2013-10-23 Jan-Benedict Glaw <jbglaw@lug-owl.de>
[official-gcc.git] / gcc / ChangeLog
blob57c7406e4546fed7416b73e9e61914816b7f811c
1 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3         * config/tilegx/tilegx.c: Include "tree.h".
5 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
7         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
8         the info, not after it.
9         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
10         dump_ssaname_info call.
11         (pp_gimple_stmt_1): Adjust caller.
12         (dump_phi_nodes): Likewise.  Don't print "# " here.
14 2013-10-22  Jan Hubicka  <jh@suse.cz>
16         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New tuning flag.
17         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
18         * i386.c (expand_small_movmem_or_setmem): New function.
19         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New function
20         (alg_usable_p): Add support for value ranges; cleanup.
21         (ix86_expand_set_or_movmem): Add support for misaligned moves.
23 2013-10-22  Sterling Augustine  <saugustine@google.com>
25         * doc/invoke.texi: Document -ggnu-pubnames.
26         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
27         logic.
28         * dwarf2out.c: Include gdb/gdb-index.h.
29         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
30         debug_generate_pub_sections.
31         (is_java, output_pubtables, output_pubname): New functions.
32         (include_pubname_in_output): Handle debug_generate_pub_sections at
33         level 2.
34         (size_of_pubnames): Use new local space_for_flags based on
35         debug_generate_pub_sections.
36         (output_pubnames): Unify pubnames and pubtypes output logic.
37         Genericize comments.  Call output_pubname.
38         (dwarf2out_finish): Move logic to output_pubnames and call it.
40 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
42         PR target/58779
43         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
44         Remove CCCmode handling.
45         <case LTU>: Return 'c' suffix for CCCmode.
46         <case GEU>: Return 'nc' suffix for CCCmode.
47         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
48         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
49         (*sub<mode>3_cc_overflow): Ditto.
50         (*subsi3_zext_cc_overflow): Ditto.
52 2013-10-22  Steve Ellcey  <sellcey@mips.com>
54         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
55         (AND (NOT OP1) (NOT OP2)).
57 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
59         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
60         meaning of merge-high and merge-low masks for little endian; avoid
61         use of vector-pack masks for little endian for mismatched modes.
63 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
65         * config/tilepro/tilepro.c: Include "tree.h".
67 2013-10-22  Andreas Schwab  <schwab@suse.de>
69         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
70         with PRE_DEC.
72 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
74         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
75         [Ed Smith-Rowland]): New entries.
76         ([Stephen M. Webb]): Update.
78 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
80         * tree-ssa-ter.h: Remove duplicate copy of file contents.
82 2013-10-21  Marek Polacek  <polacek@redhat.com>
84         PR middle-end/58809
85         * fold-const.c (fold_range_test): Return 0 if the type is not
86         an integral type.
88 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
90         * system.h: Move hwint.h include further down.
91         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
92         gcc_checking_asserts.
93         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
95 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
97         Fix volatile issues in optimize_bit_field_compare.
98         * fold-const.c (optimize_bit_field_compare): Bail out if
99         lvolatilep or rvolatilep.
101 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
103         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
104         and get_inner_reference returning different pmode for non-volatile
105         bit-field members dependent on flag_strict_volatile_bitfields.
106         * stor-layout.c (layout_decl): Remove special handling of
107         flag_strict_volatile_bitfields.
108         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
109         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
111 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
113         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
114         calculations.
116 2013-10-21  Jeff Law  <law@redhat.com>
118         * tree-ssa-threadedge.c (thread_through_normal_block): New
119         argument VISITED.  Remove VISISTED as a local variable.  When we
120         have a threadable jump, verify the destination of the jump has not
121         been visised.
122         (thread_across_edge): Allocate VISITED bitmap once at function
123         scope and use it throughout.  Make sure to set appropriate bits in
124         VISITED for E (start of jump thread path).
125         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
126         through a joiner if any edge on the path has a recorded jump thread.
128 2013-10-21  Ian Lance Taylor  <iant@google.com>
130         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
131         don't imply that attributes can solve all problems.
132         (Directory Options): Fix typo.
134 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
136         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
137         extend and extend_add.
139 2013-10-21  Richard Biener  <rguenther@suse.de>
141         PR tree-optimization/58794
142         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
143         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
145 2013-10-21  Richard Biener  <rguenther@suse.de>
147         PR middle-end/58742
148         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
149         to (T) X for sign-changing conversions (or no conversion).
151 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
153         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
155 2013-10-20  Jan Hubicka  <jh@suse.cz>
157         * config/i386/i386-tune.def: Add comment; organize into categories
159 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
161         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
162         argument.  Update function comment.
163         (expand_set_or_movmem_via_rep): New function combining
164         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
165         (expand_movmem_via_rep_mov): Remove.
166         expand_setmem_via_rep_stos): Remove.
167         (expand_movmem_epilogue): Update calls correspondingly.
168         (expand_setmem_epilogue_via_loop): Likewise.
169         (emit_memset): New.
170         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
171         (expand_set_or_movmem_prologue): New function combining
172         expand_movmem_prologue and expand_setmem_prologue.
173         (expand_movmem_prologue): Remove.
174         (expand_setmem_prologue): Remove.
175         (expand_set_or_movmem_constant_prologue): New function combining
176         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
177         (expand_constant_movmem_prologue): Remove.
178         (expand_constant_setmem_prologue): Remove.
179         (promote_duplicated_reg): Allow vector-const0 value.
180         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
181         and ix86_expand_setmem.
182         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
183         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
185 2013-10-21  Diego Novillo  <dnovillo@google.com>
187         * asan.c: Include tree.h
188         * bb-reorder.c: Likewise.
189         * cfgcleanup.c: Likewise.
190         * cfgloopmanip.c: Likewise.
191         * data-streamer-in.c: Likewise.
192         * data-streamer-out.c: Likewise.
193         * data-streamer.c: Likewise.
194         * dwarf2cfi.c: Likewise.
195         * graphite-blocking.c: Likewise.
196         * graphite-clast-to-gimple.c: Likewise.
197         * graphite-dependences.c: Likewise.
198         * graphite-interchange.c: Likewise.
199         * graphite-optimize-isl.c: Likewise.
200         * graphite-poly.c: Likewise.
201         * graphite-scop-detection.c: Likewise.
202         * graphite-sese-to-poly.c: Likewise.
203         * graphite.c: Likewise.
204         * ipa-devirt.c: Likewise.
205         * ipa-profile.c: Likewise.
206         * ipa.c: Likewise.
207         * ira.c: Likewise.
208         * loop-init.c: Likewise.
209         * loop-unroll.c: Likewise.
210         * lower-subreg.c: Likewise.
211         * lto/lto-object.c: Likewise.
212         * recog.c: Likewise.
213         * reginfo.c: Likewise.
214         * tree-loop-distribution.c: Likewise.
215         * tree-parloops.c: Likewise.
216         * tree-ssa-strlen.c: Likewise.
217         * tree-streamer.c: Likewise.
218         * value-prof.c: Likewise.
219         * target-globals.c: Likewise.
220         * expr.h: Include tree-core.h instead of tree.h.
221         * gimple.h: Likewise.
222         * ipa-prop.h: Likewise.
223         * ipa-utils.h: Likewise.
224         * lto-streamer.h: Likewise.
225         * streamer-hooks.h: Likewise.
226         * ipa-reference.h: Include cgraph.h instead of tree.h.
227         * cgraph.h: Include basic-block.h instead of tree.h.
228         * tree-streamer.h: Do not include tree.h.
229         * genattrtab.c (write_header): Generate inclusion of
230         tree.h.
231         * genautomata.c (main): Likewise.
232         * genemit.c: Likewise.
233         * genopinit.c: Likewise.
234         * genoutput.c (output_prologue): Likewise.
235         * genpeep.c: Likewise.
237 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
239         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
240         little endian.
241         (vec_unpacku_hi_v8hi): Likewise.
242         (vec_unpacku_lo_v16qi): Likewise.
243         (vec_unpacku_lo_v8hi): Likewise.
245 2013-10-20  Jan Hubicka  <jh@suse.cz>
247         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
248         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
249         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
251 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
253         * config/mips/mips.h (ISA_HAS_WSBH): Define.
254         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
255         constants.
256         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
258 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
260         PR target/58603
261         * system.h: Undef m_slot.
263 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
265         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
266         all elements for both endian flavors.
268 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
270         PR target/58792
271         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
272         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
273         and SI_REG for 64bit SYSV ABI targets.
275 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
277         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
278         to multi_reg_return.  Clarify that we are skipping USEs of multiple
279         return registers.  Use bool type where appropriate.
281 2013-10-18  Jan Hubicka  <jh@suse.cz>
283         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
284         for cold functions.
285         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
286         (X86_TUNE_PUSH_MEMORY): Likewise.
287         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
288         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
289         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
290         New.
291         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
292         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
293         Remove.
294         (ix86_option_override_internal): Update to use tune features instead
295         of variables.
297 2013-10-18  Cong Hou  <congh@google.com>
299         PR tree-optimization/58508
300         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
301         statement that contains data refs with zero-step.
303 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
305         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
306         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
307         * gimple-low.c (gimple_check_call_arg,
308         gimple_check_call_matching_types): Move to cgraph.c.
309         * gimple-low.h: Remove prototype.
310         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
311         Relocate from gimple-low.c.
312         * cgraph.h: Add prototype.  Don't include basic-block.h.
313         * gimplify.c: Add gimple-low to include list.
314         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
315         * tree-eh.c: Add gimple-low to include list.
316         * tree-nested.c: Likewise.
317         * cfgexpand.c: Add tree-ssa-address.h to include list.
318         * expr.c: Likewise.
319         * gimple-fold.c: Likewise.
320         * gimple-ssa-strength-reduction.c: Likewise.
321         * trans-mem.c: Likewise.
322         * tree-mudflap.c: Likewise.
323         * tree-ssa-loop-ivopts.c: Likewise.
324         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
325         (degenerate_phi_result): Move to tree-phinodes.c.
326         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
327         * tree-ssa-threadedge.c: Likewise.
328         * tree-vrp.c: Likewise.
329         * tree-phinodes.c (degenerate_phi_result): Relocate here.
330         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
331         * tree-phinodes.h (degenerate_phi_result): Add prototype.
332         * tree-ssa-copy.c: Include tree-ssa-dom.h.
333         * tree-ssa-forwprop.c: Likewise.
334         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
335         pass_data_cleanup_cfg_post_optimizing,
336         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
337         * tree-optimize.c: Delete File.
338         * graphite.c: Include tree-cfgcleanup.h.
339         * passes.c: Likewise.
340         * tree-cfg.c: Likewise.
341         * tree-profile.c: Likewise.
342         * tree-ssa-dse.c: Likewise.
343         * tree-ssa-loop-ivcanon.c: Likewise.
344         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
345         * tree-outof-ssa.c: Include sbitmap.h.
346         * tree-ssa-live.c: Likewise.
347         * tree-ssa-propagate.c: Likewise.
348         * tree-ssa-structalias.c: Likewise.
349         * tree-stdarg.c: Likewise.
350         * Makefile.in (OBJS): Delete tree-optimize.o.
351         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
352         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
353         * varasm.c: Include basic-block.h.
354         * cfgloop.h: Include function.h instead of basic-block.h
355         (bb_loop_depth): Move to cfgloop.c.
356         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
358 2013-10-18  Teresa Johnson  <tejohnson@google.com>
360         * predict.c (probably_never_executed): Compare frequency-based
361         count to number of training runs.
362         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
364 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
366         * config/arm/arm.c (cortexa9_extra_costs): New table.
367         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
369 2013-10-18  Jeff Law  <law@redhat.com>
371         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
373         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
374         (register_jump_thread): Add "registered_jump_thread" debug
375         counter support.
376         * dbgcnt.def (registered_jump_thread): New debug counter.
378 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
380         * config/rs6000/rs6000.c: Include cgraph.h.
382 2013-10-18  Teresa Johnson  <tejohnson@google.com>
384         * tree-ssa-tail-merge.c (replace_block_by): Update edge
385         weights during merging.
387 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
389         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
390         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
391         * tree-ssa-operands.h: Remove prototype.
392         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
393         * gimple.c (virtual_operand_p): Relocate from gimple.c.
394         * gimple.h: Add prototype.
395         * gimple-ssa.h: Include tree-ssa-operands.h.
396         * tree-dump.c: Add tree-cfg.h to include list.
397         * tree-ssa-alias.c: Add ipa-reference.h to include list.
398         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
399         * config/i386/i386.c: Don't include tree-flow.h.
400         * config/rs6000/rs6000.c: Likewise.
402 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
404         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
406 2013-10-18  Richard Biener  <rguenther@suse.de>
408         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
409         or TYPE_UNSIGNED of integral types.
410         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
411         NULL_TREE for zero-precision integral types.
413 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
415         * config/aarch64/arm_neon.h
416         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
418 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
420         * opth-gen.awk: Define target_flags_explicit.
422 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
424         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
425         fields to the reg_addr array that describes the valid addressing
426         mode for any register, general purpose registers, floating point
427         registers, and Altivec registers.
428         (FIRST_RELOAD_REG_CLASS): Likewise.
429         (LAST_RELOAD_REG_CLASS): Likewise.
430         (struct reload_reg_map_type): Likewise.
431         (reload_reg_map_type): Likewise.
432         (RELOAD_REG_VALID): Likewise.
433         (RELOAD_REG_MULTIPLE): Likewise.
434         (RELOAD_REG_INDEXED): Likewise.
435         (RELOAD_REG_OFFSET): Likewise.
436         (RELOAD_REG_PRE_INCDEC): Likewise.
437         (RELOAD_REG_PRE_MODIFY): Likewise.
438         (reg_addr): Likewise.
439         (mode_supports_pre_incdec_p): New helper functions to say whether
440         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
441         (mode_supports_pre_modify_p): Likewise.
442         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
443         print the valid address mode bits for each mode.
444         (rs6000_debug_print_mode): Likewise.
445         (rs6000_debug_reg_global): Likewise.
446         (rs6000_setup_reg_addr_masks): New function to set up the address
447         mask bits for each type.
448         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
449         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
450         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
451         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
452         PRE_MODIFY are supported.
453         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
454         registers, instead of {src,dest}_av_p.
455         (rs6000_print_options_internal): Tweak the debug output slightly.
457 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
459         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
460         isa attribute.
462 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
464         * tree-flow.h (struct omp_region): Move to omp-low.c.
465         Remove omp_ prototypes and variables.
466         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
467         (copy_var_decl): Relocate prototype from tree-flow.h.
468         * gimple.c (copy_var_decl): Relocate from omp-low.c.
469         * tree.h: Move prototype to omp-low.h.
470         * omp-low.h: New File.  Relocate prototypes here.
471         * omp-low.c (struct omp_region): Make local here.
472         (root_omp_region): Make static.
473         (copy_var_decl) Move to gimple.c.
474         (new_omp_region): Make static.
475         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
476         * tree-cfg.c: Include omp-low.h.
477         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
478         * gimplify.c: Include omp-low.h.
479         * tree-parloops.c: Likewise.
481 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
483         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
484         operands are in memory, prefer to force non-matched operand 1 to
485         the register.
487 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
489         PR target/58673
490         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
491         restrict TImode addresses to single indirect registers if both
492         -mquad-memory and -mvsx-timode are used.
493         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
494         we should emit load/store quad.  Remove using %y for quad memory
495         addresses.
497         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
498         constraints to allow load/store quad on machines where TImode is
499         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
500         for TImode to load integer constants.
502 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
504         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
506 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
508         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
509         handling of STACK_REG.
511 2013-10-17  Richard Biener  <rguenther@suse.de>
513         PR tree-optimization/58143
514         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
515         New function.
516         (rewrite_to_defined_overflow): Likewise.
517         (move_computations_dom_walker::before_dom): Rewrite stmts
518         with undefined signed overflow that are not always executed
519         into unsigned arithmetic.
521 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
523         PR target/57756
524         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
525         explicit isa flag to be an options variable, instead of using
526         global_options_set.  Remove define from rs6000.h.
527         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
529         * config/rs6000/rs6000.c (rs6000_option_override_internal):
530         Initialize rs6000_isa_flags_explicit.
531         (rs6000_function_specific_save): Add gcc_options* parameter, so
532         that the powerpc builds after the 2013-10-15 changes.
533         (rs6000_function_specific_restore): Likewise.
535 2013-10-16  DJ Delorie  <dj@redhat.com>
537         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
538         op is a REG before checking REGNO.
539         (rl78_alloc_physical_registers): Verify pattern is a SET before
540         checking SET_SRC.
542 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
544         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
545         endianness.
546         (vec_unpacks_lo_v4sf): Likewise.
547         (vec_unpacks_float_hi_v4si): Likewise.
548         (vec_unpacks_float_lo_v4si): Likewise.
549         (vec_unpacku_float_hi_v4si): Likewise.
550         (vec_unpacku_float_lo_v4si): Likewise.
552 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
554         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
555         (vsx_concat_v2sf): Likewise.
557 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
559         * config/aarch64/aarch64.md
560         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
562 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
564         PR tree-optimization/58697
565         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
566         estimated_loop_iterations_int.
567         (max_stmt_executions_int): Call get_max_loop_iterations_int.
568         (get_max_loop_iterations_int): New.  HWINT version of
569         get_max_loop_iterations.
570         * cfgloop.h: Add prototypes.
571         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
572         * loop-unroll.c (decide_peel_once_rolling): Call
573         get_estimated_loop_iterations_int.
574         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
575         * tree-ssa-loop-niter.h: Tweak prototypes.
577 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
579         * gengtype-parse.c (struct_field_seq): Ignore access-control
580         keywords ("public:" etc).
582 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
584         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
585         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
587 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
589         * config/arm/arm.opt (mlra): New option.
590         * config/arm/arm.c (arm_lra_p): New function.
591         (TARGET_LRA_P): Define.
593 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
595         * tree-core.h (tree_code_name): Remove.
596         * tree.h (get_tree_code_name): New prototype.
597         * tree.c (tree_code_name): Make static.
598         (get_tree_code_name): New function.
