.
11661         (cgraphunit.o): Likewise.
11662         (context.o): Depend on PASS_MANAGER_H.
11664         * pass_manager.h: New.
11666         * cgraphunit.c (cgraph_add_new_function): Update for moves
11667         of globals to fields of pass_manager.
11668         (analyze_function): Likewise.
11669         (expand_function): Likewise.
11670         (ipa_passes): Likewise.
11671         (compile): Likewise.
11673         * context.c (context::context): New.
11674         * context.h  (context::context): New.
11675         (context::get_passes): New.
11676         (context::passes_): New.
11678         * lto-cgraph.c (input_node): Update for moves of globals to
11679         fields of pass_manager.
11681         * passes.c (all_passes): Remove, in favor of a field of the
11682         same name within the new class pass_manager.
11683         (all_small_ipa_passes): Likewise.
11684         (all_lowering_passes): Likewise.
11685         (all_regular_ipa_passes): Likewise.
11686         (all_late_ipa_passes): Likewise.
11687         (all_lto_gen_passes): Likewise.
11688         (passes_by_id): Likewise.
11689         (passes_by_id_size): Likewise.
11690         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11691         the new class pass_manager.
11692         (set_pass_for_id): Convert to...
11693         (pass_manager::set_pass_for_id): ...method.
11694         (get_pass_for_id): Convert to...
11695         (pass_manager::get_pass_for_id): ...method.
11696         (register_one_dump_file): Move body of implementation into...
11697         (pass_manager::register_one_dump_file): ...here.
11698         (register_dump_files_1): Convert to...
11699         (pass_manager::register_dump_files_1): ...method.
11700         (register_dump_files): Convert to...
11701         (pass_manager::register_dump_files): ...method.
11702         (create_pass_tab): Update for moves of globals to fields of
11703         pass_manager.
11704         (dump_passes): Move body of implementation into...
11705         (pass_manager::dump_passes): ...here.
11706         (register_pass): Move body of implementation into...
11707         (pass_manager::register_pass): ...here.
11708         (init_optimization_passes): Convert into...
11709         (pass_manager::pass_manager): ...constructor for new
11710         pass_manager class, and initialize the pass_lists array.
11711         (check_profile_consistency): Update for moves of globals to
11712         fields of pass_manager.
11713         (dump_profile_report): Move body of implementation into...
11714         (pass_manager::dump_profile_report): ...here.
11715         (ipa_write_summaries_1): Update for moves of pass lists from
11716         being globals to fields of pass_manager.
11717         (ipa_write_optimization_summaries): Likewise.
11718         (ipa_read_summaries):  Likewise.
11719         (ipa_read_optimization_summaries): Likewise.
11720         (execute_all_ipa_stmt_fixups): Likewise.
11722         * statistics.c (statistics_fini): Update for moves of globals to
11723         fields of pass_manager.
11725         * toplev.c (general_init): Replace call to
11726         init_optimization_passes with construction of the pass_manager
11727         instance.
11729         * tree-pass.h (all_passes): Remove, in favor of a field of the
11730         same name within the new class pass_manager.
11731         (all_small_ipa_passes): Likewise.
11732         (all_lowering_passes): Likewise.
11733         (all_regular_ipa_passes): Likewise.
11734         (all_lto_gen_passes): Likewise.
11735         (all_late_ipa_passes): Likewise.
11736         (passes_by_id): Likewise.
11737         (passes_by_id_size): Likewise.
11738         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11739         the new class pass_manager.
11740         (get_pass_for_id): Remove.
11742 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11744         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11745         configs.
11747 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11749         * arm.md (mulhi3): New expand pattern.
11751 2013-07-30  Jan Hubicka  <jh@suse.cz>
11752             Martin Liska  <marxin.liska@gmail.com>
11754         * profile.c (compute_value_histograms): Do not ICE when
11755         there is mismatch only on some counters.
11757 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11759         PR rtl-optimization/57637
11760         * function.c (move_insn_for_shrink_wrap): Also check the
11761         GEN set of the LIVE problem for the liveness analysis
11762         if it exists, otherwise give up.
11764 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11766         PR tree-optimization/57993
11767         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11768         replaced statement in the candidate table.
11769         (phi_add_costs): Return infinite cost when the hidden basis does
11770         not dominate all phis on which the candidate is dependent.
11771         (replace_one_candidate): Record replaced statement in the
11772         candidate table.
11774 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11776         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11777         (ashlv2si3): New expander.
11778         (*ashlv2si3_i): New define_insn_and_split.
11779         * predicates.md (float_operation): Allow patterns with three
11780         basic sub-patterns.
11782         PR rtl-optimization/58021
11783         * mode-switching.c (create_pre_exit): Always split off preceding
11784         insns if we are not at the basic block head.
11786 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11788         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11789         (UCLIBC_DYNAMIC_LINKER): New macro.
11790         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11791         `-mnan=2008'.
11792         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11793         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11794         `-mnan=2008'.
11795         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11796         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11797         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11798         for SF and DF modes.  Use ieee_quad_format for TF mode.
11799         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11800         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11801         (mips_option_override): Handle `-mnan=legacy'.
11802         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11803         `-mabs=2008' and `-mnan=2008'.
11804         (OPTION_DEFAULT_SPECS): Add "nan" default.
11805         (ASM_SPEC): Handle `-mnan='.
11806         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11807         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11808         comment accordingly.
11809         (neg<mode>2): Likewise.
11810         * config/mips/mips.opt (mabs, mnan): New options.
11811         * doc/install.texi (Configuration): Document `--with-nan=' option.
11812         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11813         `-mnan=' options.
11814         (MIPS Options): Document them.
11815         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11816         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11817         * configure: Regenerate.
11818         * config.in: Regenerate.
11820 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11822         * config/i386/i386.md (float post-reload splitters): Do not check
11823         for subregs of SSE registers.
11825 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11826             H.J. Lu  <hongjiu.lu@intel.com>
11828         PR target/57954
11829         PR target/57988
11830         * config/i386/i386.md (post-reload splitter
11831         to avoid partial SSE reg dependency stalls): New pattern.
11833 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11835         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11837 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11839         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11840         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11842 2013-07-26  Cary Coutant  <ccoutant@google.com>
11844         * dwarf2out.c (die_checksum_ordered): Don't include template
11845         instantiations in signature.
11846         (is_template_parameter): New function.
11847         (is_template_instantiation): New function.
11848         (generate_skeleton_bottom_up): Don't include template instantiations
11849         in type unit DIE.
11850         (generate_skeleton): Likewise.
11851         (break_out_comdat_types): Move recursive call to break out nested
11852         types earlier.
11853         (prune_unused_types_mark_generic_parms_dies): Call
11854         is_template_parameter.
11856 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11858         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11859         uses vector registers.
11860         * config/aarch64/iterators.md: Add attributes rtn and vas.
11862 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11863             Richard Earnshaw  <richard.earnshaw@arm.com>
11865         * combine.c (simplify_comparison): Re-canonicalize operands
11866         where appropriate.
11867         * config/arm/arm.md (movcond_addsi): New splitter.
11869 2013-07-25  Sterling Augustine  <saugustine@google.com>
11871         * dwarf2out.c (size_of_pubnames): Move code to...
11872         (include_pubname_in_output): ...here.  New.
11873         (want_pubnames): Rearrange.
11874         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11876 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11878         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11880 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11882         PR target/38836
11883         * doc/extend.texi: Remove obsolete builtins. Fix
11884         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11886 2013-07-25  Jan Hubicka  <jh@suse.cz>
11888         * cgraph.c (release_function_body): Break out from ...
11889         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11890         and DECL_ARGUMENTS.
11891         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11892         old_tree in the map.
11893         (create_specialized_node): Update.
11894         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11895         into index.
11896         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11897         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11898         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11899         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11900         DECL_RESULT.
11902 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11904         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11905         addsi3_carryin_alt2_<optab>): Correct output template.
11907 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11909         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11910         Adjust for arm_restrict_it.
11911         Remove trailing whitespace.
11913 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11915         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11916         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11918         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11920 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11922         PR rtl-optimization/57960
11923         * lra-constraints.c (process_alt_operands): Use the right mode
11924         when checking strict_low.
11926 2013-07-25  Jan Hubicka  <jh@suse.cz>
11928         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11929         * cgraph.c (cgraph_remove_node): Do not release function body
11930         when in cgraph streaming.
11931         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11932         in other partitions are not considered reachable; fix handling of
11933         clones.
11935 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11937         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11939 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11941         PR target/19599
11942         PR target/57731
11943         PR target/57837
11944         * config/arm/arm.md ("*sibcall_insn): Replace use of
11945         Ss with US. Adjust output for v5 and v4t.
11946         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11948         * config/arm/constraints.md ("Ss"): Rename to US.
11950 2013-07-25  Terry Guo  <terry.guo@arm.com>
11952         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11953         shift_add/shift_sub0/shift_sub1 RTXs.
11955 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11956             Anton Blanchard  <anton@au1.ibm.com>
11958         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11959         (altivec_vpks<VI_char>ss): Likewise.
11960         (altivec_vpks<VI_char>us): Likewise.
11961         (altivec_vpku<VI_char>us): Likewise.
11962         (altivec_vpku<VI_char>um): Likewise.
11964 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11966         Introduce context class.
11968         * Makefile.in (CONTEXT_H): New.
11969         (OBJS): Add context.o.
11970         (toplev.o): Add CONTEXT_H to dependencies.
11971         (context.o): New.
11973         * toplev.c (general_init): Create the singleton gcc::context instance.
11975         * context.c: New.
11977         * context.h: New.
11979 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11981         PR rtl-optimization/57968
11982         * mode-switching.c (create_pre_exit): Allow instructions that
11983         don't set a return register to need a non-exit mode.
11985 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11986             Anton Blanchard  <anton@au1.ibm.com>
11988         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11989         operands to vperm for little endian.
11990         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11991         of lvsl to create the control mask for a vperm for little endian.
11993 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11994             Anton Blanchard  <anton@au1.ibm.com>
11996         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11997         two operands for little-endian.
11999 2013-07-23  Steve Ellcey  <sellcey@mips.com>
12001         * config/mips/mips.c (mips_case_values_threshold): New.
12002         (TARGET_CASE_VALUES_THRESHOLD): Define.
12004 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12005             Anton Blanchard  <anton@au1.ibm.com>
12007         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
12008         selection of field for vector splat in little endian mode.
12010 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
12012         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
12013         expanders to rs6000.md.
12014         (ior<mode>3): Likewise.
12015         (and<mode>3): Likewise.
12016         (one_cmpl<mode>2): Likewise.
12017         (nor<mode>3): Likewise.
12018         (andc<mode>3): Likewise.
12019         (eqv<mode>3): Likewise.
12020         (nand<mode>3): Likewise.
12021         (orc<mode>3): Likewise.
12023         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
12024         declaration.
12026         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
12027         to split multi-word logical operations.
12028         (rs6000_split_logical_di): Likewise.
12029         (rs6000_split_logical): Likewise.
12031         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
12032         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
12033         and allow TImode operations in 32-bit.
12034         (vsx_and<mode>3_64bit): Likewise.
12035         (vsx_ior<mode>3_32bit): Likewise.
12036         (vsx_ior<mode>3_64bit): Likewise.
12037         (vsx_xor<mode>3_32bit): Likewise.
12038         (vsx_xor<mode>3_64bit): Likewise.
12039         (vsx_one_cmpl<mode>2_32bit): Likewise.
12040         (vsx_one_cmpl<mode>2_64bit): Likewise.
12041         (vsx_nor<mode>3_32bit): Likewise.
12042         (vsx_nor<mode>3_64bit): Likewise.
12043         (vsx_andc<mode>3_32bit): Likewise.
12044         (vsx_andc<mode>3_64bit): Likewise.
12045         (vsx_eqv<mode>3_32bit): Likewise.
12046         (vsx_eqv<mode>3_64bit): Likewise.
12047         (vsx_nand<mode>3_32bit): Likewise.
12048         (vsx_nand<mode>3_64bit): Likewise.
12049         (vsx_orc<mode>3_32bit): Likewise.
12050         (vsx_orc<mode>3_64bit): Likewise.
12052         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
12053         logical types in GPRs.
12055         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
12056         logical insns to rs6000.md, and allow TImode operations in
12057         32-bit.
12058         (altivec_ior<mode>3): Likewise.
12059         (altivec_xor<mode>3): Likewise.
12060         (altivec_one_cmpl<mode>2): Likewise.
12061         (altivec_nor<mode>3): Likewise.
12062         (altivec_andc<mode>3): Likewise.
12064         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
12065         attributes for moving the 128-bit logical operations into
12066         rs6000.md.
12067         (BOOL_REGS_OUTPUT): Likewise.
12068         (BOOL_REGS_OP1): Likewise.
12069         (BOOL_REGS_OP2): Likewise.
12070         (BOOL_REGS_UNARY): Likewise.
12071         (BOOL_REGS_AND_CR0): Likewise.
12072         (one_cmpl<mode>2): Add support for DI logical operations on
12073         32-bit, splitting the operations to 32-bit.
12074         (anddi3): Likewise.
12075         (iordi3): Likewise.
12076         (xordi3): Likewise.
12077         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12078         changes to combine the 32/64-bit code, allow logical operations on
12079         TI mode in 32-bit, and to use similar match_operator patterns like
12080         scalar mode uses.  Combine the Altivec and VSX code for logical
12081         operations, and move it here.
12082         (ior<mode>3, 128-bit types): Likewise.
12083         (xor<mode>3, 128-bit types): Likewise.
12084         (one_cmpl<mode>3, 128-bit types): Likewise.
12085         (nor<mode>3, 128-bit types): Likewise.
12086         (andc<mode>3, 128-bit types): Likewise.
12087         (eqv<mode>3, 128-bit types): Likewise.
12088         (nand<mode>3, 128-bit types): Likewise.
12089         (orc<mode>3, 128-bit types): Likewise.
12090         (and<mode>3_internal): Likewise.
12091         (bool<mode>3_internal): Likewise.
12092         (boolc<mode>3_internal1): Likewise.
12093         (boolc<mode>3_internal2): Likewise.
12094         (boolcc<mode>3_internal1): Likewise.
12095         (boolcc<mode>3_internal2): Likewise.
12096         (eqv<mode>3_internal1): Likewise.
12097         (eqv<mode>3_internal2): Likewise.
12098         (one_cmpl1<mode>3_internal): Likewise.
12100 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12102         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12103         Rename flag_stack_usage to flag_stack_usage_info.
12105 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12107         * config/microblaze/sync.md: New file.
12108         * config/microblaze/microblaze.md: Include sync.md
12109         * config/microblaze/microblaze.c: Add print_operand 'y'.
12110         * config/microblaze/constraints.md: Add memory_contraint
12111         'Q' which is a single register.
12113 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12115         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12117 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12119         * reload.c (find_reloads): Exit loop once we find this operand
12120         cannot be reloaded somehow for this alternative.
12122         * reload.c (find_reloads): Exit loop once we find a hard register.
12124         * rtlanal.c (computed_jump_p): Exit loop once we find label
12125         reference is used.
12127         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12129         * cfgloopmanip.c (remove_path): Exit loop after setting
12130         irred_invalidated.
12132         * gensupport.c (subst_dup): Avoid loop if code is not
12133         MATCH_DUP nor MATCH_OP_DUP.
12135 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12137         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12139 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12141         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12142         true for SP_REGNUM if mode == ptr_mode.
12143         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12144         with value R0_REGNUM + 31.
12146 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12148         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12149         pad pointer-typed argument downward.
12151 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12153         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12154         and __ILP32__ when the ILP32 model is in use.
12156 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12158         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12159         (aarch64_load_symref_appropriately): In the case of
12160         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12161         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12162         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12163         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12164         if 'mode' doesn't equal to 'ptr_mode'.
12165         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12166         'vcall_offset'; change to call aarch64_emit_move differently depending
12167         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12168         to calculate the upper bound of 'vcall_offset'.
12169         (aarch64_cannot_force_const_mem): Change to also return true if
12170         mode != ptr_mode.
12171         (aarch64_legitimize_reload_address): In the case of large
12172         displacements, add new local variable 'xmode' and an assertion
12173         based on it; change to use 'xmode' to generate the new rtx and
12174         reload.
12175         (aarch64_asm_trampoline_template): Change to generate the template
12176         differently depending on TARGET_ILP32 or not; change to use
12177         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12178         (aarch64_trampoline_size): Removed.
12179         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12180         and replace immediate literals with it.  Change to use 'ptr_mode'
12181         instead of 'DImode' and call convert_memory_address if the mode
12182         of 'fnaddr' doesn't equal to 'ptr_mode'.
12183         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12184         to output symbol.
12185         (aarch64_elf_asm_destructor): Likewise.
12186         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12187         on TARGET_ILP32 instead of aarch64_trampoline_size.
12188         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12189         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12190         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12191         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12192         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12193         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12194         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12195         depending on the value of 'mode'.
12196         (add_losym_<mode>): New.
12197         (ldr_got_small_<mode>): New, based on ldr_got_small.
12198         (ldr_got_small): Remove.
12199         (ldr_got_small_sidi): New.
12200         * config/aarch64/iterators.md (P): New.
12201         (PTR): Change to 'ptr_mode' in the condition.
12203 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12205         * config.gcc (aarch64*-*-*): Support --with-abi.
12206         (aarch64*-*-elf): Support --with-multilib-list.
12207         (aarch64*-*-linux*): Likewise.
12208         (supported_defaults): Add abi to aarch64*-*-*.
12209         * configure.ac: Mention AArch64 for --with-multilib-list.
12210         * configure: Re-generated.
12211         * config/aarch64/biarchilp32.h: New file.
12212         * config/aarch64/biarchlp64.h: New file.
12213         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12214         (ABI_SPEC): Ditto.
12215         (MULTILIB_DEFAULTS): Ditto.
12216         (DRIVER_SELF_SPECS): Ditto.
12217         (ASM_SPEC): Update to also substitute -mabi.
12218         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12219         file whose name depends on -mabi= and -mbig-endian.
12220         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12221         TARGET_ILP32.
12222         (POINTER_SIZE): New define.
12223         (POINTERS_EXTEND_UNSIGNED): Ditto.
12224         (enum aarch64_abi_type): New enumeration tag.
12225         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12226         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12227         (TARGET_ILP32): New define.
12228         * config/aarch64/aarch64.opt (mabi): New.
12229         (aarch64_abi): New.
12230         (ilp32, lp64): New values for -mabi.
12231         * config/aarch64/t-aarch64 (comma): New define.
12232         (MULTILIB_OPTIONS): Ditto.
12233         (MULTILIB_DIRNAMES): Ditto.
12234         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12235         * doc/invoke.texi: Document -mabi for AArch64.
12237 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12239         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12241 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12242             Anton Blanchard  <anton@au1.ibm.com>
12244         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12245         endianness when selecting field to splat.
12247 2013-07-22  Eric Christopher  <echristo@gmail.com>
12249         * dwarf2out.c (die_odr_checksum): New function to use
12250         CHECKSUM_ macros and ULEB128 for DIE tag.
12251         (generate_type_signature): Use.
12253 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12255         * config.gcc (sparc*-*-*): Accept leon3 processor.
12256         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12257         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12258         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12259         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12260         (mfix-ut699): Adjust comment.
12261         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12262         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12263         (CPP_CPU_SPEC): Likewise.
12264         (ASM_CPU_SPEC): Likewise.
12265         * config/sparc/sparc.c (leon3_cost): New constant.
12266         (sparc_option_override): Add leon3 support.
12267         (mem_ref): New function.
12268         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12269         (sparc_do_work_around_errata): Look into the instruction in the delay
12270         slot and adjust accordingly.  Add fix for the data cache nullify issues
12271         of the UT699.  Change insertion position for the NOP.
12272         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12273         (leon3_load): New reservation.
12274         (leon_store): Bump latency to 2.
12275         (grfpu): New automaton.
12276         (grfpu_alu): New unit.
12277         (grfpu_ds): Likewise.
12278         (leon_fp_alu): Adjust.
12279         (leon_fp_mult): Delete.
12280         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12281         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12282         * config/sparc/sparc.md (cpu): Add leon3.
12283         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12284         (swapsi): Likewise.
12285         (atomic_test_and_set): Likewise.
12286         (ldstub): Likewise.
12288 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12290         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12291         default for R5900 targets.
12292         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12293         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12294         * config/mips/mips.c (mips_option_override): Report an error for
12295         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12296         for -march=r5900 -mhard-float.
12298 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12300         * df-problems.c (can_move_insns_across): Exit loop once we
12301         find a non-fixed, non-global register.
12303         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12304         setting can_throw.
12306         * omega.c (omega_eliminate_red): Break after setting red_found.
12307         (omega_problem_has_red_equations): Similarly after setting found.
12308         (omega_query_variable): Similarly after setting coupled.
12310 2013-07-22  Marek Polacek  <polacek@redhat.com>
12312         * gimplify.c: Don't include gimple.h twice.
12314 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12316         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12317         instead of TARGET_THUMB1.
12318         (Pz): New constraint.
12319         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12320         encodings.
12321         (compare_negsi_si): Likewise.
12322         (compare_addsi2_op0): Likewise.
12323         (compare_addsi2_op1): Likewise.
12324         (addsi3_carryin_<optab>): Likewise.
12325         (addsi3_carryin_alt2_<optab>): Likewise.
12326         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12327         for arm_restrict_it.
12328         (subsi3_carryin): Likewise.
12329         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12330         (minmax_arithsi): Disable for arm_restrict_it.
12331         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12332         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12333         (satsi_<SAT:code>_shift): Likewise.
12334         (arm_shiftsi3): Add alternative for 16-bit encoding.
12335         (arm32_movhf): Disable for arm_restrict_it.
12336         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12337         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12339 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12341         * config/arm/arm.md (attribute "insn"): Delete.
12342         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12343         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12344         (not_shiftsi): Update for attribute change.
12345         (not_shiftsi_compare0): Likewise.
12346         (not_shiftsi_compare0_scratch): Likewise.
12347         (arm_one_cmplsi2): Likewise.
12348         (thumb1_one_cmplsi2): Likewise.
12349         (notsi_compare0): Likewise.
12350         (notsi_compare0_scratch): Likewise.
12351         (thumb1_movdi_insn): Likewise.
12352         (arm_movsi_insn): Likewise.
12353         (movhi_insn_arch4): Likewise.
12354         (movhi_bytes): Likewise.
12355         (arm_movqi_insn): Likewise.
12356         (thumb1_movqi_insn): Likewise.
12357         (arm32_movhf): Likewise.
12358         (thumb1_movhf): Likewise.
12359         (arm_movsf_soft_insn): Likewise.
12360         (thumb1_movsf_insn): Likewise.
12361         (thumb_movdf_insn): Likewise.
12362         (movsicc_insn): Likewise.
12363         (movsfcc_soft_insn): Likewise.
12364         (and_scc): Likewise.
12365         (cond_move): Likewise.
12366         (if_move_not): Likewise.
12367         (if_not_move): Likewise.
12368         (if_shift_move): Likewise.
12369         (if_move_shift): Likewise.
12370         (if_shift_shift): Likewise.
12371         (if_not_arith): Likewise.
12372         (if_arith_not): Likewise.
12373         (cond_move_not): Likewise.
12374         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12375         (neon_mov<mode>): Likewise.
12376         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12377         (thumb2_movsi_vfp): Likewise.
12378         (movsf_vfp): Likewise.
12379         (thumb2_movsf_vfp): Likewise.
12380         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12381         change.
12382         (cortexa7_older_only): Likewise.
12383         (cortexa7_younger): Likewise.
12384         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12385         (1020alu_shift_op): Likewise.
12386         (1020alu_shift_reg_op): Likewise.
12387         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12388         (alu_shift_op): Likewise.
12389         (alu_shift_reg_op): Likewise.
12390         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12391         (11_alu_shift_op): Likewise.
12392         (11_alu_shift_reg_op): Likewise.
12393         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12394         (9_alu_shift_reg_op): Likewise.
12395         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12396         change.
12397         (cortex_a15_alu_shift): Likewise.
12398         (cortex_a15_alu_shift_reg): Likewise.
12399         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12400         (cortex_a5_alu_shift): Likewise.
12401         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12402         change.
12403         (cortex_a53_alu_shift): Likewise.
12404         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12405         change.
12406         (cortex_a7_alu_reg): Likewise.
12407         (cortex_a7_alu_shift): Likewise.
12408         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12409         (cortex_a8_alu_shift): Likewise.
12410         (cortex_a8_alu_shift_reg): Likewise.
12411         (cortex_a8_mov): Likewise.
12412         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12413         (cortex_a9_dp_shift): Likewise.
12414         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12415         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12416         (cortex_r4_mov): Likewise.
12417         (cortex_r4_alu_shift): Likewise.
12418         (cortex_r4_alu_shift_reg): Likewise.
12419         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12420         (526_alu_shift_op): Likewise.
12421         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12422         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12423         (626te_alu_shift_op): Likewise.
12424         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12425         (726te_alu_op): Likewise.
12426         (726te_alu_shift_op): Likewise.
12427         (726te_alu_shift_reg_op): Likewise.
12428         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12429         (mp626_alu_shift_op): Likewise.
12430         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12431         (pj4_alu_e1_conds): Likewise.
12432         (pj4_alu): Likewise.
12433         (pj4_alu_conds): Likewise.
12434         (pj4_shift): Likewise.
12435         (pj4_shift_conds): Likewise.
12436         (pj4_alu_shift): Likewise.
12437         (pj4_alu_shift_conds): Likewise.
12439 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12441         * config/arm/predicates.md (shiftable_operator_strict_it):
12442         New predicate.
12443         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12444         Disable cond_exec version for arm_restrict_it.
12445         (thumb2_smaxsi3): Convert to generate cond_exec.
12446         (thumb2_sminsi3): Likewise.
12447         (thumb32_umaxsi3): Likewise.
12448         (thumb2_uminsi3): Likewise.
12449         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12450         (thumb2_neg_abssi2): Likewise.
12451         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12452         (thumb2_movsicc_insn): Adjust alternatives.
12453         (thumb2_mov_negscc): Disable for arm_restrict_it.
12454         (thumb2_mov_negscc_strict_it): New pattern.
12455         (thumb2_mov_notscc_strict_it): New pattern.
12456         (thumb2_mov_notscc): Disable for arm_restrict_it.
12457         (thumb2_ior_scc): Likewise.
12458         (thumb2_ior_scc_strict_it): New pattern.
12459         (thumb2_cond_move): Adjust for arm_restrict_it.
12460         (thumb2_cond_arith): Disable for arm_restrict_it.
12461         (thumb2_cond_arith_strict_it): New pattern.
12462         (thumb2_cond_sub): Adjust for arm_restrict_it.
12463         (thumb2_movcond): Likewise.
12464         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12465         (thumb2_zero_extendhisi2_v6): Likewise.
12466         (thumb2_zero_extendqisi2_v6): Likewise.
12467         (orsi_notsi_si): Likewise.
12468         (orsi_not_shiftsi_si): Likewise.
12470 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12472         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12473         instruction sequence is 1 byte shorter.
12475 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12477         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12478         it is not needed after split.
12480 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12482         PR target/51784
12483         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12484         second label for nonlocal goto receivers. Don't output pic base labels
12485         unless we're producing PIC; mark that action unreachable().
12486         (ix86_save_reg): If the function contains a nonlocal label, save the
12487         PIC base reg.
12488         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12489         * config/darwin.c (emitted_pic_label_num): New GTY.
12490         (update_pic_label_number_if_needed): New.
12491         (machopic_output_function_base_name): Adjust for nonlocal receiver
12492         case.
12493         (machopic_should_output_picbase_label): New.
12494         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12495         (nonlocal_goto_receiver): New insn and split.
12497 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12499         * config/aarch64/aarch64-builtins.c
12500         (aarch64_fold_builtin): Fold abs in all modes.
12501         * config/aarch64/aarch64-simd-builtins.def
12502         (abs): Enable for all modes.
12503         * config/aarch64/arm_neon.h
12504         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12505         (vabs_f64): Add missing intrinsic.
12507 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12509         * config/aarch64/arm_neon.h (vabs_s64): New function
12511 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12513         PR target/57516
12514         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12515         * config/avr/avr.md (adjust_len): Add `round'.
12516         * config/avr/avr-protos.h (avr_out_round): New prototype.
12517         (avr_out_plus): Add `out_label' argument.
12518         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12519         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12520         Handle the case where `insn' is just a pattern.
12521         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12522         (avr_out_round): New function.
12523         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12525 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12527         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12528         Add check for flag_stack_usage to handle -fstack-usage support
12530 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12532         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12533         interaction for new Power8 flags and VSX.
12535 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12537         PR middle-end/57698
12538         * tree-inline.c (expand_call_inline): Emit errors during
12539         early_inlining only if optimization is not turned on.
12541 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12543         * passes.def: New.
12545         * passes.c (init_optimization_passes): Move the construction of
12546         the pass hierarchy into a new passes.def file.
12548         * Makefile.in (passes.o): Add dependency on passes.def.
12550 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12552         * passes.c (init_optimization_passes): Introduce macros for
12553         constructing the tree of passes (INSERT_PASSES_AFTER,
12554         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12555         TERMINATE_PASS_LIST).
12557 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12558             Wei Mi  <wmi@google.com>
12560         PR rtl-optimization/57878
12561         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12562         top.
12563         (reload_pseudo_compare_func): Check nregs first for reload
12564         pseudos.
12566 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12568         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12570 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12572         * read-rtl.c (validate_const_int): Once an invalid character is
12573         seen, quit the loop.
12575         * gengtype.c (write_roots): Similarly once we find the "deletable"
12576         or "if_marked" option.
12578 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12580         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12581         "xtab" and "sat".  Move value "clz" from here to ...
12582         (attriubte "type"): ... here.
12583         (satsi_<SAT:code>): Delete "insn" attribute.
12584         (satsi_<SAT:code>_shift): Likewise.
12585         (arm_zero_extendqisi2addsi): Likewise.
12586         (arm_extendqisi2addsi): Likewise.
12587         (clzsi2): Update for attribute changes.
12588         (rbitsi2): Likewise.
12589         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12590         attribute.
12591         (arm_usatsihi): Likewise.
12592         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12594 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12596         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12597         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12598         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12599         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12600         in alphabetical order.
12601         (attribute "core_cycles"): Update for attribute changes.
12602         (arm_addsi3): Likewise.
12603         (addsi3_compare0): Likewise.
12604         (addsi3_compare0_scratch): Likewise.
12605         (addsi3_compare_op1): Likewise.
12606         (addsi3_compare_op2): Likewise.
12607         (compare_addsi2_op0): Likewise.
12608         (compare_addsi2_op1): Likewise.
12609         (addsi3_carryin_shift_<optab>): Likewise.
12610         (subsi3_carryin_shift): Likewise.
12611         (rsbsi3_carryin_shift): Likewise.
12612         (arm_subsi3_insn): Likewise.
12613         (subsi3_compare0): Likewise.
12614         (subsi3_compare): Likewise.
12615         (arm_andsi3_insn): Likewise.
12616         (thumb1_andsi3_insn): Likewise.
12617         (andsi3_compare0): Likewise.
12618         (andsi3_compare0_scratch): Likewise.
12619         (zeroextractsi_compare0_scratch
12620         (andsi_not_shiftsi_si): Likewise.
12621         (iorsi3_insn): Likewise.
12622         (iorsi3_compare0): Likewise.
12623         (iorsi3_compare0_scratch): Likewise.
12624         (arm_xorsi3): Likewise.
12625         (thumb1_xorsi3_insn): Likewise.
12626         (xorsi3_compare0): Likewise.
12627         (xorsi3_compare0_scratch): Likewise.
12628         (satsi_<SAT:code>_shift): Likewise.
12629         (rrx): Likewise.
12630         (arm_shiftsi3): Likewise.
12631         (shiftsi3_compare0): Likewise.
12632         (not_shiftsi): Likewise.
12633         (not_shiftsi_compare0): Likewise.
12634         (not_shiftsi_compare0_scratch): Likewise.
12635         (arm_one_cmplsi2): Likewise.
12636         (thumb_one_complsi2): Likewise.
12637         (notsi_compare0): Likewise.
12638         (notsi_compare0_scratch): Likewise.
12639         (thumb1_zero_extendhisi2): Likewise.
12640         (arm_zero_extendhisi2): Likewise.
12641         (arm_zero_extendhisi2_v6): Likewise.
12642         (arm_zero_extendhisi2addsi): Likewise.
12643         (thumb1_zero_extendqisi2): Likewise.
12644         (thumb1_zero_extendqisi2_v6): Likewise.
12645         (arm_zero_extendqisi2): Likewise.
12646         (arm_zero_extendqisi2_v6): Likewise.
12647         (arm_zero_extendqisi2addsi): Likewise.
12648         (thumb1_extendhisi2): Likewise.
12649         (arm_extendhisi2): Likewise.
12650         (arm_extendhisi2_v6): Likewise.
12651         (arm_extendqisi): Likewise.
12652         (arm_extendqisi_v6): Likewise.
12653         (arm_extendqisi2addsi): Likewise.
12654         (thumb1_extendqisi2): Likewise.
12655         (thumb1_movdi_insn): Likewise.
12656         (arm_movsi_insn): Likewise.
12657         (movsi_compare0): Likewise.
12658         (movhi_insn_arch4): Likewise.
12659         (movhi_bytes): Likewise.
12660         (arm_movqi_insn): Likewise.
12661         (thumb1_movqi_insn): Likewise.
12662         (arm32_movhf): Likewise.
12663         (thumb1_movhf): Likewise.
12664         (arm_movsf_soft_insn): Likewise.
12665         (thumb1_movsf_insn): Likewise.
12666         (movdf_soft_insn): Likewise.
12667         (thumb_movdf_insn): Likewise.
12668         (arm_cmpsi_insn): Likewise.
12669         (cmpsi_shiftsi): Likewise.
12670         (cmpsi_shiftsi_swp): Likewise.
12671         (arm_cmpsi_negshiftsi_si): Likewise.
12672         (movsicc_insn): Likewise.
12673         (movsfcc_soft_insn): Likewise.
12674         (arith_shiftsi): Likewise.
12675         (arith_shiftsi_compare0
12676         (arith_shiftsi_compare0_scratch
12677         (sub_shiftsi): Likewise.
12678         (sub_shiftsi_compare0
12679         (sub_shiftsi_compare0_scratch
12680         (and_scc): Likewise.
12681         (cond_move): Likewise.
12682         (if_plus_move): Likewise.
12683         (if_move_plus): Likewise.
12684         (if_move_not): Likewise.
12685         (if_not_move): Likewise.
12686         (if_shift_move): Likewise.
12687         (if_move_shift): Likewise.
12688         (if_shift_shift): Likewise.
12689         (if_not_arith): Likewise.
12690         (if_arith_not): Likewise.
12691         (cond_move_not): Likewise.
12692         (thumb1_ashlsi3): Set type attribute.
12693         (thumb1_ashrsi3): Likewise.
12694         (thumb1_lshrsi3): Likewise.
12695         (thumb1_rotrsi3): Likewise.
12696         (shiftsi3_compare0_scratch): Likewise.
12697         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12698         (neon_mov<mode>): Likewise.
12699         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12700         attribute changes.
12701         (thumb2_movsi_insn): Likewise.
12702         (thumb2_cmpsi_neg_shiftsi): Likewise.
12703         (thumb2_extendqisi_v6): Likewise.
12704         (thumb2_zero_extendhisi2_v6): Likewise.
12705         (thumb2_zero_extendqisi2_v6): Likewise.
12706         (thumb2_shiftsi3_short): Likewise.
12707         (thumb2_addsi3_compare0_scratch): Likewise.
12708         (orsi_not_shiftsi_si): Likewise.
12709         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12710         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12711         changes.
12712         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12713         (1020alu_shift_op): Likewise.
12714         (1020alu_shift_reg_op): Likewise.
12715         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12716         (alu_shift_op): Likewise.
12717         (alu_shift_reg_op): Likewise.
12718         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12719         (11_alu_shift_op): Likewise.
12720         (11_alu_shift_reg_op): Likewise.
12721         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12722         (9_alu_shift_reg_op): Likewise.
12723         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12724         changes.
12725         (cortex_a15_alu_shift): Likewise.
12726         (cortex_a15_alu_shift_reg): Likewise.
12727         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12728         changes.
12729         (cortex_a5_alu_shift): Likewise.
12730         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12731         changes.
12732         (cortex_a53_alu_shift): Likewise.
12733         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12734         changes.
12735         (cortex_a7_alu_reg): Likewise.
12736         (cortex_a7_alu_shift): Likewise.
12737         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12738         changes.
12739         (cortex_a8_alu_shift): Likewise.
12740         (cortex_a8_alu_shift_reg): Likewise.
12741         (cortex_a8_mov): Likewise.
12742         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12743         (cortex_a9_dp_shift): Likewise.
12744         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12745         changes.
12746         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12747         changes.
12748         (cortex_r4_mov): Likewise.
12749         (cortex_r4_alu_shift): Likewise.
12750         (cortex_r4_alu_shift_reg): Likewise.
12751         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12752         (526_alu_shift_op): Likewise.
12753         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12754         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12755         (626te_alu_shift_op): Likewise.
12756         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12757         (726te_alu_op): Likewise.
12758         (726te_alu_shift_op): Likewise.
12759         (726te_alu_shift_reg_op): Likewise.
12760         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12761         (mp626_alu_shift_op): Likewise.
12762         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12763         (pj4_alu_e1_conds): Likewise.
12764         (pj4_alu): Likewise.
12765         (pj4_alu_conds): Likewise.
12766         (pj4_shift): Likewise.
12767         (pj4_shift_conds): Likewise.
12768         (pj4_alu_shift): Likewise.
12769         (pj4_alu_shift_conds): Likewise.
12770         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12771         changes.
12772         (cortexa7_older_only): Likewise.
12773         (cortexa7_younger): Likewise.
12775 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12777         * ipa-pure-const.c (generate_summary): Rename to...
12778         (pure_const_generate_summary): ... this.
12780 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12782         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12784 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12786         PR target/57909
12787         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12788         usage in HI mode.
12790 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12792         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12793         enabled without -march=zEC12.
12794         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12795         flags to be set.
12797 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12799         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12800         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12801         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12802         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12803         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12804         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12805         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12806         ISA_HAS_FP4.
12807         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12808         and ISA_HAS_NMADD3_NMSUB3.
12809         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12810         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12811         (nmsub4<mode>, nmsub3<mode>): Likewise.
12812         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12814 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12816         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12817         TARGET_MIPS5400 checking.
12819 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12820             Peter Bergner  <bergner@vnet.ibm.com>
12822         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12823         registers in the comment.
12824         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12825         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12826         rather than FIRST_PSEUDO_REGISTERS.
12828 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12830         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12831         enable extra ISA flags with TARGET_HTM.
12833 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12835         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12836         Fix comment typos.
12838 2013-07-15  Cong Hou  <congh@google.com>
12840         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12841         in compare function for sorting.
12843 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12845         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12846         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12847         * config/rs6000/rs6000.opt: Add -mhtm option.
12848         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12849         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12850         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12851         __HTM__ if the HTM instructions are available.
12852         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12853         htm_spr_reg_operand): New define_predicates.
12854         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12855         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12856         Include htm.md.
12857         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12858         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12859         HTM builtin functions.
12860         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12861         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12862         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12863         (rs6000_builtin_mask_calculate): Likewise.
12864         (rs6000_option_override_internal): Likewise.
12865         (bdesc_htm): Add new HTM builtin support.
12866         (htm_spr_num): New function.
12867         (htm_spr_regno): Likewise.
12868         (rs6000_htm_spr_icode): Likewise.
12869         (htm_expand_builtin): Likewise.
12870         (htm_init_builtins): Likewise.
12871         (rs6000_expand_builtin): Add support for HTM builtin functions.
12872         (rs6000_init_builtins): Likewise.
12873         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12874         option.
12875         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12876         (TARGET_HTM, MASK_HTM): Define macros.
12877         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12878         (FIXED_REGISTERS): Likewise.
12879         (CALL_USED_REGISTERS): Likewise.
12880         (CALL_REALLY_USED_REGISTERS): Likewise.
12881         (REG_ALLOC_ORDER): Likewise.
12882         (enum reg_class): Likewise.
12883         (REG_CLASS_NAMES): Likewise.
12884         (REG_CLASS_CONTENTS): Likewise.
12885         (REGISTER_NAMES): Likewise.
12886         (ADDITIONAL_REGISTER_NAMES): Likewise.
12887         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12888         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12889         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12890         * config/rs6000/htm.md: New file.
12891         * config/rs6000/htmintrin.h: New file.
12892         * config/rs6000/htmxlintrin.h: New file.
12894 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12896         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12897         Define SYMBOL_TINY_GOT, update comment.
12898         * config/aarch64/aarch64.c
12899         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12900         (aarch64_expand_mov_immediate): Likewise.
12901         (aarch64_print_operand): Likewise.
12902         (aarch64_classify_symbol): Likewise.
12903         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12904         (ldr_got_tiny): Define.
12906 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12908         PR tree-optimization/54094
12909         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12910           scheduling dimension for the parallelism check from the polyhedral
12911           information in the AST.
12912         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12913           in 2D + 1 form.
12915 2013-07-13  Jason Merrill  <jason@redhat.com>
12917         * print-tree.c (debug_vec_tree): Use debug_raw.
12918         (debug_raw (vec<tree, va_gc> &)): New.
12919         (debug_raw (vec<tree, va_gc> *)): New.
12920         * tree.h: Declare them.
12922 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12924         * ifcvt.c (ifcvt_after_combine): New static variable.
12925         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12926         for size.
12927         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12928         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12929         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12931 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12933         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12935 2013-07-12  Michael Matz  <matz@suse.de>
12937         PR middle-end/55771
12938         * convert.c (convert_to_real): Reject non-float inner types.
12940 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12942         * config/aarch64/aarch64-protos.h
12943         (aarch64_simd_immediate_valid_for_move): Remove.
12944         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12945         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12946         cases.
12947         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12949 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12951         * config/mips/mips.c (mips_conditional_register_usage): Do not
12952         use t[0-7] registers in MIPS16 mode when optimizing for size.
12954 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12956         * config/i386/i386.c (dispatch_function_versions): Fix array
12957         indexing of function_version_info to match actual_versions.
12959 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12961         * vec.h (struct va_gc): Move release out-of-line.
12962         (va_gc::release): Call ggc_free on released vec.
12964 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12966         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12967         Require GOT register as additional operand in UNSPEC.
12968         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12969         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12970         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12971         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12972         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12973         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12974         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12976 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12978         PR target/57631
12979         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12980         name seen by assembler/linker rather if available.
12982 2013-07-11  Andreas Schwab  <schwab@suse.de>
12984         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12986 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12988         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12990 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12992         * doc/tm.texi.in: Move hook documentation to ....
12993         * target.def: ... here.
12995         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12996         text on @hook line.
12997         * doc/tm.texi: Regenerate.
12999 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
13001         PR c++/57869
13002         * doc/invoke.texi: Document -Wconditionally-supported.
13004 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13006         PR target/57844
13007         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
13008         of my_fp.
13010 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13012         PR target/57506
13013         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
13014         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
13015         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
13016         Remove duplicate devices.
13017         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
13018         * config/avr/t-multilib: Regenerate.
13019         * config/avr/avr-tables.opt: Regenerate.
13020         * doc/avr-mmcu.texi: Regenerate.
13022 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13024         PR target/56987
13025         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
13027 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
13029         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
13030         the cost of MULT when optimizing for size.
13032 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
13034         * config/cr16/cr16-protos.h: Don't include target.h.
13036 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
13038         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
13039         adjust register size for TDmode and TFmode for VSX registers.
13041 2013-07-08  Kai Tietz  <ktietz@redhat.com>
13043         PR target/56892
13044         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
13045         hook_bool_const_tree_true.
13047 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13049         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
13050         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
13051         * config/s390/s390.md: Define FPR*_REGNUM constants.
13052         Fix FPR2_REGNUM constant (18 -> 17).
13053         ("*trunc<BFP:mode><DFP_ALL:mode>2")
13054         ("*trunc<DFP_ALL:mode><BFP:mode>2")
13055         ("trunc<BFP:mode><DFP_ALL:mode>2")
13056         ("trunc<DFP_ALL:mode><BFP:mode>2")
13057         ("*extend<BFP:mode><DFP_ALL:mode>2")
13058         ("*extend<DFP_ALL:mode><BFP:mode>2")
13059         ("extend<BFP:mode><DFP_ALL:mode>2")
13060         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
13061         FPR4_REGNUM.
13063 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
13065         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
13067 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13069         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13070         and cfun_fpr_bit_p to cfun_fpr_save_p.
13071         (s390_frame_area, s390_register_info, s390_frame_info)
13072         (s390_emit_prologue, s390_emit_epilogue)
13073         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13074         register numbers.
13075         * config/s390/s390.h: Define *_REGNUM macros for floating point
13076         register numbers.
13078 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13080         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13082 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13084         PR rtl-optimization/57786
13085         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13086         and break out of the loop when it is set to false.
13088 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13090         PR target/57819
13091         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13092         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13093         (const_int 63)) 0)).
13094         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13095         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13096         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13098         PR rtl-optimization/57829
13099         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13100         mask bits outside of mode are just sign-extension from mode to HWI.
13102 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13104         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13105         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13106         adjust_address instead of change_address to keep info about alignment.
13107         (emit_strmov): Remove.
13108         (emit_memmov): New function.
13109         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13110         (expand_movmem_epilogue): Likewise and return updated rtx for
13111         destination.
13112         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13113         destination and source.
13114         (decide_alignment): Refactor, handle vector_loop.
13115         (ix86_expand_movmem): Likewise.
13116         (ix86_expand_setmem): Likewise.
13117         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13119 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13121         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13122         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13124 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13126         * config/i386/sse.md (sse_movlhps): Change alternative 3
13127         of operand 2 to "m".
13129 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13131         PR target/57807
13132         * config/i386/sse.md (iptr): New mode attribute.
13133         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13134         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13135         (<sse>_vmmul<mode>3): Ditto.
13136         (<sse>_vmdiv<mode>3): Ditto.
13137         (sse_vmrcpv4sf2): Ditto.
13138         (<sse>_vmsqrt<mode>2): Ditto.
13139         (sse_vmrsqrtv4sf2): Ditto.
13140         (<sse>_vm<code><mode>3): Ditto.
13141         (avx_vmcmp<mode>3): Ditto.
13142         (<sse>_vmmaskcmp<mode>3): Ditto.
13143         (<sse>_comi): Ditto.
13144         (<sse>_ucomi): Ditto.
13145         (*xop_vmfrcz_<mode>): Ditto.
13146         (*fmai_fmadd_<mode>): Ditto.
13147         (*fmai_fmsub_<mode>): Ditto.
13148         (*fmai_fnmadd_<mode>): Ditto.
13149         (*fmai_fnmsub_<mode>): Ditto.
13150         (*fma4i_vmfmadd_<mode>): Ditto.
13151         (*fma4i_vmfmsub_<mode>): Ditto.
13152         (*fma4i_vmfnmadd_<mode>): Ditto.
13153         (*fma4i_vmfnmsub_<mode>): Ditto.
13154         (*xop_vmfrcz_<mode>): Ditto.
13155         (sse_cvtps2pi): Ditto.
13156         (sse_cvttps2pi): Ditto.
13157         (sse_cvtss2si): Ditto.
13158         (sse_cvtss2si_2): Ditto.
13159         (sse_cvtss2siq_2): Ditto.
13160         (sse_cvttss2si): Ditto.
13161         (sse_cvttss2siq): Ditto.
13162         (sse_cvtsd2si): Ditto.
13163         (sse_cvtsd2si_2): Ditto.
13164         (sse_cvtsd2siq_2): Ditto.
13165         (sse_cvttsd2si): Ditto.
13166         (sse_cvttsd2siq): Ditto.
13167         (sse_cvtsd2ss): Ditto.
13168         (sse_cvtss2sd): Ditto.
13169         (avx2_pbroadcast<mode>): Ditto.
13170         (avx2_pbroadcast<mode>_1): Ditto.
13171         (*avx_vperm_broadcast_v4sf): Ditto.
13173         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13174         (sse_movlhps): Ditto.
13175         (sse_storehps): Ditto.
13176         (sse_loadhps): Ditto.
13177         (sse_storelps): Ditto.
13178         (sse_loadlps): Ditto.
13179         (*vec_concatv4sf): Ditto.
13180         (*vec_interleave_highv2df): Ditto.
13181         (*vec_interleave_lowv2df): Ditto.
13182         (*vec_extractv2df_1_sse): Ditto.
13183         (*vec_extractv2df_0_sse): Ditto.
13184         (sse2_storelpd): Ditto.
13185         (sse2_loadlpd): Ditto.
13186         (sse2_movsd): Ditto.
13187         (*vec_concatv4si): Ditto.
13188         (vec_concatv2di): Ditto.
13190         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13191         for Intel asm dialect.
13192         (mmx_punpcklwd): Ditto.
13193         (mmx_punpckldq): Ditto.
13195         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13196         for intel assembler dialect.
13198 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13200         PR target/29776
13201         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13202         for BUILT_IN_C{LZ,LRSB}*.
13203         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13204         * tree-vrp.c (extract_range_basic): Handle
13205         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13206         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13207         fall thru to code calling set_value*.
13208         * builtins.c (expand_builtin): Remove *IMAX cases.
13209         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13210         if width is bigger than 2*HWI.
13212 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13214         PR rtl-optimization/55342
13215         * lra-int.h (lra_subreg_reload_pseudos): New.
13216         * lra.c: Add undoing optional reloads to the block diagram.
13217         (lra_subreg_reload_pseudos): New.
13218         (lra_optional_reload_pseudos): Change comments.
13219         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13220         lra_optional_reload_pseudos after undo transformations.
13221         * lra-assigns.c (pseudo_prefix_title): New.
13222         (lra_setup_reg_renumber): Use it.
13223         (spill_for): Ditto.  Check subreg reload pseudos too.
13224         (assign_by_spills): Consider subreg reload pseudos too.
13225         * lra-constraints.c (simplify_operand_subreg): Use
13226         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13227         (curr_insn_transform): Recognize and do optional reloads.
13228         (undo_optional_reloads): New.
13229         (lra_undo_inheritance): Call undo_optional_reloads.
13231 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13233         * tree-complex.c (expand_complex_operations_1): Fix typo.
13235 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13237         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13238         (tune_params): New member 'const vec_costs'.
13239         * config/aarch64/aarch64.c (generic_vector_cost): New.
13240         (generic_tunings): New member 'generic_vector_cost'.
13241         (aarch64_builtin_vectorization_cost): New.
13242         (aarch64_add_stmt_cost): New.
13243         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13244         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13246 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13248         PR target/57777
13249         * config/i386/predicates.md (vsib_address_operand): Disallow
13250         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13252 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13254         PR middle-end/55030
13255         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13256         expand_builtin_setjmp_receiver.
13257         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13258         with NULL for the label parameter.
13259         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13260         the frame-pointer.  Adjust comments.
13261         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13262         only if LABEL is non-NULL.
13264 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13266         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13267         (ARM_ABI_AAPCS64): Ditto.
13268         (arm_abi): Ditto.
13269         (ARM_DEFAULT_ABI): Ditto.
13271 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13273         * config/aarch64/aarch64-builtins.c
13274         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13275         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13276         (st1): Likewise.
13277         * config/aarch64/aarch64-simd.md
13278         (aarch64_ld1<VALL:mode>): New.
13279         (aarch64_st1<VALL:mode>): Likewise.
13280         * config/aarch64/arm_neon.h
13281         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13283 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13285         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13286         target ISA is AVX.
13287         (ix86_option_override_internal):Turn on all -mavx target flags by
13288         default as they are dependent on AVX anyway.
13290 2013-07-02  Cary Coutant  <ccoutant@google.com>
13292         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13293         deterministic hash.
13294         (loc_checksum_ordered): Likewise.
13295         (hash_loc_operands): Remove inline keyword.
13297 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13299         PR tree-optimization/57741
13300         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13301         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13302         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13303         Allow REAL_CST step_exprs if flag_associative_math.
13304         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13306 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13308         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13310 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13312         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13314 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13316         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13317         encoding.
13318         (iorsi3_insn): Likewise.
13319         (arm_xorsi3): Likewise.
13321 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13323         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13324         to ...
13325         (attribute "type"): ... here, and prefix with "wmmx_".
13326         (attribute "core_cycles"): Update for attribute changes.
13327         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13328         (tbcstv4hi): Likewise.
13329         (tbcstv2si): Likewise.
13330         (iwmmxt_iordi3): Likewise.
13331         (iwmmxt_xordi3): Likewise.
13332         (iwmmxt_anddi3): Likewise.
13333         (iwmmxt_nanddi3): Likewise.
13334         (iwmmxt_arm_movdi): Likewise.
13335         (iwmmxt_movsi_insn): Likewise.
13336         (mov<mode>_internal): Likewise.
13337         (and<mode>3_iwmmxt): Likewise.
13338         (ior<mode>3_iwmmxt): Likewise.
13339         (xor<mode>3_iwmmxt): Likewise.
13340         (add<mode>3_iwmmxt): Likewise.
13341         (ssaddv8qi3): Likewise.
13342         (ssaddv4hi3): Likewise.
13343         (ssaddv2si3): Likewise.
13344         (usaddv8qi3): Likewise.
13345         (usaddv4hi3): Likewise.
13346         (usaddv2si3): Likewise.
13347         (sub<mode>3_iwmmxt): Likewise.
13348         (sssubv8qi3): Likewise.
13349         (sssubv4hi3): Likewise.
13350         (sssubv2si3): Likewise.
13351         (ussubv8qi3): Likewise.
13352         (ussubv4hi3): Likewise.
13353         (ussubv2si3): Likewise.
13354         (mulv4hi3_iwmmxt): Likewise.
13355         (smulv4hi3_highpart): Likewise.
13356         (umulv4hi3_highpart): Likewise.
13357         (iwmmxt_wmacs): Likewise.
13358         (iwmmxt_wmacsz): Likewise.
13359         (iwmmxt_wmacu): Likewise.
13360         (iwmmxt_wmacuz): Likewise.
13361         (iwmmxt_clrdi): Likewise.
13362         (iwmmxt_clrv8qi): Likewise.
13363         (iwmmxt_clr4hi): Likewise.
13364         (iwmmxt_clr2si): Likewise.
13365         (iwmmxt_uavgrndv8qi3): Likewise.
13366         (iwmmxt_uavgrndv4hi3): Likewise.
13367         (iwmmxt_uavgv8qi3): Likewise.
13368         (iwmmxt_uavgv4hi3): Likewise.
13369         (iwmmxt_tinsrb): Likewise.
13370         (iwmmxt_tinsrh): Likewise.
13371         (iwmmxt_tinsrw): Likewise.
13372         (iwmmxt_textrmub): Likewise.
13373         (iwmmxt_textrmsb): Likewise.
13374         (iwmmxt_textrmuh): Likewise.
13375         (iwmmxt_textrmsh): Likewise.
13376         (iwmmxt_textrmw): Likewise.
13377         (iwmxxt_wshufh): Likewise.
13378         (eqv8qi3): Likewise.
13379         (eqv4hi3): Likewise.
13380         (eqv2si3): Likewise.
13381         (gtuv8qi3): Likewise.
13382         (gtuv4hi3): Likewise.
13383         (gtuv2si3): Likewise.
13384         (gtv8qi3): Likewise.
13385         (gtv4hi3): Likewise.
13386         (gtv2si3): Likewise.
13387         (smax<mode>3_iwmmxt): Likewise.
13388         (umax<mode>3_iwmmxt): Likewise.
13389         (smin<mode>3_iwmmxt): Likewise.
13390         (umin<mode>3_iwmmxt): Likewise.
13391         (iwmmxt_wpackhss): Likewise.
13392         (iwmmxt_wpackwss): Likewise.
13393         (iwmmxt_wpackdss): Likewise.
13394         (iwmmxt_wpackhus): Likewise.
13395         (iwmmxt_wpackwus): Likewise.
13396         (iwmmxt_wpackdus): Likewise.
13397         (iwmmxt_wunpckihb): Likewise.
13398         (iwmmxt_wunpckihh): Likewise.
13399         (iwmmxt_wunpckihw): Likewise.
13400         (iwmmxt_wunpckilb): Likewise.
13401         (iwmmxt_wunpckilh): Likewise.
13402         (iwmmxt_wunpckilw): Likewise.
13403         (iwmmxt_wunpckehub): Likewise.
13404         (iwmmxt_wunpckehuh): Likewise.
13405         (iwmmxt_wunpckehuw): Likewise.
13406         (iwmmxt_wunpckehsb): Likewise.
13407         (iwmmxt_wunpckehsh): Likewise.
13408         (iwmmxt_wunpckehsw): Likewise.
13409         (iwmmxt_wunpckelub): Likewise.
13410         (iwmmxt_wunpckeluh): Likewise.
13411         (iwmmxt_wunpckeluw): Likewise.
13412         (iwmmxt_wunpckelsb): Likewise.
13413         (iwmmxt_wunpckelsh): Likewise.
13414         (iwmmxt_wunpckelsw): Likewise.
13415         (ror<mode>3): Likewise.
13416         (ashr<mode>3_iwmmxt): Likewise.
13417         (lshr<mode>3_iwmmxt): Likewise.
13418         (ashl<mode>3_iwmmxt): Likewise.
13419         (ror<mode>3_di): Likewise.
13420         (ashr<mode>3_di): Likewise.
13421         (lshr<mode>3_di): Likewise.
13422         (ashl<mode>3_di): Likewise.
13423         (iwmmxt_wmadds): Likewise.
13424         (iwmmxt_wmaddu): Likewise.
13425         (iwmmxt_tmia): Likewise.
13426         (iwmmxt_tmiaph): Likewise.
13427         (iwmmxt_tmiabb): Likewise.
13428         (iwmmxt_tmiatb): Likewise.
13429         (iwmmxt_tmiabt): Likewise.
13430         (iwmmxt_tmiatt): Likewise.
13431         (iwmmxt_tmovmskb): Likewise.
13432         (iwmmxt_tmovmskh): Likewise.
13433         (iwmmxt_tmovmskw): Likewise.
13434         (iwmmxt_waccb): Likewise.
13435         (iwmmxt_wacch): Likewise.
13436         (iwmmxt_waccw): Likewise.
13437         (iwmmxt_waligni): Likewise.
13438         (iwmmxt_walignr): Likewise.
13439         (iwmmxt_walignr0): Likewise.
13440         (iwmmxt_walignr1): Likewise.
13441         (iwmmxt_walignr2): Likewise.
13442         (iwmmxt_walignr3): Likewise.
13443         (iwmmxt_wsadb): Likewise.
13444         (iwmmxt_wsadh): Likewise.
13445         (iwmmxt_wsadbz): Likewise.
13446         (iwmmxt_wsadhz): Likewise.
13447         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13448         (iwmmxt_wabsdiffb): Likewise.