599         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
600         tree_class_check_failed, tree_range_check_failed,
601         tree_not_class_check_failed, omp_clause_check_failed,
602         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
603         wrapper get_tree_code_name instead of calling tree_code_name directly.
604         * tree-vrp.c (dump_asserts_for): Likewise.
605         * tree-dump.c (dequeue_and_dump): Likewise.
606         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
607         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
608         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
609         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
610         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
611         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
612         dump_gimple_omp_for): Likewise.
613         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
614         * tree-ssa-pre.c (print_pre_expr): Likewise.
615         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
616         * print-tree.c (print_node_brief, print_node): Likewise.
617         * gimple.c (gimple_check_failed): Likewise.
618         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
619         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
620         * config/mep/mep.c (mep_validate_vliw): Likewise.
621         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
622         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
624 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
626         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
627         for AMD bdver3.
629 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
631         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
632         (MULTILIB_MATCHES): Add multilib for -march=v8.
634 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
636         PR target/57756
637         * optc-save-gen.awk: Add extra parameter to the save and restore
638         target calls.
639         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
640         * tree.c (build_optimization_node): New parameter.  Add extra parameter
641         to call to cl_optimization_save.
642         (build_target_option_node): New parameter. Add extra parameter
643         to call to cl_target_option_save.
644         * tree.h (build_optimization_node): New parameter.
645         (build_target_option_node): New parameter.
646         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
647         build_optimization_node and build_target_option_node.
648         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
649         (handle_pragma_push_options): Ditto.
650         * toplev.c (process_options): Ditto.
651         * opts.c (init_options_struct): Check for opts_set non-null.
652         * target.def (target_option.save): New parameter.
653         (target_option.restore): New parameter.
654         * tm.texi: Generate.
655         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
656         (ix86_pragma_target_parse): Ditto.
657         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
658         parameters.
659         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
660         to  build_optimization_node and build_target_option_node.
661         (rs6000_valid_attribute_p): Ditto.
662         (rs6000_pragma_target_parse): Ditto.
663         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
664         data.
665         * config/i386/i386.h:
666         TARGET_64BIT_P: New Macro
667         TARGET_MMX_P: New Macro.
668         TARGET_3DNOW_P: New Macro.
669         TARGET_3DNOW_A_P: New Macro.
670         TARGET_SSE_P: New Macro.
671         TARGET_SSE2_P: New Macro.
672         TARGET_SSE3_P: New Macro.
673         TARGET_SSSE3_P: New Macro.
674         TARGET_SSE4_1_P: New Macro.
675         TARGET_SSE4_2_P: New Macro.
676         TARGET_AVX_P: New Macro.
677         TARGET_AVX2_P: New Macro.
678         TARGET_AVX512F_P: New Macro.
679         TARGET_AVX512PF_P: New Macro.
680         TARGET_AVX512ER_P: New Macro.
681         TARGET_AVX512CD_P: New Macro.
682         TARGET_FMA_P: New Macro.
683         TARGET_SSE4A_P: New Macro.
684         TARGET_FMA4_P: New Macro.
685         TARGET_XOP_P: New Macro.
686         TARGET_LWP_P: New Macro.
687         TARGET_ABM_P: New Macro.
688         TARGET_BMI_P: New Macro.
689         TARGET_BMI2_P: New Macro.
690         TARGET_LZCNT_P: New Macro.
691         TARGET_TBM_P: New Macro.
692         TARGET_POPCNT_P: New Macro.
693         TARGET_SAHF_P: New Macro.
694         TARGET_MOVBE_P: New Macro.
695         TARGET_CRC32_P: New Macro.
696         TARGET_AES_P: New Macro.
697         TARGET_PCLMUL_P: New Macro.
698         TARGET_CMPXCHG16B_P: New Macro.
699         TARGET_FSGSBASE_P: New Macro.
700         TARGET_RDRND_P: New Macro.
701         TARGET_F16C_P: New Macro.
702         TARGET_RTM_P: New Macro.
703         TARGET_HLE_P: New Macro.
704         TARGET_RDSEED_P: New Macro.
705         TARGET_PRFCHW_P: New Macro.
706         TARGET_ADX_P: New Macro.
707         TARGET_FXSR_P: New Macro.
708         TARGET_XSAVE_P: New Macro.
709         TARGET_XSAVEOPT_P: New Macro.
710         TARGET_LP64_P: New Macro.
711         TARGET_X32_P: New Macro.
712         TARGET_FPMATH_DEFAULT_P: New Macro.
713         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
714         * config/i386/i386.c (ix86_option_override_internal): New parameters.
715         opts and opts_set.
716         Change ix86_tune_string to access opts->x_ix86_tune_string.
717         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
718         Change ix86_arch_string to access opts->x_ix86_arch_string.
719         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
720         Change ix86_pmode to access opts->x_ix86_pmode.
721         Change ix86_abi to access opts->x_ix86_abi.
722         Change ix86_cmodel to access opts->x_ix86_cmodel.
723         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
724         Change ix86_isa_flags_explicit to access
725         opts->x_ix86_isa_flags_explicit.
726         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
727         Change ix86_regparm to access opts->x_ix86_regparm.
728         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
729         Change ix86_preferred_stack_boundary_arg to access
730         opts->x_ix86_preferred_stack_boundary_arg.
731         Change ix86_force_align_arg_pointer to access
732         opts->x_ix86_force_align_arg_pointer.
733         Change ix86_incoming_stack_boundar_arg to access
734         opts->x_ix86_incoming_stack_boundar_arg.
735         Change ix86_fpmath to access opts->x_ix86_fpmath.
736         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
737         Change ix86_recip_name to access opts->x_ix86_recip_name.
738         Change ix86_stack_protector_guard to access
739         opts->x_ix86_stack_protector_guard.
740         Change ix86_tune_memcpy_strategy to access
741         opts->x_ix86_tune_memcpy_strategy.
742         Change ix86_tune_memset_strategy to access
743         opts->x_ix86_tune_memset_strategy.
744         Change global_options to access opts.
745         Change global_options_set to access opts_set.
746         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
747         Change TARGET_MMX to TARGET_MMX_P (opts->...).
748         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
749         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
750         Change TARGET_SSE to TARGET_SSE_P (opts->...).
751         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
752         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
753         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
754         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
755         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
756         Change TARGET_AVX to TARGET_AVX_P (opts->...).
757         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
758         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
759         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
760         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
761         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
762         Change TARGET_FMA to TARGET_FMA_P (opts->...).
763         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
764         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
765         Change TARGET_XOP to TARGET_XOP_P (opts->...).
766         Change TARGET_LWP to TARGET_LWP_P (opts->...).
767         Change TARGET_ABM to TARGET_ABM_P (opts->...).
768         Change TARGET_BMI to TARGET_BMI_P (opts->...).
769         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
770         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
771         Change TARGET_TBM to TARGET_TBM_P (opts->...).
772         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
773         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
774         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
775         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
776         Change TARGET_AES to TARGET_AES_P (opts->...).
777         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
778         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
779         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
780         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
781         Change TARGET_F16C to TARGET_F16C_P (opts->...).
782         Change TARGET_RTM to TARGET_RTM_P (opts->...).
783         Change TARGET_HLE to TARGET_HLE_P (opts->...).
784         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
785         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
786         Change TARGET_ADX to TARGET_ADX_P (opts->...).
787         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
788         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
789         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
790         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
791         Change TARGET_X32 to TARGET_X32_P (opts->...).
792         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
793         Change TARGET_FLOAT_RETURNS_IN_80387 to
794         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
795         (ix86_function_specific_save): New parameter. Use opts-> fields
796         to replace global fields.
797         (ix86_function_specific_restore): Ditto.
798         (ix86_valid_target_attribute_inner_p): New parameters.
799         Fix recursive call.
800         Fix call to ix86_handle_option and set_option.
801         (ix86_valid_target_attribute_tree): New parameters.
802         Change global_options to access opts.
803         Change global_options_set to access opts_set.
804         Fix call to ix86_valid_target_attribute_inner_p.
805         Change ix86_tune_string to access opts->x_ix86_tune_string.
806         Change ix86_arch_string to access opts->x_ix86_arch_string.
807         Change ix86_fpmath to access opts->x_ix86_fpmath
808         Fix call to ix86_option_override_internal.
809         Fix call to ix86_add_new_builtins.
810         Fix calls to build_optimization_node and build_target_option_node.
811         (ix86_valid_target_attribute_p): Remove access to global_options.
812         Use new gcc_options structure func_options.
813         Fix call to ix86_valid_target_attribute_tree.
814         Fix call to  build_optimization_node.
815         (get_builtin_code_for_version): Fix call to
816         ix86_valid_target_attribute_tree.
818 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
820         * Makefile.in (PICFLAG): New.
821         (enable_host_shared): New.
822         (INTERNAL_CFLAGS): Use PICFLAG.
823         (LIBIBERTY): Use pic build of libiberty.a if configured with
824         --enable-host-shared.
825         * configure.ac: Add --enable-host-shared, setting up new
826         PICFLAG variable.
827         * configure: Regenerate.
828         * doc/install.texi (--enable-shared): Add note contrasting it with ...
829         (--enable-host-shared): New option.
831 2013-10-15  Richard Biener  <rguenther@suse.de>
833         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
834         for built-in functions.
836 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
838         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
839         (optimize_range_tests_1): New function,
840         extracted from optimize_range_tests.
841         (optimize_range_tests_xor): Similarly.
842         (optimize_range_tests_diff): New function.
843         (optimize_range_tests): Use optimize_range_tests_1.
845 2013-10-15  Cong Hou  <congh@google.com>
847         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
848         requirement of the reduction pattern so that one operand of the
849         reduction operation can come from outside of the loop.
851 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
853         * config/arm/neon-schedgen.ml: Remove.
854         * config/arm/cortex-a9-neon.md: Remove comment regarding
855         neon-schedgen.ml.
857 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
859         * config/arm/types: Remove old neon types.
861 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
863         * config/arm/cortex-a7.md
864         (cortex_a7_neon_type): New.
865         (cortex_a7_neon_mul): Update for new types.
866         (cortex_a7_neon_mla): Likewise.
867         (cortex_a7_neon): Likewise.
869 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
871         * config/arm/cortex-a15-neon.md
872         (cortex_a15_neon_type): New,
874         (cortex_a15_neon_int_1): Remove.
875         (cortex_a15_neon_int_2): Likewise.
876         (cortex_a15_neon_int_3): Likewise.
877         (cortex_a15_neon_int_4): Likewise.
878         (cortex_a15_neon_int_5): Likewise.
879         (cortex_a15_neon_vqneg_vqabs): Likewise.
880         (cortex_a15_neon_vmov): Likewise.
881         (cortex_a15_neon_vaba): Likewise.
882         (cortex_a15_neon_vaba_qqq): Likewise.
883         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
884         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
885         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
886         Likewise.
887         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
888         (cortex_a15_neon_mla_qqq_8_16): Likewise.
889         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
890         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
891         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
892         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
893         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
894         (cortex_a15_neon_shift_1): Likewise.
895         (cortex_a15_neon_shift_2): Likewise.
896         (cortex_a15_neon_shift_3): Likewise.
897         (cortex_a15_neon_vshl_ddd): Likewise.
898         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
899         (cortex_a15_neon_vsra_vrsra): Likewise.
900         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
901         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
902         (cortex_a15_neon_bp_3cycle): Likewise.
903         (cortex_a15_neon_ldm_2): Likewise.
904         (cortex_a15_neon_stm_2): Likewise.
905         (cortex_a15_neon_mcr): Likewise.
906         (cortex_a15_neon_mrc): Likewise.
907         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
908         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
909         (cortex_a15_neon_fp_vmul_ddd): Likewise.
910         (cortex_a15_neon_fp_vmul_qqd): Likewise.
911         (cortex_a15_neon_fp_vmla_ddd): Likewise.
912         (cortex_a15_neon_fp_vmla_qqq): Likewise.
913         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
914         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
915         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
916         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
917         (cortex_a15_neon_bp_simple): Likewise.
918         (cortex_a15_neon_bp_2cycle): Likewise.
919         (cortex_a15_neon_bp_3cycle): Likewise.
920         (cortex_a15_neon_vld1_1_2_regs): Likewise.
921         (cortex_a15_neon_vld1_3_4_regs): Likewise.
922         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
923         (cortex_a15_neon_vld2_4_regs): Likewise.
924         (cortex_a15_neon_vld3_vld4): Likewise.
925         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
926         (cortex_a15_neon_vst1_3_4_regs): Likewise.
927         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
928         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
929         (cortex_a15_neon_vst3_vst4): Rename to...
930         (cortex_a15_neon_vst4): This, update for new attributes.
931         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
932         (cortex_a15_neon_vld3_vld4_lane): Likewise.
933         (cortex_a15_neon_vst1_vst2_lane): Likewise.
934         (cortex_a15_neon_vst3_vst4_lane): Likewise.
935         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
936         (cortex_a15_neon_ldm_2): Likewise.
937         (cortex_a15_neon_stm_2): Likewise.
938         (cortex_a15_neon_mcr): Likewise.
939         (cortex_a15_neon_mcr_2_mcrr): Likewise.
940         (cortex_a15_neon_mrc): Likewise.
941         (cortex_a15_neon_mrrc): Likewise.
943         (cortex_a15_neon_abd): New.
944         (cortex_a15_neon_abd_q): Likewise.
945         (cortex_a15_neon_aba): Likewise.
946         (cortex_a15_neon_aba_q): Likewise.
947         (cortex_a15_neon_acc): Likewise.
948         (cortex_a15_neon_acc_q): Likewise.
949         (cortex_a15_neon_arith_basic): Likewise.
950         (cortex_a15_neon_arith_complex): Likewise.
951         (cortex_a15_neon_multiply): Likewise.
952         (cortex_a15_neon_multiply_q): Likewise.
953         (cortex_a15_neon_mla): Likewise.
954         (cortex_a15_neon_mla_q): Likewise.
955         (cortex_a15_neon_sat_mla_long): Likewise.
956         (cortex_a15_neon_shift_acc): Likewise.
957         (cortex_a15_neon_shift_imm_basic): Likewise.
958         (cortex_a15_neon_shift_imm_complex): Likewise.
959         (cortex_a15_neon_shift_reg_basic): Likewise.
960         (cortex_a15_neon_shift_reg_basic_q): Likewise.
961         (cortex_a15_neon_shift_reg_complex): Likewise.
962         (cortex_a15_neon_shift_reg_complex_q): Likewise.
963         (cortex_a15_neon_fp_negabs): Likewise
964         (cortex_a15_neon_fp_arith): Likewise
965         (cortex_a15_neon_fp_arith_q): Likewise
966         (cortex_a15_neon_fp_cvt_int): Likewise
967         (cortex_a15_neon_fp_cvt_int_q): Likewise
968         (cortex_a15_neon_fp_cvt_16): Likewise
969         (cortex_a15_neon_fp_mul): Likewise
970         (cortex_a15_neon_fp_mul_q): Likewise
971         (cortex_a15_neon_fp_mla): Likewise
972         (cortex_a15_neon_fp_mla_q): Likewise
973         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
974         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
975         (cortex_a15_neon_bitops): Likewise.
976         (cortex_a15_neon_bitops_q): Likewise.
977         (cortex_a15_neon_from_gp): Likewise.
978         (cortex_a15_neon_from_gp_q): Likewise.
979         (cortex_a15_neon_tbl3_tbl4): Likewise.
980         (cortex_a15_neon_zip_q): Likewise.
981         (cortex_a15_neon_to_gp): Likewise.
982         (cortex_a15_neon_load_a): Likewise.
983         (cortex_a15_neon_load_b): Likewise.
984         (cortex_a15_neon_load_c): Likewise.
985         (cortex_a15_neon_load_d): Likewise.
986         (cortex_a15_neon_load_e): Likewise.
987         (cortex_a15_neon_load_f): Likewise.
988         (cortex_a15_neon_store_a): Likewise.
989         (cortex_a15_neon_store_b): Likewise.
990         (cortex_a15_neon_store_c): Likewise.
991         (cortex_a15_neon_store_d): Likewise.
992         (cortex_a15_neon_store_e): Likewise.
993         (cortex_a15_neon_store_f): Likewise.
994         (cortex_a15_neon_store_g): Likewise.
995         (cortex_a15_neon_store_h): Likewise.
996         (cortex_a15_vfp_to_from_gp): Likewise.
998 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1000         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
1002         (cortex_a9_neon_vshl_ddd): Remove.
1003         (cortex_a9_neon_vst3_vst4): Likewise.
1004         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1006         (cortex_a9_neon_bit_ops_q): New.
1008         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
1009         (cortex_a9_neon_int_2): Likewise.
1010         (cortex_a9_neon_int_3): Likewise.
1011         (cortex_a9_neon_int_4): Likewise.
1012         (cortex_a9_neon_int_5): Likewise.
1013         (cortex_a9_neon_vqneg_vqabs): Likewise.
1014         (cortex_a9_neon_vmov): Likewise.
1015         (cortex_a9_neon_vaba): Likewise.
1016         (cortex_a9_neon_vaba_qqq): Likewise.
1017         (cortex_a9_neon_shift_1): Likewise.
1018         (cortex_a9_neon_shift_2): Likewise.
1019         (cortex_a9_neon_shift_3): Likewise.
1020         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1021         (cortex_a9_neon_vsra_vrsra): Likewise.
1022         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1023         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1024         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1025         Likewise.
1026         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1027         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1028         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1029         Likewise.
1030         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1031         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1032         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1033         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1034         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1035         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1036         (cortex_a9_neon_fp_vsum): Likewise.
1037         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1038         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1039         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1040         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1041         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1042         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1043         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1044         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1045         (cortex_a9_neon_bp_simple): Likewise.
1046         (cortex_a9_neon_bp_2cycle): Likewise.
1047         (cortex_a9_neon_bp_3cycle): Likewise.
1048         (cortex_a9_neon_ldr): Likewise.
1049         (cortex_a9_neon_str): Likewise.
1050         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1051         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1052         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1053         (cortex_a9_neon_vld2_4_regs): Likewise.
1054         (cortex_a9_neon_vld3_vld4): Likewise.
1055         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1056         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1057         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1058         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1059         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1060         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1061         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1062         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1063         (cortex_a9_neon_mcr): Likewise.
1064         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1065         (cortex_a9_neon_mrc): Likewise.
1066         (cortex_a9_neon_mrrc): Likewise.
1068 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1070         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
1072         (cortex_a8_neon_vshl_ddd): Remove.
1073         (cortex_a8_neon_vst3_vst4): Likewise.
1074         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1076         (cortex_a8_neon_bit_ops_q): New.
1078         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
1079         (cortex_a8_neon_int_2): Likewise..
1080         (cortex_a8_neon_int_3): Likewise.
1081         (cortex_a8_neon_int_5): Likewise.
1082         (cortex_a8_neon_vqneg_vqabs): Likewise.
1083         (cortex_a8_neon_int_4): Likewise.
1084         (cortex_a8_neon_vaba): Likewise.
1085         (cortex_a8_neon_vaba_qqq): Likewise.
1086         (cortex_a8_neon_shift_1): Likewise.
1087         (cortex_a8_neon_shift_2): Likewise.
1088         (cortex_a8_neon_shift_3): Likewise.
1089         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1090         (cortex_a8_neon_vsra_vrsra): Likewise.
1091         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1092         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1093         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1094         Likewise.
1095         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1096         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1097         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1098         Likewise.
1099         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1100         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1101         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1102         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1103         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1104         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1105         (cortex_a8_neon_fp_vsum): Likewise.
1106         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1107         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1108         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1109         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1110         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1111         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1112         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1113         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1114         (cortex_a8_neon_bp_simple): Likewise.
1115         (cortex_a8_neon_bp_2cycle): Likewise.
1116         (cortex_a8_neon_bp_3cycle): Likewise.
1117         (cortex_a8_neon_ldr): Likewise.
1118         (cortex_a8_neon_str): Likewise.
1119         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1120         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1121         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1122         (cortex_a8_neon_vld2_4_regs): Likewise.
1123         (cortex_a8_neon_vld3_vld4): Likewise.
1124         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1125         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1126         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1127         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1128         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1129         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1130         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1131         (cortex_a8_neon_mcr): Likewise.
1132         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1133         (cortex_a8_neon_mrc): Likewise.
1134         (cortex_a8_neon_mrrc): Likewise.
1136 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1138         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1139         (fp): New.
1140         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1141         (aarch64_simd_dup<mode>): Add "type" attribute.
1142         (aarch64_dup_lane<mode>): Likewise.
1143         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1144         (*aarch64_simd_mov<mode>): Likewise.
1145         (aarch64_simd_mov_from_<mode>low): Likewise.
1146         (aarch64_simd_mov_from_<mode>high): Likewise.
1147         (orn<mode>3): Likewise.
1148         (bic<mode>3): Likewise.
1149         (add<mode>3): Likewise.
1150         (sub<mode>3): Likewise.
1151         (mul<mode>3): Likewise.
1152         (*aarch64_mul3_elt<mode>): Likewise.
1153         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1154         (*aarch64_mul3_elt_to_128df): Likewise.
1155         (*aarch64_mul3_elt_to_64v2df): Likewise.
1156         (neg<mode>2): Likewise.
1157         (abs<mode>2): Likewise.
1158         (abd<mode>_3): Likewise.
1159         (aba<mode>_3): Likewise.
1160         (fabd<mode>_3): Likewise.
1161         (*fabd_scalar<mode>3): Likewise.
1162         (and<mode>3): Likewise.
1163         (ior<mode>3): Likewise.
1164         (xor<mode>3): Likewise.
1165         (one_cmpl<mode>2): Likewise.
1166         (aarch64_simd_vec_set<mode>): Likewise.
1167         (aarch64_simd_lshr<mode>): Likewise.
1168         (aarch64_simd_ashr<mode>): Likewise.
1169         (aarch64_simd_imm_shl<mode>): Likewise.
1170         (aarch64_simd_reg_sshl<mode): Likewise.
1171         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1172         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1173         (aarch64_simd_vec_setv2di): Likewise.
1174         (aarch64_simd_vec_set<mode>): Likewise.
1175         (aarch64_mla<mode>): Likewise.
1176         (*aarch64_mla_elt<mode>): Likewise.
1177         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1178         (aarch64_mls<mode>): Likewise.
1179         (*aarch64_mls_elt<mode>): Likewise.
1180         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1181         (<su><maxmin><mode>3): Likewise.
1182         (move_lo_quad_<mode>): Likewise.
1183         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1184         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1185         (vec_pack_trunc_<mode>): Likewise.
1186         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1187         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1188         (*aarch64_<su>mlal_lo<mode>): Likewise.
1189         (*aarch64_<su>mlal_hi<mode>): Likewise.
1190         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1191         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1192         (*aarch64_<su>mlal<mode>): Likewise.
1193         (*aarch64_<su>mlsl<mode>): Likewise.
1194         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1195         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1196         (add<mode>3): Likewise.
1197         (sub<mode>3): Likewise.
1198         (mul<mode>3): Likewise.
1199         (div<mode>3): Likewise.
1200         (neg<mode>2): Likewise.
1201         (abs<mode>2): Likewise.
1202         (fma<mode>4): Likewise.
1203         (*aarch64_fma4_elt<mode>): Likewise.
1204         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1205         (*aarch64_fma4_elt_to_128df): Likewise.
1206         (*aarch64_fma4_elt_to_64v2df): Likewise.
1207         (fnma<mode>4): Likewise.
1208         (*aarch64_fnma4_elt<mode>): Likewise.
1209         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1210         (*aarch64_fnma4_elt_to_128df): Likewise.
1211         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1212         (<frint_pattern><mode>2): Likewise.
1213         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1214         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1215         (vec_unpacks_lo_v4sf): Likewise.
1216         (aarch64_float_extend_lo_v2df): Likewise.
1217         (vec_unpacks_hi_v4sf): Likewise.
1218         (aarch64_float_truncate_lo_v2sf): Likewise.
1219         (aarch64_float_truncate_hi_v4sf): Likewise.
1220         (aarch64_vmls<mode>): Likewise.
1221         (<su><maxmin><mode>3): Likewise.
1222         (<maxmin_uns><mode>3): Likewise.
1223         (reduc_<sur>plus_<mode>): Likewise.
1224         (reduc_<sur>plus_v2di): Likewise.
1225         (reduc_<sur>plus_v2si): Likewise.
1226         (reduc_<sur>plus_<mode>): Likewise.
1227         (aarch64_addpv4sf): Likewise.
1228         (clz<mode>2): Likewise.
1229         (reduc_<maxmin_uns>_<mode>): Likewise.
1230         (reduc_<maxmin_uns>_v2di): Likewise.
1231         (reduc_<maxmin_uns>_v2si): Likewise.
1232         (reduc_<maxmin_uns>_<mode>): Likewise.
1233         (reduc_<maxmin_uns>_v4sf): Likewise.
1234         (aarch64_simd_bsl<mode>_internal): Likewise.
1235         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1236         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1237         (aarch64_get_lane<mode>): Likewise.
1238         (*aarch64_combinez<mode>): Likewise.
1239         (aarch64_combine<mode>): Likewise.
1240         (aarch64_simd_combine<mode>): Likewise.
1241         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1242         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1243         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1244         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1245         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1246         (aarch64_<sur>h<addsub><mode>): Likewise.
1247         (aarch64_<sur><addsub>hn<mode>): Likewise.
1248         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1249         (aarch64_pmul<mode>): Likewise.
1250         (aarch64_<su_optab><optab><mode>): Likewise.
1251         (aarch64_<sur>qadd<mode>): Likewise.
1252         (aarch64_sqmovun<mode>): Likewise.
1253         (aarch64_<sur>qmovn<mode>): Likewise.
1254         (aarch64_s<optab><mode>): Likewise.
1255         (aarch64_sq<r>dmulh<mode>): Likewise.
1256         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1257         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1258         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1259         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1260         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1261         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1262         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1263         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1264         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1265         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1266         (aarch64_sqdmull<mode>): Likewise.
1267         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1268         (aarch64_sqdmull_n<mode>): Likewise.
1269         (aarch64_sqdmull2<mode>_internal): Likewise.
1270         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1271         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1272         (aarch64_<sur>shl<mode>): Likewise.
1273         (aarch64_<sur>q<r>shl<mode>
1274         (aarch64_<sur>shll_n<mode>): Likewise.
1275         (aarch64_<sur>shll2_n<mode>): Likewise.
1276         (aarch64_<sur>shr_n<mode>): Likewise.
1277         (aarch64_<sur>sra_n<mode>): Likewise.
1278         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1279         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1280         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1281         (aarch64_cm<optab><mode>): Likewise.
1282         (aarch64_cm<optab>di): Likewise.
1283         (aarch64_cm<optab><mode>): Likewise.
1284         (aarch64_cm<optab>di): Likewise.
1285         (aarch64_cmtst<mode>): Likewise.
1286         (aarch64_cmtstdi): Likewise.
1287         (aarch64_cm<optab><mode>): Likewise.
1288         (*aarch64_fac<optab><mode>): Likewise.
1289         (aarch64_addp<mode>): Likewise.
1290         (aarch64_addpdi): Likewise.
1291         (sqrt<mode>2): Likewise.
1292         (vec_load_lanesoi<mode>): Likewise.
1293         (vec_store_lanesoi<mode>): Likewise.
1294         (vec_load_lanesci<mode>): Likewise.
1295         (vec_store_lanesci<mode>): Likewise.
1296         (vec_load_lanesxi<mode>): Likewise.
1297         (vec_store_lanesxi<mode>): Likewise.
1298         (*aarch64_mov<mode>): Likewise.
1299         (aarch64_ld2<mode>_dreg): Likewise.
1300         (aarch64_ld2<mode>_dreg): Likewise.
1301         (aarch64_ld3<mode>_dreg): Likewise.
1302         (aarch64_ld3<mode>_dreg): Likewise.
1303         (aarch64_ld4<mode>_dreg): Likewise.
1304         (aarch64_ld4<mode>_dreg): Likewise.
1305         (aarch64_tbl1<mode>): Likewise.
1306         (aarch64_tbl2v16qi): Likewise.
1307         (aarch64_combinev16qi): Likewise.
1308         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1309         (aarch64_st2<mode>_dreg): Likewise.
1310         (aarch64_st2<mode>_dreg): Likewise.
1311         (aarch64_st3<mode>_dreg): Likewise.
1312         (aarch64_st3<mode>_dreg): Likewise.
1313         (aarch64_st4<mode>_dreg): Likewise.
1314         (aarch64_st4<mode>_dreg): Likewise.
1315         (*aarch64_simd_ld1r<mode>): Likewise.
1316         (aarch64_frecpe<mode>): Likewise.
1317         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1318         (aarch64_frecps<mode>): Likewise.
1320 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1322         * config/arm/iterators.md (V_elem_ch): New.
1323         (q): Likewise.
1324         (VQH_type): Likewise.
1325         * config/arm/arm.md (is_neon_type): New.
1326         (conds): Use is_neon_type.
1327         (anddi3_insn): Update type attribute.
1328         (xordi3_insn): Likewise.
1329         (one_cmpldi2): Likewise.
1330         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1331         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1332         (*movmisalign<mode>_neon_store): Likewise.
1333         (*movmisalign<mode>_neon_load): Likewise.
1334         (vec_set<mode>_internal): Likewise.
1335         (vec_set<mode>_internal): Likewise.
1336         (vec_setv2di_internal): Likewise.
1337         (vec_extract<mode>): Likewise.
1338         (vec_extract<mode>): Likewise.
1339         (vec_extractv2di): Likewise.
1340         (*add<mode>3_neon): Likewise.
1341         (adddi3_neon): Likewise.
1342         (*sub<mode>3_neon): Likewise.
1343         (subdi3_neon): Likewise.
1344         (fma<VCVTF:mode>4): Likewise.
1345         (fma<VCVTF:mode>4_intrinsic): Likewise.
1346         (*fmsub<VCVTF:mode>4): Likewise.
1347         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1348         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1349         (ior<mode>3): Likewise.
1350         (and<mode>3): Likewise.
1351         (orn<mode>3_neon): Likewise.
1352         (orndi3_neon): Likewise.
1353         (bic<mode>3_neon): Likewise.
1354         (bicdi3_neon): Likewise.
1355         (xor<mode>3): Likewise.
1356         (one_cmpl<mode>2): Likewise.
1357         (abs<mode>2): Likewise.
1358         (neg<mode>2): Likewise.
1359         (negdi2_neon): Likewise.
1360         (*umin<mode>3_neon): Likewise.
1361         (*umax<mode>3_neon): Likewise.
1362         (*smin<mode>3_neon): Likewise.
1363         (*smax<mode>3_neon): Likewise.
1364         (vashl<mode>3): Likewise.
1365         (vashr<mode>3_imm): Likewise.
1366         (vlshr<mode>3_imm): Likewise.
1367         (ashl<mode>3_signed): Likewise.
1368         (ashl<mode>3_unsigned): Likewise.
1369         (neon_load_count): Likewise.
1370         (ashldi3_neon_noclobber): Likewise.
1371         (ashldi3_neon): Likewise.
1372         (signed_shift_di3_neon): Likewise.
1373         (unsigned_shift_di3_neon): Likewise.
1374         (ashrdi3_neon_imm_noclobber): Likewise.
1375         (lshrdi3_neon_imm_noclobber): Likewise.
1376         (<shift>di3_neon): Likewise.
1377         (widen_ssum<mode>3): Likewise.
1378         (widen_usum<mode>3): Likewise.
1379         (quad_halves_<code>v4si): Likewise.
1380         (quad_halves_<code>v4sf): Likewise.
1381         (quad_halves_<code>v8hi): Likewise.
1382         (quad_halves_<code>v16qi): Likewise.
1383         (reduc_splus_v2di): Likewise.
1384         (neon_vpadd_internal<mode>): Likewise.
1385         (neon_vpsmin<mode>): Likewise.
1386         (neon_vpsmax<mode>): Likewise.
1387         (neon_vpumin<mode>): Likewise.
1388         (neon_vpumax<mode>): Likewise.
1389         (*ss_add<mode>_neon): Likewise.
1390         (*us_add<mode>_neon): Likewise.
1391         (*ss_sub<mode>_neon): Likewise.
1392         (*us_sub<mode>_neon): Likewise.
1393         (neon_vadd<mode>_unspec): Likewise.
1394         (neon_vaddl<mode>): Likewise.
1395         (neon_vaddw<mode>): Likewise.
1396         (neon_vhadd<mode>): Likewise.
1397         (neon_vqadd<mode>): Likewise.
1398         (neon_vaddhn<mode>): Likewise.
1399         (neon_vmul<mode>): Likewise.
1400         (neon_vfms<VCVTF:mode>): Likewise.
1401         (neon_vmlal<mode>): Likewise.
1402         (neon_vmls<mode>): Likewise.
1403         (neon_vmlsl<mode>): Likewise.
1404         (neon_vqdmulh<mode>): Likewise.
1405         (neon_vqdmlal<mode>): Likewise.
1406         (neon_vqdmlsl<mode>): Likewise.
1407         (neon_vmull<mode>): Likewise.
1408         (neon_vqdmull<mode>): Likewise.
1409         (neon_vsub<mode>_unspec): Likewise.
1410         (neon_vsubl<mode>): Likewise.
1411         (neon_vsubw<mode>): Likewise.
1412         (neon_vqsub<mode>): Likewise.
1413         (neon_vhsub<mode>): Likewise.
1414         (neon_vsubhn<mode>): Likewise.
1415         (neon_vceq<mode>): Likewise.
1416         (neon_vcge<mode>): Likewise.
1417         (neon_vcgeu<mode>): Likewise.
1418         (neon_vcgt<mode>): Likewise.
1419         (neon_vcgtu<mode>): Likewise.
1420         (neon_vcle<mode>): Likewise.
1421         (neon_vclt<mode>): Likewise.
1422         (neon_vcage<mode>): Likewise.
1423         (neon_vcagt<mode>): Likewise.
1424         (neon_vtst<mode>): Likewise.
1425         (neon_vabd<mode>): Likewise.
1426         (neon_vabdl<mode>): Likewise.
1427         (neon_vaba<mode>): Likewise.
1428         (neon_vabal<mode>): Likewise.
1429         (neon_vmax<mode>): Likewise.
1430         (neon_vmin<mode>): Likewise.
1431         (neon_vpaddl<mode>): Likewise.
1432         (neon_vpadal<mode>): Likewise.
1433         (neon_vpmax<mode>): Likewise.
1434         (neon_vpmin<mode>): Likewise.
1435         (neon_vrecps<mode>): Likewise.
1436         (neon_vrsqrts<mode>): Likewise.
1437         (neon_vqabs<mode>): Likewise.
1438         (neon_vqneg<mode>): Likewise.
1439         (neon_vcls<mode>): Likewise.
1440         (clz<mode>2): Likewise.
1441         (popcount<mode>2): Likewise.
1442         (neon_vrecpe<mode>): Likewise.
1443         (neon_vrsqrte<mode>): Likewise.
1444         (neon_vget_lane<mode>_sext_internal): Likewise.
1445         (neon_vget_lane<mode>_zext_internal): Likewise.
1446         (neon_vdup_n<mode>): Likewise.
1447         (neon_vdup_n<mode>): Likewise.
1448         (neon_vdup_nv2di): Likewise.
1449         (neon_vdup_lane<mode>_interal): Likewise.
1450         (*neon_vswp<mode>): Likewise.
1451         (neon_vcombine<mode>): Likewise.
1452         (float<mode><V_cvtto>2): Likewise.
1453         (floatuns<mode><V_cvtto>2): Likewise.
1454         (fix_trunc<mode><V_cvtto>2): Likewise.
1455         (fixuns_trunc<mode><V_cvtto>2
1456         (neon_vcvt<mode>): Likewise.
1457         (neon_vcvt<mode>): Likewise.
1458         (neon_vcvtv4sfv4hf): Likewise.
1459         (neon_vcvtv4hfv4sf): Likewise.