13449         (iwmmxt_wabsdiffh): Likewise.
13450         (iwmmxt_wabsdiffw): Likewise.
13451         (iwmmxt_waddsubhx): Likewise
13452         (iwmmxt_wsubaddhx): Likewise.
13453         (addc<mode>3): Likewise.
13454         (iwmmxt_avg4): Likewise.
13455         (iwmmxt_avg4r): Likewise.
13456         (iwmmxt_wmaddsx): Likewise.
13457         (iwmmxt_wmaddux): Likewise.
13458         (iwmmxt_wmaddsn): Likewise.
13459         (iwmmxt_wmaddun): Likewise.
13460         (iwmmxt_wmulwsm): Likewise.
13461         (iwmmxt_wmulwum): Likewise.
13462         (iwmmxt_wmulsmr): Likewise.
13463         (iwmmxt_wmulumr): Likewise.
13464         (iwmmxt_wmulwsmr): Likewise.
13465         (iwmmxt_wmulwumr): Likewise.
13466         (iwmmxt_wmulwl): Likewise.
13467         (iwmmxt_wqmulm): Likewise.
13468         (iwmmxt_wqmulwm): Likewise.
13469         (iwmmxt_wqmulmr): Likewise.
13470         (iwmmxt_wqmulwmr): Likewise.
13471         (iwmmxt_waddbhusm): Likewise.
13472         (iwmmxt_waddbhusl): Likewise.
13473         (iwmmxt_wqmiabb): Likewise.
13474         (iwmmxt_wqmiabt): Likewise.
13475         (iwmmxt_wqmiatb): Likewise.
13476         (iwmmxt_wqmiatt): Likewise.
13477         (iwmmxt_wqmiabbn): Likewise.
13478         (iwmmxt_wqmiabtn): Likewise.
13479         (iwmmxt_wqmiatbn): Likewise.
13480         (iwmmxt_wqmiattn): Likewise.
13481         (iwmmxt_wmiabb): Likewise.
13482         (iwmmxt_wmiabt): Likewise.
13483         (iwmmxt_wmiatb): Likewise.
13484         (iwmmxt_wmiatt): Likewise.
13485         (iwmmxt_wmiabbn): Likewise.
13486         (iwmmxt_wmiabtn): Likewise.
13487         (iwmmxt_wmiatbn): Likewise.
13488         (iwmmxt_wmiattn): Likewise.
13489         (iwmmxt_wmiawbb): Likewise.
13490         (iwmmxt_wmiawbt): Likewise.
13491         (iwmmxt_wmiawtb): Likewise.
13492         (iwmmxt_wmiawtt): Likewise.
13493         (iwmmxt_wmiawbbn): Likewise.
13494         (iwmmxt_wmiawbtn): Likewise.
13495         (iwmmxt_wmiawtbn): Likewise.
13496         (iwmmxt_wmiawttn): Likewise.
13497         (iwmmxt_wmerge): Likewise.
13498         (iwmmxt_tandc<mode>3): Likewise.
13499         (iwmmxt_torc<mode>3): Likewise.
13500         (iwmmxt_torvsc<mode>3): Likewise.
13501         (iwmmxt_textrc<mode>3): Likewise.
13502         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13503         (wmmxt_pack): Likewise.
13504         (wmmxt_mult_c1): Likewise.
13505         (wmmxt_mult_c2): Likewise.
13506         (wmmxt_alu_c1): Likewise.
13507         (wmmxt_alu_c2): Likewise.
13508         (wmmxt_alu_c3): Likewise.
13509         (wmmxt_transfer_c1): Likewise.
13510         (wmmxt_transfer_c2): Likewise.
13511         (wmmxt_transfer_c3): Likewise.
13512         (marvell_f_iwmmxt_wstr): Likewise.
13513         (marvell_f_iwmmxt_wldr): Likewise.
13515 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13517         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13519 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13521         Revert:
13522         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13523         * lra-constraints.c (need_for_split_p): Check call used hard regs
13524         living through calls.
13526         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13527         call used regs for call insn.
13529 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13531         PR target/57736
13532         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13533         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13535 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13537         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13539 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13541         * lra-constraints.c (need_for_split_p): Check call used hard regs
13542         living through calls.
13544 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13546         PR target/57744
13547         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13548         to tie with any other modes.  Eliminate Altivec vector mode tests,
13549         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13550         code, to return 0 if testing MODE2 for a condition, if we've
13551         already tested MODE1 for the same condition.
13553 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13555         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13556         layout.
13558 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13560         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13561         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13563 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13565         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13566         Define.
13567         (aarch64_symbolic_constant_p): Remove.
13568         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13569         static.  Fix line length and white space.
13570         (aarch64_symbolic_constant_p): Remove.
13571         * config/aarch64/predicates.md (aarch64_valid_symref):
13572         Use aarch64_classify_symbol_expression.
13574 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13576         * config/arm/constraints.md (Ts): New constraint.
13577         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13578         16-bit encodings.
13579         (compare_scc): Use "Ts" constraint for operand 0.
13580         (ior_scc_scc): Likewise.
13581         (and_scc_scc): Likewise.
13582         (and_scc_scc_nodom): Likewise.
13583         (ior_scc_scc_cmp): Likewise for operand 7.
13584         (and_scc_scc_cmp): Likewise.
13585         * config/arm/thumb2.md (thumb2_movsi_insn):
13586         Add alternatives for 16-bit encodings.
13587         (thumb2_movhi_insn): Likewise.
13588         (thumb2_movsicc_insn): Likewise.
13589         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13590         (thumb2_negscc): Use "Ts" constraint.
13591         Move mvn instruction outside cond_exec block.
13592         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13593         for 16-bit encodings.
13595 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13597         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13598         encoding.
13599         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13600         (mulsi3subsi): Likewise.
13601         (mulsidi3adddi): Likewise.
13602         (mulsidi3_v6): Likewise.
13603         (umulsidi3_v6): Likewise.
13604         (umulsidi3adddi_v6): Likewise.
13605         (smulsi3_highpart_v6): Likewise.
13606         (umulsi3_highpart_v6): Likewise.
13607         (mulhisi3tb): Likewise.
13608         (mulhisi3bt): Likewise.
13609         (mulhisi3tt): Likewise.
13610         (maddhisi4): Likewise.
13611         (maddhisi4tb): Likewise.
13612         (maddhisi4tt): Likewise.
13613         (maddhidi4): Likewise.
13614         (maddhidi4tb): Likewise.
13615         (maddhidi4tt): Likewise.
13616         (zeroextractsi_compare0_scratch): Likewise.
13617         (insv_zero): Likewise.
13618         (insv_t2): Likewise.
13619         (anddi_notzesidi_di): Likewise.
13620         (anddi_notsesidi_di): Likewise.
13621         (andsi_notsi_si): Likewise.
13622         (iordi_zesidi_di): Likewise.
13623         (xordi_zesidi_di): Likewise.
13624         (andsi_iorsi3_notsi): Likewise.
13625         (smax_0): Likewise.
13626         (smax_m1): Likewise.
13627         (smin_0): Likewise.
13628         (not_shiftsi): Likewise.
13629         (unaligned_loadsi): Likewise.
13630         (unaligned_loadhis): Likewise.
13631         (unaligned_loadhiu): Likewise.
13632         (unaligned_storesi): Likewise.
13633         (unaligned_storehi): Likewise.
13634         (extv_reg): Likewise.
13635         (extzv_t2): Likewise.
13636         (divsi3): Likewise.
13637         (udivsi3): Likewise.
13638         (arm_zero_extendhisi2addsi): Likewise.
13639         (arm_zero_extendqisi2addsi): Likewise.
13640         (compareqi_eq0): Likewise.
13641         (arm_extendhisi2_v6): Likewise.
13642         (arm_extendqisi2addsi): Likewise.
13643         (arm_movt): Likewise.
13644         (thumb2_ldrd): Likewise.
13645         (thumb2_ldrd_base): Likewise.
13646         (thumb2_ldrd_base_neg): Likewise.
13647         (thumb2_strd): Likewise.
13648         (thumb2_strd_base): Likewise.
13649         (thumb2_strd_base_neg): Likewise.
13650         (arm_negsi2): Add alternative for 16-bit encoding.
13651         (arm_one_cmplsi2): Likewise.
13653 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13655         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13656         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13657         (movdfcc): Likewise.
13658         * config/arm/vfp.md (*thumb2_movsf_vfp):
13659         Disable predication for arm_restrict_it.
13660         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13661         (*thumb2_movdfcc_vfp): Likewise.
13662         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13663         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13664         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13665         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13666         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13667         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13668         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13669         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13670         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13671         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13672         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13673         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13674         Disable predication for arm_restrict_it.
13676 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13678         * config/i386/bmiintrin.h (_bextr_u32): New.
13679         (_bextr_u64): Ditto.
13681 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13683         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13684         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13685         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13686         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13687         * config/mips/n32-elf.h: ...this new file.
13689 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13691         PR target/57224
13692         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13693         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13695 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13697         * config/mips/mips-tables.opt: Regenerate.
13698         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13699         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13700         * doc/invoke.texi: Add -m14kc.
13702 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13704         PR target/57623
13705         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13706         constraints of operand 1 and 2.
13708         PR target/57623
13709         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13710         to match RTL canonicalization.  Swap predicates and
13711         constraints of operand 1 and 2.
13713 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13715         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13716         Process OP_INOUT regs for splitting too.
13718 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13720         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13721         decl before the loop, initialize to NULL.
13722         (vectorizable_load): Initialize ptr_incr to NULL.
13724 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13726         PR lto/57208
13727         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13728         * ipa-ref.c (ipa_maybe_record_reference): New function.
13729         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13730         * ipa-cp.c (create_specialized_node): Record potential references from
13731         aggvals.
13732         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13734 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13736         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13737         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13738         'mode' to force_reg.
13739         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13740         (aarch64_expand_mov_immediate): Likewise.
13742 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13744         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13745         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13747 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13749         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13750         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13751         (struct machine_function): Add tbegin_p.
13752         (s390_canonicalize_comparison): Fold CC mode compares to
13753         conditional jump if possible.
13754         (s390_emit_jump): Return the emitted jump.
13755         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13756         Handle CCRAWmode compares.
13757         (s390_option_override): Default to -mhtm if available.
13758         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13759         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13760         FPRs instead of df_regs_ever_live_p.
13761         (s390_optimize_nonescaping_tx): New function.
13762         (s390_init_frame_layout): Extend clobbered_regs array to cover
13763         FPRs as well.
13764         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13765         (s390_expand_tbegin): New function.
13766         (enum s390_builtin): New enum definition.
13767         (code_for_builtin): New array definition.
13768         (s390_init_builtins): New function.
13769         (s390_expand_builtin): New function.
13770         (TARGET_INIT_BUILTINS): Define.
13771         (TARGET_EXPAND_BUILTIN): Define.
13772         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13773         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13774         (s390_alc_comparison): Likewise.
13775         * config/s390/s390-modes.def: Add CCRAWmode.
13776         * config/s390/s390.h (processor_flags): Add PF_TX.
13777         (TARGET_CPU_HTM): Define macro.
13778         (TARGET_HTM): Define macro.
13779         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13780         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13781         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13782         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13783         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13784         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13785         ("*cc_to_int"): Move up.
13786         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13787         constants other than 0.
13788         ("*ccraw_to_int"): New insn and splitter definition.
13789         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13790         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13791         ("tx_assist"): New expander.
13792         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13793         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13794         * config/s390/s390.opt: Add -mhtm option.
13795         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13796         * config/s390/htmxlintrin.h: New file.
13797         * config/s390/htmintrin.h: New file.
13798         * config/s390/s390intrin.h: New file.
13799         * doc/extend.texi: Document htm builtins.
13800         * config.gcc: Add the new header files to extra_headers.
13802 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13804         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13805         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13807 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13808             Pat Haugen  <pthaugen@us.ibm.com>
13809             Peter Bergner  <bergner@vnet.ibm.com>
13811         * config/rs6000/power8.md: New.
13812         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13813         setting for power8 entry.
13814         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13815         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13816         test for Power4/Power5 only.
13817         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13818         support.
13819         (force_new_group): Adjust comment.
13820         * config/rs6000/rs6000.md: Include power8.md.
13822 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13824         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13825         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13826         declaration.
13827         (tune_params): Update comment.
13828         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13829         (arm_max_conditional_execute): New function.
13830         (thumb2_final_prescan_insn): Use max_insn_skipped and
13831         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13833 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13835         PR tree-optimization/57705
13836         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13837         SSA_NAME step, provided that it is not defined inside the loop.
13838         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13839         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13841 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13843         PR middle-end/57670
13844         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13845         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13846         calls in the dump.
13847         (ipa_note_param_call): Initialize member_ptr flag.
13848         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13849         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13850         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13851         (ipa_read_indirect_edge_info): Likewise.
13853 2013-06-25  Richard Biener  <rguenther@suse.de>
13855         PR middle-end/56977
13856         * passes.c (init_optimization_passes): Move pass_fold_builtins
13857         and pass_dce earlier with -Og.
13859 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13861         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13862         <BIT_FIELD_REF>: Remove trailing TAB.
13863         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13864         remove blank line.
13866 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13868         PR tree-optimization/57358
13869         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13870         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13871         (ipa_analyze_params_uses): Generate pessimistic info when true.
13873 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13875         PR tree-optimization/57539
13876         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13877         global.inlined_to of the new node to it.  All callers changed.
13878         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13879         inlining_into, pass it to cgraph_clone_node.
13880         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13881         ipa_free_edge_args_substructures.
13882         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13883         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13884         Assert we have found an rdesc in the rdesc list.
13886 2013-06-24  Richard Biener  <rguenther@suse.de>
13888         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13889         (pointer_set_lookup): Declare.
13890         (class pointer_map): New template class implementing a
13891         generic pointer to T map.
13892         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13893         pointer_map<T>::contains, pointer_map<T>::insert,
13894         pointer_map<T>::traverse): New functions.
13895         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13896         (pointer_set_lookup): New function.
13897         (pointer_set_contains): Use pointer_set_lookup.
13898         (pointer_set_insert): Likewise.
13899         (insert_aux): Remove.
13900         (struct pointer_map_t): Embed a pointer_set_t.
13901         (pointer_map_create): Adjust.
13902         (pointer_map_destroy): Likewise.
13903         (pointer_map_contains): Likewise.
13904         (pointer_map_insert): Likewise.
13905         (pointer_map_traverse): Likewise.
13906         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13907         pointer_map<unsigned> instead of a pointer_map_t.
13908         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13909         (streamer_tree_cache_lookup): Likewise.
13910         (streamer_tree_cache_create): Likewise.
13911         (streamer_tree_cache_delete): Likewise.
13912         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13913         pointer_map<unsigned> instead of a pointer_map_t.
13914         (lto_init_tree_ref_encoder): Adjust.
13915         (lto_destroy_tree_ref_encoder): Likewise.
13916         * lto-section-out.c (lto_output_decl_index): Likewise.
13917         (lto_record_function_out_decl_state): Likewise.
13918         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13920 2013-06-24  Richard Biener  <rguenther@suse.de>
13922         PR tree-optimization/57488
13923         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13925 2013-06-24  Alan Modra  <amodra@gmail.com>
13927         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13928         (gen_easy_altivec_constant): Likewise.
13929         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13930         easy_vector_constant_msb): Likewise.
13932 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13934         PR target/57688
13935         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13936         add missing return true.
13938 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13940         PR target/52483
13941         * config/sh/predicates.md (general_extend_operand): Invoke
13942         general_movsrc_operand for memory operands.
13943         (general_movsrc_operand): Allow reg+reg addressing, do not use
13944         general_operand for memory operands.
13946 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13948         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13949         when current target options does not apply.
13950         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13951         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13952         * config/i386/bmiintrin.h: Pass appropriate target
13953         attributes to header.
13954         * config/i386/mmintrin.h: Ditto.
13955         * config/i386/nmmintrin.h: Ditto.
13956         * config/i386/avx2intrin.h: Ditto.
13957         * config/i386/fxsrintrin.h: Ditto.
13958         * config/i386/tbmintrin.h: Ditto.
13959         * config/i386/xsaveintrin.h: Ditto.
13960         * config/i386/f16cintrin.h: Ditto.
13961         * config/i386/xtestintrin.h: Ditto.
13962         * config/i386/xsaveoptintrin.h: Ditto.
13963         * config/i386/bmi2intrin.h: Ditto.
13964         * config/i386/lzcntintrin.h: Ditto.
13965         * config/i386/smmintrin.h: Ditto.
13966         * config/i386/wmmintrin.h: Ditto.
13967         * config/i386/x86intrin.h: Remove all header include guards.
13968         * config/i386/prfchwintrin.h: Ditto.
13969         * config/i386/pmmintrin.h: Ditto.
13970         * config/i386/tmmintrin.h: Ditto.
13971         * config/i386/xmmintrin.h: Ditto.
13972         * config/i386/popcntintrin.h: Ditto.
13973         * config/i386/rdseedintrin.h: Ditto.
13974         * config/i386/ammintrin.h: Ditto.
13975         * config/i386/emmintrin.h: Ditto.
13976         * config/i386/immintrin.h: Remove all header include guards.
13977         * config/i386/fma4intrin.h: Ditto.
13978         * config/i386/lwpintrin.h: Ditto.
13979         * config/i386/xopintrin.h: Ditto.
13980         * config/i386/ia32intrin.h: Ditto.
13981         * config/i386/avxintrin.h: Ditto.
13982         * config/i386/rtmintrin.h: Ditto.
13983         * config/i386/fmaintrin.h: Ditto.
13984         * config/i386/mm3dnow.h: Ditto.
13986 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13988         * common/config/i386/i386-common.c: Handle LZCNT.
13990 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13992         * doc/extend.texi: Use __atomic_store_n instead of
13993         __atomic_store in HLE example.
13995 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13997         * config/sh/sh.c: Remove <cstdlib> workaround.
13999 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14001         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
14003 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14005         * doc/extend.texi: Document that __atomic_clear and
14006           __atomic_test_and_set should only be used with bool.
14008 2013-06-20  Jan Hubicka  <jh@suse.cz>
14010         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
14011         types_same_for_odr.
14012         * tree.c (decls_same_for_odr): New function.
14013         (same_for_edr): New function.
14014         (types_same_for_odr): New function.
14015         (get_binfo_at_offset): Use it.
14016         * tree.h (types_same_for_odr): Declare.
14018 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
14019             Jason Merrill  <jason@redhat.com>
14021         * system.h: Include <cstdlib> as well as <stdlib.h>.
14023 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
14025         PR target/57655
14026         * config/i386/i386.c (construct_container): Report error if
14027         long double is used with disabled x87 float returns.
14029 2013-06-20  Jan Hubicka  <jh@suse.cz>
14031         * lto-cgraph.c (input_symtab): Do not set cgraph state.
14033 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
14035         PR rtl-optimization/57425
14036         PR rtl-optimization/57569
14037         * alias.c (write_dependence_p): Remove parameters mem_mode and
14038         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
14039         Changed all callers.
14040         (canon_anti_dependence): Get comments and semantics in sync.
14041         Add parameter mem_canonicalized.  Changed all callers.
14042         * rtl.h (canon_anti_dependence): Update prototype.
14044 2013-06-20  Richard Biener  <rguenther@suse.de>
14046         * data-streamer-in.c (streamer_read_uhwi): Optimize single
14047         byte case, inline streamer_read_uchar and defer section
14048         overrun check.
14050 2013-06-20  Richard Biener  <rguenther@suse.de>
14052         PR tree-optimization/57584
14053         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
14054         SSA names into the expanded expression that take part in
14055         abnormal coalescing.
14057 2013-06-19  Sharad Singhai  <singhai@google.com>
14059         * gcov.c (print_usage): Handle new option.
14060         (process_args): Ditto.
14061         (get_gcov_intermediate_filename): New function.
14062         (output_intermediate_file): New function.
14063         (output_gcov_file): New function
14064         (generate_results): Handle new option.
14065         (release_function): Relase demangled name.
14066         (read_graph_file): Handle demangled name.
14067         (output_lines): Ditto.
14068         * doc/gcov.texi: Document gcov intermediate format.
14070 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14072         PR bootstrap/57604
14073         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14074         (lra_emit_add): Use the functions.  Add comment about Y as an
14075         address segment.
14077 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14079         PR driver/57652
14080         * collect2.c (collect_atexit): New.
14081         (collect_exit): Delete.
14082         (main): Register collect_atexit with atexit.
14083         (collect_wait): Change collect_exit to exit.
14084         (do_wait): Same.
14085         * collect2.h (collect_exit): Delete.
14086         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14088 2013-06-19  Wei Mi  <wmi@google.com>
14090         PR rtl-optimization/57518
14091         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14092         if regno is used in paradoxical subreg.
14093         (update_equiv_regs): Check pdx_subregs[regno] before
14094         set a reg to be equivalent with a mem.
14096 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14098         PR driver/57651
14099         * file-find.h (find_a_file): Add a mode parameter.
14100         * file-find.c (find_a_file): Likewise.
14101         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14102         with X_OK for the executables.
14103         * collect2.c (main): Call find_a_file with X_OK.
14105 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14107         PR target/56942
14108         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14109         Use NEXT_INSN instead of next_real_insn.
14111 2013-06-19  Jan Hubicka  <jh@suse.cz>
14113         * cgraph.h (const_value_known_p): Replace by ...
14114         (ctor_for_folding): .. this one.
14115         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14116         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14117         * expr.c (expand_expr_real_1): Likewise.
14118         (string_constant): Likewise.
14119         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14120         * ipa.c (process_references): Likewise.
14121         (symtab_remove_unreachable_nodes): Likewise.
14122         * ipa-inline-analysis.c (param_change_prob): Likewise.
14123         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14124         (get_base_constructor): Likwise.
14125         * varpool.c (varpool_remove_node): Likewise.
14126         (varpool_remove_initializer): LIkewise.
14127         (dump_varpool_node): LIkwise.
14128         (const_value_known_p): Rewrite to ...
14129         (ctor_for_folding): ... this one.
14131 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14133         PR driver/57651
14134         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14135         PERSONALITY in $PATH derived prefixes.
14137 2013-06-19  Jeff Law  <law@redhat.com>
14139         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14140         in comment.
14142         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14143         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14144         booleans.
14146 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14148         * config/arm/vfp.md: Move VFP instruction classification documentation
14149         to ...
14150         * config/arm/arm.md: ... here.  Update instruction classification
14151         documentation.
14153 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14155         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14156         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14157         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14159 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14161         PR target/57609
14162         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14163         with NEXT_INSN.  Use tablejump_p to check for jump table data
14164         insns.
14166 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14168         PR c++/56544
14169         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14170         that now in C++ the value is correct per the C++ standards.
14172 2013-06-19  Richard Biener  <rguenther@suse.de>
14174         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14175         for global context.
14177 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14179         Revert:
14180         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14182         PR target/57609
14183         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14184         with next_active_insn.
14186 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14188         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14189         functions are inlined during failures to flag an error.
14190         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14191         in early inline pass.
14193 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14195         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14196         in comments.
14198 2013-06-18  Julian Brown  <julian@codesourcery.com>
14200         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14201         Permit virtual register pre-reload if !strict.
14202         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14203         change.
14204         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14205         prototype.
14206         * config/arm/neon.md (movmisalign<mode>): Use
14207         neon_perm_struct_or_reg_operand instead of
14208         neon_struct_or_register_operand.
14209         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14210         neon_permissive_struct_operand instead of neon_struct_operand.
14211         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14212         neon_vector_mem_operand.
14213         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14214         neon_vector_mem_operand.
14215         (neon_permissive_struct_operand): New.
14216         (neon_struct_or_register_operand): Rename to...
14217         (neon_perm_struct_or_reg_operand): This. Adjust call to
14218         neon_vector_mem_operand.
14220 2013-06-18  Richard Biener  <rguenther@suse.de>
14222         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14223         * lto-streamer.h: Include pointer-set.h.
14224         (struct lto_decl_slot): Remove.
14225         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14226         Remove next_index entry.
14227         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14228         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14229         (lto_init_tree_ref_encoder): Adjust.
14230         (lto_destroy_tree_ref_encoder): Likewise.
14231         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14232         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14233         (lto_output_decl_index): Adjust.
14234         (lto_new_out_decl_state): Likewise.
14235         (lto_record_function_out_decl_state): Likewise.
14236         * lto-streamer-out.c (copy_function): Likewise.
14238 2013-06-18  Richard Biener  <rguenther@suse.de>
14240         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14241         * cgraphunit.c: Include cfgloop.h.
14242         (init_lowered_empty_function): Initialize the loop tree.
14243         (assemble_thunk): Insert new BBs into loops.
14245 2013-06-18  Richard Biener  <rguenther@suse.de>
14247         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14248         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14249         the map from cache entry to cache index optional.
14250         (streamer_tree_cache_replace_tree): Adjust accordingly.
14251         (streamer_tree_cache_append): Likewise.
14252         (streamer_tree_cache_delete): Likewise.
14253         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14254         streamer cache map from cache entry to cache index.
14255         * lto-streamer-out.c (create_output_block): Adjust.
14257 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14259         * config/arm/arm.md (attribute "insn"): Move multiplication and
14260         division attributes to...
14261         (attribute "type"): ... here.  Remove mult.
14262         (attribute "mul32"): New attribute.
14263         (attribute "mul64"): Add umaal.
14264         (*arm_mulsi3): Update attributes.
14265         (*arm_mulsi3_v6): Likewise.
14266         (*thumb_mulsi3): Likewise.
14267         (*thumb_mulsi3_v6): Likewise.
14268         (*mulsi3_compare0): Likewise.
14269         (*mulsi3_compare0_v6): Likewise.
14270         (*mulsi_compare0_scratch): Likewise.
14271         (*mulsi_compare0_scratch_v6): Likewise.
14272         (*mulsi3addsi): Likewise.
14273         (*mulsi3addsi_v6): Likewise.
14274         (*mulsi3addsi_compare0): Likewise.
14275         (*mulsi3addsi_compare0_v6): Likewise.
14276         (*mulsi3addsi_compare0_scratch): Likewise.
14277         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14278         (*mulsi3subsi): Likewise.
14279         (*mulsidi3adddi): Likewise.
14280         (*mulsi3addsi_v6): Likewise.
14281         (*mulsidi3adddi_v6): Likewise.
14282         (*mulsidi3_nov6): Likewise.
14283         (*mulsidi3_v6): Likewise.
14284         (*umulsidi3_nov6): Likewise.
14285         (*umulsidi3_v6): Likewise.
14286         (*umulsidi3adddi): Likewise.
14287         (*umulsidi3adddi_v6): Likewise.
14288         (*smulsi3_highpart_nov6): Likewise.
14289         (*smulsi3_highpart_v6): Likewise.
14290         (*umulsi3_highpart_nov6): Likewise.
14291         (*umulsi3_highpart_v6): Likewise.
14292         (mulhisi3): Likewise.
14293         (*mulhisi3tb): Likewise.
14294         (*mulhisi3bt): Likewise.
14295         (*mulhisi3tt): Likewise.
14296         (maddhisi4): Likewise.
14297         (*maddhisi4tb): Likewise.
14298         (*maddhisi4tt): Likewise.
14299         (maddhidi4): Likewise.
14300         (*maddhidi4tb): Likewise.
14301         (*maddhidi4tt): Likewise.
14302         (divsi3): Likewise.
14303         (udivsi3): Likewise.
14304         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14305         (thumb2_mulsi_short_compare0): Likewise.
14306         (thumb2_mulsi_short_compare0_scratch): Likewise.
14307         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14308         (1020mult2): Likewise.
14309         (1020mult3): Likewise.
14310         (1020mult4): Likewise.
14311         (1020mult5): Likewise.
14312         (1020mult6): Likewise.
14313         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14314         change.
14315         (cortex_a15_mult64): Likewise.
14316         (cortex_a15_sdiv): Likewise.
14317         (cortex_a15_udiv): Likewise.
14318         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14319         (mult2): Likewise.
14320         (mult3): Likewise.
14321         (mult4): Likewise.
14322         (mult5): Likewise.
14323         (mult6): Likewise.
14324         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14325         (pj4_ir_div): Likewise.
14326         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14327         (11_mult2): Likewise.
14328         (11_mult3): Likewise.
14329         (11_mult4): Likewise.
14330         (11_mult5): Likewise.
14331         (11_mult6): Likewise.
14332         (11_mult7): Likewise.
14333         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14334         (cortex_a8_mla): Likewise.
14335         (cortex_a8_mull): Likewise.
14336         (cortex_a8_smulwy): Likewise.
14337         (cortex_a8_smlald): Likewise.
14338         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14339         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14340         (cortex_r4_mul_3): Likewise.
14341         (cortex_r4_mla_4): Likewise.
14342         (cortex_r4_mla_3): Likewise.
14343         (cortex_r4_smlald): Likewise.
14344         (cortex_r4_mull): Likewise.
14345         (cortex_r4_sdiv): Likewise.
14346         (cortex_r4_udiv): Likewise.
14347         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14348         (cortex_a7_idiv): Likewise.
14349         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14350         (9_mult2): Likewise.
14351         (9_mult3): Likewise.
14352         (9_mult4): Likewise.
14353         (9_mult5): Likewise.
14354         (9_mult6): Likewise.
14355         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14356         (cortex_a53_sdiv): Likewise.
14357         (cortex_a53_udiv): Likewise.
14358         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14359         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14360         (mp626_mult2): Likewise.
14361         (mp626_mult3): Likewise.
14362         (mp626_mult4): Likewise.
14363         * config/arm/fa526.md (526_mult1): Update attribute change.
14364         (526_mult2): Likewise.
14365         * config/arm/arm-generic.md (mult): Update attribute change.
14366         (mult_ldsched_strongarm): Likewise.
14367         (mult_ldsched): Likewise.
14368         (multi_cycle): Likewise.
14369         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14370         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14371         (606te_mult2): Likewise.
14372         (606te_mult3): Likewise.
14373         (606te_mult4): Likewise.
14374         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14375         (cortex_a9_mac16): Likewise.
14376         (cortex_a9_multiply): Likewise.
14377         (cortex_a9_mac): Likewise.
14378         (cortex_a9_multiply_long): Likewise.
14379         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14380         (626te_mult2): Likewise.
14381         (626te_mult3): Likewise.
14382         (626te_mult4): Likewise.
14384 2013-06-18  Richard Biener  <rguenther@suse.de>
14386         PR lto/57334
14387         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14389 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14391         PR target/57609
14392         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14393         with next_active_insn.
14395 2013-06-18  Alan Modra  <amodra@gmail.com>
14397         * config/rs6000/rs6000.h (enum data_align): New.
14398         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14399         (DATA_ABI_ALIGNMENT): Define.
14400         (CONSTANT_ALIGNMENT): Correct comment.
14401         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14402         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14404 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14406         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14407         ATTRIBUTE_UNUSED marking.
14409 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14411         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14412         alternative and update.
14413         (aarch64_dup_lanedi): Delete.
14414         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14415         * config/aarch64/aarch64-simd-builtins.def: Update.
14417 2013-06-17  Richard Biener  <rguenther@suse.de>
14419         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14420         (lto_input_scc): Declare.
14421         (lto_input_tree_1): Likewise.
14422         (struct lto_stats_d): Add num_tree_bodies_output and
14423         num_pickle_refs_output.
14424         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14425         (lto_read_tree_1): Split out from ...
14426         (lto_read_tree): ... this.
14427         (lto_input_scc): New function.
14428         (lto_input_tree_1): Split out from ...
14429         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14430         (lto_data_in_create): Create the streamer cache without hashes.
14431         * lto-streamer-out.c (create_output_block): Create the streamer
14432         cache with hashes when not doing WPA.
14433         (lto_write_tree_1): Split out from ...
14434         (lto_write_tree): ... this.
14435         (get_symbol_initial_value): New function.
14436         (lto_output_tree_1): Split out from ...
14437         (lto_output_tree): ... this.  Write trees as series of SCCs
14438         using a DFS walk via DFS_write_tree.
14439         (struct sccs, struct scc_entry): New types.
14440         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14441         (DFS_write_tree_body): New function.
14442         (DFS_write_tree): Likewise.
14443         (hash_tree): Likewise.
14444         (scc_entry_compare): Likewise.
14445         (hash_scc): Likewise.
14446         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14447         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14448         TREE_CHAIN as regular reference.
14449         (streamer_read_integer_cst): Remove.
14450         (streamer_get_pickled_tree): Adjust.
14451         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14452         of DECL_EXTERNALs in BLOCK_VARS for now.
14453         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14454         reference.
14455         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14456         Add hash value argument and record that if hashes are recorded
14457         in the cache.
14458         (streamer_tree_cache_insert_1): Adjust.
14459         (streamer_tree_cache_insert): Likewise.
14460         (streamer_tree_cache_insert_at): Rename to ...
14461         (streamer_tree_cache_replace_tree): ... this and adjust.
14462         (streamer_tree_cache_append): Adjust.
14463         (record_common_node): Likewise.
14464         (streamer_tree_cache_create): Add argument whether to
14465         record hash values together with trees.
14466         (streamer_tree_cache_delete): Adjust.
14467         * tree-streamer.h (struct streamer_tree_cache_d): Add
14468         vector of hashes.
14469         (streamer_read_integer_cst): Remove.
14470         (streamer_tree_cache_insert): Adjust.
14471         (streamer_tree_cache_append): Likewise.
14472         (streamer_tree_cache_insert_at): Rename to ...
14473         (streamer_tree_cache_replace_tree): ... this and adjust.
14474         (streamer_tree_cache_create): Add argument whether to record hashes.
14475         (streamer_tree_cache_get): Rename to ...
14476         (streamer_tree_cache_get_tree): ... this.
14477         (streamer_tree_cache_get_hash): New function.
14478         * tree.c (cache_integer_cst): New function.
14479         * tree.h (cache_integer_cst): Declare.
14480         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14481         * lto-symtab.c (lto_varpool_replace_node): Only release
14482         DECL_INITIAL of non-prevailing decls.
14483         * varpool.c (varpool_remove_initializer): Do not release
14484         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14486 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14488         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14489         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14490         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14491         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14492         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14493         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14494         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14495         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14496         instead of TARGET_64BIT.
14497         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14498         Require ISA_HAS_<D>DIV.
14500 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14502         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14503         (mips*-*-linux*): Move default with_llsc setting to where other
14504         defaults are set.
14505         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14506         with_arch block.
14507         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14508         Likewise.  Remove default with_tune setting.  Move default float
14509         setting to its own block.  Handle with_llsc in the same block as above.
14511 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14513         PR rtl-optimization/57425
14514         PR rtl-optimization/57569
14515         * alias.c (write_dependence_p): Add new parameters mem_mode,
14516         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14517         Changed all callers.
14518         (canon_anti_dependence): New function.
14519         * cse.c (check_dependence): Use canon_anti_dependence.
14520         * cselib.c (cselib_invalidate_mem): Likewise.
14521         * rtl.h (canon_anti_dependence): Declare.
14523 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14525         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14526         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14527         ".set mips3" for 64-bit targets.
14529 2013-06-15  Dehao Chen  <dehao@google.com>
14531         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14532         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14533         (gimple_check_call_args): Likewise.
14534         * value-prof.c (check_ic_target): Likewise.
14535         * ipa-inline.c (early_inliner): Likewise.
14536         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14537         * cgraph.c (cgraph_create_edge_1): Likewise.
14538         (cgraph_make_edge_direct): Likewise.
14540 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14542         PR target/57615
14543         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14544         rs6000_output_move_128bit to handle emitting quad memory
14545         operations.  Set attribute length to 8 bytes.
14547 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14549         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14550         New pattern.
14551         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14552         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14553         (aarch64_<su>mlsl<mode>): Likewise.
14555 2013-06-14  Mike Stump  <mikestump@comcast.net>
14557         * Makefile.in (TARGET_H): Add insn-codes.h.
14559 2013-06-14  Alan Modra  <amodra@gmail.com>
14561         PR middle-end/57134
14562         PR middle-end/57586
14563         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14564         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14565         bitfield expansion when EXPAND_MEMORY.
14566         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14568 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14570         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14571         test for clearing quad memory on 32-bit later.
14573 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14575         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14576         (fold_negate_expr): Likewise.
14577         (fold_real_zero_addition_p): Handle vectors.
14578         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14580 2013-06-14  Alan Modra  <amodra@gmail.com>
14582         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14584 2013-06-13  Jan Hubicka  <jh@suse.cz>
14586         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14587         Local comdats are not externally visible.
14588         * symtab.c (dump_symtab_base): Dump externally visible.
14589         (verify_symtab_base): Verify back links in the symtab hash.
14591 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14593         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14594         CONVERT_EXPR as equal nodes.
14596 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14598         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14600 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14602         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14603         Generalize to complex and vector.
14604         * tree.c (build_all_ones_cst): New function.
14605         * tree.h (build_all_ones_cst): Declare it.
14607 2013-06-13  Alan Modra  <amodra@gmail.com>
14609         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14610         * config/rs6000/rs6000.md (signbittf2): New insn.
14611         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14612         (abstf2_internal, cmptf_internal2): Likewise.
14613         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14615 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14616             Pat Haugen  <pthaugen@us.ibm.com>
14617             Peter Bergner  <bergner@vnet.ibm.com>
14619         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14620         power8 byte, half-word, and quad-word atomic instructions.
14621         (emit_store_conditional): Likewise.
14622         (rs6000_expand_atomic_compare_and_swap): Likewise.
14623         (rs6000_expand_atomic_op): Likewise.
14625         * config/rs6000/sync.md (larx): Add new modes for power8.
14626         (stcx): Likewise.
14627         (AINT): New mode iterator to include TImode as well as normal
14628         integer modes on power8.
14629         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14630         that VSX registers are not considered.  Use AINT mode iterator
14631         instead of INT1 to allow inclusion of quad word atomic operations
14632         on power8.
14633         (load_locked<mode>): Likewise.
14634         (store_conditional<mode>): Likewise.
14635         (atomic_compare_and_swap<mode>): Likewise.
14636         (atomic_exchange<mode>): Likewise.
14637         (atomic_nand<mode>): Likewise.
14638         (atomic_fetch_<fetchop_name><mode>): Likewise.
14639         (atomic_nand_fetch<mode>): Likewise.
14640         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14641         each type.
14642         (ATOMIC): On power8, add QImode, HImode modes.
14643         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14644         modes that promote to SImode.
14645         (load_lockedti): Convert TImode arguments to PTImode, so that we
14646         get a guaranteed even/odd register pair.
14647         (load_lockedpti): Likewise.
14648         (store_conditionalti): Likewise.
14649         (store_conditionalpti): Likewise.
14651         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14652         atomic load/store instructions.
14653         (HSI): Likewise.
14655 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14657         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14658         loads.
14659         (insn_count): New attribute, with most cases extracted from...
14660         (length): ...here.  Redefine most cases in terms of insn_count.
14661         (single_insn): Delete.
14662         (can_delay): Use insn_count to check for single instructions.
14663         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14664         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14665         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14666         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14667         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14668         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14669         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14670         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14671         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14672         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14673         rather than "length".
14674         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14675         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14676         Use "insn_count" rather than "length".
14677         * config/mips/mips-dsp.md
14678         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14679         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14680         length attributes.
14682 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14684         PR tree-optimization/57361
14685         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14687 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14689         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14690         to split.
14691         (aarch64_simd_combine<mode>): New instruction expansion.
14692         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14693         function prototype.
14694         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14695         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14697 2013-06-12  Jan Hubicka  <jh@suse.cz>
14699         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14700         decl has when in streaming stage.
14701         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14702         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14704 2013-06-12  Roland Stigge  <stigge@antcom.de>
14706         PR target/57578
14707         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14709 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14711         PR tree-optimization/57537
14712         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14713         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14715 2013-06-12  Richard Biener  <rguenther@suse.de>
14717         * data-streamer.h (streamer_write_char_stream): CSE
14718         obs->current_pointer.
14719         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14720         streamer_write_char_stream manually and optimize the resulting loop.
14721         (streamer_write_hwi_stream): Likewise.
14723 2013-06-12  Jan Hubicka  <jh@suse.cz>
14725         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14726         * cgraph.h (varpool_create_empty_node): Declare.
14727         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14728         duplicated nodes.
14729         * symtab.c (symtab_unregister_node): Be lax about missin entries
14730         in node hash.
14731         (symtab_get_node): Update comment.
14732         * varpool.c (varpool_create_empty_node): Break out from ...
14733         (varpool_node_for_decl): ... here.
14734         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14736 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14738         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14739         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14740         part.  Use straight-line flow at the end.
14741         <COMPONENT_REF>: Remove superfluous else.
14742         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14744 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14746         PR target/56564
14747         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14748         target hook even for !TREE_PUBLIC decls.  If no resolution info
14749         is available, return false for common and external decls.
14751 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14753         * config/rl78/constraints.md (U): New constraint.
14754         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14755         valloc attribute.
14757 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14759         PR target/57589
14760         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14761         to allow returning address to AT_PLATFORM name.
14763 2013-06-11  Jan Hubicka  <jh@suse.cz>
14765         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14766         * cgraph.h (symtab_node_base): Add weakref flag.
14767         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14768         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14769         (output_weakrefs): Use weakref flag.
14770         * fold-const.c (simple_operand_p): Handle WEAK.
14771         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14772         * ipa.c (varpool_externally_visible_p): Drop weakref.
14773         (function_and_variable_visibility): Update comment; fix weakref
14774         sanity checks; do not clear DECL_WEAK on them.
14775         * lto-cgraph.c (lto_output_node): update.
14776         (lto_output_varpool_node): Update.
14777         (input_overwrite_node): Update.
14778         (input_node): Update.
14779         (input_varpool_node): Update.
14780         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14781         (lto_symtab_merge_symbols): Add sanity check.
14782         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14783         * passes.c (rest_of_decl_compilation): Set static flag, too.
14784         * symtab.c (dump_symtab_base): Dump weakref.
14785         (verify_symtab_base): Sanity check weakrefs.
14786         (symtab_make_decl_local): Remove duplicated code.
14787         (symtab_alias_ultimate_target): Simplify.
14788         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14790 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14792         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14793         sequence_vect == NULL.
14795 2013-06-11  DJ Delorie  <dj@redhat.com>
14797         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14798         (rl78_unwind_word_mode): New.
14800 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14802         * final.c (debug_prefix_maps): Make static.
14804 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14806         * function.c (initial_trampoline): Remove stray copy.
14808 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14810         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14812 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14814         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14815         within bounds at the beginning of the function.
14817 2013-06-11  Alan Modra  <amodra@gmail.com>
14819         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14820         reporting.
14821         (get_named_section): Don't NULL !DECL_P decl.
14823 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14825         * doc/invoke.texi (core-avx2): Document.
14826         (slm): Likewise.
14827         (atom): Updated with MOVBE.
14829 2013-06-11  Richard Biener  <rguenther@suse.de>
14831         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14833 2013-06-11  Anton Blanchard  <anton@samba.org>
14835         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14836         correct shift value in little-endian mode.
14838 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14840         PR target/56564
14841         * varasm.c (get_variable_align): Move #endif to the right place.
14843 2013-06-10  Cary Coutant  <ccoutant@google.com>
14845         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14846         for hash so that hash table traversal order is deterministic.
14848 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14849             Pat Haugen  <pthaugen@us.ibm.com>
14850             Peter Bergner  <bergner@vnet.ibm.com>
14852         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14853         of vectors in GPRS if they are direct moves or quad word load or
14854         store moves.
14856         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14857         declaration.
14858         (direct_move_p): Likewise.
14859         (quad_load_store_p): Likewise.
14861         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14862         classes into bins based on the physical register type.
14863         (reg_class_to_reg_type): Likewise.
14864         (IS_STD_REG_TYPE): Likewise.
14865         (IS_FP_VECT_REG_TYPE): Likewise.
14866         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14867         use direct move instructions.
14868         (reload_gpr_vsx): Likewise.
14869         (reload_vsx_gpr): Likewise.
14870         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14871         information that is a simplification of register classes.  Also
14872         precalculate direct move reload helpers.
14873         (direct_move_p): New function to return true if the operation can
14874         be done as a direct move instruciton.
14875         (quad_load_store_p): New function to return true if the operation
14876         is a quad memory operation.
14877         (rs6000_legitimize_address): If quad memory, only allow register
14878         indirect for TImode addresses.
14879         (rs6000_legitimate_address_p): Likewise.
14880         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14881         (rs6000_reload_register_type): Likewise.
14882         (register_to_reg_type): Return register type.
14883         (rs6000_secondary_reload_simple_move): New helper function for
14884         secondary reload and secondary memory needed to identify anything
14885         that is a simple move, and does not need reloading.
14886         (rs6000_secondary_reload_direct_move): New helper function for
14887         secondary reload to identify cases that can be done with several
14888         instructions via the direct move instructions.
14889         (rs6000_secondary_reload_move): New helper function for secondary
14890         reload to identify moves between register types that can be done.
14891         (rs6000_secondary_reload): Add support for quad memory operations
14892         and for direct move.
14893         (rs6000_secondary_memory_needed): Likewise.
14894         (rs6000_debug_secondary_memory_needed): Change argument names.
14895         (rs6000_output_move_128bit): New function to return the move to
14896         use for 128-bit moves, including knowing about the various
14897         limitations of quad memory operations.
14899         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14900         memory operations.  call rs6000_output_move_128bit for the actual
14901         instruciton(s) to generate.
14902         (vsx_movti_64bit): Likewise.
14904         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14905         (UNSPEC_P8V_MTVSRWZ): Likewise.
14906         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14907         (UNSPEC_P8V_MTVSRD): Likewise.
14908         (UNSPEC_P8V_XXPERMDI): Likewise.
14909         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14910         (UNSPEC_FUSION_GPR): Likewise.
14911         (FMOVE128_GPR): New iterator for direct move.
14912         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14913         (f32_sv): Likewise.
14914         (f32_dm): Likewise.
14915         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14916         loads and direct move instructions.
14917         (zero_extendsidi2_lfiwzx): Likewise.
14918         (extendsidi2_lfiwax): Likewise.
14919         (extendsidi2_nocell): Likewise.
14920         (floatsi<mode>2_lfiwax): Likewise.
14921         (lfiwax): Likewise.
14922         (floatunssi<mode>2_lfiwzx): Likewise.
14923         (lfiwzx): Likewise.
14924         (fix_trunc<mode>_stfiwx): Likewise.
14925         (fixuns_trunc<mode>_stfiwx): Likewise.
14926         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14927         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14928         (parity<mode>2_cmpb): Set length/type attr.
14929         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14930         for 'mr.' to fast_compare.
14931         (bpermd_<mode>): Change type attr to popcnt.
14932         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14933         (p8_mtvsrwz_1): Likewise.
14934         (p8_mtvsrwz_2): Likewise.
14935         (reload_fpr_from_gpr<mode>): Likewise.
14936         (p8_mtvsrd_1): Likewise.
14937         (p8_mtvsrd_2): Likewise.
14938         (p8_xxpermdi_<mode>): Likewise.
14939         (reload_vsx_from_gpr<mode>): Likewise.
14940         (reload_vsx_from_gprsf): Likewise.
14941         (p8_mfvsrd_3_<mode>): LIkewise.
14942         (reload_gpr_from_vsx<mode>): Likewise.
14943         (reload_gpr_from_vsxsf): Likewise.
14944         (p8_mfvsrd_4_disf): Likewise.
14945         (multi-word GPR splits): Do not split direct moves or quad memory
14946         operations.
14948 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14950         * tree-into-ssa.c (interesting_blocks): Make static.
14952 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14954         PR target/56564
14955         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14956         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14957         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14958         (get_variable_align): New function.
14959         (get_variable_section, emit_bss, emit_common,
14960         assemble_variable_contents, place_block_symbol): Use
14961         get_variable_align instead of DECL_ALIGN.
14962         (assemble_noswitch_variable): Add align argument, use it
14963         instead of DECL_ALIGN.
14964         (assemble_variable): Adjust caller.  Use get_variable_align
14965         instead of DECL_ALIGN.
14966         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14967         caller.
14968         (DATA_ABI_ALIGNMENT): Define.
14969         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14970         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14971         opt is false, only return the psABI mandated alignment increase.
14972         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14973         (DATA_ABI_ALIGNMENT): ... this.
14974         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14975         (DATA_ABI_ALIGNMENT): ... this.
14976         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14977         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14978         (DATA_ABI_ALIGNMENT): ... this.
14979         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14980         * doc/tm.texi: Regenerated.
14982 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14984         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14985         cmp_code to construct REG_EQUAL note.
14987 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14989         PR target/57568
14990         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14991         that operands[2] doesn't overlap with operands[0].
14993 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14994             Jan Hubicka  <jh@suse.cz>
14996         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14997         hack to mark symbols as used.
14999 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
15001         PR rtl-optimization/57559
15002         * lra-constraints.c (process_alt_operands): Don't discourage
15003         memory with known offset for offsetable memory constraint.
15004         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
15006 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
15008         * varasm.c (struct oc_local_state): Reorder fields.
15009         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
15010         and adjust accordingly.
15011         (output_constructor): Reorder initialization code and adjust call to
15012         output_constructor_bitfield.
15014 2013-06-07  Jan Hubicka  <jh@suse.cz>
15016         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
15018 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
15020         * tree-object-size.c (unknown): Make const.
15022 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15024         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
15025         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
15026         for last alternative in the cpu_facility attribute.
15028 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15030         PR target/56315
15031         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
15032         (xordi3): Change operand 2 constraint to arm_xordi_operand.
15033         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
15034         * config/arm/constraints.md (Dg): New constraint.
15035         * config/arm/neon.md (xordi3_neon): Remove.
15036         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
15037         * config/arm/predicates.md (arm_xordi_operand): New predicate.
15039 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15041         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
15042         Clean up alternatives.
15044 2013-06-07  Alan Modra  <amodra@gmail.com>
15046         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
15047         va_list_gpr_size.
15049 2013-06-07  Alan Modra  <amodra@gmail.com>
15051         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
15053 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15055         * config/arm/constraints.md (Df): New constraint.
15056         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
15057         Correct length attribute for last two alternatives.
15059 2013-06-07  Alan Modra  <amodra@gmail.com>
15061         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
15062         override user -mfp-in-toc.
15063         (offsettable_ok_by_alignment): Consider just the current access
15064         rather than the whole object, unless BLKmode.  Handle
15065         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
15066         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
15067         for -mcmodel=medium.
15068         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15069         override user -mfp-in-toc or -msum-in-toc.  Default to
15070         -mno-fp-in-toc for -mcmodel=medium.
15072 2013-06-06  DJ Delorie  <dj@redhat.com>
15074         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15075         TARGET_VALID_POINTER_MODE.
15077 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15078             Pat Haugen  <pthaugen@us.ibm.com>
15079             Peter Bergner  <bergner@vnet.ibm.com>
15081         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15082         Document new power8 builtins.
15084         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15085         condition code register, to allow 128-bit logical operations to be
15086         done in the VSX or GPR registers.
15087         (nor<mode>3): Use the canonical form for nor.
15088         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15089         vclz*, and vpopcnt* vector instructions.
15090         (nand<mode>3): Likewise.
15091         (orc<mode>3): Likewise.
15092         (clz<mode>2): LIkewise.
15093         (popcount<mode>2): Likewise.
15095         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15096         that only the GPRs are recognized.
15098         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15099         support for new power8 builtins.
15101         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15102         builtin functions.
15103         (xscvdpspn): Likewise.
15104         (vclz): Likewise.
15105         (vclzb): Likewise.
15106         (vclzh): Likewise.
15107         (vclzw): Likewise.
15108         (vclzd): Likewise.
15109         (vpopcnt): Likewise.
15110         (vpopcntb): Likewise.
15111         (vpopcnth): Likewise.
15112         (vpopcntw): Likewise.
15113         (vpopcntd): Likewise.
15114         (vgbbd): Likewise.
15115         (vmrgew): Likewise.
15116         (vmrgow): Likewise.
15117         (eqv): Likewise.
15118         (eqv_v16qi3): Likewise.
15119         (eqv_v8hi3): Likewise.
15120         (eqv_v4si3): Likewise.
15121         (eqv_v2di3): Likewise.
15122         (eqv_v4sf3): Likewise.
15123         (eqv_v2df3): Likewise.
15124         (nand): Likewise.
15125         (nand_v16qi3): Likewise.
15126         (nand_v8hi3): Likewise.
15127         (nand_v4si3): Likewise.
15128         (nand_v2di3): Likewise.
15129         (nand_v4sf3): Likewise.
15130         (nand_v2df3): Likewise.
15131         (orc): Likewise.
15132         (orc_v16qi3): Likewise.
15133         (orc_v8hi3): Likewise.
15134         (orc_v4si3): Likewise.
15135         (orc_v2di3): Likewise.
15136         (orc_v4sf3): Likewise.
15137         (orc_v2df3): Likewise.
15139         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15140         allow power8 quad mode in 64-bit.
15141         (rs6000_builtin_vectorized_function): Add support to vectorize
15142         ISA 2.07 count leading zeros, population count builtins.
15143         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15144         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15145         (builtin_function_type): Add vgbbd builtin function which takes an
15146         unsigned argument.
15147         (altivec_expand_vec_perm_const): Add support for new power8 merge
15148         instructions.
15150         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15151         that does not include TImdoe for use with 32-bit.
15152         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15153         instructions.
15154         (UNSPEC_VSX_CVDPSPN): Likewise.
15155         (vsx_xscvdpspn): Likewise.
15156         (vsx_xscvspdpn): Likewise.
15157         (vsx_xscvdpspn_scalar): Likewise.
15158         (vsx_xscvspdpn_directmove): Likewise.
15159         (vsx_and<mode>3): Split logical operations into 32-bit and
15160         64-bit. Add support to do logical operations on TImode as well as
15161         VSX vector types.  Allow logical operations to be done in either
15162         VSX registers or in general purpose registers in 64-bit mode.  Add
15163         splitters if GPRs were used. For AND, add clobber of CCmode to
15164         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15165         encoding.
15166         (vsx_and<mode>3_32bit): Likewise.
15167         (vsx_and<mode>3_64bit): Likewise.
15168         (vsx_ior<mode>3): Likewise.
15169         (vsx_ior<mode>3_32bit): Likewise.
15170         (vsx_ior<mode>3_64bit): Likewise.
15171         (vsx_xor<mode>3): Likewise.
15172         (vsx_xor<mode>3_32bit): Likewise.
15173         (vsx_xor<mode>3_64bit): Likewise.
15174         (vsx_one_cmpl<mode>2): Likewise.
15175         (vsx_one_cmpl<mode>2_32bit): Likewise.
15176         (vsx_one_cmpl<mode>2_64bit): Likewise.
15177         (vsx_nor<mode>3): Likewise.
15178         (vsx_nor<mode>3_32bit): Likewise.
15179         (vsx_nor<mode>3_64bit): Likewise.
15180         (vsx_andc<mode>3): Likewise.
15181         (vsx_andc<mode>3_32bit): Likewise.
15182         (vsx_andc<mode>3_64bit): Likewise.
15183         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15184         and xxlorc instructions.
15185         (vsx_eqv<mode>3_64bit): Likewise.
15186         (vsx_nand<mode>3_32bit): Likewise.
15187         (vsx_nand<mode>3_64bit): Likewise.
15188         (vsx_orc<mode>3_32bit): Likewise.
15189         (vsx_orc<mode>3_64bit): Likewise.
15191         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15193         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15194         instruction.
15195         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15196         (p8_vmrgow): Likewise.
15197         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15198         GPRs to be split under VSX.
15199         (p8v_clz<mode>2): Add power8 count leading zero support.
15200         (p8v_popcount<mode>2): Add power8 population count support.
15201         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15202         support.
15204         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15205         instruction.
15207         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15208         builtin functions.
15209         (vec_nand): Likewise.
15210         (vec_vclz): Likewise.
15211         (vec_vclzb): Likewise.
15212         (vec_vclzd): Likewise.
15213         (vec_vclzh): Likewise.
15214         (vec_vclzw): Likewise.
15215         (vec_vgbbd): Likewise.
15216         (vec_vmrgew): Likewise.
15217         (vec_vmrgow): Likewise.
15218         (vec_vpopcnt): Likewise.
15219         (vec_vpopcntb): Likewise.
15220         (vec_vpopcntd): Likewise.
15221         (vec_vpopcnth): Likewise.
15222         (vec_vpopcntw): Likewise.
15224 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15226         PR rtl-optimization/57468
15227         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15228         spilled pseudos.
15230 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15232         PR rtl-optimization/57459
15233         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15234         type when setting live regs.
15236 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15238         * config/s390/s390.opt (mlra): New option.
15239         * config/s390/s390.c (s390_decompose_address): Check displacement
15240         for all registers for LRA.
15241         (s390_secondary_reload): Don't used secondary reloads for LRA.
15242         (s390_lra_p): New function.
15243         (TARGET_LRA_P): Define.
15244         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15245         of attribute cpu_facility to zarch for the last alternative.
15246         (*cmpmem_short): Ditto.
15248 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15250         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15251         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15252         (arm_expand_prologue): Likewise.
15254 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15256         PR c++/53743
15257         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15258         as this is now done by redirect_edge_and_branch_force.
15259         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15260         barriers, and fix interaction with splitting.
15261         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15262         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15263         reflect changes made in the routine.
15264         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15265         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15266         since this is called in cfglayout mode, and replace partition fixup
15267         with assert as that is now done by force_nonfallthru_and_redirect.
15268         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15269         already be marked with region crossing note.
15270         (insert_section_boundary_note): Make non-static, gate on flag
15271         has_bb_partition, rewrite to also check for multiple partitions.
15272         (rest_of_handle_reorder_blocks): Remove call to
15273         insert_section_boundary_note, now done later during free_cfg.
15274         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15275         * bb-reorder.h (insert_section_boundary_note): Declare.
15276         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15277         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15278         invoke insert_section_boundary_note.
15279         (try_redirect_by_replacing_jump): Remove unnecessary
15280         check for region crossing note.
15281         (fixup_partition_crossing): New function.
15282         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15283         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15284         in non-cfglayout mode.
15285         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15286         remove old code that tried to do this. Emit barrier correctly
15287         when we are in cfglayout mode.
15288         (last_bb_in_partition): New function.
15289         (rtl_split_edge): Correctly fixup partition boundaries.
15290         (commit_one_edge_insertion): Remove old code that tried to
15291         fixup region crossing edge since this is now handled in
15292         split_block, and set up insertion point correctly since
15293         block may now end in a jump.
15294         (verify_hot_cold_block_grouping): Guard against checking when not in
15295         linearized RTL mode.
15296         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15297         notes.
15298         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15299         rtl_verify_flow_info, so not called in cfglayout mode.
15300         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15301         (fixup_reorder_chain): Remove old code that attempted to fixup region
15302         crossing note as this is now handled in force_nonfallthru_and_redirect.
15303         (duplicate_insn_chain): Don't duplicate switch section notes.
15304         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15305         note.
15306         * basic-block.h (emit_barrier_after_bb): Declare.