1460         (neon_vcvt_n<mode>): Likewise.
1461         (neon_vcvt_n<mode>): Likewise.
1462         (neon_vmovn<mode>): Likewise.
1463         (neon_vqmovn<mode>): Likewise.
1464         (neon_vqmovun<mode>): Likewise.
1465         (neon_vmovl<mode>): Likewise.
1466         (neon_vmul_lane<mode>): Likewise.
1467         (neon_vmul_lane<mode>): Likewise.
1468         (neon_vmull_lane<mode>): Likewise.
1469         (neon_vqdmull_lane<mode>): Likewise.
1470         (neon_vqdmulh_lane<mode>): Likewise.
1471         (neon_vqdmulh_lane<mode>): Likewise.
1472         (neon_vmla_lane<mode>): Likewise.
1473         (neon_vmla_lane<mode>): Likewise.
1474         (neon_vmlal_lane<mode>): Likewise.
1475         (neon_vqdmlal_lane<mode>): Likewise.
1476         (neon_vmls_lane<mode>): Likewise.
1477         (neon_vmls_lane<mode>): Likewise.
1478         (neon_vmlsl_lane<mode>): Likewise.
1479         (neon_vqdmlsl_lane<mode>): Likewise.
1480         (neon_vext<mode>): Likewise.
1481         (neon_vrev64<mode>): Likewise.
1482         (neon_vrev32<mode>): Likewise.
1483         (neon_vrev16<mode>): Likewise.
1484         (neon_vbsl<mode>_internal): Likewise.
1485         (neon_vshl<mode>): Likewise.
1486         (neon_vqshl<mode>): Likewise.
1487         (neon_vshr_n<mode>): Likewise.
1488         (neon_vshrn_n<mode>): Likewise.
1489         (neon_vqshrn_n<mode>): Likewise.
1490         (neon_vqshrun_n<mode>): Likewise.
1491         (neon_vshl_n<mode>): Likewise.
1492         (neon_vqshl_n<mode>): Likewise.
1493         (neon_vqshlu_n<mode>): Likewise.
1494         (neon_vshll_n<mode>): Likewise.
1495         (neon_vsra_n<mode>): Likewise.
1496         (neon_vsri_n<mode>): Likewise.
1497         (neon_vsli_n<mode>): Likewise.
1498         (neon_vtbl1v8qi): Likewise.
1499         (neon_vtbl2v8qi): Likewise.
1500         (neon_vtbl3v8qi): Likewise.
1501         (neon_vtbl4v8qi): Likewise.
1502         (neon_vtbl1v16qi): Likewise.
1503         (neon_vtbl2v16qi): Likewise.
1504         (neon_vcombinev16qi): Likewise.
1505         (neon_vtbx1v8qi): Likewise.
1506         (neon_vtbx2v8qi): Likewise.
1507         (neon_vtbx3v8qi): Likewise.
1508         (neon_vtbx4v8qi): Likewise.
1509         (*neon_vtrn<mode>_insn): Likewise.
1510         (*neon_vzip<mode>_insn): Likewise.
1511         (*neon_vuzp<mode>_insn): Likewise.
1512         (neon_vld1<mode>): Likewise.
1513         (neon_vld1_lane<mode>): Likewise.
1514         (neon_vld1_lane<mode>): Likewise.
1515         (neon_vld1_dup<mode>): Likewise.
1516         (neon_vld1_dup<mode>): Likewise.
1517         (neon_vld1_dupv2di): Likewise.
1518         (neon_vst1<mode>): Likewise.
1519         (neon_vst1_lane<mode>): Likewise.
1520         (neon_vst1_lane<mode>): Likewise.
1521         (neon_vld2<mode>): Likewise.
1522         (neon_vld2<mode>): Likewise.
1523         (neon_vld2_lane<mode>): Likewise.
1524         (neon_vld2_lane<mode>): Likewise.
1525         (neon_vld2_dup<mode>): Likewise.
1526         (neon_vst2<mode>): Likewise.
1527         (neon_vst2<mode>): Likewise.
1528         (neon_vst2_lane<mode>): Likewise.
1529         (neon_vst2_lane<mode>): Likewise.
1530         (neon_vld3<mode>): Likewise.
1531         (neon_vld3qa<mode>): Likewise.
1532         (neon_vld3qb<mode>): Likewise.
1533         (neon_vld3_lane<mode>): Likewise.
1534         (neon_vld3_lane<mode>): Likewise.
1535         (neon_vld3_dup<mode>): Likewise.
1536         (neon_vst3<mode>): Likewise.
1537         (neon_vst3qa<mode>): Likewise.
1538         (neon_vst3qb<mode>): Likewise.
1539         (neon_vst3_lane<mode>): Likewise.
1540         (neon_vst3_lane<mode>): Likewise.
1541         (neon_vld4<mode>): Likewise.
1542         (neon_vld4qa<mode>): Likewise.
1543         (neon_vld4qb<mode>): Likewise.
1544         (neon_vld4_lane<mode>): Likewise.
1545         (neon_vld4_lane<mode>): Likewise.
1546         (neon_vld4_dup<mode>): Likewise.
1547         (neon_vst4<mode>): Likewise.
1548         (neon_vst4qa<mode>): Likewise.
1549         (neon_vst4qb<mode>): Likewise.
1550         (neon_vst4_lane<mode>): Likewise.
1551         (neon_vst4_lane<mode>): Likewise.
1552         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1553         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1554         (neon_vec_<US>mult_lo_<mode>): Likewise.
1555         (neon_vec_<US>mult_hi_<mode>): Likewise.
1556         (neon_vec_<US>shiftl_<mode>): Likewise.
1557         (neon_unpack<US>_<mode>): Likewise.
1558         (neon_vec_<US>mult_<mode>): Likewise.
1559         (vec_pack_trunc_<mode>): Likewise.
1560         (neon_vec_pack_trunc_<mode>): Likewise.
1561         (neon_vabd<mode>_2): Likewise.
1562         (neon_vabd<mode>_3): Likewise.
1564 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1566         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1567         (load_pair): Update type attribute.
1568         (store_pair): Update type attribute.
1569         * config/aarch64/iterators.md (q): New.
1571 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1573         * config/arm/types.md: Add new types for Neon insns.
1575 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1576             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1577             Sergey Lega  <sergey.s.lega@intel.com>
1578             Anna Tikhonova  <anna.tikhonova@intel.com>
1579             Ilya Tocar  <ilya.tocar@intel.com>
1580             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1581             Ilya Verbin  <ilya.verbin@intel.com>
1582             Kirill Yukhin  <kirill.yukhin@intel.com>
1583             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1585         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1586         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1587         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1588         (rcp14<mode>): New.
1589         (srcp14<mode>): Ditto.
1590         (rsqrt14<mode>): Ditto.
1591         (rsqrt14<mode>): Ditto.
1592         (avx512f_vmscalef<mode>): Ditto.
1593         (avx512f_scalef<mode>): Ditto.
1594         (avx512f_getexp<mode>): Ditto.
1595         (avx512f_sgetexp<mode>): Ditto.
1596         (avx512f_fixupimm<mode>): Ditto.
1597         (avx512f_sfixupimm<mode>): Ditto.
1598         (avx512f_rndscale<mode>): Ditto.
1599         (*avx512er_exp2<mode>): Ditto.
1600         (*avx512er_rcp28<mode>): Ditto.
1601         (avx512er_rsqrt28<mode>): Ditto.
1602         (avx512f_getmant<mode>): Ditto.
1603         (avx512f_getmant<mode>): Ditto.
1604         (avx512f_rndscale<mode>): Fix formatting.
1606 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1608         * ipa-utils.h (ipa_edge_within_scc): Declare.
1609         * ipa-cp.c (edge_within_scc): Moved...
1610         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1612 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1613             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1614             Sergey Lega  <sergey.s.lega@intel.com>
1615             Anna Tikhonova  <anna.tikhonova@intel.com>
1616             Ilya Tocar  <ilya.tocar@intel.com>
1617             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1618             Ilya Verbin  <ilya.verbin@intel.com>
1619             Kirill Yukhin  <kirill.yukhin@intel.com>
1620             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1622         * config/i386/predicates.md (const_8_to_15_operand): New.
1623         (const_16_to_31_operand): Ditto.
1624         * config/i386/sse.md (V8FI): New.
1625         (V16FI): Ditto.
1626         (reduc_splus_v8df): Ditto.
1627         (reduc_splus_v16sf): Ditto.
1628         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1629         (vec_extract_hi_<mode>): Ditto.
1630         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1631         (vec_set_lo_<mode>): Ditto.
1632         (vec_set_hi_<mode>): Ditto.
1633         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1634         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1635         (avx512f_pshufd_1): Ditto.
1636         (avx512f_broadcast<mode>): Ditto.
1637         (avx512f_broadcast<mode>): Ditto.
1638         (define_split): Split vec_extract_lo into move.
1639         (ssequartermode): Ditto.
1640         (ssedoublemode): Extened with wider modes.
1641         (vec_extract_lo_<mode>): Ditto.
1643 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1644             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1645             Sergey Lega  <sergey.s.lega@intel.com>
1646             Anna Tikhonova  <anna.tikhonova@intel.com>
1647             Ilya Tocar  <ilya.tocar@intel.com>
1648             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1649             Ilya Verbin  <ilya.verbin@intel.com>
1650             Kirill Yukhin  <kirill.yukhin@intel.com>
1651             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1653         * config/i386/predicates.md (register_or_constm1_operand): New.
1654         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1655         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1656         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1657         UNSPEC_SCATTER_PREFETCH
1658         (VI48_512): New.
1659         (avx512f_ucmp<mode>3): Ditto.
1660         (avx512f_vternlog<mode>): Ditto.
1661         (avx512f_align<mode>): Ditto.
1662         (<shift_insn><mode>3): Ditto.
1663         (avx512f_<rotate>v<mode>): Ditto.
1664         (avx512f_<rotate><mode>): Ditto.
1665         (avx512f_eq<mode>3): Ditto.
1666         (avx512f_eq<mode>3_1): Ditto.
1667         (avx512f_gt<mode>3): Ditto.
1668         (avx512f_testm<mode>3): Ditto.
1669         (avx512f_testnm<mode>3): Ditto.
1670         (avx512pf_gatherpf<mode>): Ditto.
1671         (*avx512pf_gatherpf<mode>_mask): Ditto.
1672         (*avx512pf_gatherpf<mode>): Ditto.
1673         (avx512pf_scatterpf<mode>): Ditto.
1674         (*avx512pf_scatterpf<mode>_mask): Ditto.
1675         (*avx512pf_scatterpf<mode>): Ditto.
1676         (avx512f_vec_dup_gpr<mode>): Ditto.
1677         (clz<mode>2): Ditto.
1678         (conflict<mode>): Ditto.
1679         (REDUC_SMINMAX_MODE): Extened with wider modes.
1680         (reduc_<code>_<mode>): Ditto.
1681         (vlshr<mode>3): Ditto.
1682         (vashl<mode>3): Ditto.
1684 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1685             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1686             Sergey Lega  <sergey.s.lega@intel.com>
1687             Anna Tikhonova  <anna.tikhonova@intel.com>
1688             Ilya Tocar  <ilya.tocar@intel.com>
1689             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1690             Ilya Verbin  <ilya.verbin@intel.com>
1691             Kirill Yukhin  <kirill.yukhin@intel.com>
1692             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1694         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1695         UNSPEC_SCATTER.
1696         (VI48F_512): New.
1697         (avx512fmaskmode): Ditto.
1698         (bcstscalarsuff): Ditto.
1699         (avx512f_blendm<mode>): Ditto.
1700         (cmp_imm_predicate): Ditto.
1701         (avx512f_cmp<mode>3): Ditto.
1702         (avx512f_vec_dup<mode>): Ditto.
1703         (avx512f_vec_dup_mem<mode>): Ditto.
1704         (avx512f_vpermi2var<mode>3): Ditto.
1705         (avx512f_vpermt2var<mode>3): Ditto.
1706         (vec_init<mode>): Ditto.
1707         (avx512f_gathersi<mode>): Ditto.
1708         (*avx512f_gathersi<mode>): Ditto.
1709         (*avx512f_gathersi<mode>_2): Ditto.
1710         (avx512f_gatherdi<mode>): Ditto.
1711         (*avx512f_gatherdi<mode>): Ditto.
1712         (*avx512f_gatherdi<mode>_2): Ditto.
1713         (avx512f_scattersi<mode>): Ditto.
1714         (*avx512f_scattersi<mode>): Ditto.
1715         (avx512f_scatterdi<mode>): Ditto.
1716         (*avx512f_scatterdi<mode>): Ditto.
1717         (sseintprefix): Extened with wider modes.
1718         (VEC_GATHER_IDXSI): Ditto.
1719         (VEC_GATHER_IDXDI): Ditto.
1720         (VEC_GATHER_SRCDI): Ditto.
1722 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1723             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1725         * config/arm/t-aprofile: New file.
1726         * config.gcc: Handle --with-multilib-list option.
1728 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1730         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1731         the input for r2.
1733 2013-10-15  Richard Biener  <rguenther@suse.de>
1735         * tree-loop-distribution.c (build_empty_rdg): Inline into
1736         single user.
1737         (rdg_flag_vertex): Inline into single user.
1738         (rdg_flag_vertex_and_dependent): Likewise.
1739         (build_rdg_partition_for_vertex): Remove processed bitmap.
1740         (rdg_build_partitions): Simplify.
1742 2013-10-15  Richard Biener  <rguenther@suse.de>
1744         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1745         Restructure forwarding through conversions and copies to
1746         avoid performing copy-propagation the wrong way.  Adjust
1747         recursion invocations.
1748         (forward_propagate_addr_expr): Add argument stating if we
1749         are recursing from a single-use.
1750         (ssa_forward_propagate_and_combine): Adjust.
1752 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1754         * dumpfile.h (gcc::dump_manager): New class, to hold state
1755         relating to dumpfile management.
1756         (get_dump_file_name): Remove in favor of method of dump_manager.
1757         (dump_initialized_p): Likewise.
1758         (dump_start): Likewise.
1759         (dump_finish): Likewise.
1760         (dump_switch_p): Likewise.
1761         (dump_register): Likewise.
1762         (get_dump_file_info): Likewise.
1763         * context.c (gcc::context::context): Construct the dump_manager
1764         instance.
1765         * context.h (gcc::context::get_dumps): New.
1766         (gcc::context::m_dumps): New.
1767         * coverage.c (coverage_init): Port to dump_manager API.
1768         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1769         (extra_dump_files_in_use): Likewise.
1770         (extra_dump_files_alloced): Likewise.
1771         (gcc::dump_manager::dump_manager): New.
1772         (dump_register): Convert to...
1773         (gcc::dump_manager::dump_register): ...method, replacing
1774         function-static next_dump with m_next_dump field.
1775         (get_dump_file_info): Convert to...
1776         (gcc::dump_manager::get_dump_file_info): ...method.
1777         (get_dump_file_name): Convert to...
1778         (gcc::dump_manager::get_dump_file_name): ...method.
1779         (dump_start): Convert to...
1780         (gcc::dump_manager::dump_start): ...method.
1781         (dump_finish): Convert to...
1782         (gcc::dump_manager::dump_finish): ...method.
1783         (dump_begin): Replace body with...
1784         (gcc::dump_manager::dump_begin): ...new method.
1785         (dump_phase_enabled_p): Convert to...
1786         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1787         (dump_phase_enabled_p): Convert to...
1788         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1789         (dump_initialized_p):  Convert to...
1790         (gcc::dump_manager::dump_initialized_p): ...method.
1791         (dump_flag_name): Replace body with...
1792         (gcc::dump_manager::dump_flag_name): ...new method.
1793         (dump_enable_all): Convert to...
1794         (gcc::dump_manager::dump_enable_all): ...new method.
1795         (opt_info_enable_passes): Convert to...
1796         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1797         (dump_switch_p_1): Convert to...
1798         (gcc::dump_manager::dump_switch_p_1): ...new method.
1799         (dump_switch_p):  Convert to...
1800         (gcc::dump_manager::dump_switch_p): ...new method.
1801         (opt_info_switch_p): Port to dump_manager API.
1802         (enable_rtl_dump_file): Likewise.
1803         * opts-global.c (handle_common_deferred_options): Port to new
1804         dump_manager API.
1805         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1806         (pass_manager::register_one_dump_file): Likewise.
1807         (pass_manager::register_pass): Likewise.
1808         (pass_init_dump_file): Likewise.
1809         (pass_fini_dump_file): Likewise.
1810         * statistics.c (statistics_early_init): Likewise.
1812 2013-10-14  Richard Biener  <rguenther@suse.de>
1814         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1815         iterative_hash_canonical_type, gimple_canonical_type_hash,
1816         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1817         gimple_register_canonical_type, print_gimple_types_stats,
1818         free_gimple_type_tables): Move to lto/lto.c
1819         (gt-gimple.h): Do not include.
1820         * gimple.h (gimple_register_canonical_type,
1821         print_gimple_types_stats, free_gimple_type_tables): Remove.
1822         * Makefile.in (GTFILES): Remove gimple.c.
1824 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1826         PR target/58716
1827         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1828         scanning for msp430x targets.
1830 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1832         PR bootstrap/58509
1833         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1834         (registers_ok_for_ldd_peep): Move around.
1835         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1836         * config/sparc/sparc.md (widening peepholes): Use it.
1838 2013-10-14  Richard Biener  <rguenther@suse.de>
1840         PR middle-end/58712
1841         PR middle-end/55358
1842         * gimple.c (iterative_hash_canonical_type): Make sure to
1843         record the hash into the correct hashtable slot.
1845 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1847         PR rtl-optimization/58662
1848         * combine.c (try_combine): Take into account death nodes on I2 when
1849         splitting a PARALLEL of two independent SETs.  Fix dump message.
1851 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1853         PR target/51244
1854         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1855         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1856         * config/sh/t-sh (sh_treg_combine.o): New entry.
1857         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1858         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1859         (register_sh_passes): New function.  Register sh_treg_combine pass.
1860         (sh_option_override): Invoke it.
1861         (sh_canonicalize_comparison): Handle op0_preserve_value.