15308 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15310         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15311         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15312         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15314 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15316         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15317         where appropriate.
15318         * config/arm/ldmstm.md: Regenerate.
15320 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15322         * config/arm/sync.md (atomic_loaddi_1):
15323         Disable predication for arm_restrict_it.
15324         (arm_load_exclusive<mode>): Likewise.
15325         (arm_load_exclusivesi): Likewise.
15326         (arm_load_exclusivedi): Likewise.
15327         (arm_load_acquire_exclusive<mode>): Likewise.
15328         (arm_load_acquire_exclusivesi): Likewise.
15329         (arm_load_acquire_exclusivedi): Likewise.
15330         (arm_store_exclusive<mode>): Likewise.
15331         (arm_store_exclusive<mode>): Likewise.
15332         (arm_store_release_exclusivedi): Likewise.
15333         (arm_store_release_exclusive<mode>): Likewise.
15335 2013-06-06  Richard Biener  <rguenther@suse.de>
15337         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15338         after LTO_null.
15339         (lto_tag_is_tree_code_p): Adjust.
15340         (lto_tag_is_gimple_code_p): Likewise.
15341         (lto_gimple_code_to_tag): Likewise.
15342         (lto_tag_to_gimple_code): Likewise.
15343         (lto_tree_code_to_tag): Likewise.
15344         (lto_tag_to_tree_code): Likewise.
15345         * data-streamer.h (streamer_write_hwi_in_range): Use
15346         uhwi streaming to stream the normalized range.
15347         (streamer_read_hwi_in_range): Likewise.
15349 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15351         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15352         (predicable_short_it): Likewise.
15353         (predicated): Likewise.
15354         (enabled): Handle above.
15355         (define_cond_exec): Set predicated attribute to yes.
15357 2013-06-05  Mike Stump  <mikestump@comcast.net>
15359         * gdbinit.in (__FUNCTION__): Add.
15361 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15363         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15364         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15366 2013-06-05  Jan Hubicka  <jh@suse.cz>
15368         * varasm.c (mark_decl_referenced): Revert the removal until targets
15369         are fixed.
15371 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15373         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15374         instead of mark_decl_referenced.
15376 2013-06-05  Jan Hubicka  <jh@suse.cz>
15378         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15379         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15380         and symtab_used_from_object_file_p.
15381         (cgraph_make_node_local_1): Clear forced_by_abi.
15382         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15383         * cgraph.h (symtab_node_base): Add forced_by_abi.
15384         (decide_is_variable_needed): Remove.
15385         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15386         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15387         (decide_is_symbol_needed): ... this one; handle symbols in general;
15388         always analyze virtuals; honnor forced_by_abi.
15389         (cgraph_finalize_function): Update.
15390         (varpool_finalize_decl): Update.
15391         (symbol_defined_and_needed): Remove.
15392         (analyze_functions): Update.
15393         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15394         output_refs, input_overwrite_node): Handle forced_by_abi.
15395         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15396         (address_taken_from_non_vtable_p): ... this one.
15397         (comdat_can_be_unshared_p_1): New function.
15398         (cgraph_comdat_can_be_unshared_p): Rename to ...
15399         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15400         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15401         (function_and_variable_visibility): Clear forced_by_abi as needed.
15402         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15403         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15404         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15405         * varpool.c (decide_is_variable_needed): Remove.
15407 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15409         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15410         (arm_option_override): Override arm_restrict_it where appropriate.
15411         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15412         * config/arm/arm.opt (mrestrict-it): New command-line option.
15413         * doc/invoke.texi: Document -mrestrict-it.
15415 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15417         * tsan.c (tsan_atomic_table): Make const.
15419 2013-06-05  Richard Biener  <rguenther@suse.de>
15421         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15422         index associated with the tree we are supposed to replace.
15423         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15424         TREE_ASM_WRITTEN as zero for everything but SSA names.
15426 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15428         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15430 2013-06-04  Jan Hubicka  <jh@suse.cz>
15432         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15433         (input_node, input_varpool_node): Handle correctly external same
15434         body aliases.
15435         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15436         nodes at ltrans stage.
15438 2013-06-04  Jan Hubicka  <jh@suse.cz>
15440         * ipa-inline.c (update_caller_keys): Fix availability test.
15441         (update_callee_keys): Likewise.
15442         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15443         to follow ELF standard.
15445 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15447         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15448         (mips64r5900el-*-elf*): New configurations.
15449         * config/mips/mips-cpus.def (r5900): New processor.
15450         * config/mips/mips-tables.opt: Regenerate.
15451         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15452         (mips_issue_rate): Handle PROCESSOR_R5900.
15453         (mips_reorg_process_insns): Force reorder mode for the R5900.
15454         * config/mips/mips.h (TARGET_MIPS5900): Define.
15455         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15456         TARGET_MIPS5900.
15457         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15458         TARGET_MIPS5900.
15459         * config/mips/mips.md (processor): Add r5900.
15460         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15462 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15464         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15465         into function to generate MOVI instruction.
15466         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15467         (aarch64_preferred_simd_mode): Turn into wrapper.
15468         (aarch64_output_scalar_simd_mov_immediate): New function.
15469         * config/aarch64/aarch64-protos.h: Add prototype for above.
15471 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15473         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15474         element_char member.
15475         (sizetochar): Return signed char.
15476         (aarch64_simd_valid_immediate): Remove elchar and other
15477         unnecessary variables.
15478         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15479         Calculate element_char as required.
15480         * config/aarch64/aarch64-protos.h: Update and move prototype
15481         for aarch64_output_simd_mov_immediate.
15482         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15483         Update arguments.
15485 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15487         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15488         information completed by aarch64_simd_valid_immediate.
15489         (aarch64_legitimate_constant_p): Update arguments.
15490         (aarch64_simd_valid_immediate): Work with struct rather than many
15491         pointers.
15492         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15493         (aarch64_simd_make_constant): Update arguments.
15494         (aarch64_output_simd_mov_immediate): Work with struct rather than
15495         many pointers.  Output immediate directly rather than as operand.
15496         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15497         Update prototype.
15498         * config/aarch64/constraints.md (Dn): Update arguments.
15500 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15502         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15503         longer static.
15504         (aarch64_simd_immediate_valid_for_move): Remove.
15505         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15506         (aarch64_simd_make_constant): Update call.
15507         (aarch64_output_simd_mov_immediate): Update call.
15508         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15509         Add prototype.
15510         * config/aarch64/constraints.md (Dn): Update call.
15512 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15514         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15515         return type to bool for prototype.
15516         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15517         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15518         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15520 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15522         * config/mips/mips.opt (meva): New.
15523         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15524         (ASM_SPEC): Handle -meva.
15525         * doc/invoke.texi (meva):  Document.
15527 2013-06-04  Alan Modra  <amodra@gmail.com>
15529         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15530         constant output.
15532 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15534         * rtl.def: Add extra fourth optional field to define_cond_exec.
15535         * gensupport.c (process_one_cond_exec): Process attributes from
15536         define_cond_exec.
15537         * doc/md.texi: Document fourth field in define_cond_exec.
15539 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15541         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15542         out the processing order as in store_bit_field_1.
15544 2013-06-04  Jan Hubicka  <jh@suse.cz>
15546         PR middle-end/57500
15547         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15548         non-VAR_DECL node if it does not exist yet.
15550 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15552         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15553         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15554         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15555         target_cpu_default setting.
15557 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15559         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15560         default to -fopt-info=optimized instead of all.
15561         * doc/invoke.texi: Ditto.
15562         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15563         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15564         (execute_vect_slp): Emit BB vectorization success under
15565         MSG_OPTIMIZED_LOCATIONS.
15566         * tree-vect-slp.c (vect_slp_transform_bb): Change
15567         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15568         * tree-vect-loop.c (vect_transform_loop): Ditto.
15570 2013-06-03  Jason Merrill  <jason@redhat.com>
15572         PR c++/57415
15573         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15574         Use TARGET_EXPR for C++.
15576 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15578         PR rtl-optimization/57268
15579         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15580         if DEBUG_INSN_P (insn).
15582         Reapply
15583         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15585         PR rtl-optimization/57268
15586         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15587         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15589 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15591         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15592         (ix86_avoid_lea_for_addr): Likewise.
15593         (exact_dependency_1): Likewise.
15594         (ix86_adjust_cost): Likewise.
15595         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15596         removed.
15597         (do_reorder_for_imul): Fix typo, formatting and
15598         !reload_completed check removed.
15599         (ix86_sched_reorder): Fix typo and formatting.
15600         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15601         list.
15603 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15605         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15607 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15609         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15610         <STRING_CST>: Likewise.
15611         <VECTOR_CST>: Likewise.
15613 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15614             Mikael Morin  <mikael@gcc.gnu.org>
15616         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15617         * config.in: Regenerated.
15618         * configure: Regenerated.
15620 2013-06-01  Jan Hubicka  <jh@suse.cz>
15622         PR middle-end/57366
15623         * cgraphunit.c (compile): When weakref is not supported,
15624         set up transparent aliases before final output pass.
15625         * varasm.c (assemble_alias): Do not try to do it here.
15627 2013-06-01  Jan Hubicka  <jh@suse.cz>
15629         PR middle-end/57467
15630         * passes.c (for_per_function): Skip unanalyzed functions.
15632 2013-06-01  Jan Hubicka  <jh@suse.cz>
15634         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15635         (lto_symtab_merge_symbols_1): ... this one.
15636         (lto_symtab_merge_cgraph_nodes): Rename to ...
15637         (lto_symtab_merge_symbols): ... this one; simplify.
15638         * cgraph.c (same_body_aliases_done): Rename to ...
15639         (cpp_implicit_aliases_done): ... this one.
15640         (cgraph_create_function_alias): Update.
15641         (cgraph_same_body_alias): Update.
15642         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15643         (verify_edge_corresponds_to_fndecl): Simplify.
15644         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15645         (cgraph_node): Remove same_body_alias.
15646         (varpool_node): Remove alias_of and extra_name_alias.
15647         (same_body_aliases_done): Rename to ..
15648         (cpp_implicit_aliases_done): ... this one.
15649         (symtab_alias_ultimate_target): Add default parameter.
15650         (symtab_resolve_alias): New function.
15651         (fixup_same_cpp_alias_visibility): Declare.
15652         (cgraph_function_node): Add default parameter.
15653         (cgraph_node_asm_name): Likewise.
15654         (cgraph_function_or_thunk_node): Add default parameter; do
15655         not ICE when it is NULL.
15656         (varpool_variable_node): Likewise.
15657         * tree-emutls.c (create_emultls_var): Update.
15658         (ipa_lower_emutls): Update.
15659         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15660         (cgraph_reset_node): Reset alias info.
15661         (cgraph_finalize_function): Update.
15662         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15663         (analyze_function): Simplify.
15664         (cgraph_process_same_body_aliases): Simplify.
15665         (analyze_functions): Fixup same body aliases.
15666         (handle_alias_pairs): Simplify.
15667         (assemble_thunk): Update.
15668         (assemble_thunks_and_aliases): Update.
15669         (output_weakrefs): Rewrite.
15670         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15671         (lto_output_varpool_node): Likewise.
15672         (compute_ltrans_boundary): Remve assert.
15673         (get_alias_symbol): New functoin.
15674         (input_node): Rewrite alias handling.
15675         (input_varpool_node): Likewise.
15676         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15677         * ipa.c (process_references): Handle weakrefs correctly.
15678         (symtab_remove_unreachable_nodes): Likewise.
15679         * trans-mem.c (get_cg_data): Update.
15680         (ipa_tm_create_version_alias): Update.
15681         (ipa_tm_execute): Update.
15682         * symtab.c (dump_symtab_base): Dump aliases.
15683         (verify_symtab_base): Verify aliases.
15684         (symtab_node_availability): New function.
15685         (symtab_alias_ultimate_target): Simplify.
15686         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15687         handle all the fixup cases.
15688         (symtab_resolve_alias): New function.
15689         * passes.c (ipa_write_summaries): Handle weakrefs.
15690         * varpool.c (varpool_analyze_node): Simplify.
15691         (assemble_aliases): Update.
15692         (varpool_create_variable_alias): Simplify.
15693         (varpool_extra_name_alias): Simplify.
15694         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15695         (lto_symtab_merge_symbols): ... this one.
15697 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15699         Revert
15700         PR rtl-optimization/57268
15701         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15702         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15704 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15706         Partially reverted:
15707         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15709         PR middle-end/57073
15710         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15711         further up.
15713 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15715         PR rtl-optimization/57268
15716         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15717         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15719 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15721         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15722         unordered comparison operators when -fno-trapping-math is in effect
15723         on the e500.
15724         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15725         and implement unordered comparison operators properly on the e500.
15727 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15729         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15730         for constant scalar integers.
15731         (simplify_relational_operation_1): Likewise.
15733 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15735         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15736         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15737         Fix comment.
15739 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15740             Igor Zamyatin  <igor.zamyatin@intel.com>
15742         Silvermont (SLM) architecture performance tuning.
15743         * config/i386/i386.h (enum ix86_tune_indices): Add
15744         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15745         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15747         * config/i386/i386.c (initial_ix86_tune_features)
15748         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15749         (ix86_lea_outperforms): Handle Silvermont tuning.
15750         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15751         call.
15752         (ix86_use_lea_for_mov): Likewise.
15753         (ix86_avoid_lea_for_addr): Likewise.
15754         (ix86_lea_for_add_ok): Likewise.
15755         (exact_dependency_1): New function.
15756         (exact_store_load_dependency): Likewise.
15757         (ix86_adjust_cost): Handle Silvermont tuning.
15758         (do_reoder_for_imul): Likewise.
15759         (swap_top_of_ready_list): New function.
15760         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15762         * config/i386/i386.md (peepholes that split memory operand in fp
15763         converts): New.
15765 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15767         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15768         Remove un-necessary braces.
15770 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15772         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15773         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15775 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15777         PR middle-end/57073
15778         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15780 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15782         PR target/56315
15783         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15784         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15785         * config/arm/neon.md (iordi3_neon): Remove.
15786         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15787         * config/arm/predicates.md (imm_for_neon_logic_operand):
15788         Move to earlier in the file.
15789         (neon_logic_op2): Likewise.
15790         (arm_iordi_operand_neon): New predicate.
15792 2013-05-31  Richard Biener  <rguenther@suse.de>
15794         PR tree-optimization/57478
15795         PR tree-optimization/57453
15796         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15797         are life as well.
15799 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15801         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15802         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15804 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15805             Thomas Koenig  <tkoenig@gcc.gnu.org>
15807         PR middle-end/57073
15808         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15809         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15811 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15813         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15815 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15817         * target.def (register_usage_leveling_p): New hook.
15818         * targhooks.c (default_register_usage_leveling_p): New.
15819         * targhooks.h (default_register_usage_leveling_p): New prototype.
15820         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15821         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15822         * doc/tm.texi: Update.
15823         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15825 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15827         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15828         (*insv_reg<mode>): New define_insn.
15830 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15832         PR rtl-optimization/57439
15833         * postreload.c (move2add_valid_value_p): Check that we have
15834         a zero subreg_regno_offset when accessing the register in
15835         the requested mode.
15837 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15838             Igor Zamyatin  <igor.zamyatin@intel.com>
15840         Silvermont (SLM) architecture pipeline model, tuning and
15841         insn selection.
15842         * config.gcc: Add slm config options and target.
15844         * config/i386/slm.md: New.
15846         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15848         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15849         PROCESSOR_SLM.
15850         (ix86_target_macros_internal): Likewise.
15852         * config/i386/i386.c (slm_cost): New cost.
15853         (m_SLM): New macro flag.
15854         (initial_ix86_tune_features): Set m_SLM.
15855         (x86_accumulate_outgoing_args): Likewise.
15856         (x86_arch_always_fancy_math_387): Likewise.
15857         (processor_target_table): Add slm cost.
15858         (cpu_names): Add slm cpu name.
15859         (x86_option_override_internal): Set SLM ISA.
15860         (ix86_issue_rate): New case PROCESSOR_SLM.
15861         (ia32_multipass_dfa_lookahead): Likewise.
15862         (fold_builtin_cpu): Add slm.
15864         * config/i386/i386.h (TARGET_SLM): New target macro.
15865         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15866         (processor_type): Add PROCESSOR_SLM.
15868         * config/i386/i386.md (cpu): Add new value "slm".
15869         (slm.md): Include slm.md.
15871 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15872             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15874         * config/arm/arm-protos.h: Add and update function protos.
15875         * config/arm/arm.c (use_simple_return_p): New added.
15876         (thumb2_expand_return): Check simple_return flag.
15877         * config/arm/arm.md: Add simple_return and conditional simple_return.
15878         * config/arm/iterators.md: Add iterator for return and simple_return.
15880 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15882         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15883         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15884         (arm_emit_vfp_multi_reg_pop): Likewise.
15885         (thumb2_emit_ldrd_pop): Likewise.
15886         (arm_expand_epilogue): Add misc REG_CFA notes.
15887         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15889 2013-05-29  Lawrence Crowl  <crowl@google.com>
15891         * config/arm/t-arm: Update for below.
15893         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15894         Change type to hash_table.  Update dependent calls and types.
15896         * config/i386/t-cygming: Update for below.
15898         * config/i386/t-interix: Update for below.
15900         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15901         Change type to hash_table.  Update dependent calls and types.
15902         (i386_find_on_wrapper_list::wrappers): Likewise.
15904         * config/ia64/t-ia64: Update for below.
15906         * config/ia64/ia64.c (bundle_state_table):
15907         Change type to hash_table.  Update dependent calls and types.
15909         * config/mips/mips.c (mips_reorg_process_insns::htab):
15910         Change type to hash_table.  Update dependent calls and types.
15912         * config/sol2.c (solaris_comdat_htab):
15913         Change type to hash_table.  Update dependent calls and types.
15915         * config/t-sol2: Update for above.
15917 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15919         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15920         functions are not yet marked as defined.
15922 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15923             Pat Haugen  <pthaugen@us.ibm.com>
15924             Peter Bergner  <bergner@vnet.ibm.com>
15926         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15927         instructions.
15928         (VEC_A): Likewise.
15929         (VEC_C): Likewise.
15930         (vrotl<mode>3): Likewise.
15931         (vashl<mode>3): Likewise.
15932         (vlshr<mode>3): Likewise.
15933         (vashr<mode>3): Likewise.
15935         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15936         support for power8 V2DI builtins.
15938         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15939         power8 V2DI builtins.
15940         (vupkhsw): Likewise.
15941         (vupklsw): Likewise.
15942         (vaddudm): Likewise.
15943         (vminsd): Likewise.
15944         (vmaxsd): Likewise.
15945         (vminud): Likewise.
15946         (vmaxud): Likewise.
15947         (vpkudum): Likewise.
15948         (vpksdss): Likewise.
15949         (vpkudus): Likewise.
15950         (vpksdus): Likewise.
15951         (vrld): Likewise.
15952         (vsld): Likewise.
15953         (vsrd): Likewise.
15954         (vsrad): Likewise.
15955         (vsubudm): Likewise.
15956         (vcmpequd): Likewise.
15957         (vcmpgtsd): Likewise.
15958         (vcmpgtud): Likewise.
15959         (vcmpequd_p): Likewise.
15960         (vcmpgtsd_p): Likewise.
15961         (vcmpgtud_p): Likewise.
15962         (vupkhsw): Likewise.
15963         (vupklsw): Likewise.
15964         (vaddudm): Likewise.
15965         (vmaxsd): Likewise.
15966         (vmaxud): Likewise.
15967         (vminsd): Likewise.
15968         (vminud): Likewise.
15969         (vpksdss): Likewise.
15970         (vpksdus): Likewise.
15971         (vpkudum): Likewise.
15972         (vpkudus): Likewise.
15973         (vrld): Likewise.
15974         (vsld): Likewise.
15975         (vsrad): Likewise.
15976         (vsrd): Likewise.
15977         (vsubudm): Likewise.
15979         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15980         support for power8 V2DI instructions.
15982         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15983         power8 V2DI instructions.  Combine pack and unpack insns to use an
15984         iterator for each mode.  Check whether a particular mode supports
15985         Altivec instructions instead of just checking TARGET_ALTIVEC.
15986         (UNSPEC_VPKUWUM): Likewise.
15987         (UNSPEC_VPKSHSS): Likewise.
15988         (UNSPEC_VPKSWSS): Likewise.
15989         (UNSPEC_VPKUHUS): Likewise.
15990         (UNSPEC_VPKSHUS): Likewise.
15991         (UNSPEC_VPKUWUS): Likewise.
15992         (UNSPEC_VPKSWUS): Likewise.
15993         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15994         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15995         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15996         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15997         (UNSPEC_VUPKHSB): Likewise.
15998         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15999         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
16000         (UNSPEC_VUPKHSH): Likewise.
16001         (UNSPEC_VUPKLSB): Likewise.
16002         (UNSPEC_VUPKLSH): Likewise.
16003         (VI2): Likewise.
16004         (VI_char): Likewise.
16005         (VI_scalar): Likewise.
16006         (VI_unit): Likewise.
16007         (VP): Likewise.
16008         (VP_small): Likewise.
16009         (VP_small_lc): Likewise.
16010         (VU_char): Likewise.
16011         (add<mode>3): Likewise.
16012         (altivec_vaddcuw): Likewise.
16013         (altivec_vaddu<VI_char>s): Likewise.
16014         (altivec_vadds<VI_char>s): Likewise.
16015         (sub<mode>3): Likewise.
16016         (altivec_vsubcuw): Likewise.
16017         (altivec_vsubu<VI_char>s): Likewise.
16018         (altivec_vsubs<VI_char>s): Likewise.
16019         (altivec_vavgs<VI_char>): Likewise.
16020         (altivec_vcmpbfp): Likewise.
16021         (altivec_eq<mode>): Likewise.
16022         (altivec_gt<mode>): Likewise.
16023         (altivec_gtu<mode>): Likewise.
16024         (umax<mode>3): Likewise.
16025         (smax<mode>3): Likewise.
16026         (umin<mode>3): Likewise.
16027         (smin<mode>3): Likewise.
16028         (altivec_vpkuhum): Likewise.
16029         (altivec_vpkuwum): Likewise.
16030         (altivec_vpkshss): Likewise.
16031         (altivec_vpkswss): Likewise.
16032         (altivec_vpkuhus): Likewise.
16033         (altivec_vpkshus): Likewise.
16034         (altivec_vpkuwus): Likewise.
16035         (altivec_vpkswus): Likewise.
16036         (altivec_vpks<VI_char>ss): Likewise.
16037         (altivec_vpks<VI_char>us): Likewise.
16038         (altivec_vpku<VI_char>us): Likewise.
16039         (altivec_vpku<VI_char>um): Likewise.
16040         (altivec_vrl<VI_char>): Likewise.
16041         (altivec_vsl<VI_char>): Likewise.
16042         (altivec_vsr<VI_char>): Likewise.
16043         (altivec_vsra<VI_char>): Likewise.
16044         (altivec_vsldoi_<mode>): Likewise.
16045         (altivec_vupkhsb): Likewise.
16046         (altivec_vupkhs<VU_char>): Likewise.
16047         (altivec_vupkls<VU_char>): Likewise.
16048         (altivec_vupkhsh): Likewise.
16049         (altivec_vupklsb): Likewise.
16050         (altivec_vupklsh): Likewise.
16051         (altivec_vcmpequ<VI_char>_p): Likewise.
16052         (altivec_vcmpgts<VI_char>_p): Likewise.
16053         (altivec_vcmpgtu<VI_char>_p): Likewise.
16054         (abs<mode>2): Likewise.
16055         (vec_unpacks_hi_v16qi): Likewise.
16056         (vec_unpacks_hi_v8hi): Likewise.
16057         (vec_unpacks_lo_v16qi): Likewise.
16058         (vec_unpacks_hi_<VP_small_lc>): Likewise.
16059         (vec_unpacks_lo_v8hi): Likewise.
16060         (vec_unpacks_lo_<VP_small_lc>): Likewise.
16061         (vec_pack_trunc_v8h): Likewise.
16062         (vec_pack_trunc_v4si): Likewise.
16063         (vec_pack_trunc_<mode>): Likewise.
16065         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
16066         V2DI builtins.
16067         (vec_vmaxsd): Likewise.
16068         (vec_vmaxud): Likewise.
16069         (vec_vminsd): Likewise.
16070         (vec_vminud): Likewise.
16071         (vec_vpksdss): Likewise.
16072         (vec_vpksdus): Likewise.
16073         (vec_vpkudum): Likewise.
16074         (vec_vpkudus): Likewise.
16075         (vec_vrld): Likewise.
16076         (vec_vsld): Likewise.
16077         (vec_vsrad): Likewise.
16078         (vec_vsrd): Likewise.
16079         (vec_vsubudm): Likewise.
16080         (vec_vupkhsw): Likewise.
16081         (vec_vupklsw): Likewise.
16083 2013-05-29  Jan Hubicka  <jh@suse.cz>
16085         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16086         flags; reorder rest of fields in more consistent way.
16087         (varpool_node): Remove analyzed, finalized and alias.
16088         (cgraph_ndoe): Likewise.
16089         (symtab_alias_ultimate_target): New function.
16090         (cgraph_function_node): Move offline.
16091         (cgraph_reset_node): Declare.
16092         (cgraph_comdat_can_be_unshared_p): Remove.
16093         (varpool_remove_initializer): Declare.
16094         (varpool_first_defined_variable, varpool_next_defined_variable
16095         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16096         (cgraph_function_with_gimple_body_p): Update.
16097         (varpool_all_refs_explicit_p): Update.
16098         (symtab_alias_target): New function.
16099         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16100         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16101         (cgraph_function_or_thunk_node): Simplify using
16102         symtab_alias_ultimate_target.
16103         (varpool_variable_node): Likewise.
16104         * cgraph.c (cgraph_create_function_alias): Update.
16105         (cgraph_add_thunk): Update.
16106         (cgraph_remove_node): Update.
16107         (dump_cgraph_node): Do not dump removed flags.
16108         (cgraph_function_body_availability): Update.
16109         (cgraph_propagate_frequency): Update.
16110         (verify_cgraph_node): Check sanity of local flag.
16111         (cgraph_function_node): Move here from cgraph.h; revamp for
16112         cgraph_function_or_thunk_node.
16113         * lto-symtab.c (lto_varpool_replace_node): Update.
16114         (lto_symtab_resolve_can_prevail_p): Update.
16115         (lto_symtab_merge_cgraph_nodes): Update.
16116         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16117         propagate_constants_accross_call, devirtualization_time_bonus,
16118         ipcp_propagate_stage): Update.
16119         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16120         * ipa-inline-transform.c (clone_inlined_nodes,
16121         preserve_function_body_p): Update.
16122         * ipa-reference.c (propagate): Update.
16123         (write_node_summary_p): Update.
16124         * toplev.c (wrapup_global_declaration_2): Update.
16125         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16126         (analyze_function) ... this one.
16127         (cgraph_process_new_functions): Update.
16128         (cgraph_reset_node): Export.
16129         (cgraph_finalize_function): Update.
16130         (cgraph_add_new_function): Update.
16131         (process_function_and_variable_attributes): Update.
16132         (varpool_finalize_decl): Update.
16133         (symbol_finalized): Remove.
16134         (symbol_finalized_and_needed): Rename to ...
16135         (symbol_defined_and_needed): ... update.
16136         (cgraph_analyze_functions): Update.
16137         (handle_alias_pairs): Update.
16138         (mark_functions_to_output): Update.
16139         (assemble_thunk): Update.
16140         (output_in_order): Update.
16141         (output_weakrefs): Update.
16142         (finalize_compilation_unit): Update.
16143         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16144         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16145         input_node, input_varpool_node): Update.
16146         * dbxout.c (dbxout_expand_expr): Update.
16147         * cgraphclones.c (cgraph_clone_node): Update.
16148         (cgraph_copy_node_for_versioning): Update.
16149         (cgraph_materialize_clone): Update.
16150         (cgraph_materialize_all_clones): Update.
16151         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16152         propagate_pure_const, propagate_nothrow): Update.
16153         * lto-streamer-out.c (lto_output, write_symbol): Update.
16154         * ipa-utils.c (ipa_reverse_postorder): Update.
16155         * ipa-inline.c (can_inline_edge_p): Update.
16156         (update_caller_keys, ipa_inline): Update.
16157         * dwarf2out.c (reference_to_unused,
16158         premark_types_used_by_global_vars_helper): Update.
16159         * tree-eh.c (tree_could_trap_p): Update.
16160         * ipa-split.c (consider_split, execute_split_functions): Update.
16161         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16162         has_addr_references_p): Update; move ahead in file for better
16163         readability.
16164         (process_references): Simplify.
16165         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16166         bodies are removed.
16167         (cgraph_comdat_can_be_unshared_p): Make static.
16168         (cgraph_externally_visible_p): Update.
16169         (varpool_externally_visible_p): Update.
16170         (function_and_variable_visibility): Update.
16171         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16172         ipa_tm_mark_force_output_node): Update.
16173         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16174         estimate_edge_devirt_benefit, inline_generate_summary,
16175         inline_write_summary): Update.
16176         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16177         * ipa-prop.c (ipa_compute_jump_functions): Update.
16178         (ipa_print_node_params, ipa_prop_read_section,
16179         ipa_update_after_lto_read, read_replacements_section): Update.
16180         * varasm.c (mark_decl_referenced): Update.
16181         (assemble_alias, dump_tm_clone_pairs): Update.
16182         * tree-inline.c (copy_bb): Update.
16183         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16184         Update.
16185         * symtab.c (dump_symtab_base): Print new flags.
16186         (verify_symtab_base): Verify new flags.
16187         (symtab_alias_ultimate_target): New function.
16188         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16189         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16190         Update.
16191         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16192         Update.
16193         * i386.c (ix86_get_function_versions_dispatcher,
16194         ix86_generate_version_dispatcher_body): Update.
16195         (fold_builtin_cpu): Use varpool_add_new_variable.
16196         * varpool.c (varpool_remove_initializer): Break out from ...
16197         (varpool_remove_node): ... this one.
16198         (dump_varpool_node, varpool_node_for_asm,
16199         cgraph_variable_initializer_availability, varpool_analyze_node,
16200         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16201         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16202         Update.
16204 2013-05-29  Jan Hubicka  <jh@suse.cz>
16206         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16208 2013-05-29  Easwaran Raman  <eraman@google.com>
16210         PR tree-optimization/57442
16211         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16212         when control exits the main loop.
16214 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16216         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16217         and RX600.
16218         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16219         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16220         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16221         option.
16223 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16225         PR tree-optimization/57441
16226         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16227         Don't limit size of incr_vec to number of candidates.
16229 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16231         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16232         and mips16 directories.
16233         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16234         (MULTILIB_DIRNAMES): Ditto.
16235         (MULTILIB_EXCEPTIONS): Add new exceptions.
16236         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16237         (MULTILIB_DIRNAMES): Ditto.
16238         (MULTILIB_EXCEPTIONS): Add new exceptions.
16240 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16241             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16243         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16244         SYMBOL_TINY_ABSOLUTE.
16245         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16246         SYMBOL_TINY_ABSOLUTE.
16247         (aarch64_expand_mov_immediate): Likewise.
16248         (aarch64_classify_symbol): Likewise.
16249         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16250         Permit SYMBOL_TINY_ABSOLUTE.
16251         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16253 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16254             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16256         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16257         Refactor if/switch.  Replace gcc_assert with if.
16259 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16261         * config/i386/i386.c (initial_ix86_tune_features): Enable
16262         FP Reassociation for AMD bdver1 and bdver2.
16264 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16266         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16267         and IMAGPART_EXPR do not occur within other handled_components.
16269 2013-05-29  Richard Biener  <rguenther@suse.de>
16271         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16272         access on whether the use is in the BB we currently try to
16273         vectorize.
16274         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16275         vectorize to vect_bb_slp_scalar_cost.
16277 2013-05-29  Richard Biener  <rguenther@suse.de>
16279         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16280         computing scalar cost offsetted by stmts that are kept live
16281         by scalar uses.
16282         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16283         for computation of scalar cost.
16285 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16287         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16289 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16291         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16292         array notation built-in reduction functions.
16293         * doc/passes.texi (Passes): Added documentation about changes done
16294         for Cilk Plus.
16295         * doc/invoke.texi (C Dialect Options): Added documentation about
16296         the -fcilkplus flag.
16297         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16298         (BUILTINS_DEF): Depend on cilkplus.def.
16299         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16300         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16301         * cilkplus.def: New file.
16303 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16305         PR rtl-optimization/57439
16306         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16308 2013-05-28  Easwaran Raman  <eraman@google.com>
16310         PR tree-optimization/57337
16311         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16312         (find_insert_point): Correctly identify the insertion point
16313         when two statements with the same UID is compared.
16315 2013-05-28  Richard Biener  <rguenther@suse.de>
16317         PR tree-optimization/56787
16318         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16319         from the list of data references.
16320         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16321         clobbers.
16322         (vect_analyze_loop_operations): Likewise.
16323         (vect_transform_loop): Remove clobbers.
16325 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16327         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16328         and REALPART_EXPRs have scalar type.
16330 2013-05-28  Richard Biener  <rguenther@suse.de>
16332         PR tree-optimization/57411
16333         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16334         virtual operands.
16335         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16336         virtual operand propagation.
16338 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16340         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16341         destination register for bmasksi_vis.
16342         (vector_init_bshuffle): Likewise.
16343         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16345 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16347         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16348         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16349         mode if the instruction isn't available in the original mode.
16350         * config/sparc/sparc.opt (mfix-ut699): New option.
16351         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16352         (divdf3): Turn into expander.
16353         (divdf3_nofix): New insn.
16354         (divdf3_fix): Likewise.
16355         (divsf3): Disable if -mfix-ut699.
16356         (sqrtdf2): Turn into expander.
16357         (sqrtdf2_nofix): New insn.
16358         (sqrtdf2_fix): Likewise.
16359         (sqrtsf2): Disable if -mfix-ut699.
16361 2013-05-27  Richard Biener  <rguenther@suse.de>
16363         PR middle-end/57412
16364         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16365         block for the new loop.
16367 2013-05-27  Richard Biener  <rguenther@suse.de>
16369         PR tree-optimization/57343
16370         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16371         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16372         (number_of_iterations_cond): Do not build the folded tree.
16374 2013-05-27  Richard Biener  <rguenther@suse.de>
16376         Revert
16377         PR middle-end/57381
16378         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16379         OEP_CONSTANT_ADDRESS_OF retained.
16381         PR tree-optimization/57417
16382         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16383         for unchanged base.
16384         (set_ssa_val_to): Compare addresses using
16385         get_addr_base_and_unit_offset.
16387 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16389         PR rtl-optimization/56833
16390         * postreload.c (move2add_record_mode): New function.
16391         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16392         (move2add_use_add2_insn): Use move2add_record_sym_value.
16393         (move2add_use_add3_insn): Likewise.
16394         (reload_cse_move2add): Use move2add_valid_value_p and
16395         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16396         affected regs by setting reg_mode to VOIDmode.
16397         (move2add_note_store): Don't pretend the inside of a SUBREG is
16398         the actual destination.  Invalidate single/leading registers by
16399         setting reg_mode to VOIDmode.
16400         Use move2add_record_sym_value, move2add_valid_value_p and
16401         move2add_record_mode.
16403 2013-05-27  Richard Biener  <rguenther@suse.de>
16405         PR tree-optimization/57396
16406         * tree-affine.c (double_int_constant_multiple_p): Properly
16407         return false for val == 0 and div != 0.
16409 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16411         * config/mips/mips.h: Use #elif in preprocessor conditions.
16413 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16415         PR target/53916
16416         * config/mips/constraints.md (kl): New constraint.
16417         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16418         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16419         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16420         from LO for MIPS16.
16421         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16423 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16425         PR target/55777
16426         * config/mips/mips.c (mips_can_inline_p): New function.
16427         (TARGET_CAN_INLINE_P): Define.
16429 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16431         * sched-int.h (ds_t, dw_t): Make unsigned int.
16432         Fix documentation that describes how all the ds_t bits are used.
16433         Reserve the last bit for delayed-branch scheduling.
16434         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16435         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16436         (gen_dep_weak_1): Remove prototype.
16437         * sched-deps.c (get_dep_weak_1): Make static.
16438         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16439         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16440         * doc/tm.texi: Regenerate.
16441         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16443 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16445         PR debug/56950
16446         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16448 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16449             Sandra Loosemore  <sandra@codesourcery.com>
16451         * config.gcc (powerpc-*): Allow native for with-cpu.
16453 2013-05-24  Jeff Law  <law@redhat.com>
16455         PR tree-optimization/57124
16456         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16457         conversion feeding a condition if the range has an overflow
16458         if -fstrict-overflow.  Add warnings for when we do make the
16459         transformation.
16461 2013-05-24  Dehao Chen  <dehao@google.com>
16463         * tree-cfg.c (locus_discrim_map): Fix the typo.
16464         (locus_discrim_hasher): Likewise.
16465         (locus_discrim_hasher::hash): Likewise.
16466         (locus_discrim_hasher::equal): Likewise.
16468 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16470         PR tree-optimization/57294
16471         * cgraph.h (ipa_record_stmt_references): Declare.
16472         * cgraphbuild.c (ipa_record_stmt_references): New function.
16473         (build_cgraph_edges): Use ipa_record_stmt_references.
16474         (rebuild_cgraph_edges): Likewise.
16475         (cgraph_rebuild_references): Likewise.
16476         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16477         associated with the old statement and build references from the
16478         newly built statements.
16479         * ipa-ref.c (ipa_remove_stmt_references): New function.
16480         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16482 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16484         * lra-constraints.c (emit_spill_move): Use smaller mode for
16485         mem-mem moves.
16486         (check_and_process_move): Consider mem-reg moves for secondary
16487         too.
16488         (curr_insn_transform): Don't lose insns emitted before for
16489         secondary memory moves.
16490         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16491         reg set up in the current insn.
16493 2013-05-24  Dehao Chen  <dehao@google.com>
16495         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16496         hash function.
16497         (locus_descrim_hasher::equal): Likewise.
16498         (build_gimple_cfg): New discriminator assignment algorithm.
16499         (make_edges): Likewise.
16500         (next_discriminator_for_locus): Likewise.
16501         (same_line_p): Likewise.
16502         (assign_discriminators): Likewise.
16503         (make_cond_expr_edges): Likewise.
16504         (make_gimple_switch_edges): Likewise.
16505         (make_goto_expr_edges): Likewise.
16506         (make_gimple_asm_edges): Likewise.
16508 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16510         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16511         X format specifier to only display bottom 16 bits.
16512         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16513         immediate to match for operand 2, since it will be masked.
16515 2013-05-24  Richard Biener  <rguenther@suse.de>
16517         PR tree-optimization/57287
16518         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16519         all SSA names that occur in abnormal PHIs.
16521 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16523         PR tree-ssa/57385
16524         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16525         that index is not negative.
16527 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16529         PR rtl-optimization/55177
16530         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16531         (simplify_byte_swapping_operation): New.
16532         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16533         (simplify_relational_operation_1): Deal with BSWAP.
16535 2013-05-23  Richard Henderson  <rth@redhat.com>
16537         PR target/56742
16538         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16539         (ix86_reorg): Call it.
16541 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16543         PR target/57379
16544         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16545         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16546         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16548 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16550         PR debug/57351
16551         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16553 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16554             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16556         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16557         * config/aarch64/constraints.md (Usa): Remove.
16558         * doc/md.texi (AArch64 Usa): Remove.
16560 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16561             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16563         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16564         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16565         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16566         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16568 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16570         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16571         instruction (AdvSIMD).
16572         * config/aarch64/aarch64-builtins.c
16573         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16574         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16576 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16578         PR middle-end/57347
16579         * tree.h (contains_bitfld_component_ref_p): Declare.
16580         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16581         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16582         caller.
16583         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16584         not access a bit-field.  Assert all final offsets are byte-aligned.
16586 2013-05-23  Richard Biener  <rguenther@suse.de>
16588         PR tree-optimization/57380
16589         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16590         least one invariant or re-used load.
16591         * passes.c (init_optimization_passes): Move pass_phiprop before
16592         pass_forwprop.
16594 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16596         * config/aarch64/aarch64-simd.md
16597         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16599 2013-05-23  Richard Biener  <rguenther@suse.de>
16601         PR middle-end/57381
16602         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16603         OEP_CONSTANT_ADDRESS_OF retained.
16605 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16607         PR middle-end/57344
16608         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16609         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16611 2013-05-23  Richard Biener  <rguenther@suse.de>
16613         PR rtl-optimization/57341
16614         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16615         instead of true_dependence.
16617 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16619         * bb-reorder.c (branch_threshold): Make const.
16620         (exec_threshold): Ditto.
16622 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16623             Pat Haugen  <pthaugen@us.ibm.com>
16624             Peter Bergner  <bergner@vnet.ibm.com>
16626         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16627         documentation for the power8 crypto builtins.
16629         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16631         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16632         macros for defining power8 builtin functions.
16633         (BU_P8V_AV_2): Likewise.
16634         (BU_P8V_AV_P): Likewise.
16635         (BU_P8V_VSX_1): Likewise.
16636         (BU_P8V_OVERLOAD_1): Likewise.
16637         (BU_P8V_OVERLOAD_2): Likewise.
16638         (BU_CRYPTO_1): Likewise.
16639         (BU_CRYPTO_2): Likewise.
16640         (BU_CRYPTO_3): Likewise.
16641         (BU_CRYPTO_OVERLOAD_1): Likewise.
16642         (BU_CRYPTO_OVERLOAD_2): Likewise.
16643         (XSCVSPDP): Fix typo, point to the correct instruction.
16644         (VCIPHER): Add power8 crypto builtins.
16645         (VCIPHERLAST): Likewise.
16646         (VNCIPHER): Likewise.
16647         (VNCIPHERLAST): Likewise.
16648         (VPMSUMB): Likewise.
16649         (VPMSUMH): Likewise.
16650         (VPMSUMW): Likewise.
16651         (VPERMXOR_V2DI): Likewise.
16652         (VPERMXOR_V4SI: Likewise.
16653         (VPERMXOR_V8HI: Likewise.
16654         (VPERMXOR_V16QI: Likewise.
16655         (VSHASIGMAW): Likewise.
16656         (VSHASIGMAD): Likewise.
16657         (VPMSUM): Likewise.
16658         (VPERMXOR): Likewise.
16659         (VSHASIGMA): Likewise.
16661         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16662         __CRYPTO__ if the crypto instructions are available.
16663         (altivec_overloaded_builtins): Add support for overloaded power8
16664         builtins.
16666         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16667         support for power8 crypto builtins.
16668         (builtin_function_type): Likewise.
16669         (altivec_init_builtins): Add support for builtins that take vector
16670         long long (V2DI) arguments.
16672         * config/rs6000/crypto.md: New file, define power8 crypto
16673         instructions.
16675 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16676             Pat Haugen  <pthaugen@us.ibm.com>
16677             Peter Bergner  <bergner@vnet.ibm.com>
16679         * doc/invoke.texi (Option Summary): Add power8 options.
16680         (RS/6000 and PowerPC Options): Likewise.
16682         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16683         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16684         wm, wn, wr documentation.
16686         * config/rs6000/constraints.md (wm): New constraint for VSX
16687         registers if direct move instructions are enabled.
16688         (wn): New constraint for no registers.
16689         (wq): New constraint for quad word even GPR registers.
16690         (wr): New constraint if 64-bit instructions are enabled.
16691         (wv): New constraint if power8 vector instructions are enabled.
16692         (wQ): New constraint for quad word memory locations.
16694         * config/rs6000/predicates.md (const_0_to_15_operand): New
16695         constraint for 0..15 for crypto instructions.
16696         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16697         as GPR and floating point registers.
16698         (int_reg_operand): New predicate to match only GPR registers.
16699         (base_reg_operand): New predicate to match base registers.
16700         (quad_int_reg_operand): New predicate to match even GPR registers
16701         for quad memory operations.
16702         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16703         operations in both GPR and VSX registers.
16704         (quad_memory_operand): New predicate for quad memory operations.
16705         (reg_or_indexed_operand): New predicate for direct move support.
16707         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16708         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16709         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16710         (POWERPC_MASKS): Add power8 options.
16711         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16712         various options.
16714         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16715         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16717         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16718         (-mpower8-fusion): New power8 options.
16719         (-mpower8-fusion-sign): Likewise.
16720         (-mpower8-vector): Likewise.
16721         (-mcrypto): Likewise.
16722         (-mdirect-move): Likewise.
16723         (-mquad-memory): Likewise.
16725         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16726         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16727         registers.
16728         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16729         (rs6000_debug_vector_unit): Add p8_vector.
16730         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16731         definitions.  Also print fusion state.
16732         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16733         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16734         (rs6000_option_override_internal): Add support for power8.
16735         (rs6000_common_init_builtins): Add debugging for skipped builtins
16736         if -mdebug=builtin.
16737         (rs6000_adjust_cost): Add power8 support.
16738         (rs6000_issue_rate): Likewise.
16739         (insn_must_be_first_in_group): Likewise.
16740         (insn_must_be_last_in_group): Likewise.
16741         (force_new_group): Likewise.
16742         (rs6000_register_move_cost): Likewise.
16743         (rs6000_opt_masks): Likewise.
16745         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16746         power8 capable assembler, default to power7 options.
16747         (TARGET_DIRECT_MOVE): Likewise.
16748         (TARGET_CRYPTO): Likewise.
16749         (TARGET_P8_VECTOR): Likewise.
16750         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16751         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16752         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16753         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16754         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16755         (TARGET_XSCVDPSPN): Likewise.
16756         (TARGET_XSCVSPDPN): Likewsie.
16757         (TARGET_SYNC_HI_QI): Likewise.
16758         (TARGET_SYNC_TI): Likewise.
16759         (MASK_CRYPTO): Likewise.
16760         (MASK_DIRECT_MOVE): Likewise.
16761         (MASK_P8_FUSION): Likewise.
16762         (MASK_P8_VECTOR): Likewise.
16763         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16764         temporary used by some of the direct move instructions to get two FP
16765         temporary registers does not force creation of a stack frame.
16766         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16767         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16768         that any VSX registers are tieable, even if they are also an
16769         Altivec vector mode.
16770         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16771         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16772         (RS6000_BTM_CRYPTO): Likewise.
16773         (RS6000_BTM_COMMON): Likewise.
16775         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16776         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16777         (enum rs6000_vector): Add power8 vector support.
16779 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16781         PR target/19599
16782         PR target/57340
16783         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16784         (any_sibcall_could_use_r3): this and handle indirect calls.
16785         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16787 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16789         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16791 2013-05-22  Richard Biener  <rguenther@suse.de>
16793         PR middle-end/57349
16794         * profile.c (branch_prob): Do not split blocks that are
16795         abnormally receiving from ECF_RETURNS_TWICE functions.
16797 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16799         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16800         address mode.  Move pointer mode initialization to the same place.
16802 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16804         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16805         while it has any effect.
16807 2013-05-21  Easwaran Raman  <eraman@google.com>
16809         PR tree-optimization/57322
16810         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16811         UID of the statement added to the BB to be 1.
16813 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16815         PR tree-optimization/57331
16816         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16817         of conversion from pointer type to integral type with integer.
16819 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16821         PR lto/57289
16822         * ipa-prop.c (ipa_read_node_info): Process param_used and
16823         controlled_uses in the same order as when writing.
16825 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16827         PR plugins/56754
16828         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16830 2013-05-21  Richard Biener  <rguenther@suse.de>
16832         PR tree-optimization/57318
16833         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16834         estimate stmts with side-effects as likely eliminated.
16836 2013-05-21  Richard Biener  <rguenther@suse.de>
16838         PR tree-optimization/57330
16839         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16840         preserve the call stmts fntype.
16842 2013-05-21  Richard Biener  <rguenther@suse.de>
16844         PR tree-optimization/57303
16845         * tree-ssa-sink.c (statement_sink_location): Improve killing
16846         stmt detection and properly handle self-assignments.
16848 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16850         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16851         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16852         registers. Set register size out of the PARALLEL loop.
16854 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16856         PR target/56547
16857         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16858         (*fmasf4, *fmasf4_media): New insns.
16860 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16862         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16863         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16864         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16865         (mips_idiv_insns): Update the comments to say that the returned
16866         instruction counts are in units of BASE_INSN_LENGTH.
16867         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16868         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16869         using 2 rather than 4 as the length of indirect MIPS16 and
16870         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16871         length of a NOP.  Don't divide MIPS16 lengths by 2.
16872         (mips16_split_long_branches): Assume a branch is long if the
16873         length is greater than 4 rather than 8.
16874         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16875         rather than multiplying them by 2.  Multiply instruction counts
16876         by BASE_INSN_LENGTH rather than 4.
16877         (*jump_mips16, tls_get_tp_mips16_<mode>)
16878         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16880 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16882         * config/mips/mips.md (extended_mips16): Remove branch case.
16883         (length): Remove duplicated extended_mips16 test.
16885 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16887         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16889 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16891         * recog.h (Recog_data): Rename to...
16892         (recog_data_d): ...this.
16893         (recog_data): Update accordingly.
16894         * recog.c (recog_data): Likewise.
16895         * reload.c (save_recog_data): Likewise.
16896         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16897         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16899 2013-05-17  Julian Brown  <julian@codesourcery.com>
16901         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16902         found in a REG_EQUAL note, invalidate it.
16904 2013-05-17  Easwaran Raman  <eraman@google.com>
16906         * tree-ssa-reassoc.c (find_insert_point): New function.
16907         (insert_stmt_after): Likewise.
16908         (get_def_stmt): Likewise.
16909         (ensure_ops_are_available): Likewise.
16910         (not_dominated_by): Likewise.
16911         (rewrite_expr_tree): Do not move statements beyond what is
16912         necessary. Remove call to swap_ops_for_binary_stmt...
16913         (reassociate_bb): ... and move it here.
16914         (build_and_add_sum): Assign UIDs for new statements.
16915         (linearize_expr): Likewise.
16916         (do_reassoc): Renumber gimple statement UIDs.
16918 2013-05-17  Jan Hubicka  <jh@suse.cz>
16920         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16921         weakrefs.
16922         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16923         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16924         weakrefs.
16925         (output_weakrefs): Update.
16927 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16928             Martin Jambor  <mjambor@suse.cz>
16930         PR middle-end/57276
16931         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16932         value that corresponds to the given aggval is found in values vector.
16934 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16936         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16937         sse, sse2, sse3, ssse3 and sse4a flags to options.
16939 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16941         * gengtype-state.c: (s_expr_writer): New class, to handle
16942         prettifying of output layout of s-expressions.
16943         (state_writer): New class, to write out gtype.state.
16944         (state_written_type_count): Move this variable into member data of
16945         state_writer.
16946         (s_expr_writer::s_expr_writer): New code: constructor for new class
16947         (state_writer::state_writer(): ditto
16948         (s_expr_writer::write_new_line): New function
16949         (s_expr_writer::write_any_indent): ditto
16950         (s_expr_writer::begin_s_expr): ditto
16951         (s_expr_writer::end_s_expr): ditto
16952         (write_state_fileloc): convert to method of state_writer...
16953         (state_writer:: write_state_fileloc): ...and use methods of
16954         s_expr_writer to write indentation into the gtype.state output file
16955         to visually represent the hierarchical structure of the list
16956         structures
16957         (write_state_fields): ditto, renaming to...
16958         (state_writer::write_state_fields)
16959         (write_state_a_string): ditto, renaming to...
16960         (state_writer::write_state_a_string)
16961         (write_state_string_option): ditto, renaming to...
16962         (state_writer::write_state_string_option)
16963         (write_state_type_option): ditto, renaming to...
16964         (state_writer::write_state_type_option)
16965         (write_state_nested_option): ditto, renaming to...
16966         (state_writer::write_state_nested_option)
16967         (write_state_option): ditto, renaming to...
16968         (state_writer::write_state_option)
16969         (write_state_options): ditto, renaming to...
16970         (state_writer::write_state_options)
16971         (write_state_lang_bitmap): ditto, renaming to...
16972         (state_writer::write_state_lang_bitmap)
16973         (write_state_version): ditto, renaming to...
16974         (state_writer::write_state_version)
16975         (write_state_scalar_type): ditto, renaming to...
16976         (state_writer::write_state_scalar_type)
16977         (write_state_string_type): ditto, renaming to...
16978         (state_writer::write_state_string_type)
16979         (write_state_undefined_type): ditto, renaming to...
16980         (state_writer::write_state_undefined_type)
16981         (write_state_struct_union_type): ditto, renaming to...
16982         (state_writer::write_state_struct_union_type)
16983         (write_state_struct_type): ditto, renaming to...
16984         (state_writer::write_state_struct_type)
16985         (write_state_user_struct_type): ditto, renaming to...
16986         (state_writer::write_state_user_struct_type)
16987         (write_state_lang_struct_type): ditto, renaming to...
16988         (state_writer::write_state_lang_struct_type)
16989         (write_state_param_struct_type): ditto, renaming to...
16990         (state_writer::write_state_param_struct_type)
16991         (write_state_pointer_type): ditto, renaming to...
16992         (state_writer::write_state_pointer_type)
16993         (write_state_array_type): ditto, renaming to...
16994         (state_writer::write_state_array_type)
16995         (write_state_gc_used): ditto, renaming to...
16996         (state_writer::write_state_gc_used)
16997         (write_state_common_type_content): ditto, renaming to...
16998         (state_writer::write_state_common_type_content)
16999         (write_state_type): ditto, renaming to...
17000         (state_writer::write_state_type)
17001         (write_state_pair_list): ditto, renaming to...
17002         (state_writer::write_state_pair_list)
17003         (write_state_pair): ditto, renaming to...
17004         (state_writer::write_state_pair)
17005         (write_state_typedefs): ditto, renaming to...
17006         (state_writer::write_state_typedefs)
17007         (write_state_structures): ditto, renaming to...
17008         (state_writer::write_state_structures)
17009         (write_state_param_structs): ditto, renaming to...
17010         (state_writer::write_state_param_structs)
17011         (write_state_variables): ditto, renaming to...
17012         (state_writer::write_state_variables)
17013         (write_state_srcdir): ditto, renaming to...
17014         (state_writer::write_state_srcdir)
17015         (write_state_files_list): ditto, renaming to...
17016         (state_writer::write_state_files_list)
17017         (write_state_languages): ditto, renaming to...
17018         (state_writer::write_state_languages)
17019         (write_state): create a state_writer instance and use it when
17020         writing out the state file
17022 2013-05-17  Mike Stump  <mikestump@comcast.net>
17024         PR rtl-optimization/57304
17025         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
17026         accessing DF_REF_REAL_LOC.
17028 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17030         PR rtl-optimization/57281
17031         PR rtl-optimization/57300
17032         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
17033         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
17034         what the other splitter did if the registers are dead.
17036 2013-05-17  Richard Biener  <rguenther@suse.de>
17038         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
17039         MEM_REF offsets.
17041 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17043         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
17044         linking.
17046 2013-05-17  Marek Polacek  <polacek@redhat.com>
17048         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
17049         length when doing non-zero store of storing '\0' to '\0'.
17051 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17053         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
17054         vect_external_def oprnd1 with loop_vinfo, try to emit
17055         optional cast, negation and and stmts on the loop preheader
17056         edge instead of into the pattern def seq.
17058         PR tree-optimization/57051
17059         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
17060         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
17062 2013-05-16  Nick Clifton  <nickc@redhat.com>
17064         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
17065         (rl78_is_naked_func): New function.
17066         (rl78_expand_prologue): Skip prologue generation for naked functions.
17067         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17068         * doc/extend.texi (naked): Add RL78 to the list of processors
17069         that supports this attribute.
17071 2013-05-16  Jeff Law  <law@redhat.com>
17073         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17075 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17077         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17078         cache parameters using detect_caches_amd also for CYRIX,
17079         NSC and TM2 signatures.
17081 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17082             Dzianis Kahanovich  <mahatma@eu.by>
17084         PR target/45359
17085         PR target/46396
17086         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17087         VIA/Centaur processors and determine their cache parameters
17088         using detect_caches_amd.
17090 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17092         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17093         (rtl_verify_edges): New function.
17094         (rtl_verify_bb_insns): Ditto.
17095         (rtl_verify_bb_pointers): Ditto.
17096         (rtl_verify_bb_insn_chain): Ditto.
17097         (rtl_verify_fallthru): Ditto.
17098         (rtl_verify_bb_layout): Ditto.
17099         (rtl_verify_flow_info_1): Outline checks into new functions.
17100         (rtl_verify_flow_info): Ditto.
17102 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17104         * cfghooks.c (copy_bbs): Add update_dominance argument.
17105         * cfghooks.h (copy_bbs): Update prototype.
17106         * tree-cfg.c (gimple_duplicate_sese_region):
17107         Add update_dominance argument.
17108         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17109         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17110         gimple_duplicate_sese_region call.
17111         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17112         Update copy_bbs call.
17113         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17114         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17116 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17118         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17119         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17120         vect_recog_rotate_pattern.
17121         (vect_recog_rotate_pattern): New function.
17123 2013-05-16  Jason Merrill  <jason@redhat.com>
17125         * Makefile.in (LLINKER): New variable.
17126         (mostlyclean): Remove link mutex.
17127         * configure.ac: Handle --enable-link-mutex.
17128         * lock-and-run.sh: New script.
17130 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17132         PR target/19599
17133         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17134         for NULL decl.
17136 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17138         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17140 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17142         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17143         * config/arm/arm.c (next_consecutive_mem): New function.
17144         (gen_movmem_ldrd_strd): Likewise.
17145         * config/arm/arm.md (movmemqi): Update condition and code.
17146         (unaligned_loaddi, unaligned_storedi): New patterns.
17148 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17150         * config.gcc: Obsolete *-*-solaris2.9*.
17151         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17153 2013-05-16  Richard Biener  <rguenther@suse.de>
17155         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17156         earlier, after GRAPHITE transforms and IV canonicalization.
17158 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17160         * omp-low.c (extract_omp_for_data): For collapsed loops,
17161         if at least one of the loops is known at compile time to
17162         iterate zero times, set count to 0.
17163         (expand_omp_regimplify_p): New function.
17164         (expand_omp_for_generic): For collapsed loops, if at least
17165         one of the loops isn't known to iterate at least once,
17166         add runtime check with setting count to 0.
17167         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17168         For unsigned types if it isn't known at compile time that
17169         the loop will iterate at least once, add runtime check to bypass
17170         the whole loop if initial condition isn't true.
17172 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17174         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17176 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17178         PR middle-end/57286
17179         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17180         transformations to avoid an infinite loop.
17182 2013-05-16  Marek Polacek  <polacek@redhat.com>
17184         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17186 2013-05-15  Leif Ekblad  <leif@rdos.net>
17188         * config/i386/i386.c (ix86_decompose_address): Use
17189         DEFAULT_TLS_SEG_REG to access TLS segment register.
17190         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17191         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17192         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17194 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17196         PR target/57260
17197         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17198         sibling calls to functions that would normally be lazily bound,
17199         unless $gp is call-clobbered.
17201 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17203         * config/i386/i386.c (ix86_option_override_internal): Update
17204         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17205         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17206         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17207         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17208         of TARGET_3DNOW.
17209         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17211 2013-05-15  Andreas Schwab  <schwab@suse.de>
17213         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17214         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17215         third operand.
17217 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17219         * loop-unroll.c (report_unroll_peel): Check decision before
17220         emitting unroll/peel message.
17222 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17224         * function.h (has_bb_partition): New rtl_data flag.
17225         (bb_reorder_complete): Ditto.
17226         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17227         instead of flag_reorder_blocks_and_partition.
17228         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17229         with some enhancements.
17230         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17231         * bb-reorder.c (connect_traces): Check for has_bb_partition
17232         instead of flag_reorder_blocks_and_partition.
17233         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17234         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17235         verify_hot_cold_block_grouping.
17236         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17238 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17240         PR target/19599
17241         * config/arm/predicates.md (call_insn_operand): New predicate.
17242         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17243         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17244         if insn is not a tail call.
17245         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17246         registers.
17247         * config/arm/arm.h (enum reg_class): New caller save register class.
17248         (REG_CLASS_NAMES): Likewise.
17249         (REG_CLASS_CONTENTS): Likewise.
17250         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17251         without decls.
17253 2013-05-15  Richard Biener  <rguenther@suse.de>
17255         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17256         of MSG_OPTIMIZED_LOCATIONS.
17257         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17258         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17259         message.
17260         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17261         of MSG_OPTIMIZED_LOCATIONS.
17262         (execute_vect_slp): Likewise.
17263         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17264         (vect_create_cond_for_alias_checks): Likewise.
17265         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17266         (vect_recog_widen_mult_pattern): Likewise.
17267         (vect_recog_widen_sum_pattern): Likewise.
17268         (vect_recog_over_widening_pattern): Likewise.
17269         (vect_recog_widen_shift_pattern): Likewise.
17270         (vect_recog_vector_vector_shift_pattern): Likewise.
17271         (vect_recog_divmod_pattern): Likewise.
17272         (vect_recog_mixed_size_cond_pattern): Likewise.
17273         (vect_recog_bool_pattern): Likewise.
17274         (vect_pattern_recog_1): Likewise.
17276 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17278         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17279         non-functions to builtin_unreachable.
17280         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17281         correct when new direct edges were discovered.
17283 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17285         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17286         header, print symbol order instead of node uid, print more information
17287         about indirect edge targets.
17288         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17289         uids.
17290         (ipa_make_edge_direct_to_target): Likewise.
17291         (remove_described_reference): Likewise.
17292         (propagate_controlled_uses): Likewise.
17293         (ipa_print_node_params): Also print symbol order.
17294         (ipcp_transform_function): Print symbol order instead of node uids.
17295         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17296         (cgraph_get_create_real_symbol_node): Likewise.
17297         * ipa-cp.c (print_lattice): Likewise.
17298         (print_all_lattices): Likewise.
17299         (determine_versionability): Likewise.
17300         (initialize_node_lattices): Likewise.
17301         (estimate_local_effects): Likewise.
17302         (update_profiling_info): Likewise.
17303         (create_specialized_node): Likewise.
17304         (perhaps_add_new_callers): Likewise.
17305         (decide_about_value): Likewise.
17306         (decide_whether_version_node): Likewise.
17307         (identify_dead_nodes): Likewise.
17308         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17309         (dump_inline_summary): Likewise.
17310         (estimate_node_size_and_time): Likewise.
17311         (inline_analyze_function): Likewise.
17312         * ipa-inline.c (report_inline_failed_reason): Likewise.
17313         (want_early_inline_function_p): Likewise.
17314         (edge_badness): Likewise.
17315         (update_edge_key): Likewise.
17316         (inline_small_functions): Likewise.  Add dumping of order to two other
17317         dumps.
17318         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17319         instead of node uids.
17320         (propagate_pure_const): Likewise.
17321         (propagate_pure_const): Likewise.
17322         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17323         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17324         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17325         of node uids.
17326         * tree-pretty-print.c (dump_function_header): Likewise.
17327         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17328         Print symbol order instead of node uids.
17330 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17332         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17333         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17335 2013-05-15  Richard Biener  <rguenther@suse.de>
17337         PR tree-optimization/57275
17338         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17339         return value for fail to do runtime alias checks for gather loads.
17341 2013-05-15  Jan Hubicka  <jh@suse.cz>
17343         PR lto/57038
17344         PR lto/47375
17345         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17346         weakrefs are not external.
17347         (lto_symtab_merge_decls): Fix thinko when dealing with
17348         non-lto_symtab decls.
17349         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17350         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17351         * varpool.c (dump_varpool_node): Dump more flags.
17353 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17355         * config/i386/i386.c (processor_alias_table): Add instruction
17356         FSGSBASE for AMD bdver3 architecture.
17358 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17360         * tree.c (warn_deprecated_use): Print file:line using locus color.
17361         * diagnostic.c (diagnostic_report_current_module): Print file:line
17362         and file:line:column using locus color.
17364 2013-05-14  Mike Stump  <mikestump@comcast.net>
17366         * gdbinit.in: Add __null.
17368 2013-05-14  Mike Stump  <mikestump@comcast.net>
17370         * recog.h: Rename struct recog_data to Recog_data.
17371         * recog.c: Likewise.
17372         * reload.c (can_reload_into): Likewise.
17373         * config/picochip/picochip.c: Likewise.
17375 2013-05-14  Mike Stump  <mikestump@comcast.net>
17377         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17379 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17381         * resource.h (struct resources): Remove unch_memory member.
17382         (CLEAR_RESOURCE): Don't clear unch_memory.
17383         * resource.c (mark_referenced_resources): Don't set it.
17384         (mark_set_resources): Likewise.
17385         (mark_target_live_regs): Don't clear it.
17386         (init_resource_info): Likewise.
17387         * reorg.c (resource_conflicts_p): Don't compare it.
17388         (redundant_insn): Don't set it.
17390         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17391         Remove prototypes.
17392         * emit-rtl.c (next_label): Remove unused function.
17393         (skip_consecutive_labels, link_cc0_insns): Move to ...
17394         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17395         only place where these functions are used, and make them static.
17397 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17399         * fold-const.c (fold_negate_expr): Handle vectors.
17400         (fold_truth_not_expr): Make it static.
17401         (fold_invert_truthvalue): New static function.
17402         (invert_truthvalue_loc): Handle vectors. Do not call
17403         fold_truth_not_expr directly.
17404         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17405         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17406         (fold_comparison): Handle vector constants.
17407         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17408         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17409         * tree.h (fold_truth_not_expr): Remove declaration.
17411 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17413         * config/aarch64/aarch64-simd.md
17414         (aarch64_vcond_internal<mode>): Rename to...
17415         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17416         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17417         float modes. Clarify all iterator modes.
17418         (vcond<mode><mode>): Use new name for vcond expanders.
17419         (vcond<v_cmp_result><mode>): Likewise.
17420         (vcondu<mode><mode>: Likewise.
17421         * config/aarch64/iterators.md (VDQF_COND): New.
17423 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17425         PR bootstrap/57266
17426         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17427         variable for the shift amount. Check that we shift by non-negative
17428         amounts.
17430 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17432         PR target/42017
17433         * config/arm/arm.h (EPILOGUE_USES): Only return true
17434         for LR_REGNUM after epilogue_completed.
17436 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17438         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17439         is error_mark_node.
17441 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17443         PR target/57261
17444         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17445         and Solaris 11+/x86 with gld.
17446         * configure: Regenerate.
17448 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17450         * expmed.c (expand_shift_1): Canonicalize rotates by
17451         constant bitsize / 2 to bitsize - 1.
17452         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17453         case ROTATERT>: Likewise.
17455         Revert:
17456         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17458         * config/i386/i386.md (rotateinv): New code attr.
17459         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17460         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17461         roll $31, %eax, etc.
17463 2013-05-14  Richard Biener  <rguenther@suse.de>
17465         PR middle-end/57235
17466         * tree-eh.c (sink_clobbers): Give up for successors with
17467         multiple predecessors and no virtual uses.
17469 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17471         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17472         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17474 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17476         PR middle-end/57251
17477         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17478         the case when both op0 and op1 have VOIDmode.
17480 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17482         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17483         in multiply-accumulate mode.
17485 2013-05-13  Guozhi Wei  <carrot@google.com>
17487         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17489 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17491         PR target/56975
17492         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17493         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17494         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17495         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17496         (ix86_expand_prologue): Likewise.
17497         (ix86_expand_split_stack_prologue): Likewise.
17498         (legitimate_pic_address_disp_p): Likewise.
17499         (legitimize_pic_address): Likewise.
17500         (legitimize_tls_address): Likewise.
17501         (legitimize_pe_coff_symbol): Likewise.
17502         (output_pic_addr_const): Likewise.
17503         (construct_plt_address): Likewise.
17504         (ix86_expand_call): Likewise.
17505         (x86_output_mi_thunk): Likewise.
17506         (x86_function_profiler): Likewise.
17508 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17510         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17511         similar switch cases.
17512         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17513         (aarch64_simd_mov_to_<mode>low): Delete.
17514         (aarch64_simd_mov_to_<mode>high): Delete.
17515         (move_lo_quad_<mode>): Add w<-r alternative.
17516         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17517         (aarch64_simd_mov_from_*): Update type attribute.
17518         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17519         statement.
17521 2013-05-13  Jan Hubicka  <jh@suse.cz>
17523         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17524         * config/i386/i386.c (ix86_compute_frame_layout,
17525         ix86_expand_epilogue, emit_i387_cw_initialization,
17526         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17527         ix86_local_alignment): Fix use of size/speed predicates.
17529 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17531         PR tree-optimization/45216
17532         PR tree-optimization/57157
17533         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17534         the (-Y) & (B - 1) variant if OP is |.
17535         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17536         return shifted.  Use (-op1) & (prec - 1) as other_amount
17537         instead of prec - op1.
17539 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17541         PR middle-end/42371
17542         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17543         (ipa_constant_data): New type.
17544         (ipa_jump_func): Use ipa_constant_data to hold information about
17545         constant jump functions.
17546         (ipa_get_jf_constant): Adjust to jump function type changes.
17547         (ipa_get_jf_constant_rdesc): New function.
17548         (ipa_param_descriptor): New field controlled_uses.
17549         (ipa_get_controlled_uses): New function.
17550         (ipa_set_controlled_uses): Likewise.
17551         * ipa-ref.h (ipa_find_reference): Declare.
17552         * ipa-prop.c (ipa_cst_ref_desc): New type.
17553         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17554         changes.
17555         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17556         New parameter cs.  Adjust all callers.
17557         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17558         (remove_described_reference): New function.
17559         (jfunc_rdesc_usable): Likewise.
17560         (try_make_edge_direct_simple_call): Decrement controlled use count,
17561         attempt to remove reference if it hits zero.
17562         (combine_controlled_uses_counters): New function.
17563         (propagate_controlled_uses): Likewise.
17564         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17565         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17566         (ipa_print_node_params): Print described use counter.
17567         (ipa_write_jump_function): Adjust to jump function type changes.
17568         (ipa_read_jump_function): New parameter CS, pass it to
17569         ipa_set_jf_constant.  Adjust caller.
17570         (ipa_write_node_info): Stream controlled use count
17571         (ipa_read_node_info): Likewise.
17572         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17573         asserting.
17574         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17575         count.  Remove cloning-added reference if it reaches zero.
17576         * ipa-ref.c (ipa_find_reference): New function.
17578 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17580         * config/i386/i386.c (processor_target_table): Modified default
17581         alignment values for AMD BD and BT architectures.
17583 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17585         * tree-vect-generic.c (uniform_vector_p): Move ...
17586         * tree.c (uniform_vector_p): ... here.
17587         * tree.h (uniform_vector_p): Declare it.
17588         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17589         into a scalar.
17591 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17593         PR tree-optimization/57230
17594         * tree-ssa-strlen.c (handle_char_store): Record length for
17595         array store from STRING_CST.
17597         PR tree-optimization/57230
17598         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17599         check.
17601 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17603         * config/epiphany/epiphany.c (epiphany_init): Check size of
17604         NUM_MODES_FOR_MODE_SWITCHING.
17605         (epiphany_expand_prologue):
17606         Remove CONFIG_REGNUM initial value handling code.
17607         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17608         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17609         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17610         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17611         Don't return 1 for FP_MODE_NONE.
17612         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17613         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17614         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17615         * config/epiphany/epiphany.md (save_config): New pattern.
17617 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17619         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17621 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17623         * config/i386/i386.md (memory): Handle sseishft1.
17624         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17625         (*vec_extractv2di_1): Ditto.
17627 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17629         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17630         saved registers.
17632 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17634         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17635         Add mthumb/march=armv7-a multilib.
17636         Add mthumb/march=armv7-r multilib.
17637         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17639 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17641         * config/v850/t-rtems: Add more multilibs.
17643 2013-05-10  Richard Biener  <rguenther@suse.de>
17645         PR tree-optimization/57214
17646         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17647         not propagate from SSA names that occur in abnormal PHI nodes.
17649 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17651         * stor-layout.c (element_precision): New function.
17652         * machmode.h (element_precision): Declare it.
17653         * tree.c (build_minus_one_cst): New function.
17654         (element_precision): Likewise.
17655         * tree.h (build_minus_one_cst): Declare new function.
17656         (element_precision): Likewise.
17657         * fold-const.c (operand_equal_p): Use element_precision.
17658         (fold_binary_loc): Handle vector types.
17659         * convert.c (convert_to_integer): Use element_precision.
17660         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17661         separately.
17663 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17665         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17666         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17667         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17668         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17669         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17670         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17671         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17672         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17673         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17674         (Uuw8): New constraints.
17675         (Usb4): Move into alphabetical order.
17676         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17677         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17678         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17679         previously unnamed patterns.
17680         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17681         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17682         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17683         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17684         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17685         of specific lengths.
17687 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17689         * config/i386/i386.md (rotateinv): New code attr.
17690         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17691         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17692         roll $31, %eax, etc.
17694         PR tree-optimization/45216
17695         PR tree-optimization/57157
17696         * tree-ssa-forwprop.c (simplify_rotate): New function.
17697         (ssa_forward_propagate_and_combine): Call it.
17699 2013-05-10  Richard Biener  <rguenther@suse.de>
17701         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17702         disable peeling when we version for aliasing.
17703         (vector_alignment_reachable_p): Honor explicit user alignment.
17704         (vect_supportable_dr_alignment): Likewise.
17705         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17706         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17707         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17708         then peeling to arrange for the cost-model check to come first.
17710 2013-05-10  Alan Modra  <amodra@gmail.com>
17712         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17713         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17714         * configure: Regenerate.
17716 2013-05-10  Alan Modra  <amodra@gmail.com>
17718         PR target/55033
17719         * varasm.c (default_elf_select_section): Move !DECL_P check..
17720         (get_named_section): ..to here before calling get_section_name.
17721         Adjust assertion.
17722         (default_section_type_flags): Add DECL_P check.
17723         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17724         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17726 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17728         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17729         When using gen_stack_adjust_str with a register offset, add a
17730         REG_FRAME_RELATED_EXPR note.
17732 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17734         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17735         (*vec_extractv4si_zext_mem): Ditto.
17736         (*vec_extractv2di): Add 0->x and x->x alternatives.
17737         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17738         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17740 2013-05-09  Jason Merrill  <jason@redhat.com>
17742         N3639 C++1y VLA support
17743         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17744         DECL_VALUE_EXPR.
17746         * tree.c (build_constructor_va): New.
17747         * tree.h: Declare it.
17749 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17751         PR lto/57084
17752         * gimple-fold.c (canonicalize_constructor_val): Call
17753         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17755 2013-05-09  Jan Hubicka  <jh@suse.cz>
17756             Richard Biener  <rguenther@suse.de>
17758         PR lto/54095
17759         * symtab.c (symtab_make_decl_local): Do not add private names.
17761 2013-05-09  Jan Hubicka  <jh@suse.cz>
17763         PR lto/54095
17764         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17765         (unlink_from_assembler_name_hash): Likewise.
17766         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17767         symtab_unregister_node, symtab_initialize_asm_name_hash,
17768         change_decl_assembler_name): Update.
17770 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17772         * config/aarch64/aarch64.md: New movtf split.
17773         (*movtf_aarch64): Update.
17774         (aarch64_movdi_tilow): Handle TF modes and rename to
17775         aarch64_movdi_<mode>low.
17776         (aarch64_movdi_tihigh): Handle TF modes and rename to
17777         aarch64_movdi_<mode>high
17778         (aarch64_movtihigh_di): Handle TF modes and rename to
17779         aarch64_mov<mode>high_di
17780         (aarch64_movtilow_di): Handle TF modes and rename to
17781         aarch64_mov<mode>low_di
17782         (aarch64_movtilow_tilow): Remove spurious whitespace.
17783         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17784         splits.
17785         (aarch64_print_operand): Update.
17787 2013-05-09  Alan Modra  <amodra@gmail.com>
17789         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17790         powerpc64le.
17791         * configure: Regenerate.
17793 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17795         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17796         splitter preparation statements.
17797         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17798         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17799         adjust_address_nv.
17801 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17803         * gimple-ssa-strength-reduction.c (count_candidates): Change
17804         return value to int.
17805         (analyze_candidates_and_replace): Change type of length to int.
17807 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17809         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17810         (*vec_extract<mode>): Use VI12_128 mode iterator.
17811         (*vec_extract<mode>_mem): Ditto.
17812         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17813         attribute.
17815 2013-05-08  Diego Novillo  <dnovillo@google.com>
17817         PR bootstrap/54659
17819         Revert:
17820         2012-08-17  Diego Novillo  <dnovillo@google.com>
17822         PR bootstrap/54281
17823         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17824         * config.in: Regenerate.
17825         * configure: Regenerate.
17826         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17828 2013-05-08  Jan Hubicka  <jh@suse.cz>
17830         PR lto/54095
17831         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17832         * cgraph.h (symtab_node_base): Add unique_name.
17833         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17834         input_overwrite_node, input_varpool_node): Stream unique_name.
17835         * cgraphclones.c (cgraph_create_virtual_clone,
17836         cgraph_function_versioning): Set unique_name.
17837         * ipa.c (function_and_variable_visibility): Set unique_name.
17839 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17841         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17842         (alloc_cand_and_find_basis): Restrict conditional candidate
17843         processing to CAND_MULTs.
17845 2013-05-08  Jan Hubicka  <jh@suse.cz>
17847         PR lto/54095
17848         lto-symtab.c (lto_symtab_symbol_p): New function.
17849         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17850         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17851         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17852         Skip static symbols.
17854 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17856         PR tree-optimization/57200
17857         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17858         Only call inform if the preceding warning_at returns true.
17860 2013-05-07  Han Shen  <shenhan@google.com>
17862         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17863         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17864         * common.opt (fstack-protector-strong): New option.
17865         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17866         * doc/invoke.texi (Optimization Options): Document
17867         "-fstack-protector-strong".
17868         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17870 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17872         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17874 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17876         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17877         it is not lhs of the insn.
17878         (setup_reg_equiv): Remove insn having equiv memory which it is not
17879         lhs of the insn.
17880         * lra-constraints.c (process_address): Try to improve generation
17881         code for address base + disp.
17882         (lra_constraints): Make correct the code for checking insn setting
17883         up backward equivalence.  Remove insn only if it is in the init
17884         insn list.
17885         * lra-eliminations.c (update_reg_eliminate): Change return value.
17886         (lra_eliminate): Use the result.
17888 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17890         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17891         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17892         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17893         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17894         register target operands.
17895         (*vec_extractv8hi_sse2): New pattern.
17896         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17897         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17898         (*vec_extract<mode>_mem): New insn and split pattern.
17900 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17902         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17903         (TARGET_ASAN_SHADOW_OFFSET): Define.
17904         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17905         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17907 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17909         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17910         (incr_vec_index): Return -1 if increment not found.
17911         (create_add_on_incoming_edge): Assert if increment not found.
17912         (record_increment): Limit number of increments recorded.
17913         (all_phi_incrs_profitable): Return false if an increment not found.
17914         (replace_profitable_candidates): Don't process increments that were
17915         not recorded.
17916         (analyze_candidates_and_replace): Limit size of incr_vec.
17918 2013-05-07  Richard Biener  <rguenther@suse.de>
17920         * calls.c (special_function_p): setjmp-like functions are leaf.
17921         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17922         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17924 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17926         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17927         (aarch64_simd_mov<mode>): New expander.
17928         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17929         (aarch64_simd_mov_to_<mode>high): Likewise.
17930         (aarch64_simd_mov_from_<mode>low): Likewise.
17931         (aarch64_simd_mov_from_<mode>high): Likewise.
17932         (aarch64_dup_lane<mode>): Update.
17933         (aarch64_dup_lanedi): New instruction pattern.
17934         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17935         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17937 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17939         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17940         (replace_mult_candidate): Remove unnecessary argument; remove
17941         unnecessary parameter from call to introduce_cast_before_cand.
17942         (replace_unconditional_candidate): Remove unnecessary parameter
17943         from call to replace_mult_candidate.
17944         (replace_conditional_candidate): Likewise.
17945         (insert_initializers): Use make_temp_ssa_name.
17946         (introduce_cast_before_cand): Remove unnecessary argument; use
17947         make_temp_ssa_name.
17948         (replace_one_candidate): Remove unnecessary argument; remove
17949         unnecessary parameter from calls to introduce_cast_before_cand.
17950         (replace_profitable_candidates): Remove unnecessary parameters
17951         from calls to replace_one_candidate.
17953 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17955         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17956         phi def as possibly hiding a basis for a CAND_ADD whose operands
17957         have been commuted in the analysis.
17958         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17960 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17962         * config/aarch64/aarch64.md
17963         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17964         shift value between 0-4.
17966 2013-05-07  Richard Biener  <rguenther@suse.de>
17968         * double-int.h (rshift): New overload.
17969         * double-int.c (rshift): New function.
17970         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17971         (create_reference_ops_from_ref): Remove.
17972         (vn_reference_insert): Use shared ops for constructing the
17973         reference and copy it.
17975 2013-05-07  Richard Biener  <rguenther@suse.de>
17977         PR middle-end/57190
17978         * tree-eh.c (sink_clobbers): Properly propagate
17979         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17981 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17983         PR tree-optimization/57149
17984         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17985         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17986         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17987         uninit_undefined_value_p instead of ssa_undefined_value_p.
17989         PR debug/57184
17990         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17991         for modifier == EXPAND_INITIALIZER.
17993 2013-05-07  Anton Blanchard  <anton@samba.org>
17995         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17996         for powerpc64 little endian.
17997         * configure: Regenerate.
17999 2013-05-06  Graham Stott  <grahams@btinternet.com>
18001         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
18002         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
18003         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
18004         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
18006 2013-05-06  Graham Stott  <grahams@btinternet.com>
18008         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
18009         codes which allow non-lvalues.
18011 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
18013         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
18014         components are all 1s.
18015         (integer_minus_onep): New function.
18016         * tree.h (integer_minus_onep): Declare it.
18017         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
18018         integer_minus_onep instead of integer_all_onesp.
18020 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18022         PR target/52933
18023         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
18024         variations of these patterns.
18026 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18028         * config/i386/i386.md (isa): Add x64_sse4 member.
18029         (enabled): Handle x64_sse4.
18030         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
18031         instruction for 64bit SSE4_1 targets.  Update insn attributes.
18032         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
18033         instruction for SSE4_1 targets.  Update insn attributes.
18034         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18035         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
18036         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
18037         const_1 selector.
18038         (*vec_extractv4si): Rename from *sse4_1_pextrd.
18039         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
18040         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
18042 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18044         PR target/57108
18045         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
18047 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
18049         * final.c (do_assembler_dialects): Don't handle curly braces and
18050         vertical bar escaped by % as dialect delimiters.
18051         (output_asm_insn): Print curly braces and vertical bar if escaped
18052         by % and ASSEMBLER_DIALECT defined.
18053         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
18054         * doc/tm.texi: Regenerated.
18056 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18058         * config/mips/mips.c: Include tree-pass.h.
18059         (mips_reorg): Split in pre- and post-dbr_schedule parts.
18060         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
18061         (pass_mips_machine_reorg2): New machine specific pass.
18062         (insert_pass_mips_machine_reorg2): New pass plugin definition.
18063         (mips_option_override): Register the new pass.
18064         * rtl.h (cleanup_barriers): Remove prototype.
18065         (dbr_schedule): Likewise.
18066         * jump.c (cleanup_barriers): Make static.
18067         * reorg.c (dbr_schedule): Likewise.
18069 2013-05-06  Richard Biener  <rguenther@suse.de>
18071         PR tree-optimization/57185
18072         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18073         SSA names properly.
18075 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18077         PR target/57106
18078         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18079         between "first_arg" and "insn", not "last" and "insn".
18081 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18083         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18084         (find_candidates_in_block): Re-enable slsr_process_phi.
18085         (create_phi_basis): Fix double counting of candidate adjustment.
18087 2013-05-06  Richard Biener  <rguenther@suse.de>
18089         PR middle-end/57147
18090         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18091         the edge is also fallthru, preserve it and just clear the
18092         abnormal flag.
18093         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18094         also complex, preserve that and just clear the fallthru flag.
18095         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18096         update virtual operands.
18098 2013-05-06  Alan Modra  <amodra@gmail.com>
18100         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18101         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18102         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18103         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18104         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18105         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18107 2013-05-06  Alan Modra  <amodra@gmail.com>
18109         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18110         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18111         (DEFAULT_ASM_ENDIAN): Define.
18112         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18113         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18114         Update -K PIC clause from sysv4.h.
18115         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18116         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18118 2013-05-06  Alan Modra  <amodra@gmail.com>
18120         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18121         twice for little-endian.
18122         (ashrdi3_no_power, ashrdi3): Support little-endian.
18124 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18126         PR target/55303
18127         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18128         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18129         related expanders.
18130         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18131         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18132         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18133         New predicates.
18135 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18136             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18138         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18139         * config/pa/pa.opt: Make mbig-switch a no-op.
18140         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18141         (CASE_VECTOR_MODE): Always return SImode.
18142         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18143         for the !TARGET_BIG_SWITCH case.
18144         * config/pa/pa-linux.h: Likewise.
18145         * config/pa/pa-openbsd.h: Likewise.
18146         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18147         * config/pa/pa.md (short_jump): Remove define_insn.
18148         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18149         (casesi0): Remove define_insn.
18150         (type): Remove btable_branch.
18151         (pa_combine_type): Likewise.
18152         (in_nullified_branch_delay): Likewise.
18153         (in_call_delay): Likewise.
18154         (define_delay): Likewise.
18155         (define_insn_reservation "Z3"): Likewise.
18156         (define_insn_reservation "Z4"): Likewise.
18157         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18158         (pa_adjust_insn_length): Remove adjustment for btable branches.
18159         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18160         and mno-big-switch
18162 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18164         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18165         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18166         Add m->r,x alternatives.
18167         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18168         splitters using SWI48x mode iterator.
18169         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18170         TARGET_64BIT.  Add m->x alternative.
18171         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18172         Add o->x alternative.  Enable for TARGET_SSE.
18173         (sse_storeq): Remove expander.
18174         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18175         with memory input operand.
18176         (*vec_extractv2di_1 splitter): New.
18177         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18178         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18180 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18182         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18183         (INT_LOWPART): Delete.
18184         (extract_MB): Adjust.
18185         (extract_ME): Adjust.
18186         (print_operand): Adjust.
18188 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18190         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18191         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18192         (reg_or_logical_cint_operand, easy_fp_constant,
18193         logical_const_operand): Delete "CONST_DOUBLE" case.
18194         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18195         "HOST_BITS_PER_WIDE_INT == 64" test.
18196         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18197         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18198         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18199         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18200         test.
18201         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18202         CONST_DOUBLE DImode/VOIDmode case.
18203         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18204         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18205         CONST_DOUBLE VOIDmode case.
18206         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18207         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18208         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18209         Delete CONST_DOUBLE case.
18210         (splitters for mov FMOVE64 const_double): Delete
18211         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18212         "HOST_BITS_PER_WIDE_INT >= 64" test.
18213         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18214         case.
18215         (mov DI const_double): Delete.
18217 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18219         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18220         on op shows all bits zero in mode of a lowpart subreg, return zero.
18222 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18224         PR target/57150
18225         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18226         to save TFmode registers and DImode to save TImode registers for
18227         caller save operations.
18228         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18229         mark being partially clobbered since they only use the first
18230         double word.
18232         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18233         and TDmode only use the upper 64-bits of each VSX register.
18235 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18237         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18238         (find_candidates_in_block): Disable slsr_process_phi.
18240 2013-05-03  Guozhi Wei  <carrot@google.com>
18242         * coverage.c (coverage_obj_init): Move the construction of gcov
18243         constructor to ...
18244         (build_init_ctor): ... here.
18246 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18248         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18249         (slsr_cand_d): Redefine def_phi.
18250         (stride_status, phi_adjust_status, count_phis_status): New enums.
18251         (find_phi_def): New.
18252         (find_basis_for_base_expr): New.
18253         (find_basis_for_candidate): Handle hidden bases.
18254         (alloc_cand_and_find_basis): Handle phi candidates.
18255         (slsr_process_phi): New.
18256         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18257         (create_mul_imm_cand): Likewise.
18258         (create_add_ssa_cand): Exclude phi base candidates.
18259         (create_add_imm_cand): Likewise.
18260         (slsr_process_cast): Likewise.
18261         (slsr_process_copy): Likewise.
18262         (find_candidates_in_block): Handle phi candidates.
18263         (dump_candidate): Likewise.
18264         (unconditional_cands): Delete.
18265         (unconditional_cands_with_known_stride_p): Delete.
18266         (phi_dependent_cand_p): New.
18267         (cand_increment): Handle phi-dependent candidates.
18268         (replace_dependent): Delete.
18269         (replace_mult_candidate): New.
18270         (replace_unconditional_candidate): New.
18271         (incr_vec_index): Move to avoid forward reference.
18272         (create_add_on_incoming_edge): New.
18273         (create_phi_basis): New.
18274         (replace_dependents): Delete.
18275         (replace_conditional_candidate): New.
18276         (phi_add_costs): New.
18277         (replace_uncond_cands_and_profitable_phis): New.
18278         (record_increment): Handle phi adjustments.
18279         (record_phi_increments): New.
18280         (record_increments): Handle phi adjustments.
18281         (phi_incr_cost): New.
18282         (lowest_cost_path): Handle phis.
18283         (total_savings): Likewise.
18284         (analyze_increments): Likewise.
18285         (ncd_with_phi): New.
18286         (ncd_of_cand_and_phis): New.
18287         (nearest_common_dominator_for_cands): Handle phi increments.
18288         (all_phi_incrs_profitable): New.
18289         (replace_profitable_candidates): Handle phi-dependent candidates.
18290         (analyze_candidates_and_replace): Likewise.
18292 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18294         PR bootstrap/57154
18295         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18296         do not exceed REG_BR_PROB_BASE.
18298 2013-05-03  Jeff Law  <law@redhat.com>
18300         PR tree-optimization/57144
18301         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18302         operand of the condition will bit into the new type when eliminating
18303         a cast feeding a condition.
18305 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18307         PR rtl-optimization/57130
18308         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18309         of COMPARE as in_code to the recursive call if needed.
18311 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18313         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18314         (enabled): Handle new members.
18315         * config/i386/sse.md (*vec_concatv2si): Merge from
18316         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18317         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18319 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18321         PR tree-optimization/57027
18322         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18323         for fnms opportunity, check we got the prerequisite kind
18324         of tree / gimple before using accessor functions.
18326 2013-05-03  Richard Biener  <rguenther@suse.de>
18328         * double-int.h (lshift): New overload without precision
18329         and arith argument.
18330         (operator *=, operator +=, operator -=): Move ...
18331         * double-int.c (operator *=, operator +=, operator -=): ... here
18332         and implement more efficiently.
18333         (mul_double_with_sign): Remove.
18334         (lshift_double): Adjust to take unsinged shift argument, push
18335         dispatching code to callers.
18336         (mul_double_wide_with_sign): Add early out for callers that
18337         are not interested in high parts or overflow.
18338         (lshift): New function.
18339         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18340         dispatch code here.
18341         (lrotate, rrotate): Use logical shifts.
18342         * expr.c (get_inner_reference): Use lshift.
18343         * fixed-value.c (do_fixed_divide): Likewise.
18344         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18345         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18346         (indirect_refs_may_alias_p): Likewise.
18347         (stmt_kills_ref_p_1): Likewise.
18349 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18351         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18353 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18355         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18356         scalar form of FABD instruction.
18358 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18360         * lra-constraints.c (process_alt_operands): Add checking alt
18361         number to choose the best alternative.
18363 2013-05-02  Richard Biener  <rguenther@suse.de>
18365         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18366         bitmap and its handling.
18367         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18369 2013-05-02  Richard Biener  <rguenther@suse.de>
18371         PR middle-end/57140
18372         * tree-inline.c (copy_loops): Properly handle removed loops.
18373         (copy_cfg_body): Mark destination loops for fixup if source
18374         loops needed fixup.
18376 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18378         PR target/56732
18379         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18380         generating simple_return for naked functions.
18382 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18384         PR middle-end/56988
18385         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18386         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18387         flags match.
18388         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18389         ipa_agg_replacement_value structures.
18390         (known_aggs_to_agg_replacement_list): Likewise.
18391         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18392         (read_agg_replacement_chain): Likewise.
18393         (ipcp_transform_function): Also check that by_ref flags match.
18395 2013-05-02  Richard Biener  <rguenther@suse.de>
18397         * graphds.h (struct graph): Add obstack member.
18398         * graphds.c (new_graph): Initialize obstack and allocate
18399         vertices from it.
18400         (add_edge): Allocate edge from the obstack.
18401         (free_graph): Free the obstack instead of all edges and vertices.
18403 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18405         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18406         divides.
18407         * cfg.c (update_bb_profile_for_threading): Ditto.
18408         * tree-inline.c (copy_bb): Ditto.
18409         (copy_edges_for_bb): Ditto.
18410         (initialize_cfun): Ditto.
18411         (copy_cfg_body): Ditto.
18412         (expand_call_inline): Ditto.
18413         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18414         (estimate_node_size_and_time): Ditto.
18415         (inline_merge_summary): Ditto.
18416         * cgraphclones.c (cgraph_clone_edge): Ditto.
18417         (cgraph_clone_node): Ditto.
18418         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18419         (compute_trg_info): Ditto.
18421 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18423         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18424         S reg when fp attribute set.
18425         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18427 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18429         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18430         New pattern.
18431         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18432         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18433         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18435 2013-05-02  Richard Biener  <rguenther@suse.de>
18437         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18438         (struct instantiate_cache_entry): New type.
18439         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18440         (struct instantiate_cache_type): New type.
18441         (set_instantiated_value, get_instantiated_value): Remove.
18442         (get_instantiated_value_entry): New function.
18443         (instantiate_scev_name): Use the new cache and adjust.
18444         (instantiate_scev_poly): Adjust.
18445         (instantiate_scev_binary): Likewise.
18446         (instantiate_array_ref): Likewise.
18447         (instantiate_scev_convert): Likewise.
18448         (instantiate_scev_not): Likewise.
18449         (instantiate_scev_3): Likewise.
18450         (instantiate_scev_2): Likewise.
18451         (instantiate_scev_r): Likewise.
18452         (instantiate_scev): Likewise.
18453         (resolve_mixers): Likewise.
18455 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18457         PR target/57091
18458         * lra-constraints.c (best_small_class_operands_num): Remove.
18459         (process_alt_operands): Remove small_class_operands_num.  Take
18460         small classes operands into losers and only if the operand is not
18461         matched.  Modify debugging output.
18462         (curr_insn_transform): Remove best_small_class_operands_num.
18463         Print insn name.
18465 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18467         * config/aarch64/aarch64-builtins.c
18468         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18469         * config/aarch64/aarch64-simd-builtins.def
18470         (reduc_splus_): Add new modes.
18471         (reduc_uplus_): New.
18472         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18473         (reduc_uplus_v4sf): Likewise.
18474         (reduc_splus_v4sf): Likewise.
18475         (aarch64_addv<mode>): Likewise.
18476         (reduc_uplus_<mode>): Likewise.
18477         (reduc_splus_<mode>): Likewise.
18478         (aarch64_addvv2di): Likewise.
18479         (reduc_uplus_v2di): Likewise.
18480         (reduc_splus_v2di): Likewise.
18481         (aarch64_addvv2si): Likewise.
18482         (reduc_uplus_v2si): Likewise.
18483         (reduc_splus_v2si): Likewise.
18484         (reduc_<sur>plus_<mode>): New.
18485         (reduc_<sur>plus_v2di): Likewise.
18486         (reduc_<sur>plus_v2si): Likewise.
18487         (reduc_<sur>plus_v4sf): Likewise.
18488         (aarch64_addpv4sf): Likewise.
18489         * config/aarch64/arm_neon.h
18490         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18491         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18492         add UNSPEC_SADDV, UNSPEC_UADDV.
18493         (SUADDV): New.
18494         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18496 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18498         * config/aarch64/arm_neon.h
18499         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18501 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18503         * config/aarch64/aarch64-builtins
18504         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18506 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18508         * config/aarch64/aarch64-simd-builtins.def
18509         (reduc_smax_): New.
18510         (reduc_smin_): Likewise.
18511         (reduc_umax_): Likewise.
18512         (reduc_umin_): Likewise.
18513         (reduc_smax_nan_): Likewise.
18514         (reduc_smin_nan_): Likewise.
18515         (fmax): Remove.
18516         (fmin): Likewise.
18517         (smax): Update for V2SF, V4SF and V2DF modes.
18518         (smin): Likewise.
18519         (smax_nan): New.
18520         (smin_nan): Likewise.
18521         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18522         (<su><maxmin><mode>3): ...This, refactor.
18523         (s<maxmin><mode>3): New.
18524         (<maxmin_uns><mode>3): Likewise.
18525         (reduc_<maxmin_uns>_<mode>): Refactor.
18526         (reduc_<maxmin_uns>_v4sf): Likewise.
18527         (reduc_<maxmin_uns>_v2si): Likewise.
18528         (aarch64_<fmaxmin><mode>: Remove.
18529         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18530         new builtin names.
18531         (vmin<q>_f<32,64>): Likewise.
18532         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18533         (FMAXMIN): New.
18534         (su): Add mappings for smax, smin, umax, umin.
18535         (maxmin): New.
18536         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18537         (FMAXMIN): Rename as...
18538         (FMAXMIN_UNS): ...This.
18539         (maxminv): Remove.
18540         (fmaxminv): Likewise.
18541         (fmaxmin): Likewise.
18542         (maxmin_uns): New.
18543         (maxmin_uns_op): Likewise.
18545 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18547         * config/aarch64/arm_neon.h
18548         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18549         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18550         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18552 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18554         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18555         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18557 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18559         * config/aarch64/aarch64-simd.md
18560         (vcond<mode>_internal): Handle special cases for constant masks.
18561         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18562         (vcondu<mode><mode>): Likewise.
18563         (vcond<v_cmp_result><mode>): New.
18565 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18567         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18568         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18569         * config/aarch64/aarch64-simd-builtins.def
18570         (cmeq): Update to BUILTIN_VALLDI.
18571         (cmgt): Likewise.
18572         (cmge): Likewise.
18573         (cmle): Likewise.
18574         (cmlt): Likewise.
18575         * config/aarch64/arm_neon.h
18576         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18577         to builtins or C as appropriate.
18579 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18581         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18582         (cmgeu): ...This.
18583         (cmhi): Rename to...
18584         (cmgtu): ...This.
18585         * config/aarch64/aarch64-simd.md
18586         (simd_mode): Add SF.
18587         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18588         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18589         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18590         (cstore<mode>_neg): ...This.
18591         * config/aarch64/iterators.md
18592         (VALLF): new.
18593         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18594         (COMPARISONS): New.
18595         (UCOMPARISONS): Likewise.
18596         (optab): Add missing comparisons.
18597         (n_optab): New.
18598         (cmp_1): Likewise.
18599         (cmp_2): Likewise.
18600         (CMP): Likewise.
18601         (cmp): Remove.
18602         (VCMP_S): Likewise.
18603         (VCMP_U): Likewise.
18604         (V_cmp_result): Add DF, SF modes.
18605         (v_cmp_result): Likewise.
18606         (v): Likewise.
18607         (vmtype): Likewise.
18608         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18610 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18612         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18613         define_insn to define_insn_and_split.
18614         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18615         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18616         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18617         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18618         (thumb2_negscc): Likewise.
18620 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18622         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18624 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18626         * config/arm/thumb2.md: Remove trailing whitespaces.
18628 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18630         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18631         Use gen_int_mode rather than GEN_INT.
18633 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18635         * value-prof.c (stream_in_histogram_value): Remove the strayed
18636         debug_gimple_stmt.
18638 2013-04-30  Richard Biener  <rguenther@suse.de>
18640         PR middle-end/57122
18641         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18643 2013-04-30  Richard Biener  <rguenther@suse.de>
18645         PR middle-end/57107
18646         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18648 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18650         PR rtl-optimization/56957
18651         PR rtl-optimization/57105
18652         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18653         variable.  Use just INSN_UID for determining whether an insn
18654         should be only disconnected from the insn stream.
18655         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18657 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18659         PR tree-optimization/57104
18660         * tsan.c (instrument_expr): Don't instrument accesses to
18661         DECL_HARD_REGISTER VAR_DECLs.
18663 2013-04-30  Richard Biener  <rguenther@suse.de>
18665         * function.h (loops_for_fn): New inline function.
18666         (set_loops_for_fn): Likewise.
18667         * cfgloop.h (place_new_loop): Add struct function parameter.
18668         (get_loop): Likewise.
18669         (get_loops): Likewise.
18670         (number_of_loops): Likewise.
18671         (fel_next): Adjust.
18672         (fel_init): Likewise.
18673         * cfg.c (get_loop_copy): Adjust.
18674         * cfgloop.c (flow_loops_dump): Likewise.
18675         (record_loop_exits): Likewise.
18676         (verify_loop_structure): Likewise.
18677         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18678         (estimate_reg_pressure_cost): Likewise.
18679         (mark_loop_exit_edges): Likewise.
18680         * cfgloopmanip.c (place_new_loop): Likewise.
18681         (add_loop): Likewise.
18682         (duplicate_loop): Likewise.
18683         * graph.c (draw_cfg_nodes): Likewise.
18684         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18685         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18686         (extract_affine_chrec): Likewise.
18687         (build_scop_iteration_domain): Likewise.
18688         * graphite.c (graphite_initialize): Likewise.
18689         * ira-build.c (create_loop_tree_nodes): Likewise.
18690         (more_one_region_p): Likewise.
18691         (rebuild_regno_allocno_maps): Likewise.
18692         (mark_loops_for_removal): Likewise.
18693         (mark_all_loops_for_removal): Likewise.
18694         (remove_unnecessary_regions): Likewise.
18695         (ira_build): Likewise.
18696         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18697         * loop-init.c (fix_loop_structure): Likewise.
18698         (gate_rtl_move_loop_invariants): Likewise.
18699         (gate_rtl_unswitch): Likewise.
18700         (gate_rtl_unroll_and_peel_loops): Likewise.
18701         (rtl_doloop): Likewise.
18702         * lto-streamer-in.c (input_cfg): Likewise.
18703         * lto-streamer-out.c (output_cfg): Likewise.
18704         * modulo-sched.c (sms_schedule): Likewise.
18705         * predict.c (tree_estimate_probability): Likewise.
18706         (tree_estimate_probability_driver): Likewise.
18707         (estimate_loops): Likewise.
18708         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18709         (move_sese_region_to_fn): Likewise.
18710         (debug_loop_num): Likewise.
18711         * tree-chrec.c (chrec_evaluate): Likewise.
18712         (hide_evolution_in_other_loops_than_loop): Likewise.
18713         (chrec_component_in_loop_num): Likewise.
18714         (reset_evolution_in_loop): Likewise.
18715         (evolution_function_is_invariant_rec_p): Likewise.
18716         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18717         * tree-inline.c (copy_loops): Likewise.
18718         (copy_cfg_body): Likewise.
18719         (tree_function_versioning): Likewise.
18720         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18721         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18722         Likewise.
18723         (add_to_evolution_1): Likewise.
18724         (scev_const_prop): Likewise.
18725         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18726         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18727         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18728         (tree_ssa_lim_initialize): Likewise.
18729         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18730         (verify_loop_closed_ssa): Likewise.
18731         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18732         (tree_ssa_loop_im): Likewise.
18733         (tree_ssa_loop_unswitch): Likewise.
18734         (tree_vectorize): Likewise.
18735         (check_data_deps): Likewise.
18736         (tree_ssa_loop_ivcanon): Likewise.
18737         (tree_ssa_loop_bounds): Likewise.
18738         (tree_complete_unroll): Likewise.
18739         (tree_complete_unroll_inner): Likewise.
18740         (tree_parallelize_loops): Likewise.
18741         (tree_ssa_loop_prefetch): Likewise.
18742         (tree_ssa_loop_ivopts): Likewise.
18743         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18744         * tree-vectorizer.c (vectorize_loops): Likewise.
18746 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18748         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18749         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18750         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18751         with EABI_LINK_SPEC.
18753 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18755         PR target/44578
18756         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18757         alternative.
18759 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18761         PR target/57097
18762         * lra-constraints.c (process_alt_operands): Discourage a bit more
18763         using memory for pseudos.  Print cost dump for alternatives.
18764         Modify cost values for conflicts with early clobbers.
18765         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18767 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18769         PR target/57098
18770         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18772 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18774         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18775         from/to S register.
18776         (movdi_aarch64): Support LDR/STR from/to D register.
18778 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18780         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18781         or higher by default.
18783 2013-04-29  Richard Biener  <rguenther@suse.de>
18785         PR middle-end/57075
18786         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18787         even if not adding abnormal edges for calls that can make
18788         abnormal gotos.
18790 2013-04-29  Richard Biener  <rguenther@suse.de>
18792         PR middle-end/57103
18793         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18794         TREE_BLOCK.
18795         (move_stmt_r): Remove redundant checking.
18797 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18799         PR bootstrap/57077
18800         * basic-block.h (apply_scale): New function.
18801         (apply_probability): Use apply_scale.
18802         * gimple-streamer-in.c (input_bb): Ditto.
18803         * lto-streamer-in.c (input_cfg): Ditto.
18804         * lto-cgraph.c (merge_profile_summaries): Ditto.
18805         * tree-optimize.c (execute_fixup_cfg): Ditto.
18806         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18807         (copy_edges_for_bb): Ditto.
18808         (copy_cfg_body): Ditto.
18810 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18812         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18813         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18814         (tail_merge_optimize): Handle current_loops == NULL.
18816 2013-04-26  Jeff Law  <law@redhat.com>
18818         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18819         (simplify_cond_using_ranges): Generalize code to simplify
18820         COND_EXPRs where one argument is a constant and the other
18821         is an SSA_NAME created by an integral type conversion.
18823 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18825         * config/arm/arm.md (store_minmaxsi): Use only when
18826         optimize_insn_for_size_p.
18828 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18830         PR target/57108
18831         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18833 2013-04-29  Richard Biener  <rguenther@suse.de>
18835         PR middle-end/57089
18836         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18837         loop tree make sure to schedule a fixup for the child as well.
18838         (expand_omp_for_generic): Properly add loops.
18839         (expand_omp_for_static_nochunk): Likewise.
18840         (expand_omp_for_static_chunk): Likewise.
18841         (expand_omp_for): For the degenerate case fixup loops.
18842         (expand_omp_sections): Fix default bb placement in loops.
18843         (expand_omp_atomic_pipeline): Properly add loops.
18845 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18847         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18849 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18851         * tree-ssa-tail-merge.c: Update header comment.
18853 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18855         * config/aarch64/arm_neon.h
18856         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18857         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18858         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18859         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18860         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18861         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18862         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18863         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18865 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18867         * config/aarch64/aarch64-simd.md
18868         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18869         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18870         fix_trunc, fixuns_trunc.
18871         (ftrunc<VDQF:mode>2): New.
18872         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18873         (fix_trunc_optab): New.
18875 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18877         * config/aarch64/aarch64-builtins.c
18878         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18879         iceilf, lround, iroundf.
18881 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18883         PR target/54349
18884         * config/i386/i386.h (enum ix86_tune_indices)
18885         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18886         New, split from X86_TUNE_INTER_UNIT_MOVES.
18887         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18888         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18889         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18890         (TARGET_INTER_UNIT_MOVES): Remove.
18891         * config/i386/i386.c (initial_ix86_tune_features): Update.
18892         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18893         (ix86_expand_convert_uns_didf_sse): Use
18894         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18895         (ix86_expand_vector_init_one_nonzero): Ditto.
18896         (ix86_expand_vector_init_interleave): Ditto.
18897         (inline_secondary_memory_needed): Return true for moves from SSE class
18898         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18899         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18900         * config/i386/constraints.md (Yi, Ym): Depend on
18901         TARGET_INTER_UNIT_MOVES_TO_VEC.
18902         (Yj, Yn): New constraints.
18903         * config/i386/i386.md (*movdi_internal): Change constraints of
18904         operand 1 from Yi to Yj and from Ym to Yn.
18905         (*movsi_internal): Ditto.
18906         (*movdf_internal): Ditto.
18907         (*movsf_internal): Ditto.
18908         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18909         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18910         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18911         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18912         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18913         * config/i386/sse.md (movdi_to_sse): Ditto.
18914         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18915         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18916         TARGET_INTER_UNIT_MOVES.
18917         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18918         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18919         instead of TARGET_INTER_UNIT_MOVES.
18920         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18921         operand 1 from Yi to Yj and from Ym to Yn.
18923 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18925         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18926         (float_truncate_hi_): Likewise.
18927         (float_extend_lo_): Likewise.
18928         (float_truncate_lo_): Likewise.
18929         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18930         (aarch64_float_extend_lo_v2df): Likewise.
18931         (vec_unpacks_hi_v4sf): Likewise.
18932         (aarch64_float_truncate_lo_v2sf): Likewise.
18933         (aarch64_float_truncate_hi_v4sf): Likewise.
18934         (vec_pack_trunc_v2df): Likewise.
18935         (vec_pack_trunc_df): Likewise.
18937 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18939         * config/aarch64/aarch64-builtins.c
18940         (aarch64_fold_builtin): Fold float conversions.
18941         * config/aarch64/aarch64-simd-builtins.def
18942         (floatv2si, floatv4si, floatv2di): New.
18943         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18944         * config/aarch64/aarch64-simd.md
18945         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18946         * config/aarch64/iterators.md (FLOATUORS): New.
18947         (optab): Add float, floatuns.
18948         (su_optab): Likewise.
18950 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18952         * config/aarch64/aarch64-builtins.c
18953         (aarch64_builtin_vectorized_function): Use new names for
18954         fcvt builtins.
18955         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18956         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18957         (fcvtzu): Split as...
18958         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18959         (fcvtas): Split as...
18960         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18961         (fcvtau): Split as...
18962         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18963         (fcvtps): Split as...
18964         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18965         (fcvtpu): Split as...
18966         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18967         (fcvtms): Split as...
18968         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18969         (fcvtmu): Split as...
18970         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18971         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18972         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18973         (lfrintnusf, lfrintnudf): Likewise.
18974         * config/aarch64/aarch64-simd.md
18975         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18976         define_insn.
18977         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18978         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18979         (fcvt_pattern): Likewise.
18981 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18983         * config/aarch64/aarch64-simd.md
18984         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18985         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18987 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18989         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18990         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18991         (vrnd<a,m,n,p>_f32): Implement using builtins.
18992         (vrnd<i,x><q>_f<32, 64>): New.
18994 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18996         * config/aarch64/aarch64-builtins.c
18997         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18998         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18999         (frintz): Rename to...
19000         (btrunc): ...this.
19001         (frintp): Rename to...
19002         (ceil): ...this.
19003         (frintm): Rename to...
19004         (floor): ...this.
19005         (frinti): Rename to...
19006         (nearbyint): ...this.
19007         (frintx): Rename to...
19008         (rint): ...this.
19009         (frinta): Rename to...
19010         (round): ...this.
19011         * config/aarch64/aarch64-simd.md
19012         (aarch64_frint<frint_suffix><mode>): Delete.
19013         (<frint_pattern><mode>2): Convert to insn.
19014         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
19015         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
19016         (frint_pattern): Likewise.
19017         (frint_suffix): Likewise.
19019 2013-04-29  Richard Biener  <rguenther@suse.de>
19021         PR tree-optimization/57081
19022         * loop-init.c: Include tree-flow.h.
19023         (loop_optimizer_finalize): Free number of iteration estimates.
19024         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
19026 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
19028         PR tree-optimization/57083
19029         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
19030         non-singleton shift count range, zero extend low_bound for uns case.
19032         * config/i386/predicates.md (general_vector_operand): New predicate.
19033         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
19034         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
19035         if they aren't nonimmediate operands.  If their original values
19036         satisfy const_vector_equal_evenodd_p, don't shift them.
19037         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
19038         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
19039         if not nonimmediate_operand.
19040         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
19041         instead of register_operand.
19042         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
19044 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
19046         * stor-layout.c (finalize_size_functions): Allocate a structure and
19047         reset cfun before dumping the functions.
19049 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
19051         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
19053         PR target/56866
19054         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
19055         use xop_pmacsdqh if uns_p.
19056         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
19057         the immediate rotate count.
19059 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
19061         * rtl.h (struct rtx_def): Add comment for field jump.
19062         (LRA_SUBREG_P): New macro.
19063         * recog.c (register_operand): Check LRA_SUBREG_P.
19064         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19065         stack frame.
19066         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19067         (lra_final_code_change): Skip subreg change for operators.
19068         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19069         if there are no operand changes.
19070         * lra-constraints.c (curr_insn_set): New.
19071         (match_reload): Set LRA_SUBREG_P.
19072         (emit_spill_move): Ditto.
19073         (check_and_process_move): Use curr_insn_set. Process only single
19074         set insns.  Don't initialize sec_mem_p and change_p.
19075         (simplify_operand_subreg): Use LRA_SUBREG_P.
19076         (reg_in_class_p): New function.
19077         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19078         of #ifdef.  Add code to remove cycling.
19079         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19080         non-null disp.  Reload inner instead of disp when base and index
19081         are null.  Try to put lo_sum into register.
19082         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19083         (check_and_process_move): Move code for move cost check to
19084         simple_move_p.  Remove equiv_substitution.
19085         (simple_move_p): New function.
19086         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19087         curr_insn_set.  Call check_and_process_move only for single set
19088         insns.  Use the new function.  Move call of check_and_process_move
19089         after operand equiv substitution and address process.
19091 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19093         PR go/57045
19094         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19095         with nonlocal goto receivers or returns twice calls, ignore
19096         unininitialized values from abnormal edges to nl goto receiver
19097         or returns twice call.
19099 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19101         PR tree-optimization/57051
19102         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19103         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19104         bitsize.
19106 2013-04-26  Richard Biener  <rguenther@suse.de>
19108         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19109         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19110         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19111         (fixup_loop_arrays_after_move): New function.
19112         (move_sese_region_to_fn): Properly outline the loop tree parts
19113         of the SESE region.
19115 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19117         * config/i386/i386.md (type, unit): Fix long lines.
19119 2013-04-26  Richard Biener  <rguenther@suse.de>
19121         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19122         (lto-streamer-out.o): Likewise.
19123         * cfgloop.c (init_loops_structure): Export, add struct function
19124         argument and adjust.
19125         (flow_loops_find): Adjust.
19126         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19127         (init_loops_structure): Declare.
19128         * lto-streamer-in.c: Include cfgloop.h.
19129         (input_cfg): Input the loop tree.
19130         * lto-streamer-out.c: Include cfgloop.h.
19131         (output_cfg): Output the loop tree.
19132         (output_struct_function_base): Do not drop PROP_loops.
19134 2013-03-26  Richard Biener  <rguenther@suse.de>
19136         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19137         (pass_build_cfg): Provide PROP_loops.
19138         (move_sese_region_to_fn): Remove loops that are outlined into fn
19139         for now.
19140         * tree-inline.c: Include cfgloop.h.
19141         (initialize_cfun): Do not drop PROP_loops.
19142         (copy_loops): New function.
19143         (copy_cfg_body): Copy loop structure.
19144         (tree_function_versioning): Initialize destination loop tree.
19145         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19146         (pass_parallelize_loops): Do IL verification.
19147         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19148         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19149         the CFG make sure we fixup loops as well.
19150         * tree-ssa-tail-merge.c: Include cfgloop.h.
19151         (replace_block_by): When merging loop latches mark loops for fixup.
19152         * lto-streamer-out.c (output_struct_function_base): Drop
19153         PROP_loops for now.
19154         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19155         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19156         * ipa-split.c: Include cfgloop.h.
19157         (split_function): Add the new return block to the loop tree root.
19158         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19159         whether we have removed the forwarder block.
19160         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19161         * cfgloop.h (place_new_loop): Declare.
19162         * cfgloopmanip.c (place_new_loop): Export.
19163         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19164         (tree-switch-conversion.o): Likewise.
19165         (tree-complex.o): Likewise.
19166         (tree-inline.o): Likewise.
19167         (tree-ssa-tailmerge.o): Likewise.
19168         (ipa-split.o): Likewise.
19169         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19170         (tree-ssa-copy.o): Likewise.
19171         * tree-switch-conversion.c: Include cfgloop.h
19172         (process_switch): If we emit a bit-test cascade, schedule loops
19173         for fixup.
19174         * tree-complex.c: Include cfgloop.h.
19175         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19176         * asan.c: Include cfgloop.h.
19177         (create_cond_insert_point): Properly add new basic-blocks to
19178         loops, schedule loop fixup.
19179         * cfgloop.c (verify_loop_structure): Check that looks are not
19180         marked for fixup.
19181         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19182         to loops.
19183         (expand_omp_for_generic): Likewise.
19184         (expand_omp_sections): Likewise.
19185         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19186         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19187         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19188         is initialized, not when loops are present.
19189         * tree-parloops.c (parallelize_loops): Remove checking here.
19190         * passes.c (init_optimization_passes): Schedule a copy-propagation
19191         pass before complete unrolling of inner loops.
19193 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19195         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19196         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19197         (colorize_init): Add argument to _WIN32 version.
19198         * toplev.c: Include diagnostic-color.h.
19199         (process_options): Default to -fdiagnostics-color=auto if
19200         GCC_COLORS env var is in the environment.
19201         * common.opt (fdiagnostics-color=): Add Var and Init.
19202         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19203         env var is in the environment, the default is auto rather than never.
19205         * diagnostic.h (file_name_as_prefix): Add context argument.
19206         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19207         the string as locus.
19208         * langhooks.c (lhd_print_error_function): Adjust caller.