1862         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1863         opportunities.  Canonicalize branch condition.
1864         (nott): Allow only if pseudos can be created for non-SH2A.
1866 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1868         PR target/58690
1869         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1870         (ix86_expand_movmem): Replace copy_addr_to_reg with
1871         ix86_copy_addr_to_reg.
1872         (ix86_expand_setmem): Likewise.
1874 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1876         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1877         provided by ix86_fp_compare_mode instead of CCFPUmode.
1879 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1881         * config/aarch64/arm_neon.h
1882         (vtbx<1,3>_<psu>8): Fix register constriants.
1884 2013-10-11  Jeff Law  <law@redhat.com>
1886         PR tree-optimization/58640
1887         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1888         threading paths that cross over two loop entry points.
1890 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1892         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1893         handle vector float as well.
1894         (*vsx_le_perm_load_v4si): Likewise.
1895         (*vsx_le_perm_store_v2di): Likewise.
1896         (*vsx_le_perm_store_v4si): Likewise.
1898 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1900         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1901         directly to circumvent subtract from splat{31} workaround.
1902         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1903         prototype.
1904         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1905         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1906         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1907         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1908         separate big and little endian logic.
1909         (*altivec_vperm_<mode>_internal): New define_insn.
1910         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1911         separate big and little endian logic.
1912         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1913         (vec_permv16qi): Add little endian logic.
1915 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1917         * doc/extend.texi (returns_nonnull): Remove arguments.
1919 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1920             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1921             Sergey Lega  <sergey.s.lega@intel.com>
1922             Anna Tikhonova  <anna.tikhonova@intel.com>
1923             Ilya Tocar  <ilya.tocar@intel.com>
1924             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1925             Ilya Verbin  <ilya.verbin@intel.com>
1926             Kirill Yukhin  <kirill.yukhin@intel.com>
1927             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1929         * config/i386/sse.md (VI48F_256_512): New.
1930         (avx2_permvar<mode>): Change to ...
1931         (<avx2_avx512f>_permvar<mode>): This.
1933 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1934             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1935             Sergey Lega  <sergey.s.lega@intel.com>
1936             Anna Tikhonova  <anna.tikhonova@intel.com>
1937             Ilya Tocar  <ilya.tocar@intel.com>
1938             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1939             Ilya Verbin  <ilya.verbin@intel.com>
1940             Kirill Yukhin  <kirill.yukhin@intel.com>
1941             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1943         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1944         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1945         * config/i386/sse.md (VI4_AVX): New.
1946         (sf2simodelower): Ditto.
1947         (sse2_cvtps2dq): Change to ...
1948         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1950 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1951             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1952             Sergey Lega  <sergey.s.lega@intel.com>
1953             Anna Tikhonova  <anna.tikhonova@intel.com>
1954             Ilya Tocar  <ilya.tocar@intel.com>
1955             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1956             Ilya Verbin  <ilya.verbin@intel.com>
1957             Kirill Yukhin  <kirill.yukhin@intel.com>
1958             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1960         * config/i386/sse.md (V_512): New.
1961         (VI_512): Ditto.
1962         (vcond<V_512:mode><VF_512:mode>): Ditto.
1963         (vcond<V_512:mode><VI_512:mode>): Ditto.
1964         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1966 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1967             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1968             Sergey Lega  <sergey.s.lega@intel.com>
1969             Anna Tikhonova  <anna.tikhonova@intel.com>
1970             Ilya Tocar  <ilya.tocar@intel.com>
1971             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1972             Ilya Verbin  <ilya.verbin@intel.com>
1973             Kirill Yukhin  <kirill.yukhin@intel.com>
1974             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1976         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1977         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1978         (FMAMODE): Ditto.
1979         (fma<mode>4): Removed condition.
1980         (fms<mode>4): Ditto.
1981         (fnma<mode>4): Ditto.
1982         (fnms<mode>4): Ditto.
1983         (fma4i_fmadd_<mode>): Ditto.
1984         (*fma_fmadd_<mode>): Ditto.
1985         (*fma_fmsub_<mode>): Ditto.
1986         (*fma_fnmadd_<mode>): Ditto.
1987         (*fma_fnmsub_<mode>): Ditto.
1988         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1989         (*fma_fmaddsub_<mode>): Ditto.
1990         (*fma_fmsubadd_<mode>): Ditto.
1991         (*fmai_fmadd_<mode>): Ditto.
1992         (*fmai_fmsub_<mode>): Ditto.
1993         (*fmai_fnmadd_<mode>): Ditto.
1994         (*fmai_fnmsub_<mode>): Ditto.
1996 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1997             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1998             Sergey Lega  <sergey.s.lega@intel.com>
1999             Anna Tikhonova  <anna.tikhonova@intel.com>
2000             Ilya Tocar  <ilya.tocar@intel.com>
2001             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2002             Ilya Verbin  <ilya.verbin@intel.com>
2003             Kirill Yukhin  <kirill.yukhin@intel.com>
2004             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2006         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
2007         (VI124_256): Changed to ...
2008         (VI124_256_48_512): This.
2009         (ssepackmode): Extended with wider modes.
2010         (<code><mode>3): Changed iterator.
2011         (*avx2_<code><mode>3): Ditto.
2012         (vec_pack_trunc_<mode>): Ditto.
2014 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2015             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2016             Sergey Lega  <sergey.s.lega@intel.com>
2017             Anna Tikhonova  <anna.tikhonova@intel.com>
2018             Ilya Tocar  <ilya.tocar@intel.com>
2019             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2020             Ilya Verbin  <ilya.verbin@intel.com>
2021             Kirill Yukhin  <kirill.yukhin@intel.com>
2022             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2024         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
2025         (VI8F_256_512): Ditto.
2026         (abs<mode>2): Changed iterator.
2027         (avx2_perm<mode>): Changed to ...
2028         (<avx2_avx512f>_perm<mode>): This.
2029         (avx2_perm<mode>_1): Changed to ...
2030         (<avx2_avx512f>_perm<mode>_1): This.
2032 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2033             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2034             Sergey Lega  <sergey.s.lega@intel.com>
2035             Anna Tikhonova  <anna.tikhonova@intel.com>
2036             Ilya Tocar  <ilya.tocar@intel.com>
2037             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2038             Ilya Verbin  <ilya.verbin@intel.com>
2039             Kirill Yukhin  <kirill.yukhin@intel.com>
2040             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2042         * config/i386/sse.md (VI48_AVX512F): New.
2043         (VI48_AVX2): Changed to ...
2044         (VI48_AVX2_48_AVX512F): This.
2045         (avx2_ashrv<mode>): Changed to ...
2046         (<avx2_avx512f>_ashrv<mode>): This.
2047         (avx2_<shift_insn>v<mode>): Changed to ...
2048         (<avx2_avx512f>_<shift_insn>v<mode>): This.
2050 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2051             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2052             Sergey Lega  <sergey.s.lega@intel.com>
2053             Anna Tikhonova  <anna.tikhonova@intel.com>
2054             Ilya Tocar  <ilya.tocar@intel.com>
2055             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2056             Ilya Verbin  <ilya.verbin@intel.com>
2057             Kirill Yukhin  <kirill.yukhin@intel.com>
2058             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2060         * config/i386/sse.md (VI4_AVX512F): New.
2061         (VI8_AVX2_AVX512F): Ditto.
2062         (mul<mode>3): Extended with wider modes.
2063         (*<sse4_1_avx2>_mul<mode>3): Ditto.
2064         (mul<mode>3): Ditto.
2065         (vec_widen_<s>mult_odd_<mode>): Ditto.
2067 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2068             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2069             Sergey Lega  <sergey.s.lega@intel.com>
2070             Anna Tikhonova  <anna.tikhonova@intel.com>
2071             Ilya Tocar  <ilya.tocar@intel.com>
2072             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2073             Ilya Verbin  <ilya.verbin@intel.com>
2074             Kirill Yukhin  <kirill.yukhin@intel.com>
2075             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2077         * config/i386/sse.md (VI2_AVX512F): New.
2078         (VI124_AVX512F): Ditto.
2079         (sseunpackmode): Extended with wider modes.
2080         (sseunpackfltmode): Ditto.
2081         (vec_unpacks_float_hi_<mode>): Ditto.
2082         (vec_unpacks_float_lo_<mode>): Ditto.
2083         (vec_unpacku_float_hi_<mode>): Ditto.
2084         (vec_unpacku_float_lo_<mode>): Ditto.
2085         (vec_unpacks_lo_<mode>): Ditto.
2086         (vec_unpacks_hi_<mode>): Ditto.
2087         (vec_unpacku_lo_<mode>): Ditto.
2088         (vec_unpacku_hi_<mode>): Ditto.
2090 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2091             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2092             Sergey Lega  <sergey.s.lega@intel.com>
2093             Anna Tikhonova  <anna.tikhonova@intel.com>
2094             Ilya Tocar  <ilya.tocar@intel.com>
2095             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2096             Ilya Verbin  <ilya.verbin@intel.com>
2097             Kirill Yukhin  <kirill.yukhin@intel.com>
2098             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2100         * config/i386/i386.md (multdiv): New.
2101         (multdiv_mnemonic): Ditto.
2102         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2103         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2104         (<sse>_vmdiv<mode>3): Removed.
2106 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2107             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2108             Sergey Lega  <sergey.s.lega@intel.com>
2109             Anna Tikhonova  <anna.tikhonova@intel.com>
2110             Ilya Tocar  <ilya.tocar@intel.com>
2111             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2112             Ilya Verbin  <ilya.verbin@intel.com>
2113             Kirill Yukhin  <kirill.yukhin@intel.com>
2114             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2116         * config/i386/sse.md (V): Extended with wider modes.
2117         (VF2): Ditto.
2118         (ssehalfvecmode): Ditto.
2119         (i128): Ditto.
2120         (ssepackfltmode): Ditto.
2121         (avx_vec_concat<mode>): Ditto.
2122         (V_256_512): New iterator.
2123         (VF2_512_256): Ditto.
2124         (si2dfmode): New attribute.
2125         (si2dfmodelower): Ditto.
2126         (sf2dfmode): Ditto.
2127         (concat_tg_mode): Ditto.
2128         (floatv4siv4df2): Changed to ...
2129         (float<si2dfmodelower><mode>2): This.
2130         (avx_cvtps2pd256): Changed to ...
2131         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2132         (vec_pack_trunc_v4df): Changed to ...
2133         (vec_pack_trunc_<mode>): This.
2134         (avx_vpermil<mode>): Changed to ...
2135         (<sse2_avx_avx512f>_vpermil<mode>): This.
2136         (<fixsuffix>fix_truncv8dfv8si2): New.
2137         (vec_pack_sfix_trunc_v8df): Ditto.
2138         (avx512f_rndscale<mode>): Ditto.
2139         (avx512f_roundpd512): Ditto.
2140         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2142 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2143             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2144             Sergey Lega  <sergey.s.lega@intel.com>
2145             Anna Tikhonova  <anna.tikhonova@intel.com>
2146             Ilya Tocar  <ilya.tocar@intel.com>
2147             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2148             Ilya Verbin  <ilya.verbin@intel.com>
2149             Kirill Yukhin  <kirill.yukhin@intel.com>
2150             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2152         * config/i386/i386.md (any_fix): New iterator.
2153         (fixsuffix): New attribute.
2154         * config/i386/sse.md (VF1): Extened with wider modes.
2155         (VI): Ditto.
2156         (VI_AVX2): Ditto.
2157         (VI8): Ditto.
2158         (sseintvecmodelower): Ditto.
2159         (ssescalarmode): Ditto.
2160         (ssescalarnum): Ditto.
2161         (VF1_128_256): New.
2162         (ssexmmmode): Ditto.
2163         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2164         (<sse>_rcp<mode>2): Change iterator.
2165         (rsqrt<mode>2): Ditto.
2166         (<sse>_rsqrt<mode>2): Ditto.
2167         (avx2_vec_dup<mode>): Ditto.
2168         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2169         (round<mode>2_sfix): Ditto.
2170         (avx2_pbroadcast<mode>): Ditto.
2171         (*andnot<mode>3): Handle XI mode.
2172         (*<code><mode>3): Ditto.
2173         (AVXTOSSEMODE): Removed.
2174         (avx_vpermil<mode>): Changed to ...
2175         (<sse2_avx_avx512f>_vpermil<mode>): This.
2177 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2178             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2179             Sergey Lega  <sergey.s.lega@intel.com>
2180             Anna Tikhonova  <anna.tikhonova@intel.com>
2181             Ilya Tocar  <ilya.tocar@intel.com>
2182             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2183             Ilya Verbin  <ilya.verbin@intel.com>
2184             Kirill Yukhin  <kirill.yukhin@intel.com>
2185             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2187         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2188         (<sse>_comi): Ditto.
2189         (<sse>_ucomi): Ditto.
2190         (sse_cvtss2siq_2): Ditto.
2191         (sse2_cvtsd2si): Ditto.
2192         (sse2_cvtsd2siq): Ditto.
2193         (sse2_cvttsd2si): Ditto.
2194         (sse2_cvttsd2siq): Ditto.
2195         (<shift_insn><mode>3): Ditto.
2196         (sse2_cvtsi2sdq): Update constraint and prefix.
2197         (sse_cvtsi2ss): Update prefix.
2198         (sse_cvtsi2ssq): Ditto.
2200 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2202         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2203         ignore internal calls.
2205 2013-10-11  Richard Biener  <rguenther@suse.de>
2207         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2208         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2209         TREE_OVERFLOW is set.
2211 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2213         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2215         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2216         * gimple.h: Likewise.
2218         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2220         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2221         macros.
2223         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2225         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2226         * configure: Regenerate.
2228 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2230         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2231         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2232         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2233         similarly pp_right_paren.
2234         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2235         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2236         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2237         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2238         clauses.
2239         (omp_declare_simd_clauses_equal,
2240         omp_remove_redundant_declare_simd_attrs): New functions.
2241         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2242         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2243         * tree.h (OMP_LOOP_CHECK): Define.
2244         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2245         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2246         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2247         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2248         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2249         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2250         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2251         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2252         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2253         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2254         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2255         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2256         (omp_remove_redundant_declare_simd_attrs): New prototype.
2257         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2258         GIMPLE_OMP_TEAMS): New codes.
2259         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2260         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2261         fields.
2262         (target_nesting_level): New variable.
2263         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2264         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2265         for collapse > 1 static schedule unless ordered.
2266         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2267         (determine_parallel_type): Adjust get_ws_args_for caller.
2268         (install_var_field): Handle mask & 4 for double indirection.
2269         (scan_sharing_clauses): Ignore shared clause on teams construct.
2270         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2271         (create_omp_child_function): If inside target or declare target
2272         constructs, set "omp declare target" attribute on the child function.
2273         (find_combined_for): New function.
2274         (scan_omp_parallel): Handle combined loops.
2275         (scan_omp_target, scan_omp_teams): New functions.
2276         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2277         restrictions and set ctx->cancellable for cancellable constructs.
2278         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2279         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2280         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2281         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2282         (omp_clause_aligned_alignment): New function.
2283         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2284         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2285         on teams constructs.  Handle user defined reductions and new
2286         OpenMP 4.0 clauses.
2287         (lower_reduction_clauses): Don't set placeholder to address of ref
2288         if it has already the right type.
2289         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2290         (expand_parallel_call): Use the new non-_start suffixed builtins,
2291         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2292         and GOMP_parallel_end after the call.
2293         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2294         (expand_omp_for_init_counts): Handle combined loops.
2295         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2296         loops.
2297         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2298         end of cancellable loops.
2299         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2300         Likewise.  Handle collapse > 1 loops.
2301         (expand_omp_simd): Handle combined loops.
2302         (expand_omp_for): Add inner_stmt argument, adjust callers of
2303         expand_omp_for* functions, use expand_omp_for_static*chunk even
2304         for collapse > 1 unless ordered.
2305         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2306         of cancellable sections.
2307         (expand_omp_single): Remove need_barrier variable, just rely on
2308         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2309         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2310         (expand_omp_atomic_load, expand_omp_atomic_store,
2311         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2312         (expand_omp_target): New function.
2313         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2314         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2315         (build_omp_regions_1): Immediately close region for
2316         GF_OMP_TARGET_KIND_UPDATE.
2317         (maybe_add_implicit_barrier_cancel): New function.
2318         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2319         cancellation.
2320         (lower_omp_single): Likewise.  Add clobber after the barrier.
2321         (lower_omp_taskgroup): New function.
2322         (lower_omp_for): Handle combined loops.  Adjust
2323         lower_rec_input_clauses caller.  Handle cancellation.
2324         (lower_depend_clauses): New function.
2325         (lower_omp_taskreg): Lower depend clauses.  Adjust
2326         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2327         cancellation.
2328         (lower_omp_target, lower_omp_teams): New functions.
2329         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2330         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2331         and GOMP_cancellation_point calls.
2332         (lower_omp): Fold stmts inside of target region.
2333         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2334         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2335         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2336         (BT_FN_VOID_OMPFN_PTR_UINT,
2337         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2338         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2339         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2340         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2341         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2342         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2343         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2344         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2345         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2346         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2347         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2348         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2349         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2350         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2351         BUILT_IN_GOMP_PARALLEL_END.
2352         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2353         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2354         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2355         GF_OMP_FOR_KIND_DISTRIBUTE.
2356         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2357         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2358         (dump_gimple_omp_return): Print lhs if it has any.
2359         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2360         gimple_omp_atomic_seq_cst_p.
2361         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2362         and GIMPLE_OMP_TEAMS.
2363         * langhooks.c (lhd_omp_mappable_type): New function.
2364         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2365         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2366         hook.
2367         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2368         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2369         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2370         (struct gimplify_omp_ctx): Add combined_loop field.
2371         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2372         on stmts inside of target region.
2373         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2374         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2375         ORT_TARGET and ORT_TARGET_DATA.
2376         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2377         Handle GOVD_MAP.
2378         (omp_notice_threadprivate_variable): Complain about threadprivate
2379         variables in target region.
2380         (omp_notice_variable): Complain about vars with non-mappable type
2381         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2382         (omp_check_private): Ignore ORT_TARGET* regions.