19210 2013-04-25  Lawrence Crowl  <crowl@google.com>
19212         * var-tracking.c (shared_hash_def::htab):
19213         Change type to hash_table.  Update dependent calls and types.
19215 2013-04-25  Lawrence Crowl  <crowl@google.com>
19217         * Makefile.in: Update as needed below.
19219         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19220         Move declaration to after the type's method definitons.
19222         * attribs.c (htab_t scoped_attributes::attribute_hash):
19223         Change type to hash_table.  Update dependent calls and types.
19225         * bitmap.c (htab_t bitmap_desc_hash):
19226         Change type to hash_table.  Update dependent calls and types.
19228         * cselib.c (htab_t cselib_hash_table):
19229         Change type to hash_table.  Update dependent calls and types.
19231         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19232         (hash_string_slot_node): Move implementation into lto-streamer.h
19233         struct string_slot_hasher.
19234         (eq_string_slot_node): Likewise.
19236         * data-streamer-out.c: Update output_block::string_hash_table
19237         dependent calls and types.
19239         * dwarf2cfi.c (htab_t trace_index):
19240         Change type to hash_table.  Update dependent calls and types.
19242         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19243         Change type to hash_table.  Update dependent calls and types.
19244         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19245         (htab_t optimize_external_refs::map): Likewise.
19246         (htab_t output_comp_unit::extern_map): Likewise.
19247         (htab_t output_comdat_type_unit::extern_map): Likewise.
19248         (htab_t output_macinfo::macinfo_htab): Likewise.
19249         (htab_t optimize_location_lists::htab): Likewise.
19250         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19252         * except.c (htab_t ehspec_hash_type):
19253         Change type to hash_table.  Update dependent calls and types.
19254         (assign_filter_values::ttypes): Likewise.
19255         (assign_filter_values::ehspec): Likewise.
19256         (sjlj_assign_call_site_values::ar_hash): Likewise.
19257         (convert_to_eh_region_ranges::ar_hash): Likewise.
19259         * gcse.c (htab_t pre_ldst_table):
19260         Change type to hash_table.  Update dependent calls and types.
19262         * ggc-common.c (htab_t saving_htab):
19263         Change type to hash_table.  Update dependent calls and types.
19264         (htab_t loc_hash): Likewise.
19265         (htab_t ptr_hash): Likewise.
19266         (call_count): Rename ggc_call_count.
19267         (call_alloc): Rename ggc_call_alloc.
19268         (loc_descriptor): Rename make_loc_descriptor.
19269         (add_statistics): Rename ggc_add_statistics.
19271         * ggc-common.c (saving_htab):
19272         Change type to hash_table.  Update dependent calls and types.
19274         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19275         (push_gimplify_context): Likewise.
19276         (pop_gimplify_context): Likewise.
19277         (struct gimple_temp_hash_elt): Added.
19278         (struct gimplify_hasher): Likewise.
19279         (struct gimplify_ctx.temp_htab):
19280         Change type to hash_table.  Update dependent calls and types.
19282         * gimple-fold.c: Include gimplify-ctx.h.
19284         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19285         Change type to hash_table.  Update dependent calls and types.
19286         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19287         avoid potential global name collision.
19289         * gimplify.c: Include gimplify-ctx.h.
19290         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19291         (htab_t gimplify_ctx::temp_htab):
19292         Update dependent calls and types for new type hash_table.
19293         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19294         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19296         * gimplify-ctx.h: New.
19297         (struct gimple_temp_hash_elt): Move from gimplify.c.
19298         (class gimplify_hasher): New.
19299         (struct gimplify_ctx): Move from gimple.h.
19300         (htab_t gimplify_ctx::temp_htab):
19301         Change type to hash_table.  Update dependent calls and types.
19303         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19304         (htab_t ivs_params::newivs_index):
19305         Change type to hash_table.  Update dependent calls and types.
19306         (htab_t ivs_params::params_index): Likewise.
19307         (htab_t print_generated_program::params_index): Likewise.
19308         (htab_t gloog::newivs_index): Likewise.
19309         (htab_t gloog::params_index): Likewise.
19311         * graphite.c: Include graphite-htab.h.
19312         4htab_t graphite_transform_loops::bb_pbb_mapping):
19313         Change type to hash_table.  Update dependent calls and types.
19315         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19316         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19317         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19319         * graphite-dependences.c: Include graphite-htab.h.
19320         (loop_is_parallel_p): Change hash table type of parameter.
19322         * graphite-htab.h: New.
19323         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19324         (extern find_pbb_via_hash): Move from graphite-poly.h.
19325         (extern loop_is_parallel_p): Move from graphite-poly.h.
19326         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19328         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19329         (extern loop_is_parallel_p): Move to graphite-htab.h.
19330         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19332         * haifa-sched.c (htab_t delay_htab):
19333         Change type to hash_table.  Update dependent calls and types.
19334         (htab_t delay_htab_i2): Likewise.
19336         * ira-color.c (htab_t allocno_hard_regs_htab):
19337         Change type to hash_table.  Update dependent calls and types.
19339         * ira-costs.c (htab_t cost_classes_htab):
19340         Change type to hash_table.  Update dependent calls and types.
19342         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19343         Change type to hash_table.  Update dependent calls and types.
19345         * loop-iv.c (htab_t bivs):
19346         Change type to hash_table.  Update dependent calls and types.
19348         * loop-unroll.c (htab_t opt_info::insns_to_split):
19349         Change type to hash_table.  Update dependent calls and types.
19350         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19352         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19353         (struct string_slot_hasher): New.
19354         (htab_t output_block::string_hash_table):
19355         Change type to hash_table.  Update dependent calls and types.
19357         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19358         (htab_t file_name_hash_table):
19359         Change type to hash_table.  Update dependent calls and types.
19361         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19362         calls and types.
19364         * lto-streamer.c (htab_t tree_htab):
19365         Change type to hash_table.  Update dependent calls and types.
19367         * omp-low.c: Include gimplify-ctx.h.
19369         * passes.c (htab_t name_to_pass_map):
19370         Change type to hash_table.  Update dependent calls and types.
19371         (pass_traverse): Rename to passes_pass_traverse.
19373         * plugin.c (htab_t event_tab):
19374         Change type to hash_table.  Update dependent calls and types.
19376         * postreload-gcse.c (htab_t expr_table):
19377         Change type to hash_table.  Update dependent calls and types.
19378         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19380         * sese.c (debug_rename_map_1): Make extern.
19381         (htab_t copy_bb_and_scalar_dependences::rename_map):
19382         Change type to hash_table.  Update dependent calls and types.
19384         * sese.h (extern debug_rename_map): Move to .c file.
19386         * store-motion.c (htab_t store_motion_mems_table):
19387         Change type to hash_table.  Update dependent calls and types.
19389         * trans-mem.c (htab_t tm_new_mem_hash):
19390         Change type to hash_table.  Update dependent calls and types.
19392         * tree-browser.c (htab_t TB_up_ht):
19393         Change type to hash_table.  Update dependent calls and types.
19395         * tree-cfg.c (htab_t discriminator_per_locus):
19396         Change type to hash_table.  Update dependent calls and types.
19398         * tree-complex.c: Include tree-hasher.h
19399         (htab_t complex_variable_components):
19400         Change type to hash_table.  Update dependent calls and types.
19402         * tree-eh.c (htab_t finally_tree):
19403         Change type to hash_table.  Update dependent calls and types.
19405         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19406         struct int_tree_hasher.
19407         (extern int_tree_map_eq): Likewise.
19408         (uid_decl_map_hash): Removed.
19409         (extern decl_tree_map_eq): Likewise.
19411         * tree-hasher.h: New.
19412         (struct int_tree_hasher): New.
19413         (typedef int_tree_htab_type): New.
19415         * tree-inline.c: Include gimplify-ctx.h.
19417         * tree-mudflap.c: Include gimplify-ctx.h.
19419         * tree-parloops.c: Include tree-hasher.h.
19420         (htab_t eliminate_local_variables_stmt::decl_address):
19421         Change type to hash_table.  Update dependent calls and types.
19422         (htab_t separate_decls_in_region::decl_copies): Likewise.
19424         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19425         Change type to hash_table.  Update dependent calls and types.
19427         * tree-sra.c (candidates):
19428         Change type to hash_table.  Update dependent calls and types.
19430         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19431         in tree-flow.h.
19432         (int_tree_map_hash): Likewise.
19434         * tree-ssa-dom.c (htab_t avail_exprs):
19435         Change type to hash_table.  Update dependent calls and types.
19437         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19438         Change type to hash_table.  Update dependent calls and types.
19440         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19441         Change type to hash_table.  Update dependent calls and types.
19443         * tree-ssa-phiopt.c (seen_ssa_names):
19444         Change type to hash_table.  Update dependent calls and types.
19446         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19447         Change type to hash_table.  Update dependent calls and types.
19449         * tree-ssa-uncprop.c (equiv):
19450         Change type to hash_table.  Update dependent calls and types.
19452 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19454         PR rtl-optimization/57003
19455         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19456         call note_stores with kill_clobbered_value callback again after
19457         killing regs_invalidated_by_call.
19459 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19461         * config/aarch64/aarch64-simd.md
19462         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19463         (aarch64_simd_bsl<mode>): Likewise.
19464         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19466 2013-04-25  Marek Polacek  <polacek@redhat.com>
19468         PR tree-optimization/57066
19469         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19471 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19473         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19475 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19477         * config/aarch64/aarch64-builtins.c
19478         (aarch64_fold_builtin): New.
19479         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19480         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19481         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19482         * config/aarch64/arm_neon.h
19483         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19485 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19486             Tejas Belagod  <tejas.belagod@arm.com>
19488         * config/aarch64/aarch64-builtins.c
19489         (aarch64_gimple_fold_builtin): New.
19490         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19491         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19492         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19493         (addvv4sf): Update.
19494         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19496 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19498         * config/aarch64/aarch64.md
19499         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19501 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19503         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19504         (*ngcsi_uxtw): New pattern.
19506 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19507             Julian Brown  <julian@codesourcery.com>
19509         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19510         (TB_DREG): Add T_V4HF.
19511         (v4hf_UP): New macro.
19512         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19513         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19514         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19515         built-ins.
19516         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19517         (arm_vector_mode_supported_p): Handle V4HF.
19518         (arm_mangle_map): Handle V4HFmode.
19519         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19520         * config/arm/arm_neon_builtins.def: Add entries for
19521         vcvtv4hfv4sf, vcvtv4sfv4hf.
19522         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19523         (neon_vcvtv4hfv4sf): Likewise.
19524         * config/arm/neon-gen.ml: Handle half-precision floating point
19525         features.
19526         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19527         * config/arm/arm_neon.h: Regenerate.
19528         * config/arm/neon.ml (type elts): Add F16.
19529         (type vectype): Add T_float16x4, T_floatHF.
19530         (type vecmode): Add V4HF.
19531         (type features): Add Requires_FP_bit feature.
19532         (elt_width): Handle F16.
19533         (elt_class): Likewise.
19534         (elt_of_class_width): Likewise.
19535         (mode_of_elt): Refactor.
19536         (type_for_elt): Handle F16, fix error messages.
19537         (vectype_size): Handle T_float16x4.
19538         (vcvt_sh): New function.
19539         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19540         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19541         (string_of_mode): Handle V4HF.
19542         * doc/arm-neon-intrinsics.texi: Regenerate.
19544 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19546         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19547         format specifier in 'X' case.
19549 2013-04-25  Alan Modra  <amodra@gmail.com>
19551         PR target/57052
19552         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19553         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19554         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19555         Repeat for many other rotate/shift and mask patterns using subregs.
19556         Name lshiftrt insns.
19557         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19558         on WORDS_BIG_ENDIAN.
19560 2013-04-25  Alan Modra  <amodra@gmail.com>
19562         * config.gcc: Support little-endian powerpc-linux targets.
19563         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19564         (LINK_OS_LINUX_SPEC): Define.
19565         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19566         Preserve MASK_LITTLE_ENDIAN.
19567         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19568         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19569         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19570         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19571         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19572         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19573         Correct fp word order for little-endian.  Don't shift toc entries
19574         smaller than a word for little-endian.
19575         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19576         (bswapdi2 splits): Correct low-part subreg for little-endian.
19577         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19578         low/high where such is correct only for be.
19579         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19580         little-endian for -mcall-aixdesc.
19582 2013-04-25  Alan Modra  <amodra@gmail.com>
19584         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19585         replace_equiv_address_nv.
19587 2013-04-25  Alan Modra  <amodra@gmail.com>
19589         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19591 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19593         Revert:
19594         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19595         * rtl.h (struct rtx_def): ...
19597 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19599         PR rtl-optimizations/57046
19600         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19601         for multi-reg splits.
19603 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19605         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19607 2013-04-24  Sterling Augustine  <saugustine@google.com>
19609         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19610         (comp_dir_string, debug_str_dwo_section): New.
19611         (DEBUG_STR_DWO_SECTION): Rename to ...
19612         (DEBUG_DWO_STR_SECTION): ... this.
19613         (DEBUG_NORM_STR_SECTION): Delete.
19614         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19615         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19616         (find_AT_string): Move most logic to ...
19617         (find_AT_string_in_table): ... here.  New.
19618         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19619         add_skeleton_AT_string.  Delete logic.
19620         (output_skeleton_debug_sections): Remove call to
19621         add_top_level_skeleton_die_attrs.
19622         (add_comp_dir_attribute): Move logic to comp_dir_string.
19623         (dwarf2out_init): Initialize debug_str_dwo_section.
19624         (output_indirect_string): Call find_string_form.
19625         (output_indirect_strings): Rewrite.
19626         (prune_unused_types): Empty skeleton_debug_str_hash.
19627         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19628         (dwarf2out_finish):  Call output_indirect_strings.
19630 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19632         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19634 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19636         * rtl.h (struct rtx_def): Add comment for field jump.
19637         (LRA_SUBREG_P): New macro.
19638         * recog.c (register_operand): Check LRA_SUBREG_P.
19639         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19640         stack frame.
19641         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19642         (lra_final_code_change): Skip subreg change for operators.
19643         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19644         if there are no operand changes.
19645         * lra-constraints.c (curr_insn_set): New.
19646         (match_reload): Set LRA_SUBREG_P.
19647         (emit_spill_move): Ditto.
19648         (check_and_process_move): Use curr_insn_set. Process only single
19649         set insns.  Don't initialize sec_mem_p and change_p.
19650         (simplify_operand_subreg): Use LRA_SUBREG_P.
19651         (reg_in_class_p): New function.
19652         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19653         of #ifdef.  Add code to remove cycling.
19654         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19655         non-null disp.  Reload inner instead of disp when base and index
19656         are null.  Try to put lo_sum into register.
19657         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19658         (check_and_process_move): Move code for move cost check to
19659         simple_move_p.  Remove equiv_substitution.
19660         (simple_move_p): New function.
19661         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19662         curr_insn_set.  Call check_and_process_move only for single set
19663         insns.  Use the new function.  Move call of check_and_process_move
19664         after operand equiv substitution and address process.
19666 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19668         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19669         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19670         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19672 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19674         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19676 2013-04-24  Marek Polacek  <polacek@redhat.com>
19678         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19679         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19680         (select_loops_exit_conditions): Likewise.
19681         (number_of_iterations_for_all_loops): Likewise.
19682         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19683         (scev_analysis): Likewise.
19685 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19686             Chao-ying Fu  <fu@mips.com>
19688         * config/mips/micromips.md (jraddiusp): New pattern.
19689         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19690         instruction if possible.
19692 2013-04-24  Alan Modra  <amodra@gmail.com>
19694         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19696 2013-04-24  Julian Brown  <julian@codesourcery.com>
19697             Chung-Lin Tang  <cltang@codesourcery.com>
19699         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19700         dependency behavior in enumeration type DIE generation. Add TODO note
19701         to comments about future DW_FORM_sdata/udata re-work of related code.
19703 2013-04-23  Lawrence Crowl  <crowl@google.com>
19705         * Makefile.in: Update as needed below.
19707         * hash-table.h (class hash_table):
19708         Correct many methods with parameter types compare_type to the correct
19709         value_type.  (Correct code was unlikely to notice the change.)
19710         (hash_table::elements_with_deleted) New.
19711         (class hashtable::iterator): New.
19712         (hashtable::begin()): New.
19713         (hashtable::end()): New.
19714         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19716         * statistics.c (statistics_hashes):
19717         Change type to hash_table.  Update dependent calls and types.
19719         * tree-into-ssa.c (var_infos):
19720         Change type to hash_table.  Update dependent calls and types.
19722         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19723         Change type to hash_table.  Update dependent calls and types.
19725         * tree-ssa-loop-im.c (struct mem_ref.refs):
19726         Change type to hash_table.  Update dependent calls and types.
19728         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19729         Change type to hash_table.  Update dependent calls and types.
19731         * tree-ssa-sccvn.c (vn_tables_s::nary):
19732         Change type to hash_table.  Update dependent calls and types.
19733         (vn_tables_s::phis): Likewise.
19734         (vn_tables_s::references): Likewise.
19736         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19737         (vn_reference_eq): Update parameter and return types.
19739         * tree-ssa-structalias.c (pointer_equiv_class_table):
19740         Change type to hash_table.  Update dependent calls and types.
19741         (location_equiv_class_table): Likewise.
19743         * tree-vect-data-refs.c: Consequential changes for making
19744         peeling a hash_table.
19746         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19747         (destroy_loop_vec_info): Dependent hash_table update.
19749         * tree-vectorizer.h (peeling_htab):
19750         Change type to hash_table.  Update dependent calls and types.
19752 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19754         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19755         to check the register content is equal or not.
19756         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19757         to assign register content record.
19758         * lra-eliminations.c (update_reg_eliminate): Use
19759         lra_update_reg_val_offset to update register content offset.
19760         * lra-int.h (struct lra_reg): Add offset member.
19761         (lra_reg_val_equal_p): New static inline function.
19762         (lra_update_reg_val_offset): New static inline function.
19763         (lra_assign_reg_val): New static inline function.
19764         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19765         to assign register content record.
19766         (initialize_lra_reg_info_element): Initial offset to zero.
19768 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19770         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19771         operands.  Record compression.
19773 2013-04-23  Xinliang David Li  <davidxl@google.com>
19775         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19777 2013-04-23  Richard Biener  <rguenther@suse.de>
19779         PR middle-end/57036
19780         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19781         parameter, only add abnormal goto edges from the copied body
19782         if the call could perform abnormal gotos.
19783         (copy_cfg_body): Adjust.
19785 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19787         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19789 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19791         * coretypes.h (gimple_stmt_iterator): Add struct to make
19792         compatible with C.
19794 2013-04-23  Richard Biener  <rguenther@suse.de>
19796         PR tree-optimization/57026
19797         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19798         from SSA names occuring in abnormal PHI nodes.
19800 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19802         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19804 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19806         * lto/lto.c (print_lto_report_1): Declare early.
19807         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19809 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19811         * common.opt (-flto-report-wpa): Add.
19812         * doc/invoke.texi (-flto-report-wpa): Add.
19813         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19814         (lto_main): dito.
19816 2013-04-22  Xinliang David Li  <davidxl@google.com>
19818         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19819         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19820         * Makefile.in: New dependency
19822             David Daney <ddaney.cavm@gmail.com>
19824         * configure.ac (gcc_cv_as_micromips_support):  Use the
19825         --fatal-warnings option.
19826         * configure: Regenerate.
19828 2013-04-22  Marek Polacek  <polacek@redhat.com>
19830         PR sanitizer/56990
19831         * tsan.c (instrument_expr): Don't instrument expression
19832         in case its size is zero.
19834 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19836         PR target/57032
19837         Revert:
19838         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19840         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19842 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19844         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19845         (gimple_stmt_iterator): New typedef.
19846         * gimple.h (gimple_stmt_iterator): Rename to...
19847         (gimple_stmt_iterator_d): ... This.
19848         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19849         trees be valid for GIMPLE and GENERIC.
19850         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19851         * gimple-fold.c (gimple_fold_call): Call target hook
19852         gimple_fold_builtin.
19853         * hooks.c (hook_bool_gsiptr_false): New.
19854         * hooks.h (hook_bool_gsiptr_false): New.
19855         * target.def (fold_stmt): New.
19856         * doc/tm.texi: Regenerate.
19858 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19860         PR target/57018
19861         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19862         a set sp if no stack realignment.
19864 2013-04-22  Nick Clifton  <nickc@redhat.com>
19866         * config.gcc (tilegx-linux): Extend extra_objs rather than
19867         overwriting it.
19868         (tilepro-linux): Likewise.
19870 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19872         * config/aarch64/aarch64-builtins.c
19873         (CF): Remove.
19874         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19875         (VAR<1-12>): Add MAP parameter.
19876         (BUILTIN_*): Likewise.
19877         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19878         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19879         (aarch64_ushl_n<mode>): Likewise.
19880         (aarch64_sshr_n<mode>): Likewise.
19881         (aarch64_ushr_n<mode>): Likewise.
19882         (aarch64_<maxmin><mode>): Likewise.
19883         (aarch64_sqrt<mode>): Likewise.
19884         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19885         (vshr<q>_n_*): Likewise.
19887 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19889         * config/aarch64/aarch64-builtins.c
19890         (aarch64_simd_builtin_type_mode): Handle SF types.
19891         (sf_UP): Define.
19892         (BUILTIN_GPF): Define.
19893         (aarch64_init_simd_builtins): Handle SF types.
19894         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19895         (frecps): Likewise.
19896         (frecpx): Likewise.
19897         * config/aarch64/aarch64-simd.md
19898         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19899         (aarch64_frecpe<mode>): New.
19900         (aarch64_frecps<mode>): Likewise.
19901         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19902         (v8type): Add frecp<esx>.
19903         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19904         (aarch64_frecps<mode>): Likewise.
19905         * config/aarch64/iterators.md (FRECP): New.
19906         (frecp_suffix): Likewise.
19907         * config/aarch64/arm_neon.h
19908         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19910 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19912         PR target/56995
19913         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19914         (REG_CLASS_NAMES): Idem.
19915         (REG_CLASS_CONTENTS): Idem.
19916         (REGCLASS_HAS_FP_REG): Idem.
19917         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19918         (sh_conditional_register_usage): Idem.
19920 2013-04-21  Jeff Law  <law@redhat.com>
19922         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19923         (ssa_forward_propagate_and_combine): Use it.
19925 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19927         * lra.c: Update the flow chart diagram.
19929 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19931         PR rtl-optimization/56847
19932         * lra-constraints.c (process_alt_operands): Discourage alternative
19933         with non-matche doffsettable memory constraint fro memory with
19934         known offset.
19936 2013-04-19  Richard Biener  <rguenther@suse.de>
19938         PR tree-optimization/56982
19939         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19940         function.
19941         * gimplify.c (gimplify_call_expr): Notice special calls.
19942         (gimplify_modify_expr): Likewise.
19943         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19944         abnormal control flow receivers.
19945         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19946         in the same way as cfun->has_nonlocal_labels.
19947         (gimple_purge_dead_abnormal_call_edges): Likewise.
19948         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19949         receivers start a basic-block.
19951 2013-04-19  Richard Biener  <rguenther@suse.de>
19953         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19954         member ...
19955         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19956         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19957         (SLP_TREE_LOAD_PERMUTATION): Add.
19958         (vect_transform_slp_perm_load): Adjust prototype.
19959         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19960         (vect_free_slp_instance): Likewise.
19961         (vect_create_new_slp_node): Likewise.
19962         (vect_supported_slp_permutation_p): Remove.
19963         (vect_slp_rearrange_stmts): Adjust.
19964         (vect_supported_load_permutation_p): Likewise.  Inline
19965         vect_supported_slp_permutation_p here.
19966         (vect_analyze_slp_instance): Compute load permutations per
19967         slp node instead of per instance.
19968         (vect_get_slp_defs): Adjust.
19969         (vect_transform_slp_perm_load): Likewise.
19970         (vect_schedule_slp_instance): Remove redundant code.
19971         (vect_schedule_slp): Remove hack for PR56270, add it ...
19972         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19973         CSE loads for SLP.  Adjust.
19975 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19977         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19978         spelling in two comments.
19980 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19982         PR target/56797
19983         * config/arm/arm.c (load_multiple_sequence): Require SP
19984         as base register for loads if SP is in the register list.
19986 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19988         PR tree-optimization/56718
19989         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19990         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19991         and made public.  Adjusted all callers.
19992         (ipa_intraprocedural_devirtualization): New function.
19993         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19994         (ipa_intraprocedural_devirtualization): Likewise.
19995         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19997 2013-04-19  Richard Biener  <rguenther@suse.de>
19999         PR tree-optimization/57000
20000         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
20002 2013-04-19  Terry Guo  <terry.guo@arm.com>
20004         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
20005         Replace with ...
20006         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
20007         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
20008         (cortex_m4_fmacs): Use new reservations.
20009         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
20011 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
20013         PR rtl-optimization/56999
20014         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
20015         related code.
20016         (lra_coalesce): Remove split_origin_bitmap and related code.
20017         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
20018         ranges if necessary.
20020 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
20022         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
20023         New array.
20024         (ix86_expand_call): Remove clobbered_registers array and use
20025         x86_64_ms_sysv_extra_clobbered_registers instead.
20026         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
20027         Declare here.
20028         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
20029         predicate.
20030         * config/i386/i386.md (*call_rex64_ms_sysv): Use
20031         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
20032         (*call_value_rex64_ms_sysv): Ditto.
20034 2013-04-18  Cary Coutant  <ccoutant@google.com>
20036         * dwarf2out.c (output_pubnames): Check die_perennial_p of
20037         parent instead of die_mark.
20039 2013-04-18  Diego Novillo  <dnovillo@google.com>
20041         * gimple.c (create_gimple_tmp): New.
20042         (get_expr_type): New.
20043         (build_assign): New.
20044         (build_type_cast): New.
20045         * gimple.h (enum ssa_mode): Define.
20046         (gimple_seq_set_location): New.
20047         * asan.c (build_check_stmt): Change some gimple_build_* calls
20048         to use build_assign and build_type_cast.
20050 2013-04-18  Richard Biener  <rguenther@suse.de>
20052         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
20053         handle negative step.  Remove redundant checks.
20054         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
20055         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
20056         for negative step and grouped loads fail to vectorize.
20058 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20060         * emit-rtl.c (reset_insn_used_flags): New function.
20061         (reset_all_used_flags): Use it.
20062         (verify_insn_sharing): New function.
20063         (verify_rtl_sharing): Fix verification for SEQUENCEs.
20065 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
20067         PR tree-optimization/56984
20068         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20069         and (x >> M) >= N don't register any assertion if N << M is the
20070         minimum value.
20072 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20074         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20075         do not delete_insn insns that have not yet been emitted, only
20076         unlink them with remove_insn.
20077         * df-scan.c (df_insn_delete): Revert r197492.
20079 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20081         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20082         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20084 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20086         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20087         define_insn_and_split.
20088         (and_scc,ior_scc,negscc): Likewise.
20089         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20091 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20093         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20094         can benefit from using a sequence of LDRD instructions in epilogue
20095         instead of a single LDM instruction.
20097 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20099         PR 45688
20100         * doc/extend.texi: Fix typo.
20102 2013-04-17  Richard Biener  <rguenther@suse.de>
20104         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20105         (vect_build_slp_tree): ... here.
20106         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20107         match.  Remove special-casing of mismatched complex loads.
20108         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20109         re-try the match with swapped commutative operands.
20110         (vect_supported_load_permutation_p): Remove special-casing of
20111         mismatched complex loads.
20112         (vect_analyze_slp_instance): Adjust.
20114 2013-04-17  Richard Biener  <rguenther@suse.de>
20116         PR rtl-optimization/56921
20117         * cfgloop.h (struct loop): Add simple_loop_desc member.
20118         (struct niter_desc): Mark with GTY(()).
20119         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20120         * loop-iv.c (get_simple_loop_desc): Likewise.
20121         (free_simple_loop_desc): Likewise.
20123         Revert
20124         2013-04-16  Richard Biener  <rguenther@suse.de>
20126         PR rtl-optimization/56921
20127         * loop-init.c (pass_rtl_move_loop_invariants): Add
20128         TODO_do_not_ggc_collect to todo_flags_finish.
20129         (pass_rtl_unswitch): Same.
20130         (pass_rtl_unroll_and_peel_loops): Same.
20131         (pass_rtl_doloop): Same.
20133 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20135         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20136         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20137         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20138         references.
20139         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20140         * tree-streamer.c (record_common_node): Adjust reference in comment.
20142 2013-04-17  Terry Guo  <terry.guo@arm.com>
20144         * config/arm/cortex-m4.md: Add a new bypass.
20146 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20148         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20149         New pattern.
20150         (*subs_<optab><mode>_multp2): New pattern.
20151         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20152         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20154 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20156         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20157         (*subs_mul_imm_<mode>): New pattern.
20159 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20161         PR target/56948
20162         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20163         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20164         (vsx_movti_32bit): Change j->wa to O->wa.
20166 2013-04-16  Richard Biener  <rguenther@suse.de>
20168         PR rtl-optimization/56921
20169         * loop-init.c (pass_rtl_move_loop_invariants): Add
20170         TODO_do_not_ggc_collect to todo_flags_finish.
20171         (pass_rtl_unswitch): Same.
20172         (pass_rtl_unroll_and_peel_loops): Same.
20173         (pass_rtl_doloop): Same.
20175 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20177         * config/arm/arm.c (emit_multi_reg_push): New declaration
20178         for an existing function.
20179         (arm_emit_strd_push): New function.
20180         (arm_expand_prologue): Used here.
20181         (arm_emit_ldrd_pop): New function.
20182         (arm_expand_epilogue): Used here.
20183         (arm_get_frame_offsets): Update condition.
20184         (arm_emit_multi_reg_pop): Add a special case for load of a single
20185         register with writeback.
20187 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20189         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20190         description.
20192 2013-04-16  Richard Biener  <rguenther@suse.de>
20194         PR tree-optimization/56756
20195         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20196         (first_mem_ref_loc): New.
20197         (execute_sm): Place the load temporarily before a previous
20198         access instead of in the latch edge to ensure its SSA dependencies
20199         are defined at points dominating the load.
20201 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20203         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20204         correct fix by moving header and footer insn to the footer of
20205         the merged basic block.  Clear BB_END of the merged-away block.
20207         PR middle-end/43631
20208         * emit-rtl.c (make_note_raw): New function.
20209         (link_insn_into_chain): New static inline function.
20210         (add_insn): Use it.
20211         (add_insn_before, add_insn_after): Factor insn chain linking code...
20212         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20213         using link_insn_into_chain.
20214         (note_outside_basic_block_p): New helper function for emit_note_after
20215         and emit_note_before.
20216         (emit_note_after): Use nobb variant of add_insn_after if the note
20217         should not be contained in a basic block.
20218         (emit_note_before): Use nobb variant of add_insn_before if the note
20219         should not be contained in a basic block.
20220         (emit_note_copy): Use make_note_raw.
20221         (emit_note): Likewise.
20222         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20223         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20224         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20225         the moved barrier the tail of the basic block it follows.
20226         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20228 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20230         PR tree-optimization/56962
20231         * gimple-ssa-strength-reduction.c (record_increment): Only set
20232         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20233         either rhs1 or rhs2 is equal to c->base_expr.
20235 2013-04-15  Richard Biener  <rguenther@suse.de>
20237         PR tree-optimization/56933
20238         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20239         member.
20240         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20241         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20242         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20243         dependence check ...
20244         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20245         ... here.
20246         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20247         GROUP_READ_WRITE_DEPENDENCE.
20249 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20251         * emit-rtl.c (reset_all_used_flags): New function.
20252         (verify_rtl_sharing): Call reset_all_used_flags before and after
20253         performing the checks.
20255 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20257         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20258         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20259         * config/arm/constraints.md (De): New constraint.
20260         * config/arm/neon.md (anddi3_neon): Delete.
20261         (neon_vand<mode>): Expand to standard anddi3 pattern.
20262         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20263         Move earlier in the file.
20264         (neon_inv_logic_op2): Likewise.
20265         (arm_anddi_operand_neon): New predicate.
20267 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20269         * configure.ac (gcc_cv_ld_as_needed): Set
20270         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20271         Use -z ignore, -z record on *-*-solaris2*.
20272         (HAVE_LD_AS_NEEDED): Update comment.
20273         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20274         * configure: Regenerate.
20275         * config.in: Regenerate.
20276         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20277         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20278         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20279         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20280         equivalents.  Fix markup.
20281         * doc/tm.texi: Regenerate.
20283 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20285         * config/i386/i386.opt: New option mstack-protector-guard=.
20286         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20287         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20288         TARGET_SSP_TLS_GUARD.
20289         * config/i386/i386.c (ix86_option_override_internal): Set
20290         ix86_stack_protector_guard.
20291         * config/i386/i386.md (stack_protect_set): Enable for
20292         TARGET_SSP_TLS_GUARD only.
20293         (stack_protect_set_<mode>): Ditto.
20294         (stack_protect_test): Ditto.
20295         (stack_protect_test_<mode>): Ditto.
20296         * doc/invoke.texi (i386 Option): Document.
20298 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20300         PR target/56890
20301         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20302         (S_MODES): Set H_MODE bit.
20303         (SF_MODES): Set only S_MODE and SF_MODE bits.
20304         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20305         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20306         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20307         <MODE_FLOAT>: Likewise.
20309 2013-04-15  Joey Ye  <joey.ye@arm.com>
20311         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20313 2013-04-15  Joey Ye  <joey.ye@arm.com>
20315         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20316         for real far jump.
20317         (thumb_far_jump_used_p): Count instruction size and set
20318         far_jump_used.
20320 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20322         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20323         * resource.c (mark_target_live_regs): Reformat conditional block.
20325 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20327         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20328         notes, they are emitted only just before final.
20329         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20331 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20333         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20334         * cfgrtl.c (delete_insn): Call it here instead.
20335         * lra-spills.c (lra_final_code_change): Use delete_insn.
20336         * haifa-sched.c (sched_remove_insn): Likewise.
20337         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20338         returning to the nop pool.
20339         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20340         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20342 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20344         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20345         * doc/tm.texi: Regenerated.
20347 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20349         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20350         QImode checks.
20352 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20354         * df-core.c (df_find_def): Compare register numbers.
20355         (df_find_use): Likewise.
20357 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20359         PR target/56903
20360         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20361         lra_in_progress for return.
20363 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20365         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20366         define_insn into define_insn_and_split and emit movsicc patterns.
20368 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20370         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20372 2013-04-12  Richard Biener  <rguenther@suse.de>
20374         * tree-pass.h (TODO_do_not_ggc_collect): New.
20375         * passes.c (execute_one_ipa_transform_pass): Honor
20376         TODO_do_not_ggc_collect.
20377         (execute_one_pass): Likewise.
20379         Revert
20380         2013-04-10  Richard Biener  <rguenther@suse.de>
20382         * passes.c (init_optimization_passes): Remove reload pass.
20383         * ira.c (do_reload): Merge into ...
20384         (ira): ... this.
20385         (rest_of_handle_reload): Remove.
20386         (pass_reload): Likewise.
20387         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20388         of reload for vzeroupper pass placement.
20390 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20392         PR tree-optimization/56918
20393         PR tree-optimization/56920
20394         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20395         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20396         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20397         use wide_mul_with_sign method.
20399 2013-04-12  Richard Biener  <rguenther@suse.de>
20401         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20402         not be considered a gimple constant.
20404 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20406         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20407         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20408         operations.
20410 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20411             Jakub Jelinek  <jakub@redhat.com>
20413         * opts.c: Include diagnostic-color.h.
20414         (common_handle_option): Handle OPT_fdiagnostics_color_.
20415         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20416         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20417         (diagnostic-color.o): New.
20418         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20419         (diagnostic_color_rule): New enum.
20420         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20421         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20422         the location string.
20423         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20424         either NULL, or color kind.
20425         * diagnostic-color.c: New file.
20426         * diagnostic-color.h: New file.
20427         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20428         arguments.
20429         * doc/invoke.texi (-fdiagnostics-color): Document.
20430         * pretty-print.h (pp_show_color): Define.
20431         (struct pretty_print_info): Add show_color field.
20432         * diagnostic.c: Include diagnostic-color.h.
20433         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20434         macros.  Colorize error:, warning: etc. strings and also the location
20435         string.
20436         (diagnostic_show_locus): Colorize the caret line.
20437         * pretty-print.c: Include diagnostic-color.h.
20438         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20439         inside of %< %> quotes or quoted through q format modifier.
20441 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20443         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20445 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20447         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20448         code in CC_NZ mode.
20449         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20450         pattern.
20452 2013-04-11  Marek Polacek  <polacek@redhat.com>
20454         PR tree-optimization/48184
20455         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20457 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20459         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20460         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20461         (skip_simple_arithmetic): Tidy up.
20462         * tree.h (skip_simple_constant_arithmetic): Declare.
20464 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20466         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20468 2013-04-11  Richard Biener  <rguenther@suse.de>
20470         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20471         generate vector constants.
20473 2013-04-11  Richard Biener  <rguenther@suse.de>
20475         PR tree-optimization/56878
20476         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20477         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20478         New function.
20479         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20480         Prefer to align the DR with the most invariant base address.
20482 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20484         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20485         comment.
20487 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20489         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20490         floating-point vector comparisons against 0.
20492 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20494         PR tree-optimization/56899
20495         * fold-const.c (extract_muldiv_1): Apply distributive law
20496         only if TYPE_OVERFLOW_WRAPS (ctype).
20498 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20500         PR target/56124
20501         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20502         loading has side effect.
20504 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20506         * config/sparc/sparc.c: Include tree-pass.h.
20507         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20508         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20509         head of file.  Change return type.  Split off gate function.
20510         (sparc_gate_work_around_errata): New function.
20511         (pass_work_around_errata): New pass definition.
20512         (insert_pass_work_around_errata) New pass insert definition to
20513         insert pass_work_around_errata just after delayed-branch scheduling.
20514         (sparc_option_override): Insert the pass.
20515         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20517 2013-04-10  David S. Miller  <davem@davemloft.net>
20519         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20520         or -mcpu=hypersparc.
20522         * target.def (cstore_mode): New hook.
20523         * target.h: Include insn-codes.h
20524         * targhooks.c: Likewise.
20525         (default_cstore_mode): New function.
20526         * targhooks.h: Declare it.
20527         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20528         * doc/tm.texi: Rebuild.
20529         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20530         target hook, rather than inspecting the insn_data.
20531         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20532         (TARGET_CSTORE_MODE): Redefine.
20533         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20534         result patterns.
20535         * config/sparc/predicates.md (cstore_result_operand): New special
20536         predicate.
20537         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20538         Use it for operand 0.
20539         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20540         (*snesi_special): Likewise.
20541         (*snesi_zero): Likewise.
20542         (*seqsi_zero): Likewise.
20543         (*sltu_insn): Likewise.
20544         (*sgeu_insn): Likewise.
20545         (*seqdi_special): Make operand 0 and comparison operation be of
20546         DImode.
20547         (*snedi_special): Likewise.
20548         (*snedi_special_vis3): Likewise.
20549         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20550         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20551         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20552         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20553         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20554         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20555         (*sltu_extend_sp64): Likewise.
20556         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20557         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20558         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20559         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20560         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20562 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20564         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20565         (aarch64_start_file): Use the new function.
20567 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20568             Jason Merrill  <jason@redhat.com>
20570         * common.opt: Add -gdwarf.
20571         * opts.c (common_handle_option): Handle it.
20572         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20574 2013-04-10  Richard Biener  <rguenther@suse.de>
20576         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20577         (execute_one_ipa_transform_pass): But unconditionally here.
20578         (execute_one_pass): And here.
20579         (init_optimization_passes): Remove reload pass.
20580         * tree-pass.h (TODO_ggc_collect): Remove.
20581         (pass_reload): Likewise.
20582         * ira.c (do_reload): Merge into ...
20583         (ira): ... this.
20584         (rest_of_handle_reload): Remove.
20585         (pass_reload): Likewise.
20586         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20587         of reload for vzeroupper pass placement.
20588         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20589         and todo_flags_finish of all passes.
20591 2013-04-10  Richard Biener  <rguenther@suse.de>
20593         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20594         first_const_oprnd field, rename first_def_type to first_op_type.
20595         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20596         (vect_get_and_check_slp_defs): Always use the type of the
20597         operand.  Allow mixed vect_external_def, vect_constant_def types.
20598         (vect_get_constant_vectors): Handle mixed vect_external_def,
20599         vect_constant_def types.
20601 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20603         PR tree-optimization/55524
20604         * tree-ssa-math-opts.c
20605         (convert_mult_to_fma): Don't use an fms construct
20606         when we don't have an fms operation, but fnma, and it looks
20607         likely that we'll be able to use the latter.
20609 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20611         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20612         function.
20613         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20614         inline fail caused by overwritable functions.
20616 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20618         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20619         unnecessary bits in the constant power of two case.
20621 2013-04-10  Richard Biener  <rguenther@suse.de>
20623         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20624         broken code swapping operands.
20625         (vect_build_slp_tree): Do not compute load permutations here.
20626         (vect_analyze_slp_instance): Compute load permutations here,
20627         after building the SLP tree.
20629 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20631         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20632         of next/prev_real_insn.
20634 2013-04-09  Jan Hubicka  <jh@suse.cz>
20636         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20637         Drop aliased parameter.
20638         (function_and_variable_visibility): Do not handle alias pairs.
20639         * cgraph.c (varpool_externally_visible_p): Update prototype.
20640         * varpool.c (varpool_add_new_variable): Update.
20642 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20644         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20646 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20648         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20650         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20652 2013-04-09  Marek Polacek  <polacek@redhat.com>
20654         PR tree-optimization/48762
20655         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20657 2013-04-09  Richard Biener  <rguenther@suse.de>
20659         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20660         dealing with cost.
20661         (vect_build_slp_tree): Likewise.
20662         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20663         calculating the cost of a SLP instance.
20664         (vect_analyze_slp_instance): Use it from here, after building
20665         the SLP tree.
20667 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20669         PR middle-end/56883
20670         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20671         expand_omp_for_static_chunk): Use simple_p = true in
20672         force_gimple_operand_gsi calls when assigning to addressable decls.
20674 2013-04-09  Jeff Law  <law@redhat.com>
20676         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20677         when the boolean was created by converting a wider object which
20678         had a boolean range.
20680 2013-04-09  Richard Biener  <rguenther@suse.de>
20682         * tree-vectorizer.h (slp_void_p): Remove.
20683         (slp_tree): Typedef before _slp_tree declaration.
20684         (struct _slp_tree): Use a vector of slp_tree as children.
20685         (vect_get_place_in_interleaving_chain): Remove.
20686         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20687         Move ...
20688         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20689         and make static.
20690         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20691         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20692         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20693         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20694         Use slp_node instead of slp_void_p and adjust.
20696 2013-04-09  Richard Biener  <rguenther@suse.de>
20698         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20699         work that is not necessary.
20701 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20703         PR tree-optimization/56854
20704         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20705         forward into clobber stmts if it would change MEM_REF lhs into
20706         non-MEM_REF.
20708 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20710         * tree.c (type_hash_lookup, type_hash_add): Make static.
20711         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20713 2013-04-09  Richard Biener  <rguenther@suse.de>
20715         * tree.h (unsave_expr_now): Remove.
20716         * tree-inline.c (mark_local_for_remap_r): Remove.
20717         (unsave_expr_1): Likewise.
20718         (unsave_r): Likewise.
20719         (unsave_expr_now): Likewise.
20720         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20721         (propagate_tree_value): Likewise.
20723 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20725         * doc/rtl.texi (sequence): Rewrite documentation to match the
20726         current use of SEQUENCE rtl objects.
20727         * rtl.def (SEQUENCE): Likewise.
20729         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20730         Update documentation.
20731         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20732         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20734         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20736 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20738         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20739         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20740         methods.
20741         (estimate_edge_size_and_time): Add comment to suggest using rounding
20742         methods.
20743         (estimate_node_size_and_time): Ditto.
20744         (remap_edge_change_prob): Use helper rounding divide methods.
20745         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20746         (gimple_mod_pow2_value_transform): Ditto.
20747         (gimple_mod_subtract_transform): Ditto.
20748         (gimple_ic_transform): Ditto.
20749         (gimple_stringops_transform): Ditto.
20750         * stmt.c (conditional_probability): Ditto.
20751         (emit_case_dispatch_table): Ditto.
20752         * lto-cgraph.c (merge_profile_summaries): Ditto.
20753         * tree-optimize.c (execute_fixup_cfg): Ditto.
20754         * cfgcleanup.c (try_forward_edges): Ditto.
20755         * cfgloopmanip.c (scale_loop_profile): Ditto.
20756         (loopify): Ditto.
20757         (duplicate_loop_to_header_edge): Ditto.
20758         (lv_adjust_loop_entry_edge): Ditto.
20759         * tree-vect-loop.c (vect_transform_loop): Ditto.
20760         * profile.c (compute_branch_probabilities): Ditto.
20761         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20762         * lto-streamer-in.c (input_cfg): Ditto.
20763         * gimple-streamer-in.c (input_bb): Ditto.
20764         * ipa-cp.c (update_profiling_info): Ditto.
20765         (update_specialized_profile): Ditto.
20766         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20767         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20768         rounding methods.
20769         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20770         (compute_trg_info): Ditto.
20771         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20772         (purge_dead_edges): Ditto.
20773         * loop-unswitch.c (unswitch_loop): Ditto.
20774         * cgraphclones.c (cgraph_clone_edge): Ditto.
20775         (cgraph_clone_node): Ditto.
20776         * tree-inline.c (copy_bb): Ditto.
20777         (copy_edges_for_bb): Ditto.
20778         (initialize_cfun): Ditto.
20779         (copy_cfg_body): Ditto.
20780         (expand_call_inline): Ditto.
20782 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20784         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20785         TARGET_CYGWIN64 by TARGET_64BIT.
20787 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20789         * config/epiphany/epiphany.md (GPR_1): New constant.
20790         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20791         * config/epiphany/epiphany.c (gen_compare_reg):
20792         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20793         is already in place.
20794         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20795         Don't require being called during rtl expansion; If y operlaps r0,
20796         return 0.
20797         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20798         (epiphany_expand_epilogue): Likewise.
20800         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20801         Don't use CC_FPmode for ORDERED / UNORDERED.
20802         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20804         * config/epiphany/constraints.md (CnL): New constraint.
20805         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20806         * config/epiphany/predicates.md (add_operand): Allow 1024.
20808         * config/epiphany/epiphany.md (logical_op): New code iterator.
20809         (op_mnc): New code attribute.
20810         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20811         (mov_f+1, mov_f+2): New peephole2 patterns.
20813         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20814         (cstoresi4): Also allow re-use of zero result when doing a NE
20815         comparison to a non-zero operand.
20816         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20818         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20819         Use gen_addsi3_i / gen_subsi3_i.
20821 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20823         PR c++/34949
20824         PR c++/50243
20825         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20826         contain anything but clobbers, at most one __builtin_stack_restore,
20827         optionally debug stmts and final resx, and if it has at least one
20828         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20829         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20830         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20831         which isn't defaut definition, remove them.
20832         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20833         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20834         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20835         with MEM_REF LHS with SSA_NAME address.
20837 2013-04-08  Jeff Law  <law@redhat.com>
20839         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20841 2013-04-08  Richard Biener  <rguenther@suse.de>
20843         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20844         extra newline.
20845         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20846         determined vector type.
20847         (vect_analyze_data_refs): Likewise.
20848         (vect_get_new_vect_var): Adjust.
20849         (vect_create_destination_var): Preserve SSA name versions.
20850         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20851         not dump anything here.
20853 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20855         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20856         Add member lr_slot_known.
20857         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20858         if necessary.
20859         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20860         Remove code that sets lr_slot_offset according to what a previous
20861         version of epiphany_emit_save_restore used to do.
20862         (epiphany_emit_save_restore): When doing an lr save or restore,
20863         set/verify lr_slot_known and lr_slot_offset.
20865 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20867         PR target/54338
20868         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20869         in ALL_REGS.
20871 2013-04-08  Richard Biener  <rguenther@suse.de>
20873         * alias.c (find_base_term): Fix thinko in previous change.
20875 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20877         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20878         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20879         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20880         if possible to compute val.
20881         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20882         For QImode integers don't require anything about precision.  Use
20883         const_with_all_bytes_same to find out if the constant doesn't have
20884         repeated bytes in it.
20886 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20888         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20889         within mode size.
20891 2013-04-08  Marek Polacek  <polacek@redhat.com>
20893         PR rtl-optimization/48182
20894         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20895         value to 1.
20897 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20899         PR target/55487
20900         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20901         nuses, make sure we have a label.
20903 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20905         PR target/56843
20906         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20907         (rs6000_emit_swdiv_low_precision): Remove.
20908         (rs6000_emit_swdiv): Rewrite to handle between one and four
20909         iterations of Newton-Raphson generally; modify required number of
20910         iterations for some cases.
20911         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20913 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20915         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20916         set-but-unused variable.
20918         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20919         basic blocks of released function bodies garbage-collectable.
20921         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20922         (struct rtl_opt_pass): Add TODO_df_finish.
20924         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20926 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20928         * config/arm/constraints.md (q): New constraint.
20929         * config/arm/ldrdstrd.md: New file.
20930         * config/arm/arm.md (ldrdstrd.md) New include.
20931         (arm_movdi): Use "q" instead of "r" constraint
20932         for double-word memory access.
20933         (movdf_soft_insn): Likewise.
20934         * config/arm/vfp.md (movdi_vfp): Likewise.
20935         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20936         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20937         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20938         (mem_ok_for_ldrd_strd): Likewise.
20939         (output_move_double): Update assertion.
20941 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20943         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20945 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20947         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20948         define_insn_and_split.
20949         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20951 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20953         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20954         define_insn_and_split.
20955         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20956         (shiftsi3_compare): New pattern.
20957         (rrx): New pattern.
20958         * config/arm/unspecs.md (UNSPEC_RRX): New.
20960 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20962         * config/arm/arm.md (negdi_extendsidi): New pattern.
20963         (negdi_zero_extendsidi): Likewise.
20965 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20967         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20968         define_insn_and_split.
20969         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20970         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20972 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20974         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20975         define_insn_and_split.
20976         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20977         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20979 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20981         * config/arm/arm.md (subsi3_carryin): New pattern.
20982         (subsi3_carryin_const): Likewise.
20983         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20984         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20986 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20988         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20990 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20992         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20993         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20995 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20997         * config/arm/arm.c (arm_expand_builtin): Change fcode
20998         type to unsigned int.
21000 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21002         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
21004 2013-04-04  Ian Lance Taylor  <iant@google.com>
21006         * doc/standards.texi (Standards): The Go frontend supports the Go 1
21007         language standard.
21009 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
21011         PR middle-end/56729
21012         * df-scan.c (df_insn_delete): Disable failing assert.
21014 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21016         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
21017         New function prototype.
21018         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
21019         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
21020         (arm_builtin_vectorized_function): New function.
21022 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21024         * config/arm/arm_neon_builtins.def: New file.
21025         * config/arm/arm.c (neon_builtin_data): Move contents to
21026         arm_neon_builtins.def.
21027         (enum arm_builtins): Include neon builtin definitions.
21028         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
21029         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
21031 2013-04-04  Marek Polacek  <polacek@redhat.com>
21033         PR tree-optimization/48186
21034         * predict.c (maybe_hot_frequency_p): Return false if
21035         HOT_BB_FREQUENCY_FRACTION is 0.
21036         (cgraph_maybe_hot_edge_p): Likewise.
21038 2013-04-04  Richard Biener  <rguenther@suse.de>
21040         PR tree-optimization/56826
21041         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
21042         more accurately.
21044 2013-04-04  Richard Biener  <rguenther@suse.de>
21046         PR tree-optimization/56213
21047         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
21048         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
21050 2013-04-04  Richard Biener  <rguenther@suse.de>
21052         PR tree-optimization/56837
21053         * tree-loop-distribution.c (classify_partition): For non-zero
21054         values require that the value has the same precision as its
21055         mode to be useful as memset value.
21057 2013-04-03  Nick Clifton  <nickc@redhat.com>
21059         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
21060         (fmssf4): Use fmsf.s on E3V5 architectures.
21061         (fnmasf4): Use fnmaf.s on E3V5 architectures.
21062         (fnmssf4): Use fnmsf.s on E3V5 architectures.
21064 2013-04-03  Jeff Law  <law@redhat.com>
21066         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
21067         (lra-eliminations.o): Likewise.
21069 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21071         * gcov-io.c (compute_working_sets): Moved most of body of old
21072         compute_working_sets here from profile.c.
21073         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21074         (gcov_working_set_t): Moved typedef here from basic-block.h
21075         (compute_working_set): Declare.
21076         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21077         (get_working_sets): Renamed from compute_working_set,
21078         replace most of body with call to new compute_working_sets.
21079         (get_exec_counts): Replace call to compute_working_sets
21080         to get_working_sets.
21081         * profile.h (get_working_sets): Renamed from compute_working_set.
21082         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21083         to get_working_sets.
21084         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21085         * gcov-dump.c (dump_working_sets): New function.
21087 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21089         * hwint.c (sext_hwi, zext_hwi): New functions.
21090         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21091         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21092         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21093         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21094         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21095         (sext_hwi, zext_hwi): New functions.
21097 2013-04-03  Jeff Law  <law@redhat.com>
21099         PR tree-optimization/56799
21100         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21101         back test for widening conversion erroneously dropped in prior change.
21103 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21105         PR target/56809
21106         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21107         instead of next_real_insn.
21109 2013-04-03  Marek Polacek  <polacek@redhat.com>
21111         PR sanitizer/55702
21112         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21114 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21116         PR target/56809
21117         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21118         next_real_insn.
21119         (thumb1_output_casesi): Likewise.
21120         (thumb2_output_casesi): Likewise.
21122 2013-04-03  Richard Biener  <rguenther@suse.de>
21124         PR tree-optimization/56817
21125         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21126         Split out ...
21127         (tree_unroll_loops_completely_1): ... new function to manually
21128         walk the loop tree, properly defering outer loops of unrolled
21129         loops to later iterations.
21131 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21133         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21134         (vectorizable_load): Likewise.
21135         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21136         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21138 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21140         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21141         BIT_FIELD_REF.
21143 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21145         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21147 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21149         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21151 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21153         PR tree-optimization/56790
21154         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21155         folding.
21157 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21159         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21160         Handle VEC_MERGE.
21161         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21162         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21163         equal arguments.
21165 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21167         PR c/19449
21168         * tree.h (force_folding_builtin_constant_p): New decl.
21169         * builtins.c (force_folding_builtin_constant_p): New variable.
21170         (fold_builtin_constant_p): Fold immediately also if
21171         force_folding_builtin_constant_p.
21173 2013-04-03  Richard Biener  <rguenther@suse.de>
21175         PR tree-optimization/56812
21176         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21177         DRs of the same interleaving chain are independent.
21179 2013-04-02  Jason Merrill  <jason@redhat.com>
21181         * gdbinit.in (pbb): Use debug fn.
21183 2013-04-02  Lawrence Crowl  <crowl@google.com>
21185         * sese.h (struct ivtype_map_elt_s): Remove unused.
21186         (extern debug_ivtype_map): Remove unused.
21187         (extern eq_ivtype_map_elts): Remove unused.
21188         * sese.c (debug_ivtype_map): Removed unused.
21189         (debug_ivtype_map_1): Removed unused.
21190         (debug_ivtype_elt): Remove unused.
21191         (eq_ivtype_map_elts): Remove unused.
21194 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21196         PR target/52790
21197         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21198         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21199         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21200         function.
21201         (legitimize_pe_coff_symbol): Likewise.
21202         (is_imported_p): New helper-function.
21203         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21204         for Windows x64 targets.
21205         (ix86_expand_prologue): Optimize for pe-coff targets.
21206         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21207         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21208         medium/large code-model.
21209         (legitimize_pic_address): Likewise.
21210         (legitimize_tls_address): Likewise.
21211         (ix86_expand_call): Likewise.
21212         (x86_output_mi_thunk): Likewise.
21213         (get_dllimport_decl): Add new beimport argument.
21214         (construct_plt_address): Don't assert for x64 pe-coff targets.
21215         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21216         targets.
21217         (SYMBOL_FLAG_STUBVAR): New macro.
21218         (SYMBOL_REF_STUBVAR_P): Likewise.
21219         * config/i386/winnt.c (stub_list): New structure.
21220         (stub_head): New local variable.
21221         (i386_pe_record_stub): New function.
21222         (i386_pe_file_end): Emit refptr-stubs.
21224 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21226         PR rtl-optimization/56745
21227         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21228         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21230         PR c++/34949
21231         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21232         and both of them are MEM_REFs, just compare first argument for
21233         equality and attempt to deal even with differing offsets.
21235         PR c++/34949
21236         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21237         of gimple_clobber_p to be MEM_REF.
21238         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21239         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21240         after gimplification.
21241         * asan.c (get_mem_ref_of_assignment): Don't instrument
21242         gimple_clobber_p stmts.
21243         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21244         gimple_clobber_p stmt if they have MEM_REF lhs and
21245         are dead because of another gimple_clobber_p stmt.
21246         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21247         gimple_clobber_p stmts like debug stmts.
21248         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21249         that refer to unused VAR_DECLs or uninitialized values.
21250         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21251         gimple_clobber_p stmts if they refer to removed parameters.
21252         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21253         formatting.
21255 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21257         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21258         using SWI48 mode attribute.
21260 2013-04-02  Wei Mi  <wmi@google.com>
21262         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21263         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21264         *<rotate_insn><mode>3_mask in i386.md.
21266 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21268         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21270 2013-04-02  Richard Biener  <rguenther@suse.de>
21272         PR tree-optimization/56778
21273         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21274         Runtime alias tests are not supported for gather loads.
21275         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21276         stmts referenced from SSA operands before updating SSA form.
21278 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21279             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21281         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21282         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21283         * config/arm/cortex-a53.md: New file.
21284         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21285         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21286         * config/arm/arm.c (arm_issue_rate): Likewise.
21287         * config/arm/arm-tune.md: Regenerate
21288         * config/arm/arm-tables.opt: Regenerate.
21289         * config/arm/arm-cores.def: Add cortex-a53.
21291 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21293         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21294         non-static link.
21296 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21298         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21299         scalar load/store operations using B/H registers.
21300         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21302 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21304         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21305         scalar move.
21306         * config/aarch64/aarch64.c
21307         (aarch64_simd_scalar_immediate_valid_for_move): New.
21308         * config/aarch64/aarch64-protos.h
21309         (aarch64_simd_scalar_immediate_valid_for_move): New.
21310         * config/aarch64/constraints.md (Dh, Dq): New.
21311         * config/aarch64/iterators.md (hq): New.
21313 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21315         * reorg.c (get_branch_condition): Deal with conditional returns.
21316         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21318 2013-04-01  Wei Mi  <wmi@google.com>
21320         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21321         Truncate operand 2 using %b asm operand modifier.
21322         (*<shift_insn><mode>3_mask): Ditto.
21323         (*<rotate_insn><mode>3_mask): Ditto.
21325 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21327         PR middle-end/56798
21328         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21330 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21332         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21333         of next_real_insn.
21334         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21336 2013-03-30  Lawrence Crowl  <crowl@google.com>
21338         * dse.c (clear_alias_sets): Remove never set.
21339         (disqualified_clear_alias_sets): Remove never set.
21340         (clear_alias_mode_pool): Remove never set.
21341         (dse_step0): Remove condition that is never true.
21342         (canon_address): Remove condition that is never true.
21343         (dse_step7): Remove condition that is never true.
21344         (rest_of_handle_dse): Remove condition that is never true.
21345         (rest_of_handle_dse::did_global): Remove never read from above.
21346         (dse_step2_spill): Remove never called from above.
21347         (dse_step5_spill): Remove never called from above.
21349 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21351         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21352         JUMP_TABLE_DATA changes.
21353         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21354         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21355         (Insns) <jump_table_data>: New entry.
21356         * doc/tm.texi: Regenerate.
21358         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21360         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21361         for table jump at the end of a basic block using tablejump_p.
21362         * targhooks.c (default_invalid_within_doloop): Likewise.
21363         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21364         target hook implementation that is identical to the default hook.
21365         (rs6000_invalid_within_doloop): Remove.
21367         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21368         unused variable from tablejump_p call.
21370         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21371         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21372         (INSN_DELETED_P): Likewise.
21373         (emit_jump_table_data): New prototype.
21374         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21375         after 4th as unused.
21376         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21377         * sched-vis.c (print_insn): Likewise.
21378         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21379         insn for compatibility with back ends that use next_active_insn to
21380         identify jump table data.
21381         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21382         (remove_insn): Likewise.
21383         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21384         to be emitted.
21385         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21386         (emit_jump_table_data): New function.
21388         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21389         basic block, a JUMP_TABLE_DATA never is.
21390         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21391         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21392         off from code handling real insns.
21393         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21394         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21395         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21396         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21397         is not a NONDEBUG_INSN_P.
21398         * ira-costs.c (scan_one_insn): Likewise.
21399         * jump.c (mark_all_labels): Likewise.
21400         (mark_jump_label_1): Likewise.
21401         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21402         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21403         a basic block.
21404         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21405         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21406         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21407         JUMP_TABLE_DATA_P insns.
21408         (calculate_elim_costs_all_insns): Likewise.
21409         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21410         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21411         (delete_output_reload): Code style fixups.
21412         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21413         insn flags on this non-insn.
21414         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21415         as scheduling barriers, for pre-change compatibility.
21416         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21417         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21419         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21420         redundant JUMP_TABLE_DATA_P test.
21421         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21422         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21423         (frv_for_each_packet): Likewise.
21424         * config/i386/i386.c (min_insn_size): Likewise.
21425         (ix86_avoid_jump_mispredicts): Likewise.
21426         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21427         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21428         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21429         (mips16_insn_length): Robustify.
21430         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21431         (mips16_split_long_branches): Likewise.
21432         * config/pa/pa.c (pa_combine_instructions): Likewise.
21433         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21434         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21435         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21436         as contributing to pool range lengths.
21437         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21438         Remove redundant JUMP_TABLE_DATA_P test.
21439         (sh_loop_align): Likewise.
21440         (split_branches): Likewise.
21441         (sh_insn_length_adjustment): Likewise.
21442         * config/spu/spu.c (get_branch_target): Likewise.
21444 2013-03-29  Jan Hubicka  <jh@suse.cz>
21446         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21447         gcov streaming; stream hot bb threshold to ltrans.
21448         * predict.c (get_hot_bb_threshold): Break out from ....
21449         (maybe_hot_count_p): ... here.
21450         (set_hot_bb_threshold): New function.
21451         * lto-section-in.c (lto_section_name): Add profile.
21452         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21453         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21454         and data-streamer.h
21455         (histogram_entry): New structure.
21456         (histogram, histogram_pool): New global vars.
21457         (histogram_hash): New structure.
21458         (histogram_hash::hash): New method.
21459         (histogram_hash::equal): Likewise.
21460         (account_time_size): New function.
21461         (cmp_counts): New function.
21462         (dump_histogram): New function.
21463         (ipa_profile_generate_summary): New function.
21464         (ipa_profile_write_summary): New function.
21465         (ipa_profile_read_summary): New function.
21466         (ipa_profile): Decide on threshold.
21467         (pass_ipa_profile): Add ipa_profile_write_summary and
21468         ipa_profile_read_summary.
21469         * Makefile.in (ipa.o): Update dependencies.
21470         * lto-streamer.h (LTO_section_ipa_profile): New section.
21472 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21474         * tree.h (VAR_P): New.
21476 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21478         PR lto/56777
21479         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21481 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21483         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21484         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21485         (control_flow_insn_p): Likewise.
21486         * cfgrtl.c (duplicate_insn_chain): Likewise.
21487         * final.c (get_attr_length_1): Likewise.
21488         (shorten_branches): Likewise.
21489         (final_scan_insn): Likewise.
21490         * function.c (instantiate_virtual_regs): Likewise.
21491         * gcse.c (insert_insn_end_basic_block): Likewise.
21492         * ira-costs.c (scan_one_insn): Likewise.
21493         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21494         * lra.c (check_rtl): Likewise.
21495         * reload1.c (elimination_costs_in_insn): Likewise.
21496         * reorg.c (follow_jumps): Likewise.
21498         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21499         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21500         (thumb_far_jump_used_p): Likewise.
21501         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21502         (workaround_speculation): Likewise.
21503         (add_sched_insns_for_speculation): Likewise.
21504         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21505         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21506         (frv_for_each_packet): Likewise.
21507         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21508         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21509         (final_emit_insn_group_barriers): Likewise.
21510         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21511         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21512         (mips16_insn_length): Likewise.
21513         * config/pa/pa.c (pa_reorg): Likewise.
21514         (pa_combine_instructions): Likewise.
21515         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21516         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21517         (sh_reorg): Likewise.
21518         (split_branches): Likewise.
21519         * config/spu/spu.c (get_branch_target): Likewise.
21521         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21522         JUMP_TABLE_DATA_P.
21524 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21526         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21527         Fix declaration name.
21529 2013-03-28  Lawrence Crowl  <crowl@google.com>
21531         * graphds.h (struct graph.indicies): Remove unused.
21532         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21533         (SCOP_ORIGINAL_PDDRS): Remove unused.
21534         * sese.h (extern insert_loop_close_phis): Removed unused.
21535         (extern insert_guard_phis): Removed unused.
21536         (extern ivtype_map_elt_info): Removed unused.
21537         (new_ivtype_map_elt): Removed unused.
21538         * sese.c (ivtype_map_elt_info): Removed unused.
21540 2013-03-28  Lawrence Crowl  <crowl@google.com>
21542         * Makefile.in: Add several missing include dependences.
21543         (DUMPFILE_H): New.
21544         (test-dump.o): New.  This object is not added to any executable,
21545         but is present for ad-hoc testing.
21546         * bitmap.c
21547         (debug (const bitmap_head_def &)): New.
21548         (debug (const bitmap_head_def *)): New.
21549         * bitmap.h
21550         (extern debug (const bitmap_head_def &)): New.
21551         (extern debug (const bitmap_head_def *)): New.
21552         * cfg.c
21553         (debug (edge_def &)): New.
21554         (debug (edge_def *)): New.
21555         * cfghooks.c
21556         (debug (basic_block_def &)): New.
21557         (debug (basic_block_def *)): New.
21558         * dumpfile.h
21559         (dump_node (const_tree, int, FILE *)): Correct source file.
21560         * dwarf2out.c
21561         (debug (die_struct &)): New.
21562         (debug (die_struct *)): New.
21563         * dwarf2out.h
21564         (extern debug (die_struct &)): New.
21565         (extern debug (die_struct *)): New.
21566         * gimple-pretty-print.c
21567         (debug (gimple_statement_d &)): New.
21568         (debug (gimple_statement_d *)): New.
21569         * gimple-pretty-print.h
21570         (extern debug (gimple_statement_d &)): New.
21571         (extern debug (gimple_statement_d *)): New.
21572         * ira-build.c
21573         (debug (ira_allocno_copy &)): New.
21574         (debug (ira_allocno_copy *)): New.
21575         (debug (ira_allocno &)): New.
21576         (debug (ira_allocno *)): New.
21577         * ira-int.h
21578         (extern debug (ira_allocno_copy &)): New.
21579         (extern debug (ira_allocno_copy *)): New.
21580         (extern debug (ira_allocno &)): New.
21581         (extern debug (ira_allocno *)): New.
21582         * ira-lives.c
21583         (debug (live_range &)): New.
21584         (debug (live_range *)): New.
21585         * lra-int.h
21586         (debug (lra_live_range &)): New.
21587         (debug (lra_live_range *)): New.
21588         * lra-lives.c
21589         (debug (lra_live_range &)): New.
21590         (debug (lra_live_range *)): New.
21591         * omega.c
21592         (debug (omega_pb_d &)): New.
21593         (debug (omega_pb_d *)): New.
21594         * omega.h
21595         (extern debug (omega_pb_d &)): New.
21596         (extern debug (omega_pb_d *)): New.
21597         * print-rtl.c
21598         (debug (const rtx_def &)): New.
21599         (debug (const rtx_def *)): New.
21600         * print-tree.c
21601         (debug_tree (tree): Move within file.
21602         (debug_raw (const tree_node &)): New.
21603         (debug_raw (const tree_node *)): New.
21604         (dump_tree_via_hooks (const tree_node *, int)): New.
21605         (debug (const tree_node &)): New.
21606         (debug (const tree_node *)): New.
21607         (debug_verbose (const tree_node &)): New.
21608         (debug_verbose (const tree_node *)): New.
21609         (debug_head (const tree_node &)): New.
21610         (debug_head (const tree_node *)): New.
21611         (debug_body (const tree_node &)): New.
21612         (debug_body (const tree_node *)): New.
21613         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21614         (debug (vec<tree, va_gc> &)): New.
21615         (debug (vec<tree, va_gc> *)): New.
21616         * rtl.h
21617         (extern debug (const rtx_def &)): New.
21618         (extern debug (const rtx_def *)): New.
21619         * sbitmap.c
21620         (debug_raw (simple_bitmap_def &)): New.
21621         (debug_raw (simple_bitmap_def *)): New.
21622         (debug (simple_bitmap_def &)): New.
21623         (debug (simple_bitmap_def *)): New.
21624         * sbitmap.h
21625         (extern debug (simple_bitmap_def &)): New.
21626         (extern debug (simple_bitmap_def *)): New.
21627         (extern debug_raw (simple_bitmap_def &)): New.
21628         (extern debug_raw (simple_bitmap_def *)): New.
21629         * sel-sched-dump.c
21630         (debug (vinsn_def &)): New.
21631         (debug (vinsn_def *)): New.
21632         (debug_verbose (vinsn_def &)): New.
21633         (debug_verbose (vinsn_def *)): New.
21634         (debug (expr_def &)): New.
21635         (debug (expr_def *)): New.
21636         (debug_verbose (expr_def &)): New.
21637         (debug_verbose (expr_def *)): New.
21638         (debug (vec<rtx> &)): New.
21639         (debug (vec<rtx> *)): New.
21640         * sel-sched-dump.h
21641         (extern debug (vinsn_def &)): New.
21642         (extern debug (vinsn_def *)): New.
21643         (extern debug_verbose (vinsn_def &)): New.
21644         (extern debug_verbose (vinsn_def *)): New.
21645         (extern debug (expr_def &)): New.
21646         (extern debug (expr_def *)): New.
21647         (extern debug_verbose (expr_def &)): New.
21648         (extern debug_verbose (expr_def *)): New.
21649         (extern debug (vec<rtx> &)): New.
21650         (extern debug (vec<rtx> *)): New.
21651         * sel-sched-ir.h
21652         (_list_iter_cond_expr): Make inline instead of static.
21653         * sreal.c
21654         (debug (sreal &)): New.
21655         (debug (sreal *)): New.
21656         * sreal.h
21657         (extern debug (sreal &)): New.
21658         (extern debug (sreal *)): New.
21659         * tree.h
21660         (extern debug_raw (const tree_node &)): New.
21661         (extern debug_raw (const tree_node *)): New.
21662         (extern debug (const tree_node &)): New.
21663         (extern debug (const tree_node *)): New.
21664         (extern debug_verbose (const tree_node &)): New.
21665         (extern debug_verbose (const tree_node *)): New.
21666         (extern debug_head (const tree_node &)): New.
21667         (extern debug_head (const tree_node *)): New.
21668         (extern debug_body (const tree_node &)): New.
21669         (extern debug_body (const tree_node *)): New.
21670         (extern debug (vec<tree, va_gc> &)): New.
21671         (extern debug (vec<tree, va_gc> *)): New.
21672         * tree-cfg.c
21673         (debug (struct loop &)): New.
21674         (debug (struct loop *)): New.
21675         (debug_verbose (struct loop &)): New.
21676         (debug_verbose (struct loop *)): New.
21677         * tree-dump.c: Add header dependence.
21678         * tree-flow.h
21679         (extern debug (struct loop &)): New.
21680         (extern debug (struct loop *)): New.
21681         (extern debug_verbose (struct loop &)): New.
21682         (extern debug_verbose (struct loop *)): New.
21683         * tree-data-ref.c
21684         (debug (data_reference &)): New.
21685         (debug (data_reference *)): New.
21686         (debug (vec<data_reference_p> &)): New.
21687         (debug (vec<data_reference_p> *)): New.
21688         (debug (vec<ddr_p> &)): New.
21689         (debug (vec<ddr_p> *)): New.
21690         * tree-data-ref.h
21691         (extern debug (data_reference &)): New.
21692         (extern debug (data_reference *)): New.
21693         (extern debug (vec<data_reference_p> &)): New.
21694         (extern debug (vec<data_reference_p> *)): New.
21695         (extern debug (vec<ddr_p> &)): New.
21696         (extern debug (vec<ddr_p> *)): New.
21697         * tree-ssa-alias.c
21698         (debug (pt_solution &)): New.
21699         (debug (pt_solution *)): New.
21700         * tree-ssa-alias.h
21701         (extern debug (pt_solution &)): New.
21702         (extern debug (pt_solution *)): New.
21703         * tree-ssa-alias.c
21704         (debug (_var_map &)): New.
21705         (debug (_var_map *)): New.
21706         (debug (tree_live_info_d &)): New.
21707         (debug (tree_live_info_d *)): New.
21708         * tree-ssa-alias.h
21709         (extern debug (_var_map &)): New.
21710         (extern debug (_var_map *)): New.
21711         (extern debug (tree_live_info_d &)): New.
21712         (extern debug (tree_live_info_d *)): New.
21714 2013-03-28  Jan Hubicka  <jh@suse.cz>
21716         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21718 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21720         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21721         record only when desired or required.
21723 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21725         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21726         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21728 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21730         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21731         (*andsi3_compare0_uxtw): New pattern.
21732         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21733         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21735 2013-03-28  Jan Hubicka  <jh@suse.cz>
21737         * data-streamer-in.c (streamer_read_gcov_count): New function.
21738         * gimple-streamer-out.c: Include value-prof.h.
21739         (output_gimple_stmt): Output histogram.
21740         (output_bb): Use streamer_write_gcov_count.
21741         * value-prof.c: Include data-streamer.h
21742         (dump_histogram_value): Add HIST_TYPE_MAX.
21743         (stream_out_histogram_value): New function.
21744         (stream_in_histogram_value): New function.
21745         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21746         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21747         * data-streamer-out.c (streamer_write_gcov_count): New function.
21748         (streamer_write_gcov_count_stream): New function.
21749         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21750         (lto_output_node): Likewise.
21751         (input_node, input_edge): Likewise.
21752         * lto-streamer-out.c (output_cfg): Update streaming.
21753         * lto-streamer-in.c (input_cfg): Likewise.
21754         * data-streamer.h (streamer_write_gcov_count,
21755         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21756         * gimple-streamer-in.c: Include value-prof.h
21757         (input_gimple_stmt): Input histograms.
21758         (input_bb): Update profile streaming.
21760 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21762         * genmodes.c (emit_max_int): New function.
21763         (emit_insn_modes_h): Added call to emit_max_function.
21764         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21765         Added doc.
21766         * machmode.def: Fixed comment.
21768 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21770         * combine.c (try_combine): Removed useless assert.
21771         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21773 2013-03-28  Marek Polacek  <polacek@redhat.com>
21774             Richard Biener  <rguenther@suse.de>
21776         PR tree-optimization/56695
21777         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21778         build signed result of a vector comparison.
21779         * tree-cfg.c (verify_gimple_comparison): Check that a result
21780         of a vector comparison has signed type.
21782 2013-03-28  Richard Biener  <rguenther@suse.de>
21784         PR tree-optimization/37021
21785         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21786         do not restrict gaps between groups.
21787         * tree-vect-stmts.c (vectorizable_load): Properly account for
21788         a gap between groups.
21790 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21792         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21793         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21794         is not enabled.
21796 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21798         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21799         * doc/extend.texi (Named Address Spaces): Ditto.
21800         (Variable Attributes): Ditto.
21802 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21804         * config.build: Add support for cygwin x64 target.
21805         * config.gcc: Likewise.
21806         * config.host: Likewise.
21807         * configure.ac: Likewise
21808         * configure: Regenerated.
21810 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21812         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21813         * config/i386/t-cygwin-w64: New file.
21814         * config/i386/cygwin-w64.h: New file.
21815         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21816         and add support for x64-cygwin target.
21817         (CPP_SPEC): Likewise.
21818         (CXX_WRAP_SPEC_LIST): Undefine before define.
21819         (LIBGCJ_SONAME): Use 15 as version.
21821 2013-03-27  Richard Biener  <rguenther@suse.de>
21823         PR tree-optimization/56716
21824         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21825         dumping for ref nodes.
21827 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21829         PR tree-optimization/55334
21830         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21831         restricted pointers to arrays.
21833 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21835         * Makefile.in (.SUFFIXES): Add .cc.
21836         (.c.o): Apply same recipe for implicit rule .cc.o.
21838 2013-03-27  Richard Biener  <rguenther@suse.de>
21840         PR tree-optimization/37021
21841         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21842         REALPART/IMAGPART_EXPRs around the supported refs.
21843         * tree-ssa-structalias.c (find_func_aliases): Assume that
21844         floating-point values are not used to transfer pointers.
21846 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21848         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21849         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21850         * doc/tm.texi: Regenerate.
21851         * targhooks.h (default_has_ifunc_p): New.
21852         * targhooks.c (default_has_ifunc_p): Ditto.
21853         * config/linux-protos.h: New file.
21854         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21855         hook for linux which disables support of indirect functions in android.
21856         * config/linux-android.c: New file.
21857         * config/t-linux-android.c: Ditto.
21858         * config.gcc: Added new object file linux-android.o.
21859         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21860         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21861         * varasm.c (do_assemble_alias): Likewise.
21862         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21863         doesn't support indirect functions.
21864         * configure: Regenerate.
21866 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21868         PR target/56102
21869         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21870         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21871         mult-word mode.
21873 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21875         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21877 2013-03-27  Terry Guo  <terry.guo@arm.com>
21879         * config/arm/arm-cores.def: Added core cortex-r7.
21880         * config/arm/arm-tune.md: Regenerated.
21881         * config/arm/arm-tables.opt: Regenerated.
21882         * doc/invoke.texi: Added entry for core cortex-r7.
21884 2013-03-27  Walter Lee  <walt@tilera.com>
21886         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21887         double-decrement of next_scratch_regno.
21889 2013-03-27  Walter Lee  <walt@tilera.com>
21891         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21892         input operands.
21893         (insn_v1mulus): Ditto.
21894         (insn_v2muls): Ditto.
21896 2013-03-27  Walter Lee  <walt@tilera.com>
21898         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21899         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21901 2013-03-27  Walter Lee  <walt@tilera.com>
21903         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21904         (*sibcall_value): Ditto.
21906 2013-03-27  Walter Lee  <walt@tilera.com>
21908         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21909         (insn_mnz_v8qi): ... this ...
21910         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21911         vector equivalent.
21912         (insn_v<n>mnz): Replaced by ...
21913         (insn_v1mnz): ... this ...
21914         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21915         equivalent.
21916         (insn_mz_<mode>): Replaced by ...
21917         (insn_mz_v8qi): ... this ...
21918         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21919         vector equivalent.
21920         (insn_v<n>mz): Replaced by ...
21921         (insn_v1mz): ... this ...
21922         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21923         equivalent.
21925 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21927         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21929 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21931         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21932         than fprintf with a non-constant, non-format string.
21934 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21936         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21937         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21938         operand 0 predicate.
21939         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21940         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21941         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21942         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21943         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21944         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21945         (mov<mode>_insv_1): Remove expander.  Merge insn with
21946         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21947         Use general_x64nomem_operand as operand 1 predicate.
21948         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21949         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21950         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21951         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21952         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21953         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21954         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21955         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21956         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21957         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21958         (general_x64nomem_operand): Ditto.
21960 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21962         * config/rtems.opt: Add -pthread option.
21964 2013-03-26  Richard Biener  <rguenther@suse.de>
21966         * alias.c (find_base_term): Avoid redundant and not used recursion.
21967         (base_alias_check): Get the initial base term from the caller.
21968         (true_dependence_1): Compute and pass base terms to base_alias_check.
21969         (write_dependence_p): Likewise.
21970         (may_alias_p): Likewise.
21972 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21974         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21975         PC-relative load in SI modes and above only.
21977 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21979         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21980         * config/arm/iwmmxt.md (WCGR0): Update.
21981         (WCGR1, WCGR2, WCGR3): Likewise.
21983 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21985         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21986         Use x64 and nox64 isa attributes.
21988 2013-03-26  Richard Biener  <rguenther@suse.de>
21990         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21991         alignment computations and rely on get_object_alignment_1
21992         for the !TYPE_P case.
21993         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21995 2013-03-26  Walter Lee  <walt@tilera.com>
21997         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21998         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
22000 2013-03-25  Jeff Law  <law@redhat.com>
22002         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
22003         check for INTEGRAL_TYPE_P that was missing due to checking in
22004         wrong version of prior patch.
22006 2013-03-25  Walter Lee  <walt@tilera.com>
22008         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
22009         TILEGX_INSN_SHUFFLEBYTES1.
22010         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
22011         shufflebytes1.
22012         (tilegx_builtins): Ditto.
22013         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
22015 2013-03-25  Walter Lee  <walt@tilera.com>
22017         * config/tilegx/tilegx.md (floatsisf2): New pattern.
22018         (floatunssisf2): New pattern.
22019         (floatsidf2): New pattern.
22020         (floatunssidf2): New pattern.
22022 2013-03-25  Walter Lee  <walt@tilera.com>
22024         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
22025         tests for constraint J, K, N, P.
22027 2013-03-25  Walter Lee  <walt@tilera.com>
22029         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
22030         Use indirect/pcrel encoding.
22031         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
22032         Ditto.
22034 2013-03-25  Steve Ellcey  <sellcey@mips.com>
22036         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
22037         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
22038         * config/mips/mips.c (mips_option_override): Set IMADD default.
22039         * config/mips/mips.h (PTF_AVOID_IMADD): New.
22040         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
22041         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
22042         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
22043         * doc/invoke.texi (-mimadd/-mno-imadd): New.
22045 2013-03-25  Jeff Law  <law@redhat.com>
22047         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
22048         slightly to avoid creating and folding useless trees.  Simplify
22049         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
22051 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
22053         * config/i386/i386.md (*zero_extendsidi2): Merge with
22054         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
22055         * config/i386/predicates.md (x86_64_zext_operand): Rename from
22056         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
22057         targets.  Clarify comment.
22059 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22061         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
22062         pass-through jump functions differently.
22063         (ipa_read_jump_function): Likewise.  Also use setter functions to set
22064         up jump functions.
22066 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22068         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22069         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22070         process it.
22071         (ipa_get_indirect_edge_target): New function.
22072         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22073         ipa_get_indirect_edge_target.  Update all callers.
22074         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22075         ipa_get_indirect_edge_target_1 instead of calling
22076         ipa_get_indirect_edge_target.
22077         (create_specialized_node): Pass aggvlas to
22078         ipcp_discover_new_direct_edges.
22080 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22082         * config/arm/arm.md (f_sels, f_seld): New types.
22083         (*cmov<mode>): New pattern.
22084         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22085         predicate.
22087 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22089         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22090         POSIX-printf for mingw-hosted  builds.
22092 2013-03-25  Richard Biener  <rguenther@suse.de>
22094         PR middle-end/56694
22095         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22096         must-not-throw stmt location.
22098 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22100         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22101         Emit load-acquire versions when acq is true.
22102         (arm_emit_store_exclusive): Add rel parameter.
22103         Emit store-release versions when rel is true.
22104         (arm_split_compare_and_swap): Use acquire-release instructions
22105         instead.
22106         of barriers when appropriate.
22107         (arm_split_atomic_op): Likewise.
22108         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22109         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22110         (VUNSPEC_SLX): Likewise.
22111         (VUNSPEC_LDA): Likewise.
22112         (VUNSPEC_STL): Likewise.
22113         * config/arm/sync.md (atomic_load<mode>): New pattern.
22114         (atomic_store<mode>): Likewise.
22115         (arm_load_acquire_exclusive<mode>): Likewise.
22116         (arm_load_acquire_exclusivesi): Likewise.
22117         (arm_load_acquire_exclusivedi): Likewise.
22118         (arm_store_release_exclusive<mode>): Likewise.
22120 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22122         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22123         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22124         * config/mip/predicates.md (lwsp_swsp_operand,
22125         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22126         sb16_operand, db4_operand, db7_operand, ib3_operand,
22127         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22128         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22129         andi16_operand): New predicates.
22130         * config/mips/mips.md (compression): New attribute.
22131         (enabled): New attribute.
22132         (length): Consider compression in computing length.
22133         (shift_compression): New code attribute.
22134         (*add<mode>3): New operands. Record compression.
22135         (sub<mode>3): Likewise.
22136         (one_cmpl<mode>2): Likewise.
22137         (*and<mode>3): Likewise.
22138         (*ior<mode>3): Likewise.
22139         (unnamed pattern for xor): Likewise.
22140         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22141         (*<optab><mode>3): Likewise.
22142         (*mov<mode>_internal: Likewise.
22143         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22144         (mips_unsigned_immediate_p): New.
22145         (umips_lwsp_swsp_address_p): New.
22146         (m16_based_address_p): New.
22147         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22148         (mips_unsigned_immediate_p): New prototype.
22149         (lwsp_swsp_address_p): New prototype.
22150         (m16_based_address_p): New prototype.
22151         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22152         (mips_signed_immediate_p): New function.
22153         (m16_based_address_p): New function.
22154         (lwsp_swsp_address_p): New function.
22155         (mips_print_operand_punctuation): Recognize short delay slot insns
22156         for microMIPS.add<mode>3"
22158 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22160         PR target/56720
22161         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22162         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22164 2013-03-25  Richard Biener  <rguenther@suse.de>
22166         PR tree-optimization/56689
22167         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22168         any edge.
22170 2013-03-25  Richard Biener  <rguenther@suse.de>
22172         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22173         of bitmap.
22174         (memory_references): Likewise.
22175         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22176         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22177         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22178         (gather_mem_refs_in_loops): Fold into ...
22179         (analyze_memory_references): ... this.  Move initialization
22180         to tree_ssa_lim_initialize.
22181         (fill_always_executed_in): Rename to ...
22182         (fill_always_executed_in_1): ... this.
22183         (fill_always_executed_in): Move contains_call computation to
22184         this new function from ...
22185         (tree_ssa_lim_initialize): ... here.
22186         (tree_ssa_lim): Call fill_always_executed_in.
22188 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22190         * postreload.c (reload_combine): Fix code detecting returns.
22192 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22194         * function.c (emit_use_return_register_into_block): On cc0 targets,
22195         do not emit the sequence between cc0 setter and user.
22197 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22199         * config/i386/predicates.md (local_symbolic_operand): Interpret
22200         dll-imported symbols as none-local.
22202 2013-03-25  Richard Biener  <rguenther@suse.de>
22204         * tree-ssa-loop-im.c (struct depend): Remove.
22205         (struct lim_aux_data): Make depends a vec of gimples.
22206         (free_lim_aux_data): Adjust.
22207         (add_dependency): Likewise.
22208         (set_level): Likewise.
22210 2013-03-25  Richard Biener  <rguenther@suse.de>
22212         PR middle-end/56434
22213         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22214         the pointer returned by calls with ECF_MALLOC set.
22216 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22218         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22220 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22222         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22223         using MMXMODE mode iterator.
22224         (*move<mode>_internal): Merge with *movv2sf_internal and
22225         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22227 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22229         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22230         (record_last_mem_set_info): Likewise.
22232         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22233         of XNEWVEC followed by memset.
22234         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22236 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22238         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22239         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22240         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22241         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22242         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22243         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22244         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22245         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22246         BARRIER_P instead of GET_CODE.
22248 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22250         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22251         inaccuracy in the probing code.
22253         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22254         (ctrapdi4): Likewise.
22256 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22258         * calls.c (expand_call): Add missing guard to code handling return
22259         of non-BLKmode structures in MSB.
22260         * function.c (expand_function_end): Likewise.
22262 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22264         * combine.c (try_combine): Adjust comment.  Do not add the set of
22265         insn #0 if the destination indirectly is set or dies in insn #2.
22266         Tidy up code to distribute a new note.
22268 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22270         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22271         also for alternatives 16 and 17.
22273 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22275         * config/i386/sse.md (*mov<mode>_internal): Merge with
22276         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22277         Emit insn template depending on type attribute.  Use
22278         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22279         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22280         attribute calculation.  Remove unit attribute calculation.
22281         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22282         Set prefix_data16 attribute for DImode ssemov types.
22283         Use Ym instead of y for SSE-MMX conversion alternatives.
22284         Reorder operand constraints.
22286 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22288         * df.h (df_insn_delete): Adjust prototype.
22289         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22290         and let it decide whether mark the basic block dirty.
22291         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22292         * df-scan.c (df_insn_info_delete): New helper function, split
22293         off from df_insn_delete.
22294         (df_scan_free_bb_info): Use it.
22295         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22296         Likewise.
22297         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22298         that the insn is actually an insn and it has a non-NULL basic block.
22299         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22301 2013-03-22  Richard Biener  <rguenther@suse.de>
22303         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22304         dep_ref members.
22305         (mem_ref_alloc): Do not allocate them.
22306         (refs_independent_p): Do not query or maintain a cache.
22308 2013-03-22  Richard Biener  <rguenther@suse.de>
22310         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22311         (gather_mem_refs_in_loops): Do not compute it.
22312         (analyze_memory_references): Do not allocate it.
22313         (tree_ssa_lim_finalize): Do not free it.
22314         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22316 2013-03-22  Richard Biener  <rguenther@suse.de>
22318         * is-a.h (as_a): Use gcc_checking_assert.
22320 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22322         * config/aarch64/aarch64.c (aarch64_print_operand): New
22323         format specifier for printing a constant in hex.
22324         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22325         format specifier for printing second operand.
22327 2013-03-22  Richard Biener  <rguenther@suse.de>
22329         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22330         bitmaps.
22331         (gather_mem_refs_in_loops): Perform store accumulation here.
22332         (create_vop_ref_mapping_loop): Remove.
22333         (create_vop_ref_mapping): Likewise.
22334         (analyze_memory_references): Initialize refs_stored_in_loop.
22335         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22336         (record_indep_loop): Remove.
22337         (record_dep_loop): New function.
22338         (ref_indep_loop_p_1): Adjust to only walk over references
22339         in the loop, not its subloops.
22340         (ref_indep_loop_p): Rename to ...
22341         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22342         maintaining a more fine-grained cache.
22343         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22344         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22346 2013-03-22  Richard Biener  <rguenther@suse.de>
22348         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22349         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22350         aggregate mem_ref_loc.
22351         (free_mem_ref_locs): Inline into ...
22352         (memref_free): ... this and adjust.
22353         (mem_ref_alloc): Adjust.
22354         (mem_ref_locs_alloc): Remove.
22355         (record_mem_ref_loc): Adjust.
22356         (get_all_locs_in_loop): Rewrite into ...
22357         (for_all_locs_in_loop): ... this iterator.
22358         (rewrite_mem_ref_loc): New functor.
22359         (rewrite_mem_refs): Use for_all_locs_in_loop.
22360         (sm_set_flag_if_changed): New functor.
22361         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22362         (ref_always_accessed): New functor.
22363         (ref_always_accessed_p): Use for_all_locs_in_loop.
22365 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22367         * tree-pass.h (PROP_gimple_lvec): New.
22368         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22369         (init_optimization_passes): Move pass_lower_vector.
22370         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22371         PROP_gimple_lvec.
22372         (pass_lower_vector): Provide PROP_gimple_lvec.
22373         (pass_lower_vector_ssa): Likewise.
22374         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22376 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22378         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22380 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22382         * config/i386/i386.md (*movdi_internal): Disparage slightly
22383         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22384         conversion alternatives.
22386 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22388         PR middle-end/48087
22389         * diagnostic.def (DK_WERROR): New kind.
22390         * diagnostic.h (werrorcount): Define.
22391         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22392         promoted to DK_ERROR, increment DK_WERROR counter instead of
22393         DK_ERROR counter.
22394         * toplev.c (toplev_main): Call print_ignored_options even if
22395         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22396         even if just werrorcount is non-zero.
22398         PR debug/55608
22399         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22400         on failure.
22401         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22402         (string_cst_pool_decl): New function.
22403         (optimize_one_addr_into_implicit_ptr): New function.
22404         (resolve_addr_in_expr): Optimize DWARF location expression
22405         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22406         which doesn't live in memory, but has DW_AT_location or
22407         DW_AT_const_value, or refers to a string literal, into
22408         DW_OP_GNU_implicit_pointer.
22409         (optimize_location_into_implicit_ptr): New function.
22410         (resolve_addr): If removing DW_AT_location of a variable because
22411         it was DW_OP_addr of address of the variable, but the variable doesn't
22412         live in memory, try to emit const value attribute for the initializer.
22414 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22416         * tree.h (VECTOR_TYPE_P): New macro.
22417         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22418         TYPE_MODE): Use it.
22419         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22420         VEC_COND_EXPR cannot be lvalues.
22421         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22423 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22425         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22426         Restrict the transformation to equal modes.
22428 2013-03-21  Richard Biener  <rguenther@suse.de>
22430         PR tree-optimization/39326
22431         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22432         (MEM_ANALYZABLE): Adjust.
22433         (record_mem_ref_loc): Move bitmap ops ...
22434         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22435         unanalyzable refs, do not record locations for it.
22436         (analyze_memory_references): Allocate ref zero as shared
22437         unanalyzable ref.
22438         (refs_independent_p): Do not test for unanalyzed mems here.
22439         (ref_indep_loop_p_1): Special-case disambiguation against
22440         the unanalyzed ref.
22441         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22443 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22445         * config/arm/arm-protos.h (tune_params): Add
22446         prefer_neon_for_64bits field.
22447         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22448         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22449         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22450         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22451         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22452         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22453         (arm_option_override): Handle -mneon-for-64bits new option.
22454         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22455         (prefer_neon_for_64bits): Declare new variable.
22456         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22457         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22458         (arch_enabled): Handle new arch types. Remove support for onlya8
22459         and nota8.
22460         (one_cmpldi2): Use new arch names.
22461         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22462         * config/arm/arm.opt (mneon-for-64bits): Add option.
22463         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22464         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22465         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22466         of onlya8.
22467         * doc/invoke.texi (-mneon-for-64bits): Document.
22469 2013-03-21  Richard Biener  <rguenther@suse.de>
22471         PR tree-optimization/39326
22472         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22473         (sort_bbs_in_loop_postorder_cmp): New function.
22474         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22476 2013-03-21  Richard Biener  <rguenther@suse.de>
22478         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22479         (vect_insert_into_interleaving_chain): Likewise.
22480         (vect_drs_dependent_in_basic_block): Inline ...
22481         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22482         split out from ...
22483         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22484         (vect_check_interleaving): Simplify.
22485         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22486         (vect_slp_analyze_data_ref_dependences): ... this new function.
22487         (dr_group_sort_cmp): New function.
22488         (vect_analyze_data_ref_accesses): Compute data-reference groups
22489         here instead of in vect_analyze_data_ref_dependence.  Use
22490         a more efficient algorithm.
22491         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22492         vect_slp_analyze_data_ref_dependences.  Call
22493         vect_analyze_data_ref_accesses earlier.
22494         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22495         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22496         (vect_slp_analyze_data_ref_dependences): New prototype.
22498 2013-03-21  Richard Biener  <rguenther@suse.de>
22500         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22501         ref is stored in the loop.
22502         (find_refs_for_sm): Walk only over all stores.
22503         (store_motion_loop): Allocate from lim_bitmap_obstack.
22504         (store_motion): Likewise.
22506 2013-03-21  Richard Biener  <rguenther@suse.de>
22508         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22509         Update virtual SSA form.
22511 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22513         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22514         * configure: Regenerate.
22515         * config.in: Regenerate.
22516         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22517         if !HAVE_LD_EH_FRAME_CIEV3.
22519 2013-03-21  Richard Biener  <rguenther@suse.de>
22521         * tree-cfg.c (verify_expr_no_block): New function.
22522         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22523         nor DECL_VALUE_EXPR have locations with associated blocks.
22524         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22525         (clear_unused_block_pointer): Remove code dealing with
22526         blocks in DECL_DEBUG_EXPR locations.
22528 2013-03-21  Richard Biener  <rguenther@suse.de>
22530         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22531         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22532         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22533         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22534         instead of DECL_DEBUG_EXPR_IS_FROM.
22535         * gimplify.c (gimplify_modify_expr): Likewise.
22536         * tree-cfg.c (verify_expr_location_1): Likewise.
22537         * tree-complex.c (create_one_component_var): Likewise.
22538         * tree-sra.c (create_access_replacement): Likewise.
22539         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22540         (clear_unused_block_pointer): Likewise.
22541         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22542         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22543         * var-tracking.c (var_debug_decl): Likewise.
22544         (track_expr_p): Likewise.
22545         * tree-inline.c (add_local_variables): Likewise.  Set
22546         DECL_HAS_DEBUG_EXPR_P after copying it.
22547         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22548         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22550 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22552         PR bootstrap/56656
22553         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22554         * configure: Regenerate.
22555         * config.in: Regenerate.
22556         * config/i386/i386.md (*movdf_internal): Use
22557         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22558         movd instead of movq mnemonic for interunit moves.
22559         (*movdi_internal): Ditto.
22561 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22563         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22564         (abd<mode>_3): New pattern.
22565         (aba<mode>_3): New pattern.
22566         (fabd<mode>_3): New pattern.
22568 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22570         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22571         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22572         occurrence of REGISTER_PREFIX as its empty string.
22574 2013-03-20  Jeff Law  <law@redhat.com>
22576         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22577         addititional equivalences for equality comparisons between an SSA_NAME
22578         and a constant where the SSA_NAME was set from a widening conversion.
22580 2013-03-20  Walter Lee  <walt@tilera.com>
22582         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22584 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22586         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22587         depending on type attribute.
22588         (*movti_internal): Ditto.
22589         (*movtf_internal): Ditto.
22590         (*movxf_internal): Ditto.
22591         (*movdf_internal): Ditto.
22592         (*movsf_internal): Ditto.
22594 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22596         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22597         maybe_vex for sselog1 and ssemov types.
22598         (*movdi_internal): Reorder operand constraints.
22599         (*movsi_internal): Ditto.  Set prefix attribute to
22600         maybe_vex for sselog1 and ssemov types.
22601         (*movtf_internal): Set prefix attribute to maybe_vex
22602         for sselog1 and ssemov types.
22603         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22604         DImode ssemov types.  Reorder operand constraints.
22605         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22606         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22607         attribute for SImode ssemov types.  Reorder operand constraints.
22609 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22611         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22612         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22614 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22616         * config/rs6000/predicates.md (indexed_address, update_address_mem
22617         update_indexed_address_mem): New predicates.
22618         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22619         attribute for load/store instructions.
22620         * config/rs6000/dfp.md (movsd_store): Likewise.
22621         (movsd_load): Likewise.
22622         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22623         (unnamed HI->DI extend define_insn): Likewise.
22624         (unnamed SI->DI extend define_insn): Likewise.
22625         (unnamed QI->SI extend define_insn): Likewise.
22626         (unnamed QI->HI extend define_insn): Likewise.
22627         (unnamed HI->SI extend define_insn): Likewise.
22628         (unnamed HI->SI extend define_insn): Likewise.
22629         (extendsfdf2_fpr): Likewise.
22630         (movsi_internal1): Likewise.
22631         (movsi_internal1_single): Likewise.
22632         (movhi_internal): Likewise.
22633         (movqi_internal): Likewise.
22634         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22635         attribute for load/store instructions.
22636         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22637         instructions.
22638         (mov<mode>_softfloat): Likewise.
22639         (mov<mode>_hardfloat32): Likewise.
22640         (mov<mode>_hardfloat64): Likewise.
22641         (mov<mode>_softfloat64): Likewise.
22642         (movdi_internal32): Likewise.
22643         (movdi_internal64): Likewise.
22644         (probe_stack_<mode>): Likewise.
22646 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22648         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22649         floating point, and decimal floating point to reload iterator.
22651         * config/rs6000/constraints.md (wl constraint): New constraints to
22652         return FLOAT_REGS if certain options are used to reduce the number
22653         of separate patterns that exist in the file.
22654         (wx constraint): Likewise.
22655         (wz constraint): Likewise.
22657         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22658         -mdebug=reg, print wg, wl, wx, and wz constraints.
22659         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22660         Initialize the reload functions for 64-bit binary/decimal floating
22661         point types.
22662         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22663         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22664         create the buffer on the stack to overcome not having a 32-bit
22665         load and store.
22666         (rs6000_emit_move): Likewise.
22667         (rs6000_secondary_memory_needed_rtx): Likewise.
22668         (rs6000_alloc_sdmode_stack_slot): Likewise.
22669         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22670         via xxlxor, just like DFmode 0.0.
22672         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22673         define as 1 if we are running on a power7 or newer.
22674         (enum r6000_reg_class_enum): Add new constraints.
22676         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22677         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22678         with other moves by using conditional constraits (wg).  Use LFIWZX
22679         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22680         (movsd splitter): Likewise.
22681         (movsd_hardfloat): Likewise.
22682         (movsd_softfloat): Likewise.
22684         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22685         binary and decimal floating point moves.
22686         (fmove_ok): New attributes to combine binary and decimal floating
22687         point moves, and to combine power6x (mfpgpr) moves along normal
22688         floating moves.
22689         (real_value_to_target): Likewise.
22690         (f32_lr): Likewise.
22691         (f32_lm): Likewise.
22692         (f32_li): Likewise.
22693         (f32_sr): Likewise.
22694         (f32_sm): Likewise.
22695         (f32_si): Likewise.
22696         (movsf): Combine binary and decimal floating point moves.  Combine
22697         power6x (mfpgpr) moves with other moves by using conditional
22698         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22699         (mov<mode> for SFmode/SDmode); Likewise.
22700         (SFmode/SDmode splitters): Likewise.
22701         (movsf_hardfloat): Likewise.
22702         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22703         (movsf_softfloat): Likewise.
22704         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22706         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22707         wx and wz constraints.
22709         * config/rs6000/constraints.md (wg constraint): New constraint to
22710         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22712         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22713         constraint.
22715         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22716         -mdebug=reg, print wg, wl, wx, and wz constraints.
22717         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22718         Initialize the reload functions for 64-bit binary/decimal floating
22719         point types.
22720         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22721         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22722         create the buffer on the stack to overcome not having a 32-bit
22723         load and store.
22724         (rs6000_emit_move): Likewise.
22725         (rs6000_secondary_memory_needed_rtx): Likewise.
22726         (rs6000_alloc_sdmode_stack_slot): Likewise.
22727         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22728         via xxlxor, just like DFmode 0.0.
22730         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22731         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22732         with other moves by using conditional constraits (wg).  Use LFIWZX
22733         and STFIWX for loading SDmode on power7.
22734         (movdd splitters): Likewise.
22735         (movdd_hardfloat32): Likewise.
22736         (movdd_softfloat32): Likewise.
22737         (movdd_hardfloat64_mfpgpr): Likewise.
22738         (movdd_hardfloat64): Likewise.
22739         (movdd_softfloat64): Likewise.
22741         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22742         64-bit binary and decimal floating point moves.
22743         (FMOVE64X): Likewise.
22744         (movdf): Combine 64-bit binary and decimal floating point moves.
22745         Combine power6x (mfpgpr) moves with other moves by using
22746         conditional constraits (wg).
22747         (mov<mode> for DFmode/DDmode): Likewise.
22748         (DFmode/DDmode splitters): Likewise.
22749         (movdf_hardfloat32): Likewise.
22750         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22751         (movdf_softfloat32): Likewise.
22752         (movdf_hardfloat64_mfpgpr): Likewise.
22753         (movdf_hardfloat64): Likewise.
22754         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22755         (movdf_softfloat64): Likewise.
22756         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22757         (reload_<mode>_load): Move to later in the file so they aren't in
22758         the middle of the floating point move insns.
22759         (reload_<mode>_store): Likewise.
22761         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22762         constraint.
22764         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22765         constraint if -mdebug=reg.
22766         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22767         Enable using dd reload support if needed.
22769         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22770         binary and decimal floating point moves in rs6000.md.
22771         (movtd_internal): Likewise.
22773         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22774         decimal floating point moves.
22775         (movtf): Likewise.
22776         (movtf_internal): Likewise.
22777         (mov<mode>_internal, TDmode/TFmode): Likewise.
22778         (movtf_softfloat): Likewise.
22779         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22781         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22782         movdi_internal64, using wg constraint for move direct operations.
22783         (movdi_internal64): Likewise.
22785         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22786         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22787         the various virtual registers. Use GPR/FPR first/last values,
22788         instead of hard coding the register numbers.  Print which modes
22789         have reload functions registered.
22790         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22791         settings before/after setting cpu, target and subtarget settings.
22792         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22793         and for secondary reload failures in rs6000_secondary_reload_inner.
22794         (rs6000_secondary_reload_fail): Likewise.
22795         (rs6000_secondary_reload_inner): Likewise.
22797         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22798         macros for first/last GPR and FPR registers.
22799         (LAST_GPR_REGNO): Likewise.
22800         (FIRST_FPR_REGNO): Likewise.
22801         (LAST_FPR_REGNO): Likewise.
22803         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22804         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22805         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22806         (vcond<mode><mode>): Likewise.
22807         (vcondu<mode><mode>): Likewise.
22808         (vector_gtu<mode>): Likewise.
22809         (vector_gte<mode>): Likewise.
22810         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22811         to prevent the compiler from converting DImode operations to TImode.
22812         (ior<mode>3): Likewise.
22813         (and<mode>3): Likewise.
22814         (one_cmpl<mode>2): Likewise.
22815         (nor<mode>3): Likewise.
22816         (andc<mode>3): Likewise.
22818         * config/rs6000/constraints.md (wt constraint): New constraint
22819         that returns VSX_REGS if TImode is allowed in VSX registers.
22821         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22822         constant under VSX.
22824         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22825         similar to TImode, but it is restricted to being in the GPRs.
22827         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22828         TImode to occupy a single VSX register.
22830         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22831         -mvsx-timode for power7/power8.
22832         (power7 cpu): Likewise.
22833         (power8 cpu): Likewise.
22835         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22836         sure that TFmode/TDmode take up two registers if they are ever
22837         allowed in the upper VSX registers.
22838         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22839         registers.
22840         (rs6000_init_hard_regno_mode_ok): Likewise.
22841         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22842         constraint.  Print if LRA is turned on.
22843         (rs6000_option_override_internal): Give an error if -mvsx-timode
22844         and VSX is not enabled.
22845         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22846         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22847         to reg+offset addressing.  Use PTImode when checking offset
22848         addresses for validity.
22849         (reg_offset_addressing_ok_p): Likewise.
22850         (rs6000_legitimate_offset_address_p): Likewise.
22851         (rs6000_legitimize_address): Likewise.
22852         (rs6000_legitimize_reload_address): Likewise.
22853         (rs6000_legitimate_address_p): Likewise.
22854         (rs6000_eliminate_indexed_memrefs): Likewise.
22855         (rs6000_emit_move): Likewise.
22856         (rs6000_secondary_reload): Likewise.
22857         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22858         reloads to fpr registers to continue to use reg+offset addressing,
22859         but 64-bit reloads to altivec registers need reg+reg addressing.
22860         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22861         it.  Treat LO_SUM like a PLUS operation.
22862         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22863         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22864         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22865         registers to share a register with a smaller sized type, since VSX
22866         puts scalars in the upper 64-bits.
22867         (print_operand): Add support for PTImode.
22868         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22869         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22870         registers, but don't have arithmetic support.
22871         (rs6000_memory_move_cost): Add test for VSX.
22872         (rs6000_opt_masks): Add -mvsx-timode.
22874         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22875         for TImode.
22876         (VSs): Likewise.
22877         (VSr): Use wt constraint for TImode.
22878         (VSv): Drop TImode support.
22879         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22880         (vsx_movti_64bit): Likewise.
22881         (vsx_movti_32bit): Likewise.
22882         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22883         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22884         one '?' on the appropriate output constraint.  Do not allow TImode
22885         logical operations on 32-bit systems.
22886         (vsx_ior<mode>3): Likewise.
22887         (vsx_xor<mode>3): Likewise.
22888         (vsx_one_cmpl<mode>2): Likewise.
22889         (vsx_nor<mode>3): Likewise.
22890         (vsx_andc<mode>3): Likewise.
22891         (vsx_concat_<mode>): Likewise.
22892         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22894         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22895         OPTION_MASK_VSX_TIMODE.
22896         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22897         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22899         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22900         (TI2 iterator): New iterator for TImode, PTImode.
22901         (wd mode attribute): Add values for vector types.
22902         (movti_string): Replace TI move operations with operations for TImode
22903         and PTImode.  Add support for TImode being allowed in VSX registers.
22904         (mov<mode>_string, TImode/PTImode): Likewise.
22905         (movti_ppc64): Likewise.
22906         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22907         (TI mode splitters): Likewise.
22909         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22910         constraint.
22912 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22914         PR tree-optimization/56355
22915         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22916         Also handle integers with undefined overflow.
22918 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22919             Maciej W. Rozycki  <macro@codesourcery.com>
22920             Tom de Vries  <tom@codesourcery.com>
22921             Nathan Sidwell  <nathan@codesourcery.com>
22922             Iain Sandoe  <iain@codesourcery.com>
22923             Nathan Froyd  <froydnj@codesourcery.com>
22924             Chao-ying Fu  <fu@mips.com>
22926         * doc/extend.texi: (micromips, nomicromips, nocompression):
22927         Document new function attributes.
22928         * doc/invoke.texi (minterlink-compressed, mmicromips,
22929         m14k, m14ke, m14kec): Document new options.
22930         (minterlink-mips16): Update documentation.
22931         * doc/md.texi (ZC, ZD): Document new constraints.
22932         * configure.ac (gcc_cv_as_micromips): Check if linker
22933         supports the .set micromips directive.
22934         * configure: Regenerate.
22935         * config.in: Regenerate.
22936         * config/mips/mips-tables.opt: Regenerate.
22937         * config/mips/micromips.md: New file.
22938         * constraints.md (ZC, ZD): New constraints.
22939         * config/mips/predicates.md (movep_src_register): New predicate.
22940         (movep_src_operand): New predicate.
22941         (non_volatile_mem_operand): New predicate.
22942         * config/mips/mips.md (multimem): New type.
22943         (length): Differentiate between 17-bit and 18-bit branch offsets.
22944         (MOVEP1, MOVEP2): New mode iterator.
22945         (mov_<load>l): Use ZC constraint.
22946         (mov_<load>r): Likewise.
22947         (mov_<store>l): Likewise.
22948         (mov_<store>r): Likewise.
22949         (*branch_equality<mode>_inverted): Add microMIPS support.
22950         (*branch_equality<mode>): Likewise.
22951         (*jump_absolute): Likewise.
22952         (indirect_jump_<mode>): Likewise.
22953         (tablejump_<mode>): Likewise.
22954         (<optab>_internal): Likewise.
22955         (sibcall_internal): Likewise.
22956         (sibcall_value_internal): Likewise.
22957         (prefetch): Use constraint ZD.
22958         * config/mips/mips.opt (minterlink-compressed): New option.
22959         (minterlink-mips16): Now an alias for minterlink-compressed.
22960         (mmicromips): New option.
22961         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22962         (compare_and_swap_12): Likewise.
22963         (sync_add<mode>): Likewise.
22964         (sync_<optab>_12): Likewise.
22965         (sync_old_<optab>_12): Likewise.
22966         (sync_new_<optab>_12): Likewise.
22967         (sync_nand_12): Likewise.
22968         (sync_old_nand_12): Likewise.
22969         (sync_new_nand_12): Likewise.
22970         (sync_sub<mode>): Likewise.
22971         (sync_old_add<mode>): Likewise.
22972         (sync_old_sub<mode>): Likewise.
22973         (sync_new_add<mode>): Likewise.
22974         (sync_new_sub<mode>): Likewise.
22975         (sync_<optab><mode>): Likewise.
22976         (sync_old_<optab><mode>): Likewise.
22977         (sync_new_<optab><mode>): Likewise.
22978         (sync_nand<mode>): Likewise.
22979         (sync_old_nand<mode>): Likewise.
22980         (sync_new_nand<mode>): Likewise.
22981         (sync_lock_test_and_set<mode>): Likewise.
22982         (test_and_set_12): Likewise.
22983         (atomic_compare_and_swap<mode>): Likewise.
22984         (atomic_exchange<mode>_llsc): Likewise.
22985         (atomic_fetch_add<mode>_llsc): Likewise.
22986         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22987         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22988         (umips_save_restore_pattern_p): Likewise.
22989         (umips_load_store_pair_p): Likewise.
22990         (umips_output_load_store_pair): Likewise.
22991         (umips_movep_target_p): Likewise.
22992         (umips_12bit_offset_address_p): Likewise.
22993         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22994         (mips_base_mips16): Rename this...
22995         (mips_base_compression_flags): ...to this. Update all uses.
22996         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22997         (mips_mips16_decl_p): Delete.
22998         (mips_nomips16_decl_p): Delete.
22999         (mips_get_compress_on_flags): New function.
23000         (mips_get_compress_off_flags): New function.
23001         (mips_get_compress_mode): New function.
23002         (mips_get_compress_on_name): New function.
23003         (mips_get_compress_off_name): New function.
23004         (mips_insert_attributes): Support multiple compression types.
23005         (mips_merge_decl_attributes): Likewise.
23006         (umips_12bit_offset_address_p): New function.
23007         (mips_start_function_definition): Emit .set micromips directive.
23008         (mips_call_may_need_jalx_p): New function.
23009         (mips_function_ok_for_sibcall): Add microMIPS support.
23010         (mips_print_operand_punctuation): Support short delay slots and
23011         compact jumps.
23012         (umips_swm_mask, umips_swm_encoding): New.
23013         (umips_build_save_restore): New function.
23014         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23015         (was_mips16_p): Remove.
23016         (old_compression_mode): New.
23017         (mips_set_compression_mode): New function.
23018         (mips_set_current_function): Add microMIPS support.
23019         (mips_option_override): Likewise.
23020         (umips_save_restore_pattern_p): New function.
23021         (umips_output_save_restore): New function.
23022         (umips_load_store_pair_p_1): New function.
23023         (umips_load_store_pair_p): New function.
23024         (umips_output_load_store_pair_1): New function.
23025         (umips_output_load_store_pair): New function.
23026         (umips_movep_target_p) New function.
23027         (mips_prepare_pch_save): Add microMIPS support.
23028         * config/mips/mips.h (TARGET_COMPRESSION): New.
23029         (TARGET_CPU_CPP_BUILTINS): Update macro
23030         to use new compression flags and to support microMIPS.
23031         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23032         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23033         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23034         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23035         (ASM_SPEC): Support mmicromips and mno-micromips.
23036         (M16STORE_REG_P): New macro.
23037         (MIPS_CALL): Support TARGET_MICROMIPS.
23038         (MICROMIPS_J): New macro.
23039         (mips_base_mips16): Rename this...
23040         (mips_base_compression_flags): ...to this.
23041         (UMIPS_12BIT_OFFSET_P): New macro.
23042         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23043         (MULTILIB_DIRNAMES): Likewise.
23044 2013-03-20  Richard Biener  <rguenther@suse.de>
23046         PR tree-optimization/56661
23047         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
23048         the result does not have to be distinct.
23050 2013-03-20  Richard Biener  <rguenther@suse.de>
23052         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
23053         remap_gimple_op_r.
23055 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23056             Steven Bosscher  <steven@gcc.gnu.org>
23058         PR rtl-optimization/56605
23059         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
23061 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23063         PR bootstrap/56656
23064         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
23065         that require movd instead of movq.
23067 2013-03-20  Richard Biener  <rguenther@suse.de>
23069         * tree-ssa-structalias.c (struct variable_info): Add pointer
23070         to the first field of an aggregate with sub-vars.  Make
23071         this and the pointer to the next subfield its ID.
23072         (vi_next): New function.
23073         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23074         storedanything_id, integer_id): Increment by one.
23075         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23076         get_call_clobber_vi): Adjust.
23077         (solution_set_expand): Simplify and speedup.
23078         (solution_set_add): Inline into ...
23079         (set_union_with_increment): ... this.  Adjust accordingly.
23080         (do_sd_constraint): Likewise.
23081         (do_ds_constraint): Likewise.
23082         (do_complex_constraint): Simplify.
23083         (build_pred_graph): Adjust.
23084         (solve_graph): Likewise.  Simplify and speedup.
23085         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23086         get_constraint_for_component_ref, get_constraint_for_1,
23087         first_vi_for_offset, first_or_preceding_vi_for_offset,
23088         create_function_info_for, create_variable_info_for_1,
23089         create_variable_info_for, intra_create_variable_infos): Adjust.
23090         (init_base_vars): Push NULL for ID zero.
23091         (compute_points_to_sets): Adjust.
23093 2013-03-20  Richard Biener  <rguenther@suse.de>
23095         * cfgloop.c (verify_loop_structure): Streamline and avoid
23096         ICEing on corrupt loop tree.
23097         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23098         loop tree.
23100 2013-03-20  Richard Biener  <rguenther@suse.de>
23102         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23103         check whether an SSA update is needed.
23105 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23107         * config/mips/constraints.md (T): Rename to...
23108         (Yf): ...this.
23109         (U): Rename to...
23110         (Yd): ...this.
23111         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23112         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23114 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23116         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23117         (*subsi3_carryin_uxtw): Likewise.
23119 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23121         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23122         (*rorsi3_insn_uxtw): Likewise.