2383         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2384         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2385         (find_combined_omp_for): New function.
2386         (gimplify_omp_for): Handle gimplification of combined loops.
2387         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2388         OMP_TEAMS.
2389         (gimplify_omp_target_update): New function.
2390         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2391         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2392         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2393         (gimplify_body): If fndecl has "omp declare target" attribute, add
2394         implicit ORT_TARGET context around it.
2395         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2396         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2397         * tree-nested.c (convert_nonlocal_reference_stmt,
2398         convert_local_reference_stmt, convert_gimple_call): Handle
2399         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2400         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2401         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2402         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2403         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2404         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2405         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2406         BUILT_IN_GOMP_LOOP_END_CANCEL,
2407         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2408         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2409         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2410         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2411         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2412         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2413         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2414         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2415         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2416         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2417         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2418         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2419         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2420         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2421         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2422         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2423         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2424         gimple_build_omp_teams): New functions.
2425         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2426         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2427         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2428         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2429         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2430         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2431         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2432         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2433         GF_OMP_ATOMIC_SEQ_CST): New.
2434         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2435         gimple_build_omp_teams): New prototypes.
2436         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2437         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2438         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2439         GIMPLE_OMP_SINGLE as end of range.
2440         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2441         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2442         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2443         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2444         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2445         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2446         gimple_omp_target_kind, gimple_omp_target_set_kind,
2447         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2448         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2449         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2450         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2451         gimple_omp_teams_set_clauses): New inlines.
2452         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2453         and GIMPLE_OMP_TASKGROUP.
2454         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2455         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2456         enum omp_clause_proc_bind_kind): New.
2457         (union omp_clause_subcode): Add depend_kind, map_kind and
2458         proc_bind_kind fields.
2459         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2460         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2461         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2462         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2463         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2465 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2467         * predict.c (tree_estimate_probability): Add new parameter
2468         for estimate_bb_frequencies.
2469         (estimate_bb_frequencies): Add new parameter to force estimation.
2470         (rebuild_frequencies): When max frequency in function is small,
2471         recompute counts from frequencies.
2472         * predict.h (estimate_bb_frequencies): New parameter.
2474 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2476         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2477         optimizations are disabled.
2479 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2481         * coverage.c (coverage_finish): Fix leak of da_file_name.
2483 2013-10-10  Jan Hubicka  <jh@suse.cz>
2485         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2486         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2487         Bobcat and generic.
2489 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2491         PR middle-end/58670
2492         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2493         if any labels are in FALLTHRU_BB, use a special label emitted
2494         immediately after the asm goto insn rather than label_rtx
2495         of the LABEL_DECL.
2496         (expand_asm_stmt): Adjust caller.
2497         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2498         edge if the last insn in predecessor is a jump with single successor,
2499         but it isn't simplejump_p.
2501 2013-10-10  Richard Biener  <rguenther@suse.de>
2503         PR tree-optimization/58656
2504         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2506 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2508         * gimplify.c: Include expr.h and tm_p.h for targets with special
2509         va-arg padding requirements.
2511 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2513         * tree-flow.h: Move some prototypes to gimple.h.
2514         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2515         * gimple.h: Relocate some prototypes from tree-flow.h
2516         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2517         Move to gimplify.c.
2518         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2519         (build_va_arg_indirect_ref): Relocate and make static.
2520         (std_gimplify_va_arg_expr): Relocate here.
2521         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2522         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2524 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2526         * doc/md.texi: Document the mnemonic attribute.
2528 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2530         PR target/57377
2531         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2532         (set_attr_alternative x ...) when searching for user defined
2533         mnemonic attribute.
2535 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2537         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2539 2013-10-09  Easwaran Raman <eraman@google.com>
2541         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2542         * cfgexpand.c (defer_stack_allocation): ...use here
2543         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2545 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2547         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2548         (operand_equal_for_value_replacement): New function, extracted from
2549         value_replacement and enhanced to catch more cases.
2550         (value_replacement): Use operand_equal_for_value_replacement.
2552 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2554         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2555         get_max_loop_iterations.
2557 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2559         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2560         Place comment above function.
2562 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2564         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2565         are not related to tree-cfg.c.
2566         * tree-ssa-address.h: New file.  Relocate prototypes.
2567         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2568         (addr_for_mem_ref): New.  Combine call to get_address_description and
2569         return addr_for_mem_ref.
2570         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2571         * tree-ssa-live.h: Adjust prototypes.
2572         * passes.c: Include tree-ssa-live.h.
2573         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2574         * graphite.c (graphite_transform_loops): Make static.
2575         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2576         make_pass_graphite, pass_data_graphite_transforms,
2577         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2578         * ipa-pure-const.c (warn_function_noreturn): Make static.
2579         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2580         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2581         Relocate from tree-cfg.c
2582         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2583         static.
2584         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2585         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2586         Move to ipa-pure-const.c.
2587         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2588         Relocate from tree-optimize.c.
2589         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2590         make_pass_fixup_cfg): Move to tree-cfg.c.
2591         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2592         Relocate some prototypes.
2593         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2594         * tree-dump.c (dump_function_to_file): Remove prototype.
2595         Add tree-flow.h to the include file.
2596         * tree-dump.h: Remove prototype.
2597         * tree-parloops.h: New File.  Add prototypes.
2598         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2599         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2600         from tree-ssa-loop.c.
2601         * tree-predcom.c (run_tree_predictive_commoning,
2602         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2603         Relocate here from tree-ssa-loop.c.
2604         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2605         ssa_name_values.release ().
2606         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2607         (ssa_name_values): Relocate from tree-flow.h.
2608         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2609         * tree-ssa-loop.c (run_tree_predictive_commoning,
2610         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2611         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2612         make_pass_graphite, pass_data_graphite_transforms,
2613         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2614         tree_parallelize_loops, pass_data_parallelize_loops,
2615         make_pass_parallelize_loops): Move to other files.
2616         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2617         moved here.
2618         * tree.h: Remove prototypes from tree-address.c.
2620 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2622         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2623         (struct int_tree_map): Move to tree-hasher.h
2624         (SCALE, LABEL, PERCENT): Move to gimple.h
2625         * tree-flow-inline.h: Delete.  Move functions to other files.
2626         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2627         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2628         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2629         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2630         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2631         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2632         * tree-hasher.h: Don't include tree-flow.h.
2633         (struct int_tree_map): Relocate from tree-flow.h.
2634         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2635         tree-flow-inline.h and make static.
2636         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2637         tree-flow-inline.h.
2638         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2639         tree-flow-inline.h and make static.
2640         * tree.h (is_global_var, may_be_aliased): Relocate from
2641         tree-flow-inline.h.
2642         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2643         * value-prof.c: No longer include tree-flow-inline.h.
2644         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2646 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2648         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2649         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2650         (enum move_pos): Move to tree-ssa-loop-im.h
2651         * cfgloop.h: Move some prototypes.
2652         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2653         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2654         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2655         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2656         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2657         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2658         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2659         (enum move_pos): Relocate here.
2660         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2661         tree-ssa-loop.c.
2662         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2663         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2664         make_pass_lim): Relocate here from tree-ssa-loop.c.
2665         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2666         tree-ssa-loop.c.
2667         (loop_edge_to_cancel, unloop_loops): Make static.
2668         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2669         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2670         (tree_complete_unroll, gate_tree_complete_unroll,
2671         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2672         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2673         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2674         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2675         (stmt_invariant_in_loop_p): Remove unused function.
2676         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2677         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2678         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2679         (gcov_type_to_double_int): Move to cfgloop.h.
2680         (double_int_cmp, bound_index,
2681         estimate_numbers_of_iterations_loop): Make static.
2682         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2683         (max_loop_iterations): Factor out get_max_loop_iterations.
2684         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2685         cfgloop.c.
2686         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2687         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2688         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2689         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2690         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2691         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2692         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2693         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2694         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2695         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2696         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2697         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2698         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2699         pass_data_complete_unroll, make_pass_complete_unroll,
2700         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2701         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2702         tree-ssa-loop-ivcanon.c.
2703         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2704         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2705         tree-ssa-loop-prefetch.c.
2706         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2707         tree-ssa-loop-im.c.
2708         (get_lsm_tmp_name): Relocate and add suffix parameter.
2709         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2710         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2711         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2712         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2713         (get_estimated_loop_iterations): Factor out accessor from
2714         estimated_loop_iterations in tree-ssa-loop-niter.c.
2715         (get_max_loop_iterations): Factor out accessor from
2716         _max_loop_iterations in tree-ssa-niter.c.
2717         * loop-unroll.c (decide_unroll_constant_iterations,
2718         decide_unroll_runtime_iterations, decide_peel_simple,
2719         decide_unroll_stupid): Use new get_* accessors.
2721 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2723         PR tree-optimization/20318
2724         * doc/extend.texi (returns_nonnull): New function attribute.
2725         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2726         attribute.
2727         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2728         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2730 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2732         PR middle-end/58570
2733         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2734         false if both components are bitfields.
2736 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2738         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2739         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2740         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2741         * config/aarch64/aarch64.h
2742         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2743         * config/aarch64/aarch64-simd-builtins.def
2744         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2746 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2748         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2750 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2752         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2754 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2756         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2757         (vneg_f64): New intrinsic.
2758         (vneg_s8): Asm replaced with C.
2759         (vneg_s16): Likewise.
2760         (vneg_s32): Likewise.
2761         (vneg_s64): New intrinsic.
2762         (vnegq_f32): Asm replaced with C.
2763         (vnegq_f64): Likewise.
2764         (vnegq_s8): Likewise.
2765         (vnegq_s16): Likewise.
2766         (vnegq_s32): Likewise.
2767         (vnegq_s64): Likewise.
2769 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2771         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2773 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2775         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2776         packed stack special handling.
2777         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2778         back to fixed stack slots for FPRs saved due to stdarg.
2780 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2782         * config/s390/s390.c (s390_frame_info): Restructure function.
2784 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2786         * config/s390/s390.c (struct s390_frame_layout): New field
2787         gpr_save_slots.
2788         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2789         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2790         regs_ever_clobbered to char*.
2791         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2792         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2793         (s390_register_info): Enable FPR save slots.  Move/Copy some
2794         functionality into ...
2795         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2796         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2797         function.
2798         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2799         not imply a stack frame.
2800         (s390_init_frame_layout): Remove variable clobbered_regs.
2801         (s390_update_register_info): Remove function.
2802         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2803         cannot be used for register renaming.
2804         (s390_hard_regno_scratch_ok): New function.
2805         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2806         (s390_initial_elimination_offset): Change offset calculation of
2807         the return address pointer.
2808         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2809         (restore_gprs): Set frame related flag.
2810         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2811         (s390_emit_prologue): Call s390_register_info instead of
2812         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2813         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2814         (s390_optimize_prologue): Call s390_optimize_register_info.
2815         Try to remove also FPR slot save/restore INSNs.  Remove frame
2816         related flags from restore INSNs.
2818 2013-10-08  DJ Delorie  <dj@redhat.com>
2820         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2821         (movhi): Likewise.
2823         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2824         avoid conflict with the MI use of %c.
2825         * config/rl78/rl78-real.md: change %c to %C throughout.
2826         * config/rl78/rl78-virt.md: Likewise.
2828 2013-10-08  Jan Hubicka  <jh@suse.cz>
2830         * config/i386/i386.c (ix86_option_override_internal): Switch
2831         to SSE math for -ffast-math when target ISA supports SSE2.
2833 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2835         * tree-flow.h: Remove some prototypes.
2836         * tree.h: Remove some protypes, add a couple.
2837         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2838         using_eh_for_cleanups_p): Add interface routines for front ends.
2839         * tree-eh.h: New file.  Add protoptyes.
2840         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2841         (add_stmt_to_eh_lp_fn): Make static.
2842         (lower_try_finally): Use new using_eh_for_cleanups_p.
2843         * emit-rtl.c: Include tree-eh.h.
2844         * gimple.h: Include tree-eh.h.
2846 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2848         PR tree-optimization/58480
2849         * tree-vrp.c (infer_nonnull_range): New function.
2850         (infer_value_range): Call infer_nonnull_range.
2852 2013-10-08  Dehao Chen  <dehao@google.com>
2854         PR tree-optimization/58619
2855         * tree-inline.c (copy_phis_for_bb): Combine location data
2856         only if non-null.
2858 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2860         PR target/58423
2861         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2862         RTX_FRAME_RELATED_P on INSN.
2864 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2866         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2867         (altivec_expand_vec_perm_const): Call it.
2869 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2871         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2872         sequences for LE VSX loads and stores at expand time.
2873         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2874         prototype.
2875         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2876         (rs6000_gen_le_vsx_permute): New.
2877         (rs6000_gen_le_vsx_load): New.
2878         (rs6000_gen_le_vsx_store): New.
2879         (rs6000_gen_le_vsx_move): New.
2880         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2881         (*vsx_le_perm_load_v4si): New.
2882         (*vsx_le_perm_load_v8hi): New.
2883         (*vsx_le_perm_load_v16qi): New.
2884         (*vsx_le_perm_store_v2di): New.
2885         (*vsx_le_perm_store_v4si): New.
2886         (*vsx_le_perm_store_v8hi): New.
2887         (*vsx_le_perm_store_v16qi): New.
2888         (*vsx_xxpermdi2_le_<mode>): New.
2889         (*vsx_xxpermdi4_le_<mode>): New.
2890         (*vsx_xxpermdi8_le_V8HI): New.
2891         (*vsx_xxpermdi16_le_V16QI): New.
2892         (*vsx_lxvd2x2_le_<mode>): New.
2893         (*vsx_lxvd2x4_le_<mode>): New.
2894         (*vsx_lxvd2x8_le_V8HI): New.
2895         (*vsx_lxvd2x16_le_V16QI): New.
2896         (*vsx_stxvd2x2_le_<mode>): New.
2897         (*vsx_stxvd2x4_le_<mode>): New.
2898         (*vsx_stxvd2x8_le_V8HI): New.
2899         (*vsx_stxvd2x16_le_V16QI): New.
2901 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2903         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2905 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2907         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2908         loop to work also for 31bit ABI.
2909         Save the stack pointer for frame_size > 0.
2911 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2913         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2914         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2915         constraint letters from expanders.
2916         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2917         retry count to general_operand.
2918         ("tabort"): Give operand 0 a mode.
2919         ("tabort_1"): Add mode and constraint letter for operand 0.
2920         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2922 2013-10-04  Jeff Law  <law@redhat.com>
2924         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2926         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2927         out of ...
2928         (thread_across_edge): Here.  Call it.
2930 2013-10-04  Cary Coutant  <ccoutant@google.com>
2932         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2933         discarding a location list expression (or a piece of one).
2935 2013-10-03  Jan Hubicka  <jh@suse.cz>
2937         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2938         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2939         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2941 2013-10-03  Jan Hubicka  <jh@suse.cz>
2943         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2944         accumulate-outgoing-args when producing unwind info.
2946 2013-10-03  Wei Mi  <wmi@google.com>
2948         * lra-constraints.c (insert_move_for_subreg): New function
2949         extracted from simplify_operand_subreg.
2950         (simplify_operand_subreg): Add reload for paradoxical subreg.
2952 2013-10-03  Rong Xu  <xur@google.com>
2954         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2955         the candidate of builtin_expect such that we should fix the
2956         size/time estimation.
2957         (estimate_function_body_sizes): Do the acutally size/time fix-up
2958         for builtin_expect.
2960 2013-10-03  Rong Xu  <xur@google.com>
2962         * predict.c (tree_predict_by_opcode): Get the probability
2963         for builtin_expect from param builtin_expect_probability.
2964         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2965         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2966         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2968 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2970         PR c++/19476
2971         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2972         * calls.c (alloca_call_p): Use get_callee_fndecl.
2973         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2974         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2975         Likewise.
2976         (vrp_visit_stmt): Remove duplicated code.
2978 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2980         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2981         ceildf2, btruncdf2, instead of vsx_* name.
2983         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2984         iterators to only do V2DF and V4SF here.  Move the DF code to
2985         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2986         just 'v' since only vector operations are handled with these insns
2987         after moving the DF support to rs6000.md.
2988         (vsx_sub<mode>3): Likewise.
2989         (vsx_mul<mode>3): Likewise.
2990         (vsx_div<mode>3): Likewise.
2991         (vsx_fre<mode>2): Likewise.
2992         (vsx_neg<mode>2): Likewise.
2993         (vsx_abs<mode>2): Likewise.
2994         (vsx_nabs<mode>2): Likewise.
2995         (vsx_smax<mode>3): Likewise.
2996         (vsx_smin<mode>3): Likewise.
2997         (vsx_sqrt<mode>2): Likewise.
2998         (vsx_rsqrte<mode>2): Likewise.
2999         (vsx_fms<mode>4): Likewise.
3000         (vsx_nfma<mode>4): Likewise.
3001         (vsx_copysign<mode>3): Likewise.
3002         (vsx_btrunc<mode>2): Likewise.
3003         (vsx_floor<mode>2): Likewise.
3004         (vsx_ceil<mode>2): Likewise.
3005         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
3006         (vsx_sminsf3): Likewise.
3007         (vsx_fmadf4): Likewise.
3008         (vsx_fmsdf4): Likewise.
3009         (vsx_nfmadf4): Likewise.
3010         (vsx_nfmsdf4): Likewise.
3011         (vsx_cmpdf_internal1): Likewise.
3013         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
3014         simpler to select whether a target has SPE or traditional floating
3015         point support in iterators.
3016         (TARGET_DF_SPE): Likewise.
3017         (TARGET_SF_FPR): Likewise.
3018         (TARGET_DF_FPR): Likewise.
3019         (TARGET_SF_INSN): Macros to say whether floating point support
3020         exists for a given operation for expanders.
3021         (TARGET_DF_INSN): Likewise.
3023         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
3024         combining of SF/DF mode operations, using both traditional and VSX
3025         registers.
3026         (Fvsx): Likewise.
3027         (Ff): Likewise.
3028         (Fv): Likewise.
3029         (Fs): Likewise.
3030         (Ffre): Likewise.
3031         (FFRE): Likewise.
3032         (abs<mode>2): Combine SF/DF modes using traditional floating point
3033         instructions.  Add support for using the upper DF registers with
3034         VSX support, and SF registers with power8-vector support.  Update
3035         expanders for operations supported by both the SPE and traditional
3036         floating point units.
3037         (abs<mode>2_fpr): Likewise.
3038         (nabs<mode>2): Likewise.
3039         (nabs<mode>2_fpr): Likewise.
3040         (neg<mode>2): Likewise.
3041         (neg<mode>2_fpr): Likewise.
3042         (add<mode>3): Likewise.
3043         (add<mode>3_fpr): Likewise.
3044         (sub<mode>3): Likewise.
3045         (sub<mode>3_fpr): Likewise.
3046         (mul<mode>3): Likewise.
3047         (mul<mode>3_fpr): Likewise.
3048         (div<mode>3): Likewise.
3049         (div<mode>3_fpr): Likewise.
3050         (sqrt<mode>3): Likewise.
3051         (sqrt<mode>3_fpr): Likewise.
3052         (fre<Fs>): Likewise.
3053         (rsqrt<mode>2): Likewise.
3054         (cmp<mode>_fpr): Likewise.
3055         (smax<mode>3): Likewise.
3056         (smin<mode>3): Likewise.
3057         (smax<mode>3_vsx): Likewise.
3058         (smin<mode>3_vsx): Likewise.
3059         (negsf2): Delete SF operations that are merged with DF.
3060         (abssf2): Likewise.
3061         (addsf3): Likewise.
3062         (subsf3): Likewise.
3063         (mulsf3): Likewise.
3064         (divsf3): Likewise.
3065         (fres): Likewise.
3066         (fmasf4_fpr): Likewise.
3067         (fmssf4_fpr): Likewise.
3068         (nfmasf4_fpr): Likewise.
3069         (nfmssf4_fpr): Likewise.
3070         (sqrtsf2): Likewise.
3071         (rsqrtsf_internal1): Likewise.
3072         (smaxsf3): Likewise.
3073         (sminsf3): Likewise.
3074         (cmpsf_internal1): Likewise.
3075         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
3076         (negdf2): Delete DF operations that are merged with SF.
3077         (absdf2): Likewise.
3078         (nabsdf2): Likewise.
3079         (adddf3): Likewise.
3080         (subdf3): Likewise.
3081         (muldf3): Likewise.
3082         (divdf3): Likewise.
3083         (fred): Likewise.
3084         (rsqrtdf_internal1): Likewise.
3085         (fmadf4_fpr): Likewise.
3086         (fmsdf4_fpr): Likewise.
3087         (nfmadf4_fpr): Likewise.
3088         (nfmsdf4_fpr): Likewise.
3089         (sqrtdf2): Likewise.
3090         (smaxdf3): Likewise.
3091         (smindf3): Likewise.
3092         (cmpdf_internal1): Likewise.
3093         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3094         (btrunc<mode>2): Delete separate expander, and combine with the
3095         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3096         (btrunc<mode>2_fpr): Likewise.
3097         (ceil<mode>2): Likewise.
3098         (ceil<mode>2_fpr): Likewise.
3099         (floor<mode>2): Likewise.
3100         (floor<mode>2_fpr): Likewise.
3101         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3102         Add support for using the upper registers with VSX and
3103         power8-vector.  Move insns to be closer to the define_expands. On
3104         VSX systems, prefer the traditional form of FMA over the VSX
3105         version, since the traditional form allows the target not to
3106         overlap with the inputs.
3107         (fms<mode>4_fpr): Likewise.
3108         (nfma<mode>4_fpr): Likewise.
3109         (nfms<mode>4_fpr): Likewise.
3111 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3112             Richard Earnshaw  <richard.earnshaw@arm.com>
3114         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3115         (struct mult_cost_table): Likewise.
3116         (struct mem_cost_table): Likewise.
3117         (struct fp_cost_table): Likewise.
3118         (struct vector_cost_table): Likewise.
3119         (cpu_cost_table): Likewise.
3120         * config/arm/arm.opt (mold-rts-costs): New option.
3121         (mnew-generic-costs): Likewise.
3122         * config/arm/arm.c (generic_extra_costs): New table.
3123         (cortexa15_extra_costs): Likewise.
3124         (arm_slowmul_tune): Use NULL as new costs.
3125         (arm_fastmul_tune): Likewise.
3126         (arm_strongarm_tune): Likewise.
3127         (arm_xscale_tune): Likewise.
3128         (arm_9e_tune): Likewise.
3129         (arm_v6t2_tune): Likewise.
3130         (arm_cortex_a5_tune): Likewise.
3131         (arm_cortex_a9_tune): Likewise.
3132         (arm_v6m_tune): Likewise.
3133         (arm_fa726te_tune): Likewise.
3134         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3135         (arm_cortex_tune): Use generict_extra_costs.
3136         (shifter_op_p): New function.
3137         (arm_unspec_cost): Likewise.
3138         (LIBCALL_COST): Define.
3139         (arm_new_rtx_costs): New function.
3140         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3141         table is available. Use old costs otherwise unless mnew-generic-costs
3142         is specified.
3143         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3144         (cpu_cost_table): Declare.
3146 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3148         PR target/58460
3149         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3150         (*subs_mul_imm_<mode>)
3151         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3152         (*sub_<shift>_<mode>)
3153         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3154         Remove k constraint.
3156 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3158         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3159         code.
3160         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3162 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3164         * predict.c (probably_never_executed): New function.
3165         (probably_never_executed_bb_p): Invoke probably_never_executed.
3166         (probably_never_executed_edge_p): Ditto.
3167         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3168         Treat profile insanities conservatively.
3170 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3172         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3174 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3176         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3177         use it.  Use smaller increase for scratch.  Don't increase reject
3178         for early clobber scratch.
3179         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3180         setting eliminated regs except setting fp from hfp.
3181         (lra_eliminate): Check lra_insn_recog_data on NULL.
3183 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3185         PR target/58587
3186         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3187         setting -mvsx-timode by default until the underlying problem is fixed.
3188         (RS6000_CPU, power7 defaults): Likewise.
3190 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3192         * config/x-linux (host-linux.o): Remove header dependencies.
3193         Use $(COMPILE) and $(POSTCOMPILE).
3194         * config/t-linux-android (linux-android.o): Ditto.
3196 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3198         * Makefile.in (expmed.o-warn): Remove.
3200 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3202         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3203         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3205 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3207         * dojump.c (do_jump_1): Divide probability between
3208         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3210 2013-10-02  Tom Tromey  <tromey@redhat.com>
3212         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3214 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3216         * tree-flow.h: Remove some prototypes.
3217         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3218         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3219         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3220         mark_virtual_phi_result_for_renaming): Relocate here.
3221         * tree-into-ssa.h: Add prototypes.
3222         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3223         single_pred_before_succ_order.
3224         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3225         (nonfreeing_call_p) Move to gimple.c.
3226         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3227         tree-ssa-phiopt.c.
3228         * basic-block.h (single_pred_before_succ_order): Add prototype.
3229         * gimple.c (nonfreeing_call_p): Relocate here.
3230         * gimple.h: Add prototype.
3231         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3232         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3233         * tree-ssa.h: Include tree-ssa-dom.h.
3235 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3237         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3238         Use $(COMPILE) and $(POSTCOMPILE).
3240         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3242 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3244         * tree-flow.h: Remove some prototypes.
3245         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3246         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3247         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3248         propagate_tree_value*): Move from here to...
3249         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3250         propagate_tree_value*): Relocate here.
3251         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3252         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3253         * gimple-fold.c: Remove gimple-fold.h from include list.
3254         * tree-vrp.c: Remove gimple-fold.h from include list.
3255         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3256         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3257         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3258         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3259         * sese.c: Add tree-ssa-propagate.h to include list.
3261 2013-10-02  Richard Biener  <rguenther@suse.de>
3263         * tree-loop-distribution.c: Include tree-vectorizer.h for
3264         find_loop_location.
3265         (enum partition_kind): Remove PKIND_REDUCTION.
3266         (struct partition_s): Remove has_writes member, add reduction_p member.
3267         (partition_alloc): Adjust.
3268         (partition_builtin_p): Likewise.
3269         (partition_has_writes): Remove.
3270         (partition_reduction_p): New function.
3271         (partition_merge_into): Likewise.
3272         (generate_code_for_partition): Commonize builtin partition
3273         handling tail.
3274         (rdg_cannot_recompute_vertex_p): Remove.
3275         (already_processed_vertex_p): Likewise.
3276         (rdg_flag_vertex): Do not set has_writes.
3277         (classify_partition): Adjust.
3278         (rdg_build_partitions): Do not set has_writes, treat all
3279         partitions as useful.
3280         (distribute_loop): Record number of library calls generated.  Adjust.
3281         (tree_loop_distribution): Report number of loops and library
3282         calls generated as opt-info.
3284 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3286         * tree-flow.h: Include new .h files.  Move prototypes.
3287         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3288         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3289         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3290         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3291         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3292         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3293         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3294         * tree.h (get_ref_base_and_extent): Move prototype out.
3295         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3296         tree-dfa.h.
3297         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3298         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3299         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3300         * tree-scalar-evolution.c: Include tree.h.
3301         * sese.c: Include tree.h.
3302         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3303         * dwarf2out.c: Include tree-dfa.h.
3304         * tree-chrec.c: Include tree.h.
3305         * tree-data-ref.c: Include tree.h.
3307 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3309         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3310         Fix whitespace.
3312 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3314         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3315         Use $(COMPILE) and $(POSTCOMPILE).
3316         (sol2-cxx.o): Likewise.
3317         (sol2-stubs.o): Likewise.
3318         (sol2.o): Likewise.
3319         * config/x-solaris (host-solaris.o): Likewise.
3321         * config/sparc/t-sparc (sparc.o): Remove.
3322         (sparc-c.o): Remove header dependencies.
3323         Use $(COMPILE) and $(POSTCOMPILE).
3324         * config/sparc/x-sparc: Likewise.
3326 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3328         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3329         * config/arc/arc700.md: Likewise.
3330         * config/arc/arc-modes.def: Likewise.
3331         * config/arc/arc-simd.h: Likewise.
3332         * config/arc/t-arc-uClibc: Likewise.
3333         * config/arc/t-arc-newlib: Likewise.
3335 2013-10-02  Renlin Li  <renlin.li@arm.com>
3337         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3338         plus_constant.
3339         (aarch64_expand_epilogue): Likewise.
3341 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3342             Yufeng Zhang  <yufeng.zhang@arm.com>
3344         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3345         declaration.
3346         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3347         'base_in' represent a conversion and legal_cast_p_1 holds; set
3348         'base_in' with the returned value from get_unwidened.
3350 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3352         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3353         plus_constant is not used.
3355 2013-10-01  Wei Mi  <wmi@google.com>
3357         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3358         * config/i386/i386.md: Add define_peephole2 to
3359         break partial reg stall for cvtss2sd/cvtsd2ss.
3361 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3363         * config/arc/arc.c (pass_arc_ifcvt::clone):
3364         Update for ctxt_ -> m_ctxt change.
3366 2013-10-01  Jeff Law  <law@redhat.com>
3368         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3369         outgoing_edge and intermediate_edge fields.  Instead store the path.
3370         (redirection_data::hash): Hash on the last edge's destination index.
3371         (redirection_data::equal): Check the entire thread path.
3372         (lookup_redirectio_data): Corresponding changes.
3373         (create_edge_and_update_destination_phis): Likewise.
3374         (thread_single_edge): Likewise.
3376 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3377             Diego Novillo <dnovillo@google.com>
3379         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3380         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3381         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3382         versions of these patterns.
3384         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3385         (__builtin_arc_aligned): Likewise.
3387         * config/arc/arc.md: Expand adc_0 comment stating the intended
3388         purpose and why it isn't ready.
3389         Replace commented out call_value_via_label_mixed with a
3390         plain comment about bl_s.
3392         * config/arc/arc.c (stdio.h): Don't include directly.
3393         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3394         Remove if (1) condition.
3395         (arc_encode_section_info): Fix comment.
3397 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3399         * config/arc/arc.c (arc_conditional_register_usage):
3400         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3401         Also set reg_alloc_order for DMA config regs.
3403 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3404             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3406         * doc/install.texi (--with-cpu): Mention ARC.
3407         (arc-*-elf32): New paragraph.
3408         (arc-linux-uclibc): Likewise.
3409         * doc/md.texi (Machine Constraints): Add ARC part.
3410         * doc/invoke.texi: (menu): Add ARC Options.
3411         (Machine Dependent Options) <ARC Options>: Add synopsis.
3412         (node ARC Options): Add.
3413         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3414         (ARC Built-in Functions): New section defining
3415         generic ARC built-in functions.
3416         (ARC SIMD Built-in Functions): New section defining SIMD specific
3417         built-in functions.
3418         (Declaring Attributes of Functions): Extended
3419         description of short_call and long_call attributes for ARC and
3420         added index entries.
3422 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3423             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3424             Joern Rennecke  <joern.rennecke@embecosm.com>
3425             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3426             Brendan Kehoe  <brendan@zen.org>
3427             Michael Eager  <eager@eagercon.com>
3428             Simon Cook  <simon.cook@embecosm.com>
3429             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3431         * config/arc, common/config/arc: New directories.
3433 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3434             Brendan Kehoe  <brendan@zen.org>
3435             Simon Cook  <simon.cook@embecosm.com>
3437         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3439 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3441         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3442         * tree-ssa-coalesce.h: New. Move prototype to here.
3443         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3444         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3445         (gimple_can_coalesce_p): Move to...
3446         * gimple.c (gimple_can_coalesce_p): Here.
3448 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3450         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3451         (dump_decl_set): Move to gimple.c.
3452         * gimple.h: Don't include tree-ssa-operands.h.
3453         (dump_decl_set): Add prototype.
3454         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3455         Move to gimple-ssa.h.
3456         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3457         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3458         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3459         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3460         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3461         than PHI_ARG_DEF.
3462         (dump_decl_set): Relocate here.
3463         * gimple-ssa.h: New file.
3464         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3465         Relocate from gimple.h.
3466         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3467         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3468         swap_tree_operands and remove non-ssa path.
3469         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3470         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3471         swap_ssa_operands.
3472         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3473         vect_is_simple_reduction_1): Use swap_ssa_operands.
3474         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3475         (enum need_phi_state): Move to tree-into-ssa.c.
3476         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3477         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3478         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3479         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3480         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3481         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3482         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3483         num_imm_uses): Move to ssa-iterators.h.
3484         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3485         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3486         tree-phinodes.h.
3487         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3488         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3489         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3490         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3491         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3492         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3493         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3494         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3495         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3496         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3497         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3498         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3499         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3500         (set_phi_nodes): Move to tree-phinodes.h.
3501         * tree-ssa-operands.h (enum ssa_op_iter_type,
3502         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3503         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3504         (dump_decl_set): Remove prototype.
3505         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3506         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3507         (set_phi_nodes): Relocate from tree-flow-inline.h.
3508         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3509         tree-flow-inline.h
3510         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3511         include list.  Temporarily add gimple.h to include list.
3512         * ssa-iterators.h: New file.
3513         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3514         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3515         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3516         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3517         Relocate from tree-ssa-operands.h.
3518         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3519         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3520         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3521         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3522         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3523         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3524         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3525         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3526         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3527         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3528         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3529         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3530         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3531         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3532         Relocate from tree-flow-inline.h.
3533         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3535 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3537         * aarch64-simd.md
3538         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3539         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3540         Insert '\t' to output template.
3541         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3542         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3543         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3544         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3546 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3548         * doc/install.texi (Host/target specific installation notes for GCC):
3549         Put @anchor before @heading.
3550         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3551         Use @email for email addresses.
3553 2013-10-01  Jeff Law  <law@redhat.com>
3555         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3556         a vec.  Only delete the path if we create one without successfully
3557         registering a jump thread.
3558         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3559         as a pointer.
3560         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3561         (paths): New vector of jump threading paths.
3562         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3563         (THREAD_PATH): New accessor macro for the entire thread path.
3564         (lookup_redirection_data): Get intermediate and final outgoing edge
3565         from the thread path.
3566         (create_edge_and_update_destination_phis): Copy the threading path.
3567         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3568         jump threading path.
3569         (ssa_redirect_edges): Get edges and block types from the jump threading
3570         path.  Free the path vector.
3571         (thread_block): Get edges from the jump threading path.  Look at the
3572         entire path to see if we thread to a loop exit.  If we cancel a jump
3573         thread request, then free the path vector.
3574         (thread_single_edge): Get edges and block types from the jump threading
3575         path.  Free the path vector.
3576         (thread_through_loop_header): Get edges and block types from the jump
3577         threading path.  Free the path vector.
3578         (mark_threaded_blocks): Iterate over the vector of paths and store
3579         the path on the appropriate edge.  Get edges and block types from the
3580         jump threading path.
3581         (mark_threaded_blocks): Get edges and block types from the jump
3582         threading path.  Free the path vector.
3583         (thread_through_all_blocks): Use the vector of paths rather than
3584         a vector of 3-edge sets.
3585         (register_jump_thread): Accept pointer to a path vector rather
3586         than the path vector itself.  Store the path vector for later use.
3587         Simplify.
3589 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3590             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3592         PR target/58574
3593         * config/s390/s390.c (s390_split_branches): Modify check for table
3594         jump insns.
3595         (s390_chunkify_start): Rearrange table jump insn check in order to
3596         deal with compare and branch insns correctly.
3598 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3600         PR target/58578
3601         Revert
3602         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3603         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3604         define_insn_and_split.
3605         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3606         (shiftsi3_compare): New pattern.
3607         (rrx): New pattern.
3608         * config/arm/unspecs.md (UNSPEC_RRX): New.
3610 2013-10-01  Alan Modra  <amodra@gmail.com>
3612         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3613         casing inout operands.
3615 2013-10-01  Richard Biener  <rguenther@suse.de>
3617         PR tree-optimization/58553
3618         * tree-loop-distribution.c (struct partition_s): Add niter member.