23124 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23126         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23127         (*extrsi5_insn_uxtw): Likewise.
23129 2013-03-19  Richard Biener  <rguenther@suse.de>
23131         PR tree-optimization/56273
23132         * passes.c (init_optimization_passes): Move second VRP after DOM.
23134 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23136         * config/i386/i386.md (*movti_internal): Merge from
23137         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23138         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23139         nox64 isa attributes.
23141 2013-03-18  Richard Biener  <rguenther@suse.de>
23143         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23144         (unite): Likewise.
23145         (merge_node_constraints): Likewise.
23146         (build_succ_graph): Likewise.
23147         (valid_graph_edge): Inline into single caller.
23148         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23149         and cache varinfo.
23150         (scc_visit): Fix formatting and variable use.
23151         (do_sd_constraint): Use gcc_checking_assert.
23152         (do_ds_constraint): Likewise.
23153         (do_complex_constraint): Likewise.
23154         (condense_visit): Likewise.  Cleanup.
23155         (dump_pred_graph): New function.
23156         (perform_var_substitution): Dump the pred-graph before
23157         variable substitution.
23158         (find_equivalent_node): Use gcc_checking_assert.
23159         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23161 2013-03-18  Richard Biener  <rguenther@suse.de>
23163         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23164         Remove cond_expr_stmt_list argument and do not gimplify the
23165         built expression.
23166         (vect_loop_versioning): Adjust.
23167         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23168         Cleanup to use less temporaries.
23169         (vect_create_data_ref_ptr): Cleanup.
23171 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23173         PR tree-optimization/56635
23174         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23175         require types_compatible_p types.
23177 2013-03-18  Nick Clifton  <nickc@redhat.com>
23179         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23180         spurious backslash.
23182         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23183         Add missing line to comment describing function.
23185 2013-03-18  Richard Biener  <rguenther@suse.de>
23187         PR tree-optimization/56210
23188         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23189         Handle string / character search functions.
23190         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23192 2013-03-18  Richard Biener  <rguenther@suse.de>
23194         PR middle-end/56483
23195         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23196         and implement properly.
23197         * gimple.h (gimple_cond_single_var_p): Remove.
23199 2013-03-18  Richard Biener  <rguenther@suse.de>
23201         * tree-data-ref.h (find_data_references_in_loop): Declare.
23202         * tree-data-ref.c (get_references_in_stmt): Use a stack
23203         vector pre-allocated in the callers.
23204         (find_data_references_in_stmt): Adjust.
23205         (graphite_find_data_references_in_stmt): Likewise.
23206         (create_rdg_vertices): Likewise.
23207         (find_data_references_in_loop): Export.
23208         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23209         Compute dependences here...
23210         (vect_analyze_data_refs): ...not here.  When we encounter
23211         a non-vectorizable data reference in basic-block vectorization
23212         truncate the data reference vector.  Do not bother to
23213         fixup data-dependence information for gather loads.
23214         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23215         of data references, as reported.
23217 2013-03-18  Richard Biener  <rguenther@suse.de>
23219         PR tree-optimization/3713
23220         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23221         has_constants and expr.
23222         (stmt_has_constants): Properly valueize SSA names when deciding
23223         whether the stmt has constants.
23225 2013-03-18  Richard Biener  <rguenther@suse.de>
23227         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23228         whole function when there is nothing to do.
23229         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23230         * tree-vectorizer.c (vectorize_loops): Update virtual and
23231         loop-closed SSA once.
23232         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23234 2013-03-18  Richard Biener  <rguenther@suse.de>
23236         PR middle-end/56113
23237         * domwalk.c (bb_postorder): New global static.
23238         (cmp_bb_postorder): New function.
23239         (walk_dominator_tree): Replace scheme imposing an order for
23240         visiting dominator sons by one sorting them at the time they
23241         are pushed on the stack.
23243 2013-03-18  Richard Biener  <rguenther@suse.de>
23245         PR tree-optimization/39326
23246         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23247         (struct mem_ref): Replace mem member with ao_ref typed member.
23248         (MEM_ANALYZABLE): Adjust.
23249         (memref_eq): Likewise.
23250         (mem_ref_alloc): Likewise.
23251         (gather_mem_refs_stmt): Likewise.
23252         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23253         (execute_sm_if_changed_flag_set): Adjust.
23254         (execute_sm): Likewise.
23255         (ref_always_accessed_p): Likewise.
23256         (refs_independent_p): Likewise.
23257         (can_sm_ref_p): Likewise.
23259 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23261         PR c/56566
23262         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23263         return 1 even for !unsignedp.
23265 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23267         * config/i386/i386.md (isa): Add x64 and nox64.
23268         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23269         (*pushtf): Enable *roF alternative for x64 isa only.
23270         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23271         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23272         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23273         (*movtf_internal): Merge from *movtf_internal_rex64 and
23274         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23275         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23276         nox64 isa attributes.
23277         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23278         nox64 isa attributes.
23279         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23281 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23283         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23285 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23287         PR target/56640
23288         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23289         class names.  Remove trailing comma after "ALL_REGS".
23291 2013-03-16  Jan Hubicka  <jh@suse.cz>
23293         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23294         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23295         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23296         of cgraph_get_create_node.
23297         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23299 2013-03-16  Jason Merrill  <jason@redhat.com>
23301         PR debug/49090
23302         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23303         with DW_AT_default_value.
23305 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23307         * BASE-VER: Set to 4.9.0.
23309 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23311         PR target/56619
23312         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23313         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23314         Document _x* TSX intrinsics.
23316 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23317             David Holsgrove  <david.holsgrove@xilinx.com>
23319         * configure.ac: Add MicroBlaze TLS support detection.
23320         * configure: Regenerate.
23321         * config/microblaze/microblaze-protos.h
23322         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23323         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23324         * config/microblaze/microblaze.c (microblaze_address_type): Add
23325         ADDRESS_TLS and tls_reloc address types.
23326         (microblaze_address_info): Add tls_reloc.
23327         (TARGET_HAVE_TLS): Define.
23328         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23329          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23330          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23331         load_tls_operand,  microblaze_call_tls_get_addr,
23332         microblaze_legitimize_tls_address): New functions.
23333         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23334         (get_base_reg): Use microblaze_tls_symbol_p.
23335         (microblaze_classify_address): Handle TLS.
23336         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23337         label_mentioned_p and microblaze_tls_referenced_p.
23338         (microblaze_legitimize_address): Handle TLS.
23339         (microblaze_address_insns): Handle ADDRESS_TLS.
23340         (pic_address_needs_scratch): Handle TLS.
23341         (print_operand_address): Handle TLS.
23342         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23343         (microblaze_expand_move): Handle TLS.
23344         (microblaze_legitimate_constant_p): Check
23345         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23346         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23347         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23348         (PIC_OFFSET_TABLE_REGNUM): Set.
23349         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23350         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23351         (addsi3, movsi_internal2, movdf_internal): Update constraints
23352         * config/microblaze/predicates.md (arith_plus_operand): Define
23353         (move_operand): Redefine as move_src_operand,
23354         check microblaze_tls_referenced_p.
23356 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23358         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23359         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23361 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23363         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23364         CC mode for AND.
23366 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23368         PR tree-optimization/53265
23369         * common.opt (Waggressive-loop-optimizations): New option.
23370         * tree-ssa-loop-niter.c: Include tree-pass.h.
23371         (do_warn_aggressive_loop_optimizations): New function.
23372         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23373         if number_of_latch_executions returned constant.
23374         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23375         early.  If number_of_latch_executions returned constant, set
23376         nb_iterations_upper_bound back to it.
23377         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23378         field.
23379         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23380         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23382         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23383         (MULTILIB_OSDIRNAMES): Set.
23384         * genmultilib: If defaultosdirname doesn't start with :: , set
23385         defaultosdirname2 instead, clear it and emit two . multilib_raw
23386         entries instead of just one.
23388 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23390         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23391         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23392         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23393         (SUBTARGET_OVERRIDE_OPTIONS): New.
23395 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23397         PR target/49880
23398         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23399         (musermode): Convert to Var(TARGET_USERMODE).
23400         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23401         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23402         * config/sh/sh.c (sh_option_override): Use
23403         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23404         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23405         condition.
23406         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23407         TARGET_SH4.
23408         (udivsi3_i4_single, divsi3_i4_single): Use
23409         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23411 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23413         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23414         default setting.
23416 2013-03-13  Richard Biener  <rguenther@suse.de>
23418         PR tree-optimization/56608
23419         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23420         calls when vectorizing basic-blocks.
23422 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23424         PR plugins/45078
23425         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23426         tm_file.
23428 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23430         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23432 2013-03-11  Jan Hubicka  <jh@suse.cz>
23434         PR lto/56557
23435         * lto-streamer-out.c (output_symbol_p): Skip references from
23436         constructors of external variables.
23438 2013-03-11  Jan Hubicka  <jh@suse.cz>
23440         PR middle-end/56571
23441         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23442         from pseudos.
23443         * emit-rtl.c (verify_rtx_sharing): Likewise.
23444         (copy_insn_1): Likewise.
23445         * rtl.c (copy_rtx): Likewise.
23447 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23449         PR target/56591
23450         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23451         output_operand_lossage message.
23453 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23455         PR target/56470
23456         * arm.c (shift_op): Validate RTL pattern on the fly.
23457         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23458         the RTL.
23460 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23462         PR target/56347
23463         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23464         new call patterns that clobber %fr12.
23465         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23466         split and postreload patterns.
23467         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23468         registers %fr12 and %fr12R as call used.
23470 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23472         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23473         (canon_address, record_store, replace_read, check_mem_read_rtx,
23474         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23475         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23476         rest_of_handle_dse): Likewise.
23478 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23480         PR middle-end/56524
23481         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23482         Add base_optabs.
23483         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23484         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23485         (save_optabs_if_changed): Replace with...
23486         (init_tree_optimization_optabs): ...this.
23487         * optabs.c (save_optabs_if_changed): Rename to...
23488         (init_tree_optimization_optabs): ...this.  Take the optimization node
23489         as argument.  Do nothing if the base optabs are already correct.
23490         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23491         to recompute optabs.
23492         * function.h (function): Remove optabs field.
23493         * function.c (invoke_set_current_function_hook): Call
23494         init_tree_optimization_optabs.  Use the result to initialize
23495         this_fn_optabs.
23497 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23499         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23500         if GTMA_HAS_NO_INSTRUMENTATION.
23501         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23502         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23503         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23504         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23505         GTMA_HAS_NO_INSTRUMENTATION.
23507 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23509         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23510         libasan_preinit.o.
23512 2013-03-08  Marek Polacek  <polacek@redhat.com>
23513             Jakub Jelinek  <jakub@redhat.com>
23515         PR tree-optimization/56478
23516         * predict.c (is_comparison_with_loop_invariant_p): Change the
23517         type of loop_step to tree.
23518         (predict_loops): Adjust.
23519         (predict_iv_comparison): Perform the computations on double_ints.
23521 2013-03-08  Richard Biener  <rguenther@suse.de>
23523         PR tree-optimization/56570
23524         * tree-cfg.c (verify_expr_location_1): Verify locations for
23525         DECL_DEBUG_EXPR.
23526         * tree-sra.c (create_access_replacement): Strip locations
23527         from DECL_DEBUG_EXPRs.
23529 2013-03-08  Richard Biener  <rguenther@suse.de>
23531         * tree-inline.c (expand_call_inline): Do not associate
23532         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23533         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23535 2013-03-08  Richard Biener  <rguenther@suse.de>
23537         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23538         or block changes with -Og.  Fix for location / block encoding
23539         changes and PHI arguments with locations.
23541 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23543         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23544         for all counters.
23545         (struct output_info): Likewise.
23546         (register_overhead): Remove bad gcc_assert.
23547         (bitmap_find_bit): If there is only a single bitmap element, do not
23548         count a miss as a search.
23549         (print_statistics): Update for counter type changes.
23550         (dump_bitmap_statistics): Likewise.  Print headers such that they
23551         are properly lined up with the printed counters.
23553 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23555         PR tree-optimization/56559
23556         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23557         check that it has only a single use.
23559 2013-03-07  Richard Biener  <rguenther@suse.de>
23561         * doc/invoke.texi (fwhole-program): Discourage use in combination
23562         with -flto.
23564 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23566         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23568         PR tree-optimization/56539
23569         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23570         instead of GSI_CONTINUE_LINKING as last argument to
23571         force_gimple_operand_gsi.  Adjust function comment.
23573         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23574         aarch64-cores.def.
23576         PR middle-end/56548
23577         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23578         promoted mode, convert the result back to the original mode.
23580 2013-03-06  Richard Biener  <rguenther@suse.de>
23582         PR middle-end/56294
23583         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23584         (insert_updated_phi_nodes_compare_uids): New function.
23585         (update_ssa): Sort symbols_to_rename after UID before
23586         traversing it to insert PHI nodes.
23588 2013-03-06  Richard Biener  <rguenther@suse.de>
23590         PR middle-end/50494
23591         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23592         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23594         Revert
23595         2013-02-13  Richard Biener  <rguenther@suse.de>
23597         PR lto/50494
23598         * varasm.c (output_constant_def_1): Get the decl representing
23599         the constant as argument.
23600         (output_constant_def): Wrap output_constant_def_1.
23601         (make_decl_rtl): Use output_constant_def_1 with the decl
23602         representing the constant.
23603         (build_constant_desc): Optionally re-use a decl already
23604         representing the constant.
23605         (tree_output_constant_def): Adjust.
23607 2013-03-06  Joey Ye  <joey.ye@arm.com>
23609         PR lto/50293
23610         * gcc.c (convert_white_space): New function.
23611         (main): Handles white space in function name.
23613 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23615         PR target/56529
23616         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23617         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23618         to SH_DIV_CALL_TABLE for TARGET_SH2.
23619         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23620         list.
23621         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23622         call-table options.
23624 2013-03-05  Sterling Augustine  <saugustine@google.com>
23625             Cary Coutant  <ccoutant@google.com>
23627         PR debug/55364
23628         * dwarf2out.c (resolve_addr): Don't call
23629         remove_loc_list_addr_table_entries a second time for the same
23630         expression.
23632 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23634         PR debug/56510
23635         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23636         (avoid_complex_debug_insns): New function.
23637         (expand_debug_locations): Call it.
23639         PR rtl-optimization/56484
23640         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23641         lifetimes of hard registers on small register class machines.
23643 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23645         * config/microblaze/microblaze-protos.h: Rename
23646         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23647         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23648         fast_interrupt.
23649         (microblaze_fast_interrupt_function_p): New function.
23650         (microblaze_is_interrupt_handler): Rename to
23651         microblaze_is_interrupt_variant and add fast_interrupt check.
23652         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23653         (save_restore_insns): Likewise.
23654         (compute_frame_size): Likewise.
23655         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23656         (microblaze_globalize_label): Likewise.
23657         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23658         * config/microblaze/microblaze.md: Use wrapper
23659         microblaze_is_interrupt_variant.
23661 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23663         * sdbout.c (sdbout_one_type): Switch to current function's section
23664         supporting cold/hot.
23666 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23668         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23669         -mxl-reorder.
23671 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23673         PR middle-end/56461
23674         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23675         if VALGRIND_GET_VBITS is defined, temporarily make object
23676         memory all defined, and restore previous valgrind addressability
23677         and definability afterwards.  Free this_object at the end.
23679         PR middle-end/56461
23680         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23681         right before calling lra_create_live_ranges, also call it
23682         when clearing live_p.  Only call lra_clear_live_ranges
23683         at the end if live_p.
23685         PR middle-end/56461
23686         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23688 2013-03-05  Richard Biener  <rguenther@suse.de>
23690         PR tree-optimization/56521
23691         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23692         value-id.
23694 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23696         PR c++/55135
23697         * except.h (remove_unreachable_eh_regions): New prototype.
23698         * except.c (remove_eh_handler_splicer): New function, split out
23699         of remove_eh_handler.
23700         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23701         warning about running it on many EH regions one at a time.
23702         (remove_unreachable_eh_regions_worker): New function, walk the
23703         EH tree in depth-first order and remove non-marked regions.
23704         (remove_unreachable_eh_regions): New function.
23705         * tree-eh.c (mark_reachable_handlers): New function, split out
23706         from remove_unreachable_handlers.
23707         (remove_unreachable_handlers): Use mark_reachable_handlers and
23708         remove_unreachable_eh_regions.
23709         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23710         and remove_unreachable_eh_regions.
23712 2013-03-05  Richard Biener  <rguenther@suse.de>
23714         PR middle-end/56525
23715         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23716         not freeing them until the end.
23718 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23720         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23722 2013-03-05  Richard Biener  <rguenther@suse.de>
23724         PR tree-optimization/56270
23725         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23726         of loads after scheduling an SLP instance.
23728 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23730         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23731         tic6x.exp.
23732         (check_gcc_parallelize): Run guality.exp as a separate job from
23733         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23734         struct-layout-1.exp with stackalign.exp.
23736         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23738         PR middle-end/56461
23739         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23740         load_index sbitmap even if some bit in it isn't set.
23742         PR middle-end/56461
23743         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23744         (discover_iteration_bound_by_body_walk): Change queues to
23745         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23746         spelling in comment.  Call safe_push on queues[bound_index] directly.
23747         Release queues[queue_index] in every iteration unconditionally.
23748         Release bounds vector.
23750         PR middle-end/56461
23751         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23752         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23753         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23754         inner_phis vector.
23756 2013-03-05  Richard Biener  <rguenther@suse.de>
23758         PR lto/56515
23759         * tree-inline.c (remap_blocks_to_null): New function.
23760         (expand_call_inline): When expanding a call stmt without
23761         an associated block inline remap all callee blocks to NULL.
23763 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23765         PR rtl-optimization/56494
23766         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23767         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23768         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23770         PR middle-end/56461
23771         * sel-sched-ir.c (free_sched_pools): Release
23772         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23773         if succs_info_pool.max_top isn't -1.
23775         PR bootstrap/56509
23776         * opts.c (opts_obstack, opts_concat): Moved to...
23777         * opts-common.c (opts_obstack, opts_concat): ... here.
23779 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23781         PR middle-end/56461
23782         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23784 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23786         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23787         all appropriate places.
23789 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23791         PR tree-optimization/56424
23792         * ipa-split.c (split_function): Do not set the RSO flag if result is
23793         not by reference and its type is a register type.
23795 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23797         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23798         (microblaze_legitimate_pic_operand): Likewise
23799         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23800         new function microblaze_legitimate_pic_operand
23801         * config/microblaze/microblaze-protos.h
23802         (microblaze_legitimate_pic_operand): Declare.
23804 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23806         * config/microblaze/predicates.md (call_insn_simple_operand):
23807         New predicate for supported rtx code types.
23808         * config/microblaze/microblaze.md (call_internal1): Use
23809         call_insn_simple_operand predicate.
23811 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23813         PR middle-end/56461
23814         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23815         partitions.ordered_remove.
23817         PR middle-end/56461
23818         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23819         vec_oprnds0.create (1) for modifier == NONE.
23821         PR middle-end/56461
23822         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23823         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23824         vec_oprnds1 right before pushing anything to it for
23825         scalar_shift_arg.
23827         PR middle-end/56461
23828         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23829         set nbbs to 0 instead of having separate code path.
23830         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23831         instead of false as last argument if returning NULL.
23833 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23835         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23836         the attribute is now called "target" instead of "option".
23837         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23838         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23839         attribute/pragma name for TARGET_OPTION_VALID_P and
23840         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23841         * doc/tm.texi: Regenerated.
23843 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23845         * config/microblaze/microblaze.c:
23846         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23847         * config/microblaze/microblaze.h: Add -mxl-reorder to
23848         DRIVER_SELF_SPECS.
23849         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23850         instructions emitted if TARGET_REORDER.
23851         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23852         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23853         separately.
23855 2013-03-01  Xinliang David Li  <davidxl@google.com>
23857         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23858         walk length.
23860 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23862         PR middle-end/56461
23863         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23864         vector even when returning true.  Fix up function comment formatting.
23866         PR middle-end/56461
23867         * ira-build.c (ira_loop_nodes_count): New variable.
23868         (create_loop_tree_nodes): Initialize it.
23869         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23871         PR middle-end/56461
23872         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23873         method on dr_chain and result_chain.
23874         * tree-vect-stmts.c (vectorizable_store): Only call
23875         result_chain.create if j == 0.
23877         PR middle-end/56461
23878         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23879         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23880         before overwriting it.
23882 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23884         * doc/extended.texi (C Extensions): Change order in @menu
23885         to match @node.
23886         (Other MIPS Built-in Functions): Move last MIPS entry before
23887         "picoChip Built-in Functions".
23888         (SH Built-in Functions): Move after RX Built-in Functions.
23889         * doc/gcc.texi (Introduction): Change order in @menu
23890         to match @node.
23891         * doc/md.texi (Constraints): Ditto.
23892         * gty.texi (Type Information): Ditto.
23893         (User-provided marking routines for template types): Make
23894         subsection.
23895         * doc/invoke.texi (AArch64 Options): Move before
23896         "Adapteva Epiphany Options".
23898 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23899             Jakub Jelinek  <jakub@redhat.com>
23901         PR sanitizer/56454
23902         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23903         no_address_safety_analysis attribute.
23904         * doc/extend.texi (no_address_safety_attribute): Rename to
23905         no_sanitize_address attribute, mention no_address_safety_analysis
23906         attribute as deprecated alias.
23908 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23910         PR middle-end/56461
23911         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23912         type to vec<vec<tree> > *.
23913         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23914         to be vec<tree> instead of vec<tree> *, set vec_defs
23915         to vNULL and call vec_defs.create (number_of_vects), adjust other
23916         uses of vec_defs.
23917         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23918         vectorizable_condition): Adjust vect_get_slp_defs callers.
23920 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23922         * config/aarch64/aarch64.c
23923         (aarch64_float_const_representable): Remove unused variable.
23925 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23927         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23929 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23931         * config/aarch64/aarch64-builtins.c
23932         (aarch64_init_simd_builtins): Make static.
23934 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23936         * config/aarch64/aarch64.c
23937         (aarch64_simd_make_constant): Make static.
23939 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23941         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23942         with no initialization to the RHS of debug statements.
23944 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23946         PR tree-optimization/56294
23947         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23948         Adjust dumping.
23949         (get_access_replacement): Do not call create_access_replacement.
23950         Assert a replacement exists.
23951         (get_repl_default_def_ssa_name): Create the replacement declaration
23952         itself.
23954 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23956         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23957         final_end_function.
23959 2013-02-28  Marek Polacek  <polacek@redhat.com>
23961         PR rtl-optimization/56466
23962         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23963         if we're changing a loop.
23964         (peel_loops_completely): Likewise.
23966 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23968         PR c++/55813
23969         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23971 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23973         PR target/56445
23974         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23975         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23976         INTX_FTYPE_FX, FX_FTYPE_INTX.
23977         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23979 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23981         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23982         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23983         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23984         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23985         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23986         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23987         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23988         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23989         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23990         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23991         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23992         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23993         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23994         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23995         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23996         (avrxmega6): Increase max flash segments from 5 to 6.
23997         * config/avr/t-multilib: Regenerate.
23998         * config/avr/avr-tables.opt: Regenerate.
23999         * doc/avr-mmcu.texi: Regenerate.
24001 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24003         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
24004         (avr_device_to_arch): Rename to avr_device_to_ld.
24005         (avr_device_to_as): New prototype.
24006         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
24007         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
24008         * config/avr/driver-avr.c (avr_device_to_as): New.
24009         (avr_device_to_arch): Rename to avr_device_to_ld.
24011 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24013         PR middle-end/56461
24014         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
24015         method on dr_chain and result_chain.
24017         PR middle-end/56461
24018         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
24019         pointer_set_destroy on not_executed_last_iteration.
24021         PR middle-end/56461
24022         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
24024         PR middle-end/56461
24025         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
24026         FOR_EACH_DEFINED_FUNCTION when freeing state.
24028         PR middle-end/56461
24029         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
24030         pool_free.
24031         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
24032         overwriting it.
24034         PR middle-end/56461
24035         * ipa-cp.c (decide_whether_version_node): Call vec_free on
24036         known_aggs[i].items and release known_aggs vector.
24038         PR middle-end/56461
24039         * ipa-reference.c (propagate): Free node_info even for alias nodes.
24041 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24043         * config/microblaze/microblaze.c (microblaze_emit_compare):
24044         Use xor for EQ/NE comparisions.
24045         * config/microblaze/microblaze.md (cstoresf4): Add constraints
24046         (cbranchsf4): Adjust operator to comparison_operator.
24048 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24050         PR middle-end/56461
24051         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
24052         vector.
24053         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
24054         vec_safe_push, always update *slot.
24055         (redirect_edge_var_map_clear): Use vec_free.
24056         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
24057         (free_var_map_entry): Use vec_free.
24058         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
24059         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
24061 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
24063         PR middle-end/45472
24064         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
24065         when the may_trap_p bit of the exprs being merged differs.
24066         Reorder tests for speculativeness in the logical and operator.
24068 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24070         * incpath.c (add_standard_paths): Use reconcat instead of concat
24071         where appropriate and avoid leaking memory.
24073         * opts.h: Include obstack.h.
24074         (opts_concat): New prototype.
24075         (opts_obstack): New declaration.
24076         * opts.c (opts_concat): New function.
24077         (opts_obstack): New variable.
24078         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24079         (finish_options): Use opts_concat instead of concat
24080         and XOBNEWVEC instead of XNEWVEC.
24081         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24082         generate_option): Likewise.
24083         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24084         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24086         PR target/56455
24087         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24088         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24090 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24092         PR middle-end/56461
24093         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24095 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24097         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24098         (arm_block_move_unaligned_straight): Likewise.
24099         (arm_adjust_block_mem): Likewise.
24101 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24103         PR target/48901
24104         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24105         temp, cond and label.
24106         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24108         PR target/52500
24109         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24110         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24112         PR target/52501
24113         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24114         of prologue/epilogue functions.
24116         PR target/52550
24117         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24118         Remove unused variable cfa_offset.
24119         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24121         PR target/54639
24122         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24123         type promotion to unsigned.
24125         PR target/54640
24126         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24127         for HOST_WIDE_INT of 32 bit / same size as int.
24128         (arm_block_move_unaligned_straight): Likewise.
24129         (arm_adjust_block_mem): Likewise.
24131         PR target/54662
24132         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24133         ALL_CFLAGS.
24135 2013-02-26  Marek Polacek  <polacek@redhat.com>
24137         PR tree-optimization/56426
24138         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24140 2013-02-26  Richard Biener  <rguenther@suse.de>
24142         PR target/56444
24143         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24144         unused variable loops.
24146 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24148         PR tree-optimization/56448
24149         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24150         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24151         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24152         later operands of the references, or even first operand for
24153         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24155         PR tree-optimization/56443
24156         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24157         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24158         to type_for_mode langhook.
24160 2013-02-25  Matt Turner  <mattst88@gmail.com>
24162         * doc/invoke.texi: Document r4700.
24164 2013-02-25  Richard Biener  <rguenther@suse.de>
24166         PR tree-optimization/56175
24167         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24168         split out from ...
24169         (simplify_bitwise_binary): ... here.  Also guard the conversion
24170         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24172 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24174         Revert:
24175         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24176             Maciej W. Rozycki  <macro@codesourcery.com>
24177             Tom de Vries  <tom@codesourcery.com>
24178             Nathan Sidwell  <nathan@codesourcery.com>
24179             Iain Sandoe  <iain@codesourcery.com>
24180             Nathan Froyd  <froydnj@codesourcery.com>
24181             Chao-ying Fu  <fu@mips.com>
24183         * doc/extend.texi: (micromips, nomicromips, nocompression):
24184         Document new function attributes.
24185         * doc/invoke.texi (minterlink-compressed, mmicromips,
24186         m14k, m14ke, m14kec): Document new options.
24187         (minterlink-mips16): Update documentation.
24188         * doc/md.texi (ZC, ZD): Document new constraints.
24189         * configure.ac (gcc_cv_as_micromips): Check if linker
24190         supports the .set micromips directive.
24191         * configure: Regenerate.
24192         * config.in: Regenerate.
24193         * config/mips/mips-tables.opt: Regenerate.
24194         * config/mips/micromips.md: New file.
24195         * constraints.md (ZC, AD): New constraints.
24196         * config/mips/predicates.md (movep_src_register): New predicate.
24197         (movep_src_operand): New predicate.
24198         (non_volatile_mem_operand): New predicate.
24199         * config/mips/mips.md (multimem): New type.
24200         (length): Differentiate between 17-bit and 18-bit branch offsets.
24201         (MOVEP1, MOVEP2): New mode iterator.
24202         (mov_<load>l): Use ZC constraint.
24203         (mov_<load>r): Likewise.
24204         (mov_<store>l): Likewise.
24205         (mov_<store>r): Likewise.
24206         (*branch_equality<mode>_inverted): Add microMIPS support.
24207         (*branch_equality<mode>): Likewise.
24208         (*jump_absolute): Likewise.
24209         (indirect_jump_<mode>): Likewise.
24210         (tablejump_<mode>): Likewise.
24211         (<optab>_internal): Likewise.
24212         (sibcall_internal): Likewise.
24213         (sibcall_value_internal): Likewise.
24214         (prefetch): Use constraint ZD.
24215         * config/mips/mips.opt (minterlink-compressed): New option.
24216         (minterlink-mips16): Now an alias for minterlink-compressed.
24217         (mmicromips): New option.
24218         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24219         (compare_and_swap_12): Likewise.
24220         (sync_add<mode>): Likewise.
24221         (sync_<optab>_12): Likewise.
24222         (sync_old_<optab>_12): Likewise.
24223         (sync_new_<optab>_12): Likewise.
24224         (sync_nand_12): Likewise.
24225         (sync_old_nand_12): Likewise.
24226         (sync_new_nand_12): Likewise.
24227         (sync_sub<mode>): Likewise.
24228         (sync_old_add<mode>): Likewise.
24229         (sync_old_sub<mode>): Likewise.
24230         (sync_new_add<mode>): Likewise.
24231         (sync_new_sub<mode>): Likewise.
24232         (sync_<optab><mode>): Likewise.
24233         (sync_old_<optab><mode>): Likewise.
24234         (sync_new_<optab><mode>): Likewise.
24235         (sync_nand<mode>): Likewise.
24236         (sync_old_nand<mode>): Likewise.
24237         (sync_new_nand<mode>): Likewise.
24238         (sync_lock_test_and_set<mode>): Likewise.
24239         (test_and_set_12): Likewise.
24240         (atomic_compare_and_swap<mode>): Likewise.
24241         (atomic_exchange<mode>_llsc): Likewise.
24242         (atomic_fetch_add<mode>_llsc): Likewise.
24243         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24244         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24245         (umips_save_restore_pattern_p): Likewise.
24246         (umips_load_store_pair_p): Likewise.
24247         (umips_output_load_store_pair): Likewise.
24248         (umips_movep_target_p): Likewise.
24249         (umips_12bit_offset_address_p): Likewise.
24250         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24251         (mips_base_mips16): Rename this...
24252         (mips_base_compression_flags): ...to this. Update all uses.
24253         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24254         (mips_mips16_decl_p): Delete.
24255         (mips_nomips16_decl_p): Delete.
24256         (mips_get_compress_on_flags): New function.
24257         (mips_get_compress_off_flags): New function.
24258         (mips_get_compress_mode): New function.
24259         (mips_get_compress_on_name): New function.
24260         (mips_get_compress_off_name): New function.
24261         (mips_insert_attributes): Support multiple compression types.
24262         (mips_merge_decl_attributes): Likewise.
24263         (umips_12bit_offset_address_p): New function.
24264         (mips_start_function_definition): Emit .set micromips directive.
24265         (mips_call_may_need_jalx_p): New function.
24266         (mips_function_ok_for_sibcall): Add microMIPS support.
24267         (mips_print_operand_punctuation): Support short delay slots and
24268         compact jumps.
24269         (umips_swm_mask, umips_swm_encoding): New.
24270         (umips_build_save_restore): New function.
24271         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24272         (was_mips16_p): Remove.
24273         (old_compression_mode): New.
24274         (mips_set_compression_mode): New function.
24275         (mips_set_current_function): Add microMIPS support.
24276         (mips_option_override): Likewise.
24277         (umips_save_restore_pattern_p): New function.
24278         (umips_output_save_restore): New function.
24279         (umips_load_store_pair_p_1): New function.
24280         (umips_load_store_pair_p): New function.
24281         (umips_output_load_store_pair_1): New function.
24282         (umips_output_load_store_pair): New function.
24283         (umips_movep_target_p) New function.
24284         (mips_prepare_pch_save): Add microMIPS support.
24285         * config/mips/mips.h (TARGET_COMPRESSION): New.
24286         (TARGET_CPU_CPP_BUILTINS): Update macro
24287         to use new compression flags and to support microMIPS.
24288         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24289         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24290         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24291         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24292         (ASM_SPEC): Support mmicromips and mno-micromips.
24293         (M16STORE_REG_P): New macro.
24294         (MIPS_CALL): Support TARGET_MICROMIPS.
24295         (MICROMIPS_J): New macro.
24296         (mips_base_mips16): Rename this...
24297         (mips_base_compression_flags): ...to this.
24298         (UMIPS_12BIT_OFFSET_P): New macro.
24299         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24300         (MULTILIB_DIRNAMES): Likewise.
24302 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24304         PR rtl-optimization/56131
24305         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24306         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24307         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24309 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24311         * doc/invoke.texi (-fsanitize=): Move from optimization
24312         to debugging options.
24314 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24316         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24318 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24319             Alexander Monakov  <amonakov@ispras.ru>
24321         PR middle-end/56077
24322         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24323         flush pending lists also on non-jumps.  Adjust comment.
24325 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24326             Maciej W. Rozycki  <macro@codesourcery.com>
24327             Tom de Vries  <tom@codesourcery.com>
24328             Nathan Sidwell  <nathan@codesourcery.com>
24329             Iain Sandoe  <iain@codesourcery.com>
24330             Nathan Froyd  <froydnj@codesourcery.com>
24331             Chao-ying Fu  <fu@mips.com>
24333         * doc/extend.texi: (micromips, nomicromips, nocompression):
24334         Document new function attributes.
24335         * doc/invoke.texi (minterlink-compressed, mmicromips,
24336         m14k, m14ke, m14kec): Document new options.
24337         (minterlink-mips16): Update documentation.
24338         * doc/md.texi (ZC, ZD): Document new constraints.
24339         * configure.ac (gcc_cv_as_micromips): Check if linker
24340         supports the .set micromips directive.
24341         * configure: Regenerate.
24342         * config.in: Regenerate.
24343         * config/mips/mips-tables.opt: Regenerate.
24344         * config/mips/micromips.md: New file.
24345         * constraints.md (ZC, AD): New constraints.
24346         * config/mips/predicates.md (movep_src_register): New predicate.
24347         (movep_src_operand): New predicate.
24348         (non_volatile_mem_operand): New predicate.
24349         * config/mips/mips.md (multimem): New type.
24350         (length): Differentiate between 17-bit and 18-bit branch offsets.
24351         (MOVEP1, MOVEP2): New mode iterator.
24352         (mov_<load>l): Use ZC constraint.
24353         (mov_<load>r): Likewise.
24354         (mov_<store>l): Likewise.
24355         (mov_<store>r): Likewise.
24356         (*branch_equality<mode>_inverted): Add microMIPS support.
24357         (*branch_equality<mode>): Likewise.
24358         (*jump_absolute): Likewise.
24359         (indirect_jump_<mode>): Likewise.
24360         (tablejump_<mode>): Likewise.
24361         (<optab>_internal): Likewise.
24362         (sibcall_internal): Likewise.
24363         (sibcall_value_internal): Likewise.
24364         (prefetch): Use constraint ZD.
24365         * config/mips/mips.opt (minterlink-compressed): New option.
24366         (minterlink-mips16): Now an alias for minterlink-compressed.
24367         (mmicromips): New option.
24368         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24369         (compare_and_swap_12): Likewise.
24370         (sync_add<mode>): Likewise.
24371         (sync_<optab>_12): Likewise.
24372         (sync_old_<optab>_12): Likewise.
24373         (sync_new_<optab>_12): Likewise.
24374         (sync_nand_12): Likewise.
24375         (sync_old_nand_12): Likewise.
24376         (sync_new_nand_12): Likewise.
24377         (sync_sub<mode>): Likewise.
24378         (sync_old_add<mode>): Likewise.
24379         (sync_old_sub<mode>): Likewise.
24380         (sync_new_add<mode>): Likewise.
24381         (sync_new_sub<mode>): Likewise.
24382         (sync_<optab><mode>): Likewise.
24383         (sync_old_<optab><mode>): Likewise.
24384         (sync_new_<optab><mode>): Likewise.
24385         (sync_nand<mode>): Likewise.
24386         (sync_old_nand<mode>): Likewise.
24387         (sync_new_nand<mode>): Likewise.
24388         (sync_lock_test_and_set<mode>): Likewise.
24389         (test_and_set_12): Likewise.
24390         (atomic_compare_and_swap<mode>): Likewise.
24391         (atomic_exchange<mode>_llsc): Likewise.
24392         (atomic_fetch_add<mode>_llsc): Likewise.
24393         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24394         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24395         (umips_save_restore_pattern_p): Likewise.
24396         (umips_load_store_pair_p): Likewise.
24397         (umips_output_load_store_pair): Likewise.
24398         (umips_movep_target_p): Likewise.
24399         (umips_12bit_offset_address_p): Likewise.
24400         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24401         (mips_base_mips16): Rename this...
24402         (mips_base_compression_flags): ...to this. Update all uses.
24403         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24404         (mips_mips16_decl_p): Delete.
24405         (mips_nomips16_decl_p): Delete.
24406         (mips_get_compress_on_flags): New function.
24407         (mips_get_compress_off_flags): New function.
24408         (mips_get_compress_mode): New function.
24409         (mips_get_compress_on_name): New function.
24410         (mips_get_compress_off_name): New function.
24411         (mips_insert_attributes): Support multiple compression types.
24412         (mips_merge_decl_attributes): Likewise.
24413         (umips_12bit_offset_address_p): New function.
24414         (mips_start_function_definition): Emit .set micromips directive.
24415         (mips_call_may_need_jalx_p): New function.
24416         (mips_function_ok_for_sibcall): Add microMIPS support.
24417         (mips_print_operand_punctuation): Support short delay slots and
24418         compact jumps.
24419         (umips_swm_mask, umips_swm_encoding): New.
24420         (umips_build_save_restore): New function.
24421         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24422         (was_mips16_p): Remove.
24423         (old_compression_mode): New.
24424         (mips_set_compression_mode): New function.
24425         (mips_set_current_function): Add microMIPS support.
24426         (mips_option_override): Likewise.
24427         (umips_save_restore_pattern_p): New function.
24428         (umips_output_save_restore): New function.
24429         (umips_load_store_pair_p_1): New function.
24430         (umips_load_store_pair_p): New function.
24431         (umips_output_load_store_pair_1): New function.
24432         (umips_output_load_store_pair): New function.
24433         (umips_movep_target_p) New function.
24434         (mips_prepare_pch_save): Add microMIPS support.
24435         * config/mips/mips.h (TARGET_COMPRESSION): New.
24436         (TARGET_CPU_CPP_BUILTINS): Update macro
24437         to use new compression flags and to support microMIPS.
24438         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24439         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24440         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24441         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24442         (ASM_SPEC): Support mmicromips and mno-micromips.
24443         (M16STORE_REG_P): New macro.
24444         (MIPS_CALL): Support TARGET_MICROMIPS.
24445         (MICROMIPS_J): New macro.
24446         (mips_base_mips16): Rename this...
24447         (mips_base_compression_flags): ...to this.
24448         (UMIPS_12BIT_OFFSET_P): New macro.
24449         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24450         (MULTILIB_DIRNAMES): Likewise.
24452 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24454         PR target/52555
24455         * target-globals.c (save_target_globals): For init_reg_sets and
24456         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24458 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24460         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24461         * config/aarch64/t-aarch64
24462         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24464 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24466         PR inline-asm/56148
24467         * lra-constraints.c (process_alt_operands): Reload operand
24468         conflicting with earlier clobber only if no more other conflicting
24469         operands.
24471 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24473         PR sanitizer/56393
24474         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24475         if not linking a shared library.
24477 2013-02-22  Seth LaForge  <sethml@google.com>
24479         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24481 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24483         * config/arm/arm.md (split for extendsidi): Update condition.
24484         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24485         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24486         (qhs_zextenddi_cstr): Likewise.
24488 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24490         PR middle-end/56420
24491         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24492         avoid signed wrapping.
24493         (expand_mult): Handle properly multiplication by
24494         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24495         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24496         in the compiler if coeff is HOST_WIDE_INT_MIN.
24497         (expand_divmod): Don't make ext_op1 static, change it's type to
24498         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24500         PR rtl-optimization/50339
24501         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24502         field.
24503         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24504         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24505         into splitting_ashiftrt field.
24506         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24507         ASHIFTRT.
24508         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24509         choices.
24511 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24513         PR middle-end/56108
24514         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24515         are sure to go irrevocable.
24517 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24519         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24520         scalars are valid operands.
24522 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24524         PR tree-optimization/56310
24525         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24526         only matching indices and non-negative final offsets.
24527         (intersect_aggregates_with_edge): Pass src_idx to
24528         agg_replacements_to_vector.  Pass src_idx insstead of index to
24529         intersect_with_agg_replacements.
24531 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24533         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24534         instead of hard-wired defaults.
24536 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24538         * doc/invoke.texi (MIPS Options): Update documentation of the
24539         floating-point multiply-accumulate instruction restrictions.
24541 2013-02-21  Kostya Serebryany  <kcc@google.com>
24543         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24544         asan_shadow_offset on x86_64 linux.
24546 2013-02-21  Richard Biener  <rguenther@suse.de>
24548         PR tree-optimization/56415
24549         Revert
24550         2013-02-11  Richard Biener  <rguenther@suse.de>
24552         PR tree-optimization/56273
24553         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24554         first VRP run.
24556 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24558         PR bootstrap/56258
24559         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24560         instead of @itemx.
24562         PR inline-asm/56405
24563         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24564         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24566 2013-02-20  Jan Hubicka  <jh@suse.cz>
24568         PR tree-optimization/56265
24569         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24570         when target is referenced for first time.
24572 2013-02-20  Richard Biener  <rguenther@suse.de>
24574         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24575         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24576         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24577         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24578         not return anything.
24579         (rename_ssa_copies): Do not remove unused locals.
24580         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24581         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24582         * passes.c (execute_function_todo): Do not schedule unused locals
24583         removal if cleanup_tree_cfg did something.
24584         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24585         about the number of removed locals.
24587 2013-02-20  Richard Biener  <rguenther@suse.de>
24589         PR tree-optimization/56398
24590         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24592 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24594         PR tree-optimization/55334
24595         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24596         restricted pointers to arrays.
24598 2013-02-20  Richard Biener  <rguenther@suse.de>
24599             Jakub Jelinek  <jakub@redhat.com>
24601         PR tree-optimization/56396
24602         * tree-ssa-ccp.c (n_const_val): New static variable.
24603         (get_value): Return NULL for SSA names we don't have a lattice
24604         entry for.
24605         (ccp_initialize): Initialize n_const_val.
24606         * tree-ssa-copy.c (n_copy_of): New static variable.
24607         (init_copy_prop): Initialize n_copy_of.
24608         (get_value): Return NULL_TREE for SSA names we don't have a
24609         lattice entry for.
24611 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24613         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24615 2013-02-20  Richard Biener  <rguenther@suse.de>
24617         * genpreds.c (write_lookup_constraint): Do not compare first
24618         letter of the constraint again.
24620 2013-02-20  Richard Biener  <rguenther@suse.de>
24622         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24623         and ceil_log2.
24624         (get_use_iv_cost): Terminate hashtable walk when coming across
24625         an empty entry.
24627 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24629         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24630         reassociation for avx2 targets.
24632 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24634         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24635         Add version check for v8.10.a to enable microblaze_has_clz
24636         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24637         version and TARGET_PATTERN_COMPARE check
24638         * config/microblaze/microblaze.md: New clzsi2 instruction
24640 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24642         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24643         function before branching.
24645 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24647         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24648         DUMP_INSN_RTX_UID.
24649         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24651 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24653         PR middle-end/55889
24654         * sel-sched.c: Include ira.h.
24655         (implicit_clobber_conflict_p): New function.
24656         (moveup_expr): Use it.
24657         * Makefile.in (sel-sched.o): Depend on ira.h.
24659 2013-02-19  Richard Biener  <rguenther@suse.de>
24661         PR tree-optimization/56384
24662         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24663         (vn_hash_type): Split out from ...
24664         (vn_hash_constant_with_type): ... here.
24665         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24666         (vn_phi_eq): Compare types from vn_phi_s structure.
24667         (vn_phi_lookup): Populate vn_phi_s type.
24668         (vn_phi_insert): Likewise.
24670 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24672         PR tree-optimization/56350
24673         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24674         if haven't found reduction or nested cycle operand, rather than
24675         asserting we must find it.
24677         PR tree-optimization/56381
24678         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24679         to fold_build3.
24681 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24682             Jakub Jelinek  <jakub@redhat.com>
24684         PR target/52555
24685         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24686         (swap_optab_enable): Same.
24687         (init_all_optabs): Use argument instead of global.
24688         * tree.h (struct tree_optimization_option): New field target_optabs.
24689         * expr.h (init_all_optabs): Add argument to prototype.
24690         (TREE_OPTIMIZATION_OPTABS): New.
24691         (save_optabs_if_changed): Protoize.
24692         * optabs.h: Declare this_fn_optabs.
24693         * optabs.c (save_optabs_if_changed): New.
24694         Declare this_fn_optabs.
24695         (init_optabs): Add argument to init_all_optabs() call.
24696         * function.c (invoke_set_current_function_hook): Handle per
24697         function optabs.
24698         * function.h (struct function): New field optabs.
24699         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24700         optimization_current_node has changed.
24701         * target-globals.h (save_target_globals_default_opts): Protoize.
24702         * target-globals.c (save_target_globals_default_opts): New.
24704 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24706         PR target/56347
24707         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24708         registers %fr12 and %fr12R as call used.
24710         PR target/56214
24711         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24712         and HImode, require all displacements to be an integer multiple of
24713         their mode size.
24714         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24715         only allow QImode and HImode when reload is in progress and strict is
24716         true.  Likewise for symbolic addresses.  Use base14_operand to check
24717         displacements in REG+BASE addresses.
24719 2013-02-18  Richard Biener  <rguenther@suse.de>
24721         PR tree-optimization/56366
24722         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24723         handle sign-conversion of outer-loop initial induction value.
24725 2013-02-18  Richard Biener  <rguenther@suse.de>
24727         PR middle-end/56349
24728         * cfghooks.c (merge_blocks): If we merge a latch into another
24729         block adjust references to it.
24730         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24731         (verify_loop_structure): Verify that a recorded latch is in fact
24732         a latch.
24734 2013-02-18  Richard Biener  <rguenther@suse.de>
24736         PR tree-optimization/56321
24737         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24738         order SSA name release and virtual operand unlinking.
24740 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24742         * config/microblaze/microblaze.md (save_stack_block): Define.
24743         (restore_stack_block): Likewise.
24745 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24747         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24748         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24749         * config/microblaze/microblaze.c (microblaze_option_override):
24750         Bail out early for PIC modes when target does not support PIC.
24752 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24754         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24755         Replace with a microblaze version.
24756         (microblaze_trampoline_init): Adapt for microblaze.
24757         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24758         microblaze.
24760 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24761             Dodji Seketeli  <dodji@redhat.com>
24763         PR asan/56330
24764         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24765         (instrument_mem_region_access): Do not forget to always put
24766         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24767         0) statement, even for cases where either 'base' or 'base + len'
24768         are not instrumented -- because they have been previously
24769         instrumented.  Simplify the logic by putting all the statements
24770         instrument 'base + len' inside a sequence, and then insert that
24771         sequence right before the current insertion point.  Then, to
24772         instrument 'base + len', just get an iterator on that statement.
24773         And do not forget to update the pointer to iterator the function
24774         received as argument.
24776 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24778         PR rtl-optimization/56348
24779         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24781 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24783         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24784         (clean_graph_dump_file): Pass base to start_graph_dump.
24786 2013-02-14  Richard Henderson  <rth@redhat.com>
24788         PR target/55941
24789         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24791 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24793         * collect2-aix.h: Define F_LOADONLY.
24795 2013-02-14  Richard Biener  <rguenther@suse.de>
24797         PR lto/50494
24798         * varasm.c (output_constant_def_1): Get the decl representing
24799         the constant as argument.
24800         (output_constant_def): Wrap output_constant_def_1.
24801         (make_decl_rtl): Use output_constant_def_1 with the decl
24802         representing the constant.
24803         (build_constant_desc): Optionally re-use a decl already
24804         representing the constant.
24805         (tree_output_constant_def): Adjust.
24807 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24809         Fix an asan crash
24810         * asan.c (instrument_builtin_call):  Really put the length of the
24811         second source argument into src1_len.
24813 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24815         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24816         argument.  If it is false, don't create edge from then_bb to
24817         fallthru_bb.
24818         (insert_if_then_before_iter): Pass true to it.
24819         (build_check_stmt): Pass false to it.
24820         (transform_statements): Flush hash table only on extended basic
24821         block boundaries, rather than at the beginning of every bb.
24822         Don't flush hash table on nonfreeing_call_p calls.
24823         * tree-flow.h (nonfreeing_call_p): New prototype.
24824         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24826 2013-02-13  David S. Miller  <davem@davemloft.net>
24828         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24830 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24832         PR target/56184
24833         * ira.c (max_regno_before_ira): Move from ...
24834         (ira): ... here.
24835         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24836         vec_safe_length.
24838 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24840         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24842 2013-02-13  Richard Biener  <rguenther@suse.de>
24844         PR lto/56295
24845         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24846         globals in MEM_REFs.
24848 2013-02-13  Richard Biener  <rguenther@suse.de>
24850         * loop-init.c (loop_optimizer_init): Clear loop state when
24851         re-initializing preserved loops.
24852         * loop-unswitch.c (unswitch_single_loop): Return whether
24853         we unswitched the loop.  Do not verify loop state here.
24854         (unswitch_loops): When we unswitched a loop discover new loops.
24856 2013-02-13  Kostya Serebryany  <kcc@google.com>
24858         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24859         on x86_64 linux.
24860         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24862 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24864         Avoid instrumenting duplicated memory access in the same basic block
24865         * Makefile.in (asan.o): Add new dependency on hash-table.h
24866         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24867         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24868         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24869         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24870         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24871         (get_mem_ref_of_assignment): New functions.
24872         (get_mem_refs_of_builtin_call): Extract from
24873         instrument_builtin_call and tweak a little bit to make it fit with
24874         the new signature.
24875         (instrument_builtin_call): Use the new
24876         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24877         of is_gimple_builtin_call.
24878         (instrument_derefs, instrument_mem_region_access): Insert the
24879         instrumented memory reference into the hash table.
24880         (maybe_instrument_assignment): Renamed instrument_assignment into
24881         this, and change it to advance the iterator when instrumentation
24882         actually happened and return true in that case.  This makes it
24883         homogeneous with maybe_instrument_assignment, and thus give a
24884         chance to callers to be more 'regular'.
24885         (transform_statements): Clear the memory reference hash table
24886         whenever we enter a new BB, when we cross a function call, or when
24887         we are done transforming statements.  Use
24888         maybe_instrument_assignment instead of instrumentation.  No more
24889         need to special case maybe_instrument_assignment and advance the
24890         iterator after calling it; it's now handled just like
24891         maybe_instrument_call.  Update comment.
24893 2013-02-13  Richard Biener  <rguenther@suse.de>
24895         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24896         Fix loop discovery code.
24898 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24900         PR inline-asm/56148
24901         * lra-constraints.c (process_alt_operands): Match early clobber
24902         operand with itself.  Check conflicts with earlyclobber only if
24903         the operand is not reloaded.  Prefer to reload conflicting operand
24904         if earlyclobber and matching operands are the same.
24906 2013-02-12  Richard Biener  <rguenther@suse.de>
24908         PR lto/56297
24909         * lto-streamer-out.c (write_symbol): Do not output symbols
24910         for hard register variables.
24912 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24914         PR target/54222
24915         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24916         (umulsidi3_insn, mulsidi3_insn): New insns.
24918 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24920         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24921         (struct tune_params): Add vec_costs field.
24922         * config/arm/arm.c (arm_builtin_vectorization_cost)
24923         (arm_add_stmt_cost): New functions.
24924         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24925         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24926         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24927         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24928         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24929         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24930         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24932 2013-02-12  Richard Biener  <rguenther@suse.de>
24934         PR lto/56295
24935         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24936         decls again if possible.
24938 2013-02-12  Richard Biener  <rguenther@suse.de>
24940         PR middle-end/56288
24941         * tree-ssa.c (verify_ssa_name): Fix check, move
24942         SSA_NAME_IN_FREE_LIST check up.
24944 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24945             Steven Bosscher   <steven@gcc.gnu.org>
24947         PR rtl-optimization/56151
24948         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24949         equal to op0 or op1, and last_insn pattern is CODE operation
24950         with MEM dest and one of the operands matches that MEM.
24952 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24954         * doc/extend.texi: Document Function Multiversioning and "default"
24955         parameter string to target attribute.
24956         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24957         target attribute parameter is "default".
24958         (ix86_compare_version_priority): Remove checks for target attribute.
24959         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24960         Remove check for target attribute equal to NULL. Add assert.
24961         (ix86_generate_version_dispatcher_body): Change error to sorry.
24963 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24964             Jack Howarth  <howarth@bromo.med.uc.edu>
24965             Patrick Marlier  <patrick.marlier@gmail.com>
24967         PR libitm/55693
24968         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24969         define ENDFILE_SPEC as TM_DESTRUCTOR.
24970         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24972 2013-02-11  Alexander Potapenko  <glider@google.com>
24973             Jack Howarth  <howarth@bromo.med.uc.edu>
24974             Jakub Jelinek  <jakub@redhat.com>
24976         PR sanitizer/55617
24977         * config/darwin.c (cdtor_record): Rename ctor_record.
24978         (sort_cdtor_records): Rename sort_ctor_records.
24979         (finalize_dtors): New routine to sort destructors by
24980         priority before use in assemble_integer.
24981         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24983 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24985         PR rtl-optimization/56275
24986         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24987         offset is non-negative and less than cmode size before
24988         calling simplify_subreg.
24990 2013-02-11  Richard Biener  <rguenther@suse.de>
24992         PR tree-optimization/56264
24993         * cfgloop.h (fix_loop_structure): Adjust prototype.
24994         * loop-init.c (fix_loop_structure): Return the number of
24995         newly discovered loops.
24996         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24997         are discovered, do a full loop-closed SSA rewrite.
24999 2013-02-11  Richard Biener  <rguenther@suse.de>
25001         PR tree-optimization/56273
25002         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
25003         first VRP run.
25004         (check_array_ref): Fix missing newline in dumps.
25005         (search_for_addr_array): Likewise.
25007 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
25009         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
25011 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
25013         PR target/56256
25014         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
25016 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
25018         PR rtl-optimization/56246
25019         * lra-constraints.c (simplify_operand_subreg): Try to reuse
25020         reload pseudo.
25021         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
25022         constraints are satisfied.
25024 2013-02-08  Jeff Law  <law@redhat.com>
25026         PR debug/53948
25027         * emit-rtl.c (reg_is_parm_p): New function.
25028         * regs.h (reg_is_parm_p): New prototype.
25029         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
25030         callee-clobbered registers.
25032 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
25034         PR target/56043
25035         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
25036         If there is no implicit builtin declaration, just return NULL.
25038 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25040         * config/i386/sse.md (FMAMODEM): New mode iterator.
25041         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
25042         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
25044 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25046         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
25047         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
25048         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
25050 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25052         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
25053         (microblaze*-*-elf): Likewise.
25054         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
25055         LINK_SPEC.
25056         * config/microblaze/microblaze-c.c: Add builtin defines for
25057         _LITTLE_ENDIAN and _BIG_ENDIAN.
25058         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
25059         add to TARGET_DEFAULT flags.
25060         Expand ASM_SPEC and LINK_SPEC.
25061         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
25062         * config/microblaze/microblaze.md: Update extendsidi2 and
25063         movdi_internal instructions to use low-order / high-order reg
25064         print_operands.
25065         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
25066         options and inversemask / mask of LITTLE_ENDIAN.
25067         * config/microblaze/t-microblaze: Expand multilib options to
25068         include mlittle-endian (le) and update exceptions patterns.
25070 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25072         PR rtl-optimization/56195
25073         * lra-constraints.c (get_reload_reg): Don't reuse regs
25074         if they have smaller mode than requested, if they have
25075         wider mode than requested, try to return a SUBREG.
25077         PR tree-optimization/56250
25078         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25079         if type is unsigned and code isn't MULT_EXPR.
25081 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25083         PR tree-optimization/56064
25084         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25085         bits according to mode.
25086         * fixed-value.h (fixed_from_double_int)
25087         (const_fixed_from_double_int): Adjust comments.
25089 2013-02-08  Richard Biener  <rguenther@suse.de>
25091         PR lto/56231
25092         * lto-streamer.h (struct data_in): Remove current_file, current_line
25093         and current_col members.
25094         * lto-streamer-out.c (lto_output_location): Stream changed bits
25095         en-block for efficiency.
25096         * lto-streamer-in.c (clear_line_info): Remove.
25097         (lto_input_location): Cache current file, line and column
25098         globally via local statics.  Read changed bits en-block.
25099         (input_function): Do not call clear_line_info.
25100         (lto_read_body): Likewise.
25101         (lto_input_toplevel_asms): Likewise.
25103 2013-02-08  Michael Matz  <matz@suse.de>
25105         PR tree-optimization/52448
25106         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25107         (nt_call_phase): New static.
25108         (add_or_mark_expr): Only mark accesses with newer phase than any
25109         call seen.
25110         (nonfreeing_call_p): New.
25111         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25112         (nt_fini_block): Keep blocks marked as visited.
25113         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25115 2013-02-08  Richard Biener  <rguenther@suse.de>
25117         * ira.c (ira): Free broken dominator information.
25119 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25121         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25123 2013-02-08  Marek Polacek  <polacek@redhat.com>
25125         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25127 2013-02-08  Richard Biener  <rguenther@suse.de>
25129         PR middle-end/56181
25130         * cfgloop.h (flow_loops_find): Adjust.
25131         (bb_loop_header_p): Declare.
25132         * cfgloop.c (bb_loop_header_p): New function split out from ...
25133         (flow_loops_find): ... here.  Adjust function signature,
25134         support incremental loop structure update.
25135         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25136         * cfgloopmanip.c (fix_loop_structure): Move ...
25137         * loop-init.c (fix_loop_structure): ... here.
25138         (apply_loop_flags): Split out from ...
25139         (loop_optimizer_init): ... here.
25140         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25141         in incremental mode, only remove dead loops here.
25143 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25145         PR target/54222
25146         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25147         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25148         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25149         (*round<mode>3.libgcc): New insns for fixed-modes.
25150         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25151         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25152         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25153         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25154         implementations.  Define to __builtin_avr_absFX,
25155         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25156         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25157         __builtin_avr_countlsFX, respectively.
25158         * config/avr/avr-c.c (target.h): Include it.
25159         (enum avr_builtin_id): New enum.
25160         (avr_resolve_overloaded_builtin): New static function.
25161         (avr_register_target_pragmas): Use it to set
25162         targetm.resolve_overloaded_builtin.
25163         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25164         tree nodes used by DEF_BUILTIN.
25165         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25166         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25167         <AVR_BUILTIN_xxBITS>: Same.
25169 2013-02-08  Richard Biener  <rguenther@suse.de>
25171         * cfgloop.c (verify_loop_structure): Properly handle
25172         a loop exiting to another loop header.
25173         * ira-int.h (ira_loops): Remove.
25174         * ira.c (ira_loops): Remove.
25175         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25176         (do_reload): Use loop_optimizer_finalize.
25177         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25178         number_of_loops to access the loop tree.
25179         (more_one_region_p): Likewise.
25180         (finish_loop_tree_nodes): Likewise.
25181         (rebuild_regno_allocno_maps): Likewise.
25182         (mark_loops_for_removal): Likewise.
25183         (mark_all_loops_for_removal): Likewise.
25184         (remove_unnecessary_regions): Likewise.
25185         (ira_build): Likewise.
25186         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25188 2013-02-08  Richard Biener  <rguenther@suse.de>
25190         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25191         * ipa-pure-const.c (analyze_function): Avoid calling
25192         mark_irreducible_loops twice.
25193         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25195 2013-02-07  David S. Miller  <davem@davemloft.net>
25197         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25198         on 'reg'.
25199         * var-tracking.c (vt_add_function_parameter): Test the presence of
25200         HAVE_window_save properly and do not remap argument registers when
25201         we have a leaf function.
25203 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25205         PR bootstrap/56227
25206         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25207         instead of "ll".
25208         * config/i386/i386.c (ix86_print_operand): Ditto.
25210 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25212         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25214 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25216         PR rtl-optimization/56225
25217         * lra-constraints.c (process_alt_operands): Check that reload hard
25218         reg can hold value for strict_low_part.
25220 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25222         PR debug/56154
25223         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25224         dwarf2out_end_function.
25225         (in_first_function_p, maybe_at_text_label_p,
25226         first_loclabel_num_not_at_text_label): New variables.
25227         (dwarf2out_var_location): In the first function find out
25228         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25229         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25230         functions.
25232 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25234         PR rtl-optimization/56178
25235         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25236         SUBREG of a register.  Tidy up related block of code.
25237         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25238         note if the source is a register or a SUBREG of a register.
25240 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25242         PR target/56228
25243         * config/rs6000/rs6000.md (ptrm): New mode attr.
25244         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25245         call_value_indirect_aix<pttrsize>,
25246         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25247         m in constraints.
25249 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25251         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25252         if -bnortl. Convert to strcmp and strncmp.
25254 2013-02-07  Alan Modra  <amodra@gmail.com>
25256         PR target/54009
25257         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25258         addresses won't wrap when offsetting.
25259         (rs6000_secondary_reload): Provide secondary reloads needed for
25260         wrapping LO_SUM addresses.
25262 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25264         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25265         MACH, just __MACH__.
25267 2013-02-06  Richard Biener  <rguenther@suse.de>
25269         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25270         instead of calling fix_loop_structure.
25272 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25274         PR middle-end/56217
25275         * omp-low.c (use_pointer_for_field): Return false if
25276         lower_send_shared_vars doesn't generate any copy-out code.
25278 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25280         PR rtl-optimization/56131
25281         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25282         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25283         of the label is NULL.  Add comment.
25285 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25287         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25289         PR sanitizer/55374
25290         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25291         (STATIC_LIBTSAN_LIBS): Likewise.
25292         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25293         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25294         is defined, don't add anything else beyond that.
25295         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25296         (LINK_COMMAND_SPEC): Use them.
25298         PR tree-optimization/56205
25299         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25300         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25301         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25303 2013-02-05  Richard Biener  <rguenther@suse.de>
25305         PR tree-optimization/53342
25306         PR tree-optimization/53185
25307         * tree-vectorizer.h (vect_check_strided_load): Remove.
25308         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25309         not disallow peeling for vectorized strided loads.
25310         (vect_check_strided_load): Make static and simplify.
25311         (vect_analyze_data_refs): Adjust.
25312         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25313         correctly when vectorizing strided loads.
25315 2013-02-05  Richard Biener  <rguenther@suse.de>
25317         * doc/install.texi: Refer to ISL, not PPL.
25319 2013-02-05  Jan Hubicka  <jh@suse.cz>
25321         PR tree-optimization/55789
25322         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25324 2013-02-05  Jan Hubicka  <jh@suse.cz>
25326         PR tree-optimization/55789
25327         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25328         the dead call anyway.
25330 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25332         PR sanitizer/55374
25333         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25335 2013-02-04  Alexander Potapenko  <glider@google.com>
25336             Jack Howarth  <howarth@bromo.med.uc.edu>
25337             Jakub Jelinek  <jakub@redhat.com>
25339         PR sanitizer/55617
25340         * config/darwin.c (sort_ctor_records): Stabilized qsort
25341         on constructor priority by using original position.
25342         (finalize_ctors): New routine to sort constructors by
25343         priority before use in assemble_integer.
25344         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25346 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25348         PR libstdc++/54314
25349         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25350         about visibility on artificial decls.
25351         * config/sol2.c (solaris_assemble_visibility): Likewise.
25353 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25355         PR target/56186
25356         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25357         argument and improve checking of return-argument types for 16-byte
25358         modes.
25359         (ix86_function_value_1): Add additional valtype argument on call
25360         of function_value_64.
25361         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25362         handling infunction_value_64 function.
25364 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25366         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25368 2013-02-04  Richard Biener  <rguenther@suse.de>
25370         PR tree-optimization/56188
25371         * tree-ssa-structalias.c (label_visit): Consider case with
25372         initially non-empty points-to set.
25373         (perform_var_substitution): Dump node mapping and clean up.
25375 2013-02-04  Richard Guenther  <rguenther@suse.de>
25377         PR lto/56168
25378         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25379         node prevail as last resort.
25380         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25381         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25383 2013-02-04  Richard Biener  <rguenther@suse.de>
25385         PR tree-optimization/56113
25386         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25387         Merge into ...
25388         (equiv_class_lookup_or_add): ... this.
25389         (label_visit): Adjust and fix error in previous patch.
25390         (perform_var_substitution): Adjust.
25392 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25394         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25395         * config/sh/sh4-300.md: Likewise.
25396         * config/sh/sh4a.md: Likewise.
25397         * config/sh/constraints.md: Likewise.
25398         * config/sh/sh.md: Likewise.
25399         * config/sh/netbsd-elf.h: Likewise.
25400         * config/sh/predicates.md: Likewise.
25401         * config/sh/sh-protos.h: Likewise.
25402         * config/sh/ushmedia.h: Likewise.
25403         * config/sh/linux.h: Likewise.
25404         * config/sh/sh.c: Likewise.
25405         * config/sh/superh.h: Likewise.
25406         * config/sh/elf.h: Likewise.
25407         * config/sh/sh4.md: Likewise.
25408         * config/sh/sh.h: Likewise.
25410 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25412         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25413         constraint to match_test floating_point_store_memory_operand().
25414         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25415         (base14_operand): New.
25416         (floating_point_store_memory_operand): New.
25417         (integer_store_memory_operand): Revise to use base14_operand and
25418         reg_plus_base_memory_operand.
25419         (move_dest_operand): Allow symbolic_memory_operands.
25420         (symbolic_memory_operand): Check for LO_SOM.
25421         (symbolic_operand): Change default case to break.
25422         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25423         CONST_DOUBLE values to be reloaded by putting them into memory when
25424         the destination is a floating point register.
25425         (movdf): Remove code to handle CONST_DOUBLE.
25426         (movsf): Likewise.
25427         (reload_indf_r1): New.
25428         (reload_insf_r1): New.
25429         Consistently use "Q" and "T" constraints with integer and floating
25430         point move instructions, respectively.
25431         (movdi): Remove FAIL.
25432         Change predicate for source operand unamed DImode move from
25433         general_operand to move_src_operand.
25434         (umulsidi3): Change predicate for destination operand to
25435         register_operand.
25436         Likewise for similar unamed patterns.
25437         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25438         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25439         (hppa_legitimize_address): Simplify mask calculation.
25440         (pa_emit_move_sequence): Revised handling of secondary reloads from
25441         REG+D addresses for floating point loads and stores.  Directly handle
25442         loading CONST0_RTX (mode) to a floating point register.
25443         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25444         to floating point registers.  Don't restrict secondary reloads to
25445         floating point registers to integer modes.  Revise some comments and
25446         cleanup some code.
25447         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25448         (pa_legitimate_address_p): New.
25449         (pa_legitimize_reload_address): New.
25450         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25451         (STRICT_REG_OK_FOR_BASE_P): New.
25452         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25453         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25455 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25456             Andrew Dixie  <andrewd@gentrack.com>
25458         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25459         flag set.
25461 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25463         * expmed.c (extract_bit_field_1): Pass the full width of the
25464         structure to get_best_reg_extraction_insn.
25466 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25468         PR target/54601
25469         * configure.ac (use_cxa_atexit): Add AIX.
25470         * configure: Regenerate.
25472         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25474 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25476         PR debug/54793
25477         * final.c (need_profile_function): New variable.
25478         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25479         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25480         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25481         notes, targetm.asm_out.function_prologue doesn't emit anything,
25482         HAVE_prologue and profiler should be emitted before prologue,
25483         set need_profile_function instead of emitting it.
25484         (final_scan_insn): If need_profile_function, emit
25485         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25486         NOTE_INSN_FUNCTION_BEG note.
25488 2013-02-01  Richard Henderson  <rth@redhat.com>
25490         * config/rs6000/rs6000.md (smulditi3): New.
25491         (umulditi3): New.
25493         * config/alpha/alpha.md (umulditi3): New.
25495 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25497         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25498         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25500 2013-02-01  Richard Biener  <rguenther@suse.de>
25502         PR tree-optimization/56113
25503         * tree-ssa-structalias.c (label_visit): Reduce work for
25504         single-predecessor nodes.
25506 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25508         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25509         range isn't testing for zero.
25511 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25513         PR middle-end/56113
25514         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25516 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25517             Nick Clifton  <nickc@redhat.com>
25519         * config/v850/constraints.md (Q): Define as a memory constraint.
25520         * config/v850/predicates.md (label_ref_operand): New predicate.
25521         (e3v5_shift_operand): New predicate.
25522         (ior_operator): New predicate.
25523         * config/v850/t-v850: Add e3v5 multilib.
25524         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25525         (v850_gen_movdi): Prototype.
25526         * config/v850/v850.c: Add support for e3v5 architecture.
25527         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25528         TARGET_V850E_UP.
25529         (construct_save_jarl): Add e3v5 long JARL support.
25530         (v850_adjust_insn_length): New function.  Adjust length of call
25531         insns when using e3v5 instructions.
25532         (v850_gen_movdi): New function: Generate instructions to move a
25533         DImode value.
25534         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25535         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25536         (TARGET_USE_FPU): Enable for e3v5.
25537         (CONST_OK_FOR_W): New macro.
25538         (ADJUST_INSN_LENGTH): Define.
25539         * config/v850/v850.md (UNSPEC_LOOP): Define.
25540         (attr cpu): Add v850e3v5.
25541         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25542         (movdi): New pattern.
25543         (movdi_internal): New pattern.
25544         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25545         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25546         (cstoresf4): Likewise.
25547         (cstoredf4): Likewise.
25548         (insv): New pattern.
25549         (rotlso3_a): New pattern.
25550         (rotlsi3_b): New pattern
25551         (rotlsi3_v850e3v5): New pattern.
25552         (doloop_begin): New pattern.
25553         (fix_loop_counter): New pattern.
25554         (doloop_end): New pattern.
25555         (branch_normal): Add e3v5 long branch support.
25556         (branch_invert): Likewise.
25557         (branch_z_normal): Likewise.
25558         (branch_z_invert): Likewise.
25559         (branch_nz_normal): Likewise.
25560         (branch_nz_invert): Likewise.
25561         (call_internal_short): Add e3v5 register-indirect JARL support.
25562         (call_internal_long): Likewise.
25563         (call_value_internal_short): Likewise.
25564         (call_value_internal_long): Likewise.
25565         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25566         (mloop): New option.
25567         * config.gcc: Add support for configuring v840e3v5 target.
25568         * doc/invoke.texi: Document new v850 specific command line options.
25570 2013-01-31  Paul Koning  <ni1d@arrl.net>
25572         PR debug/55059
25573         PR debug/54508
25574         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25575         children if parent is a class.
25576         (prune_unused_types_prune): Don't add DW_AT_declaration.
25578 2013-01-31  Richard Biener  <rguenther@suse.de>
25580         PR tree-optimization/56157
25581         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25582         match up operand with SLP child.
25584 2013-01-31  Jason Merrill  <jason@redhat.com>
25586         PR debug/54410
25587         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25588         parameters the first time.
25589         (gen_scheduled_generic_parms_dies): Check completeness here.
25591 2013-01-31  Richard Biener  <rguenther@suse.de>
25593         PR middle-end/53073
25594         * common.opt (faggressive-loop-optimizations): New flag,
25595         enabled by default.
25596         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25597         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25598         infer_loop_bounds_from_undefined by it.
25600 2013-01-31  Richard Biener  <rguenther@suse.de>
25602         PR tree-optimization/56150
25603         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25604         visit virtual operands.
25605         (find_uses_to_rename_bb): Likewise.
25607 2013-01-31  Richard Biener  <rguenther@suse.de>
25609         PR tree-optimization/56150
25610         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25611         mixed store non-store stmts.
25613 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25615         PR sanitizer/55374
25616         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25617         LIBASAN_EARLY_SPEC is defined.
25618         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25619         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25620         before %o.
25621         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25623         PR c++/55742
25624         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25625         invalid args instead of ICEing on it.
25626         (ix86_valid_target_attribute_tree): Return error_mark_node if
25627         ix86_valid_target_attribute_inner_p failed.
25628         (ix86_valid_target_attribute_p): Return false only if
25629         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25630         target("default") attribute.
25631         (sorted_attr_string): Change argument from const char * to tree,
25632         merge in all target attribute arguments rather than just one.
25633         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25634         instead of free.  Avoid using strcat.
25635         (ix86_mangle_function_version_assembler_name): Mangle
25636         target("default") as if no target attribute is present.  Adjust
25637         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25638         instead of xmalloc and XDELETEVEC instead of free.
25639         (ix86_function_versions): Don't return true if one of the decls
25640         doesn't have target attribute.  If they don't and one of the decls
25641         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25642         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25643         (ix86_supports_function_versions): Remove.
25644         (make_name): Fix up formatting.
25645         (make_dispatcher_decl): Remove resolver_name and its initialization.
25646         Avoid leaking memory.
25647         (is_function_default_version): Return true if there is
25648         target("default") attribute rather than no target attribute at all.
25649         (make_resolver_func): Avoid leaking memory.
25650         (ix86_generate_version_dispatcher_body): Likewise.
25651         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25652         * target.def (supports_function_versions): Remove.
25653         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25654         * doc/tm.texi: Regenerated.
25656 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25658         PR rtl-optimization/56144
25659         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25660         for values with side effects.
25662 2013-01-30  Richard Biener  <rguenther@suse.de>
25664         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25665         (sparseset_pop): Likewise.
25666         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25667         to be able to use quick_push in the worker loop.
25669 2013-01-30  Marek Polacek  <polacek@redhat.com>
25671         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25673 2013-01-30  Richard Biener  <rguenther@suse.de>
25675         PR lto/56147
25676         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25678 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25680         PR tree-optimization/56064
25681         * fixed-value.c (fixed_from_double_int): New function.
25682         * fixed-value.h (fixed_from_double_int): New prototype.
25683         (const_fixed_from_double_int): New static inline function.
25684         * fold-const.c (native_interpret_fixed): New static function.
25685         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25686         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25687         (native_encode_fixed): New static function.
25688         (native_encode_expr) <FIXED_CST>: Use it.
25689         (native_interpret_int): Move double_int worker code to...
25690         * double-int.c (double_int::from_buffer): ...this new static method.
25691         * double-int.h (double_int::from_buffer): Prototype it.
25693 2013-01-30  Richard Biener  <rguenther@suse.de>
25695         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25696         New pointer-map and obstack.
25697         (init_alias_vars): Allocate pointer-map and obstack.
25698         (delete_points_to_sets): Free them.
25699         (find_what_var_points_to): Cache result.
25700         (find_what_p_points_to): Adjust for changed interface of
25701         find_what_var_points_to.
25702         (compute_points_to_sets): Likewise.
25703         (ipa_pta_execute): Likewise.
25705 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25707         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25708         * configure: Regenerate.
25709         * config.in: Regenerate.
25710         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25711         #nobits/#progbits if supported.
25713 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25715         PR target/56121
25716         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25717         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25718         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25720 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25722         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25723         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25725 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25727         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25728         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25730 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25732         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25733         declaration.
25734         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25735         * config/arm/cortex-a7.md: New bypasses using
25736         arm_mac_accumulator_is_result.
25738 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25740         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25741         (cortex_a7_neon_mla): Likewise.
25742         (cortex_a7_fpfmad): New reservation.
25743         (cortex_a7_fpmacs): Use ffmas and update required units.
25744         (cortex_a7_fpmuld): Update required units and latency.
25745         (cortex_a7_fpmacd): Likewise.
25746         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25747         (cortex_a7_neon). Likewise.
25748         (bypass) Update participating units.
25750 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25752         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25753         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25754         from fmac to ffma.
25755         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25756         (vfp_fmul): Use ffmad.
25757         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25758         (cortex_r4_fmacd): Use ffmad.
25759         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25760         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25761         (cortex_a9_fmacd): Use ffmad.
25762         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25763         (cortex_a8_vfp_macd): Use ffmad.
25764         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25765         (cortex_a5_fpmacd): Use ffmad.
25766         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25767         (cortex_a15_vfp_macd): Use ffmad.
25768         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25770 2013-01-29  Jason Merrill  <jason@redhat.com>
25772         PR libstdc++/54314
25773         * varasm.c (default_assemble_visibility): Don't warn about
25774         visibility on artificial decls.
25776 2013-01-29  Richard Biener  <rguenther@suse.de>
25778         PR tree-optimization/56113
25779         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25780         the bitmap leader.
25781         (label_visit): Free duplicate bitmaps and record the leader instead.
25782         (perform_var_substitution): Adjust.
25784 2013-01-29  Richard Biener  <rguenther@suse.de>
25786         PR tree-optimization/55270
25787         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25788         the CFG, schedule loops for fixup.
25790 2013-01-29  Nick Clifton  <nickc@redhat.com>
25792         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25793         SP_REG.
25795 2013-01-28  Leif Ekblad  <leif@rdos.net>
25797         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25798         * config/i386/i386.h (TARGET_RDOS): New macro.
25799         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25800         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25801         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25802         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25803         DEFAULT_LARGE_SECTION_THRESHOLD.
25804         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25805         * config/i386/rdos.h: New file.
25806         * config/i386/rdos64.h: New file.
25808 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25810         PR other/54814
25811         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25812         TEST_HARD_REG_BIT.
25814 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25816         PR rtl-optimization/56117
25817         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25818         call cselib_lookup_from_insn on the MEM before calling
25819         add_insn_mem_dependence.
25821 2013-01-28  Richard Biener  <rguenther@suse.de>
25823         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25824         to a stmt that didn't have one.
25825         (copy_phis_for_bb): Likewise for PHI arguments.
25826         (copy_debug_stmt): Likewise for debug stmts.
25828 2013-01-28  Richard Biener  <rguenther@suse.de>
25830         PR tree-optimization/56034
25831         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25832         (partition_builtin_p): Adjust.
25833         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25834         it is the last partition.
25835         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25836         up the vertex for the definition.
25837         (classify_partition): Classify whether a partition is a
25838         PKIND_REDUCTION, thus has uses outside of the loop.
25839         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25840         Merge all PKIND_REDUCTION partitions into the last partition.
25841         (tree_loop_distribution): Seed partitions from reductions as well.
25843 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25845         PR tree-optimization/56125
25846         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25847         pow(x,c) into sqrt(x) * powi(x, n/2) or
25848         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25849         optimizing for size.
25850         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25851         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25852         integer.
25854         PR tree-optimization/56094
25855         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25856         to UNKNOWN_LOCATION while gimplifying expr.
25858 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25860         PR target/56114
25861         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25862         operand 0 in movabs insn template for -masm=intel asm alternative.
25863         (*movabs<mode>_2): Ditto for operand 1.
25865 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25867         PR target/54663
25868         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25869         of microblaze-c.o
25871 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25873         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25874         tm_file.
25876 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25878         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25879         Undef to avoid warning.
25881 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25883         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25884         * configure: Regenerate.
25886 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25888         PR tree-optimization/56098
25889         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25890         for stmts with volatile ops.
25891         (cond_store_replacement): Don't optimize if assign has volatile ops.
25892         (cond_if_else_store_replacement_1): Don't optimize if either
25893         then_assign or else_assign have volatile ops.
25894         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25895         volatile ops.
25897 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25899         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25901 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25903         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25904         missing ':' in asm example.
25906 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25908         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25909         entries into lane and laneq entries.
25910         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25911         Remove AdvSIMD scalar modes.
25912         (aarch64_sq<r>dmulh_laneq<mode>): New.
25913         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25914         modes.
25915         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25916         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25917         * config/aarch64/iterators.md (VCOND): New.
25918         (VCONQ): New.
25920 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25922         PR target/54222
25923         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25924         Add NULL LIBNAME argument to existing definitions.
25925         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25926         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25927         * config/avr/avr.c (DEF_BUILTIN): Same.
25928         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25929         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25930         implementation is available (DECL_ASSEMBLER_NAME is set).
25931         (avr_fold_absfx): New static function.
25932         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25933         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25934         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25935         AVR_BUILTIN_ABSLLK.
25936         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25937         (abshk, absk, abslk, absllk): Provide as static inline functions.
25939 2013-01-25  Marek Polacek  <polacek@redhat.com>
25941         PR tree-optimization/56035
25942         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25944 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25946         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25947         (*movtf_internal_rex64): Add (!o,C) alternative
25948         (*movxf_internal_rex64): Ditto.
25949         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25951 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25953         * doc/invoke.texi: fix typo.
25954         * doc/objc.texi: fix typo.
25956 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25958         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25959         for the first two alternatives.
25961 2013-01-24  Diego Novillo  <dnovillo@google.com>
25963         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25964         (ggc-zone.o): Remove.
25965         * configure.ac: Remove option --with-gc.
25966         * configure: Re-generate.
25967         * doc/install.texi: Remove documentation for --with-gc.
25968         * gengtype.c (write_enum_defn): Remove.  Update all users.
25969         (write_Types_process_field): Remove generation of gt_e_* argument.
25970         (output_type_enum): Remove.  Update all users.
25971         (write_enum_defn): Remove.  Update all users.
25972         (enum alloc_zone): Remove.  Update all users.
25973         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25974         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25975         Update all callers.
25976         (struct ptr_data): Remove field TYPE.  Update all users.
25977         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25978         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25979         Update all users.
25980         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25981         (struct alloc_zone): Remove.
25982         (ggc_internal_alloc_zone_stat): Remove.
25983         (ggc_internal_cleared_alloc_zone_stat): Remove.
25984         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25985         (ggc_pch_count_object): Remove last argument.  Update all users.
25986         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25987         (struct alloc_zone): Remove.
25988         * ggc-zone.c: Remove.
25989         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25990         (struct alloc_zone): Remove.
25991         (ggc_alloc_typed_stat): Remove.
25992         (ggc_alloc_typed): Remove.
25993         (ggc_splay_alloc): Remove first argument.
25994         (rtl_zone): Remove.  Update all users.
25995         (tree_zone): Remove.  Update all users.
25996         (tree_id_zone): Remove.  Update all users.
25997         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25998         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25999         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
26000         * tree-ssanames.c: Remove references to zone allocator in comments.
26002 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
26004         * config/avr/avr.c (avr_out_fract): Make register numbers that
26005         might be outside of source operand signed.
26007 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
26009         * config/i386/constraints.md (Yf): New constraint.
26010         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
26011         of f constraint to conditionaly disable x87 register preferences.
26012         (*movdf_internal): Ditto.
26013         (*movsf_internal): Ditto.
26015 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
26017         PR inline-asm/55934
26018         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
26019         that have operands with impossible constraints.
26020         Add a FIXME for a speed-up opportunity.
26021         * lra-constraints.c (process_alt_operands): Verify that a class
26022         selected from constraints on asms is valid for the operand mode.
26023         (curr_insn_transform): Remove incorrect comment.
26025 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
26027         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
26028         TOC operand is a valid symbol ref in the constant pool.
26030 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26032         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
26034 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
26036         PR target/54222
26037         * config/avr/stdfix.h: New file.
26038         * t-avr (stdfix-gcc.h): New rule to build it.
26039         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
26041 2013-01-23  Kostya Serebryany  <kcc@google.com>
26043         * config/darwin.h: remove dependency on
26044         CoreFoundation (asan on Mac OS).
26046 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
26048         PR target/49069
26049         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
26050         instead of cmpdi_operand for first comparison operand.
26051         Don't assert that comparison operands aren't both constants.
26053 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26055         * doc/install.texi (Downloading the Source): Update references to
26056         downloading separate components.
26058 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26060         * doc/extend.texi (__int128): Improve grammar.
26062 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
26064         PR target/56028
26065         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
26066         alternative to (o,r).
26067         (*movdi_internal_rex64): Remove (!o,n) alternative.
26068         (DImode immediate->memory splitter): Remove.
26069         (DImode immediate->memory peephole2): Remove.
26070         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26071         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26072         alternative to (!o,*r).
26073         (*movtf_internal_sse): New pattern.
26074         (*movxf_internal_rex64): New pattern.
26075         (*movxf_internal): Disable for TARGET_64BIT.
26076         (*movdf_internal_rex64): Remove (!o,F) alternative.
26078 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26080         PR middle-end/56074
26081         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26082         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26083         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26084         stmt locations where LOCATION_LOCUS of the stmt location is
26085         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26087         PR target/55686
26088         * config/i386/i386.md (UNSPEC_STOS): New.
26089         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26090         *strsetqi_1): Add UNSPEC_STOS.
26092 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26094         PR c++/56067
26095         * doc/invoke.texi: Remove left over -Wsynth example.
26097 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26099         PR tree-optimization/56051
26100         * fold-const.c (fold_binary_loc): Don't fold
26101         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26102         a narrowing conversion, or widening conversion from signed
26103         to unsigned.
26105 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26107         PR rtl-optimization/56023
26108         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26109         dependent on debug instruction.
26111 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26113         PR middle-end/56022
26114         * function.c (allocate_struct_function): Call
26115         invoke_set_current_function_hook earlier.
26117 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26119         * reload1.c (init_reload): Only initialize reload_obstack
26120         during the first call.
26122 2013-01-21  Marek Polacek  <polacek@redhat.com>
26124         * cfgloop.c (verify_loop_structure): Fix up grammar.
26126 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26128         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26129         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26131 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26133         PR target/56058
26134         * config/arm/marvell-pj4.md: Update copyright year.
26135         Fix up use of alu to alu_reg and simple_alu_imm.
26137 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26139         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26141 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26143         PR target/55433
26144         * lra-constraints.c (curr_insn_transform): Don't reuse original
26145         insn for secondary memory move when memory mode should be different.
26147 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26149         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26150         atomic_storedi_1): New patterns.
26152 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26154         btver2 pipeline descriptions.
26155         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26156         descriptions.
26157         * config/i386/i386.md (btver2_decode): New type attributes.
26158         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26159         type attributes.
26160         * config/i386/btver2.md: New file describing btver2 pipelines.
26162 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26164         PR tree-optimization/52631
26165         * tree-ssa-sccvn (visit_use): Before looking up the original
26166         statement, try looking up the simplified expression.
26168 2013-01-19  Anthony Green  <green@moxielogic.com>
26170         * config/moxie/moxie.c (moxie_expand_prologue): Set
26171         current_function_static_stack_size.
26173 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26175         PR tree-optimization/56029
26176         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26177         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26179 2013-01-18  Sharad Singhai  <singhai@google.com>
26181         PR tree-optimization/55995
26182         * dumpfile.c (dump_loc): Print location only if available.
26183         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26185 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26187         PR target/55433
26188         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26189         secondary memory move.
26190         (inherit_reload_reg): Use rclass instead of cl for
26191         check_secondary_memory_needed_p.
26193 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26195         PR middle-end/56015
26196         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26197         the case where writing real complex part of target modifies op1.
26199 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26201         * config/aarch64/aarch64-simd.md
26202         (aarch64_vcond_internal<mode>): Handle unordered cases.
26203         * config/aarch64/iterators.md (v_cmp_result): New.
26205 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26206             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26208         * config/arm/marvell-pj4.md: New file.
26209         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26210         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26211         (generic_vfp): Likewise.
26212         * config/arm/arm-cores.def: Add marvell-pj4.
26213         * config/arm/arm-tune.md: Regenerate.
26214         * config/arm/arm-tables.opt: Regenerate.
26215         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26216         * doc/invoke.texi: Document marvell-pj4.
26218 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26220         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26222 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26224         PR debug/54114
26225         PR debug/54402
26226         PR debug/49888
26227         * var-tracking.c (negative_power_of_two_p): New.
26228         (global_get_addr_cache, local_get_addr_cache): New.
26229         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26230         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26231         heading comment.
26232         (vt_stack_offset_p): Remove.
26233         (vt_canon_true_dep): Always canonicalize loc's address.
26234         (clobber_overlapping_mems): Make sure we have a MEM.
26235         (local_get_addr_clear_given_value): New.
26236         (val_reset): Clear local cached entries.
26237         (compute_bb_dataflow): Create and release the local cache.
26238         Disable duplicate MEMs clobbering.
26239         (emit_notes_in_bb): Clobber MEMs likewise.
26240         (vt_emit_notes): Create and release the local cache.
26241         (vt_initialize, vt_finalize): Create and release the global
26242         cache, respectively.
26243         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26245 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26247         PR libmudflap/53359
26248         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26249         not found in the symtab.
26251 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26253         PR debug/56006
26254         PR rtl-optimization/55547
26255         PR rtl-optimization/53827
26256         PR debug/53671
26257         PR debug/49888
26258         * alias.c (offset_overlap_p): New, factored out of...
26259         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26260         the conservative special case for symbolic constants.  Don't
26261         adjust zero sizes on alignment.
26263 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26265         PR rtl-optimization/52573
26266         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26267         REG_UNUSED for the same register.
26269 2013-01-17  Richard Biener  <rguenther@suse.de>
26270             Marek Polacek  <polacek@redhat.com>
26272         PR rtl-optimization/55833
26273         * loop-unswitch.c (unswitch_loops): Move loop verification...
26274         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26275         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26276         Set it to true when we're removing a loop from hierarchy tree in
26277         an irreducible region.
26278         (fix_bb_placements): Adjust caller.
26279         (fix_loop_placements): Likewise.
26281 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26283         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26284         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26285         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26286         Remove ID.  Adjust comments.
26287         * config/avr/avr-c.c (avr_builtin_name): Remove.
26288         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26289         * config/avr/avr.c (avr_tolower): New static function.
26290         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26291         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26292         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26293         default expansion.
26295 2013-01-17  Jan Hubicka  <jh@suse.cz>
26297         PR tree-optimization/55273
26298         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26300 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26302         PR target/55981
26303         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26304         store through atomic_store<mode>_1.
26305         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26307 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26309         PR tree-optimizations/55264
26310         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26311         for virtual methods.
26312         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26313         virtual methods before inlining is over.
26314         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26315         virtual functions.
26316         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26317         non-virtual.
26319 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26321         PR rtl-optimization/56005
26322         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26323         pending reads for prefetch.
26325 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26327         * config/aarch64/aarch64.md
26328         (*cstoresi_neg_uxtw): New pattern.
26329         (*cmovsi_insn_uxtw): New pattern.
26330         (*<optab>si3_uxtw): New pattern.
26331         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26332         (*<optab>si3_insn_uxtw): New pattern.
26333         (*bswapsi2_uxtw): New pattern.
26335 2013-01-16  Richard Biener  <rguenther@suse.de>
26337         * tree-inline.c (tree_function_versioning): Remove set but
26338         never used variable.
26340 2013-01-16  Richard Biener  <rguenther@suse.de>
26342         PR tree-optimization/55964
26343         * tree-flow.h (rename_variables_in_loop): Remove.
26344         (rename_variables_in_bb): Likewise.
26345         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26346         (copy_loop_before): Adjust and delete update-ssa status.
26347         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26348         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26349         (rename_variables_in_loop): Remove.
26350         (slpeel_update_phis_for_duplicate_loop): Likewise.
26351         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26352         use available cfg machinery instead of duplicating it.
26353         Update PHI nodes and perform poor-mans SSA update here.
26354         (slpeel_tree_peel_loop_to_edge): Adjust.
26356 2013-01-16  Richard Biener  <rguenther@suse.de>
26358         PR tree-optimization/54767
26359         PR tree-optimization/53465
26360         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26361         (vrp_visit_phi_node): For PHI arguments coming via backedges
26362         drop all symbolical range information.
26363         (execute_vrp): Compute backedges.
26365 2013-01-16  Richard Biener  <rguenther@suse.de>
26367         * doc/install.texi: Update CLooG and ISL requirements to
26368         0.18.0 and 0.11.1.
26370 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26372         PR target/55301
26373         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26374         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26375         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26377 2013-01-16  DJ Delorie  <dj@redhat.com>
26379         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26380         (UNSPECV_SP_SWITCH_E): New.
26381         (sp_switch_1): Change to an unspec.
26382         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26383         replace $r15.
26385 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26387         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26388         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26389         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26390         (expand_mem_thread_fence): Ditto.
26391         (expand_mem_signal_fence): Ditto.
26392         (expand_atomic_load): Ditto.
26393         (expand_atomic_store): Ditto.
26395 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26397         PR rtl-optimization/55547
26398         PR rtl-optimization/53827
26399         PR debug/53671
26400         PR debug/49888
26401         * alias.c (memrefs_conflict_p): Set sizes to negative after
26402         AND adjustments.
26404 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26406         PR target/55940
26407         * function.c (thread_prologue_and_epilogue_insns): Always
26408         add crtl->drap_reg to set_up_by_prologue.set, even if
26409         stack_realign_drap is false.
26411 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26413         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26414         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26415         *call): Fix indention.
26417 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26419         PR target/55876
26420         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26421         Update comment.
26423 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26425         PR rtl-optimization/55153
26426         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26428 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26430         PR tree-optimization/55920
26431         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26432         accesses as grp_to_be_debug_replaced.
26434 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26436         PR tree-optimization/55920
26437         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26438         there is non-useless type conversion needed from debug rhs to lhs,
26439         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26441 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26442             Mikael Pettersson  <mikpe@it.uu.se>
26444         PR target/43961
26445         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26446         Thumb.
26447         (ASM_OUTPUT_CASE_LABEL): Remove.
26448         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26449         * final.c (shorten_branches): Update alignment of labels before
26450         jump tables if CASE_VECTOR_SHORTEN_MODE.
26452 2013-01-15  Richard Biener  <rguenther@suse.de>
26454         PR bootstrap/55961
26455         * system.h: Do not include gmp.h for building host tools.
26457 2013-01-15  Richard Biener  <rguenther@suse.de>
26459         PR middle-end/55882
26460         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26461         account for bitpos when computing alignment.
26463 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26465         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26466         (ix86_target_macros_internal): Likewise.
26468         * config/i386/i386.c (m_CORE2I7): Removed.
26469         (m_CORE_HASWELL): New macro.
26470         (m_CORE_ALL): Likewise.
26471         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26472         (initial_ix86_arch_features): Likewise.
26473         (processor_target_table): Initializations for Core avx2.
26474         (cpu_names): New names "core-avx2".
26475         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26476         PROCESSOR_CORE_HASWELL.
26477         (ix86_issue_rate): New case.
26478         (ia32_multipass_dfa_lookahead): Likewise.
26479         (ix86_sched_init_global): Likewise.
26481         * config/i386/i386.h (TARGET_HASWELL): New macro.
26482         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26483         (processor_type): New PROCESSOR_HASWELL.
26485 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26487         PR tree-optimization/55955
26488         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26489         *SHIFT_EXPR and *ROTATE_EXPR codes.
26491         PR tree-optimization/48766
26492         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26493         -ftrapv disable -fwrapv.
26495 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26497         PR target/55974
26498         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26499         etc. to 1 and not to __flash.
26500         Use LL suffix for __INT24_MAX__ with -mint8.
26501         Use ULL suffix for __UINT24_MAX__ with -mint8.
26503 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26505         * config/avr/avr-arch.h
26506         (struct base_arch_s): Use typedef avr_arch_t instead.
26507         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26508         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26509         * config/avr/avr.c: Same.
26510         * config/avr/avr-devices.c: Same.
26511         * config/avr/driver-avr.c: Same.
26512         * config/avr/gen-avr-mmcu-texi.c: Same.
26513         * config/avr/avr-mcus.def: Adjust comment.
26515 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26517         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26518         * config/aarch64/iterators.md (VALLDI): New.
26520 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26521             Andi Kleen  <ak@linux.intel.com>
26523         PR target/55948
26524         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26525         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26526         memmodel flag.
26528 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26530         * config/avr/avr-stdint.h: Remove trailing blanks.
26531         * config/avr/avr-log.h: Same.
26532         * config/avr/avr-arch.h: Same.
26533         * config/avr/avr-devices.c: Same.
26534         * config/avr/avr-dimode.md: Same.
26535         * config/avr/predicates.md: Same.
26536         * config/avr/avr-c.c: Same.  And fix typo.
26538         * config/avr/avr-protos.h: Same.  And:
26539         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26540         (init_cumulative_args): Rename to avr_init_cumulative_args.
26541         (expand_prologue): Rename to avr_expand_prologue.
26542         (expand_epilogue): Rename to avr_expand_epilogue.
26543         (adjust_insn_length): Rename to avr_adjust_insn_length.
26544         (notice_update_cc): Rename to avr_notice_update_cc.
26545         (final_prescan_insn): Rename to avr_final_prescan_insn.
26546         * config/avr/avr.c: Same.
26547         * config/avr/avr.h: Same.
26548         * config/avr/avr.md: Remove trailing blanks.
26549         (prologue): Use avr_expand_prologue.
26550         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26552 2013-01-14  Richard Biener  <rguenther@suse.de>
26554         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26555         verify_location, collect_subblocks): New functions.
26556         (verify_gimple_in_cfg): Verify that locations only reference
26557         BLOCKs in the functions BLOCK tree.
26559 2013-01-14  Richard Biener  <rguenther@suse.de>
26561         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26562         PHI argument.
26563         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26564         unshare reference.
26565         (insert_out_of_ssa_copy_on_edge): Likewise.
26566         (rewrite_close_phi_out_of_ssa): Likewise.
26567         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26568         debug expressions.
26569         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26570         propagated constants.
26571         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26572         can not be shared.
26574 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26576         * config/avr/avr-modes.def: Add GPL copyright notice.
26578 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26580         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26581         MEMMODEL_MASK to determine memory model.
26582         (atomic_store<mode>): Ditto from operands[2].
26583         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26585 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26587         PR fortran/55935
26588         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26589         (fold_gimple_assign): Don't call unshare_expr here.
26590         (fold_ctor_reference): Call unshare_expr.
26592 2013-01-13  Terry Guo  <terry.guo@arm.com>
26594         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26595         * doc/fragments.texi: Document MULTILIB_REUSE.
26596         * gcc.c (multilib_reuse): New internal spec.
26597         (set_multilib_dir): Also search multilib from multilib_reuse.
26598         * genmultilib (tmpmultilib3): Refactor code.
26599         (tmpmultilib4): Ditto.
26600         (multilib_reuse): New multilib argument.
26602 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26604         * Makefile.in: Update copyright.
26606 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26608         PR middle-end/55890
26609         * calls.c (expand_call): Check if arg_nr is valid.
26611 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26613         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26614         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26615         documentation.  Add missing '__' in front of
26616         __builtin_ia32_packssdw256.
26618 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26620         PR target/55719
26621         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26622         NO_REGS for larl operands.
26623         (s390_reload_larl_operand): Use s390_load_address instead of
26624         emit_move_insn.
26626 2013-01-11  Richard Biener  <rguenther@suse.de>
26628         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26629         (verify_node_sharing): ... here.
26630         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26632 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26634         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26635         Substitute TREECHECKING.
26636         * configure: Regenerate.
26637         * Makefile.in (TREECHECKING): New.
26639 2013-01-11  Richard Guenther  <rguenther@suse.de>
26641         PR tree-optimization/44061
26642         * tree-vrp.c (extract_range_basic): Compute zero as
26643         value-range for __builtin_constant_p of function parameters.
26645 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26647         Update copyright years.
26649 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26651         PR rtl-optimization/55672
26652         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26653         const to be eliminable.
26655 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26657         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26658         * configure: Regenerate.
26660 2013-01-10  Richard Biener  <rguenther@suse.de>
26662         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26664 2013-01-10  Richard Biener  <rguenther@suse.de>
26666         PR bootstrap/55792
26667         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26668         locations for virtual PHI arguments.
26669         (rewrite_update_phi_arguments): Likewise.
26671 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26673         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26674         on to assembler.
26676 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26678         PR tree-optimization/55921
26679         * tree-complex.c (expand_complex_asm): New function.
26680         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26682 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26684         PR target/55718
26685         * config/s390/s390.c (s390_symref_operand_p)
26686         (s390_loadrelative_operand_p): Merge the two functions.
26687         (s390_check_qrst_address, print_operand_address): Add parameters
26688         to s390_loadrelative_operand_p invokation.
26689         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26690         (s390_reload_larl_operand, s390_secondary_reload): Use
26691         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26692         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26694 2013-01-09  Mike Stump  <mikestump@comcast.net>
26696         * dse.c (record_store): Remove unnecessary assert.
26698 2013-01-09  Jan Hubicka  <jh@suse.cz>
26700         PR tree-optimization/55569
26701         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26702         * cfgloop.h (scale_loop_profile): Likewise.
26704 2013-01-09  Jan Hubicka  <jh@suse.cz>
26706         PR lto/45375
26707         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26708         functions.
26709         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26711 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26713         PR middle-end/55114
26714         * expr.h (maybe_emit_group_store): Declare.
26715         * expr.c (maybe_emit_group_store): New function.
26716         * builtins.c (expand_builtin_int_roundingfn): Call it.
26717         (expand_builtin_int_roundingfn_2): Likewise.
26719 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26721         PR rtl-optimization/55829
26722         * lra-constraints.c (match_reload): Add code for absent output.
26723         (curr_insn_transform): Add code for reloads of matched inputs
26724         without output.
26726 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26728         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26729         attribute of movddup insn to DF.
26730         (*vec_interleave_lowv2df): Ditto.
26731         (vec_dupv2df): Ditto.
26733 2013-01-09  Jan Hubicka  <jh@suse.cz>
26735         PR tree-optimiation/55875
26736         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26737         EVERY_ITERATION parameter.
26738         (number_of_iterations_exit): Check if exit is executed every iteration.
26739         (idx_infer_loop_bounds): Similarly here.
26740         (n_of_executions_at_most): Simplify
26741         to only test for cases where statement is dominated by the
26742         particular bound; handle correctly the "postdominance" test.
26743         (scev_probably_wraps_p): Use max loop iterations info
26744         as a global bound first.
26746 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26747             Nick Clifton  <nickc@redhat.com>
26749         * config/v850/v850.md (cbranchsf4): New pattern.
26750         (cstoresf4): New pattern.
26751         (cbranchdf4): New pattern.
26752         (cstoredf4): New pattern.
26753         (movsicc): Disallow floating point comparisons.
26754         (cmpsf_le_insn): Fix order of operators.
26755         (cmpsf_lt_insn): Likewise.
26756         (cmpsf_eq_insn): Likewise.
26757         (cmpdf_le_insn): Likewise.
26758         (cmpdf_lt_insn): Likewise.
26759         (cmpdf_eq_insn): Likewise.
26760         (cmpsf_ge_insn): Use LE comparison.
26761         (cmpdf_ge_insn): Likewise.
26762         (cmpsf_gt_insn): Use LT comparison.
26763         (cmpdf_gt_insn): Likewise.
26764         (cmpsf_ne_insn): Delete pattern.
26765         (cmpdf_ne_insn): Delete pattern.
26766         * config/v850/v850.c (v850_gen_float_compare): Use
26767         gen_cmpdf_eq_insn for NE comparison.
26768         (v850_float_z_comparison_operator)
26769         (v850_float_nz_comparison_operator): Move from here ...
26770         * config/v850/predicates.md: ... to here.  Move GT and GE
26771         comparisons into v850_float_z_comparison_operator.
26772         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26773         Delete prototype.
26774         (v850_float_nz_comparison_operator): Likewise.
26776 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26778         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26779         with calls to gen_insvsi/gen_insvdi.
26781 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26783         * config/i386/i386.c (initial_ix86_tune_features): Set up
26784         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26786 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26787             Jakub Jelinek  <jakub@redhat.com>
26789         PR tree-optimization/48189
26790         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26791         If nitercst is 0, don't predict the exit edge.
26793 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26795         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26796         in asm_fprintf with reg_names.
26797         (aarch64_print_operand_address): Likewise.
26798         (aarch64_return_addr): Likewise.
26799         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26801 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26803         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26804         (INT_U6_BITS): Likewise.
26805         * config/pa/predicates.md (uint6_operand): New predicate.
26806         (shift5_operand, shift6_operand): Likewise.
26807         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26808         arith32_operand.
26809         (lshrdi3): Use shift6_operand.
26810         (shrpsi4, shrpdi4): New insn patterns.
26811         (extzv): Delete expander.
26812         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26813         predicates in unamed zero extract patterns.  Tighten common constraint.
26814         (extv): Delete expander.
26815         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26816         predicates in unamed sign extract patterns.  Tighten common constraint.
26817         (insv): Delete expander.
26818         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26819         predicates in unamed insert patterns.  Tighten common constraint.
26820         Change uint32_operand predicate to uint6_operand predicate in unamed
26821         DImode pattern to insert constant values of type 1...1xxxx.
26823 2013-01-04  Jan Hubicka  <jh@suse.cz>
26825         PR tree-optimization/55823
26826         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26827         issue.
26829 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26830             Uros Bizjak  <ubizjak@gmail.com>
26832         PR rtl-optimization/55845
26833         * df-problems.c (can_move_insns_across): Stop scanning at
26834         volatile_insn_p source instruction or give up if
26835         across_from .. across_to range contains any volatile_insn_p
26836         instructions.
26838 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26840         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26841         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26842         Declare.
26843         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26844         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26846 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26848         PR fortran/55341
26849         * asan.c (asan_clear_shadow): New function.
26850         (asan_emit_stack_protection): Use it.
26852 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26854         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26855         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26856         with tab instead of space.
26858 2013-01-08  Nick Clifton  <nickc@redhat.com>
26860         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26861         register bank 0 at the start of an interrupt handler.
26862         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26863         MDBH registers.
26865 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26867         * config/aarch64/aarch64-simd.md
26868         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26869         (aarch64_simd_bsl): Likewise.
26870         (aarch64_vcond_internal<mode>): Likewise.
26871         (vcond<mode><mode>): Likewise.
26872         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26873         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26875 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26877         * config/aarch64/aarch64-builtins.c
26878         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26880 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26882         PR debug/55579
26883         * tree-sra.c (analyze_access_subtree): Return true also after
26884         potentially creating a debug-only replacement.
26886 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26888         PR middle-end/55890
26889         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26891         PR tree-optimization/54120
26892         * tree-vrp.c (range_fits_type_p): Don't allow
26893         src_precision < precision from signed vr to unsigned_p
26894         if vr->min or vr->max is negative.
26895         (simplify_float_conversion_using_ranges): Test can_float_p
26896         against CODE_FOR_nothing.
26898 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26899             Richard Biener  <rguenther@suse.de>
26901         PR middle-end/55851
26902         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26903         types instead of just INTEGER_TYPE types.
26905 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26907         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26908         TF_SIZE): Define.
26910 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26912         PR target/42661
26913         * config/mips/mips.opt: Change mad to mmad to match documentation.
26915 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26917         PR target/55897
26918         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26919         .progmemx.data now.
26921 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26923         PR target/55897
26924         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26925         (avr_addrspace_t): Add .section_name field.
26926         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26927         array size.
26928         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26929         NULL entry.  Put __memx into .progmemx.data.
26930         (progmem_section_prefix): Remove.
26931         (avr_asm_init_sections): No need to initialize progmem_section.
26932         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26933         section name prefix.
26934         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26935         retrieve the progmem section.
26936         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26937         boundary to run over avr_addrspace[].
26938         (avr_register_target_pragmas): Ditto.
26940 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26942         * varasm.c (output_constant_def_contents): For asan_protect_global
26943         protected strings, adjust DECL_ALIGN if needed, before testing for
26944         anchored symbols.
26945         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26946         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26947         normal decls.
26948         (output_object_block): For asan protected decls, emit asan padding
26949         after their contents.
26950         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26951         (asan_finish_file): Test it here instead.
26953 2013-01-07  Nick Clifton  <nickc@redhat.com>
26954             Matthias Klose  <doko@debian.org>
26955             Doug Kwan  <dougkwan@google.com>
26956             H.J. Lu  <hongjiu.lu@intel.com>
26958         PR driver/55470
26959         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26961         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26963         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26965         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26967         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26969 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26971         PR target/54461
26972         * doc/install.texi (Cross-Compiler-Specific Options): Document
26973         --with-avrlibc.
26975 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26977         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26978         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26979         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26980         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26981         vqmovun_high_s64): Fix source operand number and update copyright.
26983 2013-01-07  Richard Biener  <rguenther@suse.de>
26985         PR middle-end/55890
26986         * gimple.h (gimple_call_builtin_p): New overload.
26987         * gimple.c (validate_call): New function.
26988         (gimple_call_builtin_p): Likewise.
26989         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26990         Use gimple_call_builtin_p.
26991         (find_func_clobbers): Likewise.
26992         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26993         (strlen_optimize_stmt): Likewise.
26995 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26997         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26998         (vld1q_dup_*): Likewise.
26999         (vld1_*): Likewise.
27000         (vld1q_*): Likewise.
27001         (vld1_lane_*): Likewise.
27002         (vld1q_lane_*): Likewise.
27004 2013-01-07  Richard Biener  <rguenther@suse.de>
27006         * lto-streamer.h (LTO_minor_version): Bump to 2.
27008 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27010         * config/aarch64/aarch64-protos.h
27011         (aarch64_const_double_zero_rtx_p): Rename to...
27012         (aarch64_float_const_zero_rtx_p): ...this.
27013         (aarch64_float_const_representable_p): New.
27014         (aarch64_output_simd_mov_immediate): Likewise.
27015         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
27016         move immediate case.
27017         * config/aarch64/aarch64.c
27018         (aarch64_const_double_zero_rtx_p): Rename to...
27019         (aarch64_float_const_zero_rtx_p): ...this.
27020         (aarch64_print_operand): Allow printing of new constants.
27021         (aarch64_valid_floating_const): New.
27022         (aarch64_legitimate_constant_p): Check for valid floating-point
27023         constants.
27024         (aarch64_simd_valid_immediate): Likewise.
27025         (aarch64_vect_float_const_representable_p): New.
27026         (aarch64_float_const_representable_p): Likewise.
27027         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
27028         (aarch64_output_simd_mov_immediate): New.
27029         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
27030         (*movdf_aarch64): Likewise.
27031         * config/aarch64/constraints.md (Ufc): New.
27032         (Y): call aarch64_float_const_zero_rtx.
27033         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
27035 2013-01-07  Richard Biener  <rguenther@suse.de>
27037         PR tree-optimization/55888
27038         PR tree-optimization/55862
27039         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
27040         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
27041         not if it is contained therein.
27043 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27045         * config/avr/t-avr: Typo.
27047 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27049         PR55243
27050         * config/avr/t-avr: Don't automatically rebuild
27051         $(srcdir)/config/avr/t-multilib
27052         $(srcdir)/config/avr/avr-tables.opt
27053         $(srcdir)/doc/avr-mmcu.texi
27054         (avr-mcus): New phony target to build them on request.
27055         (s-avr-mlib, s-avr-mmcu-texi): Remove.
27056         * avr/avr-mcus.def: Adjust comments.
27058 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
27060         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
27062 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27064         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
27066 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27068         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27070 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27072         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27073         to generate profiling.
27074         * config/rs6000/aix64.h (LIB_SPEC): Same.
27076 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27078         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27079         New function.
27080         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27082 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27084         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27085         unconditionally.
27086         (ix86_expand_move): Ditto.
27087         (ix86_zero_extend_to_Pmode): Ditto.
27088         (ix86_expand_call): Ditto.
27089         (ix86_expand_special_args_builtin): Ditto.
27090         (ix86_expand_builtin): Ditto.
27092 2013-01-04  Richard Biener  <rguenther@suse.de>
27094         PR tree-optimization/55862
27095         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27096         translating them through PHI nodes.
27098 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27100         PR tree-optimization/55755
27101         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27102         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27104 2013-01-04  Marek Polacek  <polacek@redhat.com>
27106         PR middle-end/55859
27107         * opts.c (default_options_optimization): Clarify error message.
27109 2013-01-04  Richard Biener  <rguenther@suse.de>
27111         PR middle-end/55863
27112         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27113         reassociation.
27115 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27117         PR target/53789
27118         * config/pa/pa.md (movsi): Revert previous change.
27119         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27120         references.
27122 2013-01-03  Richard Henderson  <rth@redhat.com>
27124         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27125         after eliminating TLS symbols.
27127 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27129         PR bootstrap/50167
27130         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27131         * graphite-poly.c (debug_gmp_value): Likewise.
27133 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27135         PR target/55712
27136         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27137         selected code model, define __code_mode_small__, __code_model_medium__,
27138         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27139         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27140         xchg temporary register with %k.  Declare temporary register as
27141         early clobbered.
27142         [__x86_64__]: For medium and large code models, preserve %rbx register.
27144 2013-01-03  Richard Biener  <rguenther@suse.de>
27146         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27147         (dump_subscript): Adjust.
27148         (finalize_ddr_dependent): Do not dump redundant info.
27149         (analyze_siv_subscript): Adjust.
27150         (subscript_dependence_tester): Likewise.
27151         (compute_affine_dependence): Likewise.
27153 2013-01-03  Richard Biener  <rguenther@suse.de>
27155         Revert
27156         2013-01-03  Richard Biener  <rguenther@suse.de>
27158         PR tree-optimization/55857
27159         * tree-vect-stmts.c (vectorizable_load): Do not setup
27160         re-alignment for invariant loads.
27162         2013-01-02  Richard Biener  <rguenther@suse.de>
27164         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27165         invariant load do not generate a vector load from the scalar location.
27167 2013-01-03  Richard Biener  <rguenther@suse.de>
27169         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27170         for not vectorizing.
27171         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27172         not build INDIRECT_REFs, call get_name once only.
27173         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27174         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27176 2013-01-03  Richard Biener  <rguenther@suse.de>
27178         PR tree-optimization/55857
27179         * tree-vect-stmts.c (vectorizable_load): Do not setup
27180         re-alignment for invariant loads.
27182 2013-01-03  Richard Biener  <rguenther@suse.de>
27184         PR lto/55848
27185         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27186         prefer a built-in decl.
27188 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27190         * gcc.c (process_command): Update copyright notice dates.
27191         * gcov.c (print_version): Likewise.
27192         * gcov-dump.c (print_version): Likewise.
27194         PR rtl-optimization/55838
27195         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27196         iv0.step, iv1.step and step.
27198 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27199             Marc Glisse  <marc.glisse@inria.fr>
27201         PR tree-optimization/55832
27202         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27203         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27204         integer_{one,zero}_node.
27206 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27208         PR debug/54402
27209         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27210         * var-tracking.c (reverse_op): Don't add reverse ops to
27211         VALUEs that have already
27212         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27214 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27216         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27218 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27220         * dumpfile.c (dump_loc): Print filename with location.
27221         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27222         new location_t parameter to emit complete unroll message with
27223         new dump framework.
27224         (canonicalize_loop_induction_variables): Compute loops location
27225         and pass to try_unroll_loop_completely.
27226         * loop-unroll.c (report_unroll_peel): New function.
27227         (peel_loops_completely): Use new dump format with location
27228         for main dumpfile message, and invoke report_unroll_peel on success.
27229         (decide_unrolling_and_peeling): Ditto.
27230         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27231         by report_unroll_peel.
27232         (decide_peel_completely): Ditto.
27233         (decide_unroll_constant_iterations): Ditto.
27234         (decide_unroll_runtime_iterations): Ditto.
27235         (decide_peel_simple): Ditto.
27236         (decide_unroll_stupid): Ditto.
27237         * cfgloop.c (get_loop_location): New function.
27238         * cfgloop.h (get_loop_location): Declare.
27240 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27242         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27243         NULL.
27245 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27247         PR middle-end/55198
27248         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27249         BLKmode objects when EXPAND_MEMORY is specified.
27251 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27253         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27254         in loop predicate.
27255         (fold_builtin_cpu): Do not share cpu model decls across statements.
27257 2013-01-02  Jason Merrill  <jason@redhat.com>
27259         PR c++/55804
27260         * tree.c (build_array_type_1): Revert earlier change.
27262 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27264         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27265         "cortex-a57".
27266         * config/aarch64/aarch64-tune.md: Re-generate.
27268 2013-01-02  Richard Biener  <rguenther@suse.de>
27270         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27271         invariant load do not generate a vector load from the scalar location.
27273 2013-01-02  Richard Biener  <rguenther@suse.de>
27275         PR bootstrap/55784
27276         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27277         * configure: Regenerate.
27279 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27281         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27282         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27283         (expand_builtin_int_roundingfn_2): Keep the original target around
27284         for the fallback case.
27286 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27288         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27289         to be clear for sign changes.
27291 2013-01-01  Jan Hubicka  <jh@suse.cz>
27293         * ipa-inline-analysis.c: Fix formatting.
27295 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27297         PR tree-optimization/55831
27298         * tree-vect-loop.c (get_initial_def_for_induction): Use
27299         gsi_after_labels instead of gsi_start_bb.
27301 Copyright (C) 2013 Free Software Foundation, Inc.
27303 Copying and distribution of this file, with or without modification,
27304 are permitted in any medium without royalty provided the copyright
27305 notice and this notice are preserved.