3619         (classify_partition): Populate niter member for the partition
3620         and properly identify whether the relevant store happens before
3621         or after the loop exit.
3622         (generate_memset_builtin): Use niter member from the partition.
3623         (generate_memcpy_builtin): Likewise.
3625 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3627         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3628         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3630 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3632         * basic-block.h (edge_list): Prefix member names with "m_".
3633         * context.h (context): Likewise.
3634         * domwalk.h (dom_walker): Likewise.
3635         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3636         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3637         * hash-table.h (hash_table): Likewise.
3638         * machmode.h (bit_field_mode_iterator): Likewise.
3639         * pass_manager.h (pass_list): Likewise.
3640         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3641         * tree-pass.h (pass_data): Likewise.
3642         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3643         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3644         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3645         * asan.c (pass_data_asan): Update accordingly.
3646         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3647         (control_dependences::control_dependences): Likewise.
3648         (control_dependences::~control_dependences): Likewise.
3649         (control_dependences::~control_dependences): Likewise.
3650         (control_dependences::get_edges_dependent_on): Likewise.
3651         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3652         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3653         * context.c (gcc::context::context): Likewise.
3654         * cprop.c (pass_rtl_cprop::clone): Likewise.
3655         * domwalk.c (dom_walker::walk): Likewise.
3656         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3657         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3658         * mode-switching.c (pass_mode_switching::clone): Likewise.
3659         * passes.c (opt_pass::opt_pass): Likewise.
3660         (pass_manager::pass_manager): Likewise.
3661         * predict.c (pass_strip_predict_hints::clone): Likewise.
3662         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3663         (pass_split_all_insns::clone): Likewise.
3664         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3665         Likewise.
3666         (bit_field_mode_iterator::next_mode): Likewise.
3667         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3668         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3669         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3670         * tree-complex.c (pass_lower_complex::clone): Likewise.
3671         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3672         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3673         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3674         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3675         (pass_fold_builtins::clone): Likewise.
3676         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3677         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3678         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3679         (pass_cd_dce::clone): Likewise.
3680         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3681         (pass_phi_only_cprop::clone): Likewise.
3682         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3683         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3684         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3685         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3686         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3687         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3688         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3689         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3690         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3691         * tree-vrp.c (pass_vrp::clone): Likewise.
3692         * tsan.c (pass_tsan::clone): Likewise.
3694 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3696         PR middle-end/58564
3697         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3698         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3700         PR middle-end/58564
3701         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3702         optimization, punt if sign_bit_p looked through any zero extension.
3704 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3706         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3707         Update redirected out edge count in joiner case.
3708         (ssa_redirect_edges): Common the joiner and non-joiner cases
3709         so that joiner case gets profile updates.
3711 2013-09-30  Richard Biener  <rguenther@suse.de>
3713         PR tree-optimization/58554
3714         * tree-loop-distribution.c (classify_partition): Require
3715         unconditionally executed stores for memcpy and memset recognition.
3716         (tree_loop_distribution): Calculate dominance info.
3718 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3720         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3721         (NO_PROFILE_COUNTERS): Likewise.
3722         (PROFILE_HOOK): Likewise.
3723         (FUNCTION_PROFILER): Likewise.
3724         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3726 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3728         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3729         calls and defines in TARGET_MACHO conditional.
3730         (load_macho_picbase_di): Likewise.
3731         (reload_macho_picbase): Likewise.
3732         (reload_macho_picbase_si): Likewise.
3733         (reload_macho_picbase_di): Likewise.
3734         (nonlocal_goto_receiver): Likewise.
3736 2013-09-30  Nick Clifton  <nickc@redhat.com>
3738         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3739         that use the MSP430X ISA.
3740         (msp430_option_override): Scan -mmcu command line option for any
3741         MCU name that supports the MSP430X ISA.
3742         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3743         -mmcu options which enable the MSP430X ISA.
3745 2013-09-30  Richard Biener  <rguenther@suse.de>
3747         PR middle-end/58532
3748         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3749         before looking for setjmp-like calls.
3751 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3753         PR target/10901
3754         * config/darwin-protos.h (machopic_get_function_picbase): New.
3755         * config/darwin.c (machopic_get_function_picbase): New.
3756         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3757         label for a new func.  (load_macho_picbase_di): Likewise.
3758         (reload_macho_picbase): New expand.
3759         (reload_macho_picbase_si): New insn.
3760         (reload_macho_picbase_di): New insn.
3761         (nonlocal_goto_receiver): New define and split.
3762         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3763         (unspecv enum): Add UNSPECV_NLGR.
3765 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3767         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3768         that altivec registers are correctly sized on Darwin.
3770 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3772         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3773         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3774         * config/x-darwin (host-darwin.o): Likewise.
3775         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3776         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3777         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3779 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3781         * doc/invoke.texi: Fix usage of @tie{} command.
3783 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3785         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3787 2013-09-28  Jan Hubicka  <jh@suse.cz>
3789         * config/i386/x86-tune.def: Add documentation for each of the options;
3790         add whitespace.
3792 2013-09-28  Jan Hubicka  <jh@suse.cz>
3794         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3795         generic.
3796         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3797         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3798         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3799         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3800         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3802 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3804         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3805         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3806         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3807         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3808         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3809         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3810         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3811         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3812         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3813         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3814         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3815         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3816         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3817         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3818         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3819         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3820         gimple.h, godump.c, graphite-clast-to-gimple.c,
3821         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3822         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3823         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3824         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3825         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3826         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3827         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3828         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3829         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3830         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3831         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3832         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3833         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3834         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3835         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3836         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3837         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3838         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3839         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3840         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3841         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3842         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3843         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3844         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3845         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3846         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3847         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3848         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3849         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3850         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3851         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3852         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3853         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3854         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3855         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3856         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3857         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3858         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3859         whitespace before "(".
3861 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3863         * expr.h (extract_bit_field): Remove packedp parameter.
3864         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3865         from forward declaration.
3866         (store_split_bit_field): Remove packedp arg from calls to
3867         extract_fixed_bit_field.
3868         (extract_bit_field_1): Remove packedp parameter and packedp
3869         argument from recursive calls and calls to extract_fixed_bit_field.
3870         (extract_bit_field): Remove packedp parameter and corresponding
3871         arg to extract_bit_field_1.
3872         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3873         to issue warnings.
3874         (extract_split_bit_field): Remove packedp arg from call to
3875         extract_fixed_bit_field.
3876         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3877         (copy_blkmode_from_reg): Likewise.
3878         (copy_blkmode_to_reg): Likewise.
3879         (read_complex_part): Likewise.
3880         (store_field): Likewise.
3881         (expand_expr_real_1): Likewise.
3882         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3883         to extract_bit_field.
3884         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3885         call to extract_bit_field.
3886         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3887         call to extract_bit_field.
3888         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3889         and special packedp behavior from -fstrict-volatile-bitfields
3890         documentation.
3892 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3894         * lra-eliminations.c (init_elim_table): Guard value_p.
3896 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3898         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3899         DFmode, DImode, and SFmode in the upper VSX registers based on the
3900         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3901         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3902         of the VSX register handling.
3904         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3905         (f32_sv): Likewise.
3906         (zero_extendsidi2_lfiwzx): Add support for loading into the
3907         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3908         only do VSX memory options on Altivec registers.
3909         (extendsidi2_lfiwax): Likewise.
3910         (extendsfdf2_fpr): Likewise.
3911         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3912         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3913         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3914         (movdi_internal64): Likewise.
3916 2013-09-27  Xinliang David Li  <davidxl@google.com>
3918         * opts.c (finish_options): Adjust parameters
3919         according to vect cost model.
3920         (common_handle_option): Set dynamic vect cost
3921         model for FDO.
3922         targhooks.c (default_add_stmt_cost): Compute stmt cost
3923         unconditionally.
3924         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3925         Use helper function.
3926         * tree-vectorizer.h (unlimited_cost_model): New function.
3927         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3928         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3929         function.
3930         (vect_enhance_data_refs_alignment): Ditto.
3931         * flag-types.h: New enum.
3932         * common/config/i386/i386-common.c (ix86_option_init_struct):
3933         No need to initialize vect_cost_model flag.
3934         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3935         unconditionally.
3937 2013-09-27  Diego Novillo  <dnovillo@google.com>
3939         * gimple.h (enum ssa_mode): Remove.
3941 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3943         * cfgloop.h (number_of_loops): Fix typo in check for null.
3945 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3947         PR middle-end/58551
3948         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3949         are children of outermost saved_cfun's loop, and set it up to
3950         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3951         if loop != loop0 and SESE region contains bbs that belong to loop0.
3953 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3955         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3956         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3957         (set_address_segment, set_address_base, set_address_index)
3958         (set_address_disp): Accept the argument unconditionally.
3959         (baseness): Remove must_be_base_p and must_be_index_p checks.
3960         (decompose_normal_address): Classify as much as possible in the
3961         main loop.
3963 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3965         * cse.c (count_reg_usage): Handle INT_LIST.
3966         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3967         * reginfo.c (reg_scan_mark_refs): Likewise.
3968         * reload1.c (eliminate_regs_1): Likewise.
3970 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3972         PR middle-end/58547
3973         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3974         signed.
3976 2013-09-27  Richard Biener  <rguenther@suse.de>
3978         PR tree-optimization/58459
3979         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3980         restriction not propagating into loops.
3982 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3984         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3985         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3986         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3988 2013-09-26  Richard Biener  <rguenther@suse.de>
3990         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3992 2013-09-26  Richard Biener  <rguenther@suse.de>
3994         * alias.h (component_uses_parent_alias_set): Rename to ...
3995         (component_uses_parent_alias_set_from): ... this.
3996         * alias.c (component_uses_parent_alias_set): Rename to ...
3997         (component_uses_parent_alias_set_from): ... this and return
3998         the desired parent.
3999         (reference_alias_ptr_type_1): Use the result from
4000         component_uses_parent_alias_set_from instead of stripping
4001         components one at a time.
4002         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
4004 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
4006         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
4007         Move prototypes to...
4008         * tree-ssa-ter.h: New File.  Move prototypes here.
4009         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
4010         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
4011         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
4012         from is_replaceable_p.
4013         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
4014         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
4015         (process_replaceable): Use ter_is_replaceable_p.
4016         (find_replaceable_in_bb): Use ter_is_replaceable_p.
4017         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
4018         newly refactored ssa_is_replaceable_p.
4019         * cfgexpand.c: Include tree-outof-ssa.h.
4020         * ssaexpand.h: Delete.
4022 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
4024         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
4025         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
4026         tree-ssa.c
4027         (create_gimple_tmp): Delete.
4028         (get_expr_type, build_assign, build_type_cast): Move to...
4029         * gimple-builder.c: New File.
4030         (get_expr_type): Relocate from gimple.c.
4031         (build_assign, build_type_cast): Change to only create ssanames.
4032         * gimple.h: Move prototypes to...
4033         * gimple-builder.h: New File. Here.
4034         * tree-ssa.h: And here.
4035         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
4036         count_uses_and_derefs): Relocate from gimple.c.
4037         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
4038         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
4039         * tree-ssa-math-opts (execute_cse_reciprocals): Use
4040         gimple_replace_ssa_lhs.
4041         * asan.c: Include gimple-builder.h.
4042         * Makefile.in: Add gimple-builder.o.
4044 2013-09-26  Richard Biener  <rguenther@suse.de>
4046         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
4047         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
4048         (loe_visit_block): Use gcc_checking_assert.
4049         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
4050         gimple_assign_ssa_name_copy_p.
4051         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
4052         change.
4054 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
4056         * config/rs6000/t-rs6000 (rs6000.o): Remove.
4057         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
4059 2013-09-26  Richard Biener  <rguenther@suse.de>
4061         PR tree-optimization/58539
4062         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
4063         the fact that debug statements are not taking part in loop-closed
4064         SSA construction.
4066 2013-09-26  Nick Clifton  <nickc@redhat.com>
4068         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4069         time warning message.
4070         (msp430_print_operand_raw): Delete unused letter parameter.
4071         (TARGET_PRINT_OPERAND_ADDRESS): Define.
4072         (msp430_print_operand_address): New function.
4073         (msp430_print_operand): Move address printing code from here to
4074         new function.
4075         * config/msp430/msp430.md (movsipsi2): Add comment in generated
4076         assembler.
4077         (zero_extendpsisi2): Likewise.
4078         (extendpsisi2): New pattern.
4079         (andneghi3): New pattern.
4081 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4083         * config/aarch64/aarch64.opt (mlra): New option.
4084         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4085         (TARGET_LRA_P): Define.
4087 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4089         * expr.c (expand_assignment): Remove obsolete comment.
4091 2013-09-25  Jeff Law  <law@redhat.com>
4093         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4094         tree-ssa-threadupdate.h.
4095         (register_jump_thread): Similarly.
4096         * tree-ssa-threadupdate.h: New header file.
4097         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4098         * tree-vrp.c: Likewise.
4099         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4100         (thread_around_empty_blocks): Change type of path vector argument to
4101         an edge,type pair from just an edge.  Initialize both elements when
4102         appending to a jump threading path.  Tweak references to elements
4103         appropriately.
4104         (thread_across_edge): Similarly.  Release memory for the elements
4105         as needed.
4106         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4107         (dump_jump_thread_path): New function broken out from
4108         register_jump_thread.
4109         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4110         path vector entries.  Search the path for NULL edges and dump
4111         the path if one is found.  Tweak the conversion of path to 3-edge
4112         form to use the block copy type information embedded in the path.
4114 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4116         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4118 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4119             Vladimir Makarov  <vmakarov@redhat.com>
4121         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4122         from the least significant bit.
4123         (strip_address_mutations): Add bitfield operations handling.
4124         (must_be_index_p): Add shifting and rotate operations handling.
4125         (set_address_base): Use must_be_base_p predicate.
4126         (set_address_index): Use must_be_index_p predicate.
4128 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4129             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4130             Sergey Lega  <sergey.s.lega@intel.com>
4131             Anna Tikhonova  <anna.tikhonova@intel.com>
4132             Ilya Tocar  <ilya.tocar@intel.com>
4133             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4134             Ilya Verbin  <ilya.verbin@intel.com>
4135             Kirill Yukhin  <kirill.yukhin@intel.com>
4136             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4138         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4139         Use new names.
4140         (ix86_expand_vector_move_misalign): Support new unaligned load and
4141         stores and use new names.
4142         (CODE_FOR_sse2_storedqu): Rename to ...
4143         (CODE_FOR_sse2_storedquv16qi): ... this.
4144         (CODE_FOR_sse2_loaddqu): Rename to ...
4145         (CODE_FOR_sse2_loaddquv16qi): ... this.
4146         (CODE_FOR_avx_loaddqu256): Rename to ...
4147         (CODE_FOR_avx_loaddquv32qi): ... this.
4148         (CODE_FOR_avx_storedqu256): Rename to ...
4149         (CODE_FOR_avx_storedquv32qi): ... this.
4150         * config/i386/i386.md (fpint_logic): New.
4151         * config/i386/sse.md (VMOVE): Extend for AVX512.
4152         (VF): Ditto.
4153         (VF_128_256): New.
4154         (VF_512): Ditto.
4155         (VI_UNALIGNED_LOADSTORE): Ditto.
4156         (sse2_avx_avx512f): Ditto.
4157         (sse2_avx2): Extend for AVX512.
4158         (sse4_1_avx2): Ditto.
4159         (avx2_avx512f): New.
4160         (sse): Extend for AVX512.
4161         (sse2): Ditto.
4162         (sse4_1): Ditto.
4163         (avxsizesuffix): Ditto.
4164         (sseintvecmode): Ditto.
4165         (ssePSmode): Ditto.
4166         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4167         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4168         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4169         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4170         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4171         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4172         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4173         (STORENT_MODE): Extend for AVX512.
4174         (*absneg<mode>2): Replace constraint "x" with "v".
4175         (*mul<mode>3): Ditto.
4176         (*ieee_smin<mode>3): Ditto.
4177         (*ieee_smax<mode>3): Ditto.
4178         (avx_cmp<mode>3): Replace VF with VF_128_256.
4179         (*<sse>_maskcmp<mode>3_comm): Ditto.
4180         (<sse>_maskcmp<mode>3): Ditto.
4181         (<sse>_andnot<mode>3): Extend for AVX512.
4182         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4183         (<code><mode>3, fpint_logic): New.
4184         (*<code><mode>3): Extend for AVX512.
4185         (avx512flogicsuff): New.
4186         (avx512f_<logic><mode>): Ditto.
4187         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4188         VF_128_256.
4189         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4190         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4191         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4192         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4193         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4194         (xop_vpermil2<mode>3): Ditto.
4195         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4196         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4197         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4198         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4200 2013-09-25  Tom Tromey  <tromey@redhat.com>
4202         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4203         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4204         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4205         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4206         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4207         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4208         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4209         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4210         (GRAPHITE_HTAB_H): Remove.
4212 2013-09-25  Tom Tromey  <tromey@redhat.com>
4214         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4216 2013-09-25  Tom Tromey  <tromey@redhat.com>
4218         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4220 2013-09-25  Tom Tromey  <tromey@redhat.com>
4222         * config/i386/t-i386 (i386.o): Remove.
4223         (i386-c.o): Use COMPILE and POSTCOMPILE.
4225 2013-09-25  Tom Tromey  <tromey@redhat.com>
4227         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4229 2013-09-25  Tom Tromey  <tromey@redhat.com>
4231         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4232         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4233         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4234         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4235         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4236         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4237         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4238         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4239         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4240         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4241         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4242         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4243         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4244         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4245         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4246         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4247         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4248         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4249         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4250         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4251         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4252         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4253         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4254         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4255         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4256         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4257         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4258         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4259         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4260         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4261         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4262         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4263         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4264         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4265         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4266         (graphite-clast-to-gimple.o, graphite-dependences.o)
4267         (graphite-interchange.o, graphite-poly.o)
4268         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4269         (graphite-optimize-isl.o, tree-vect-loop.o)
4270         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4271         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4272         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4273         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4274         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4275         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4276         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4277         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4278         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4279         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4280         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4281         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4282         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4283         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4284         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4285         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4286         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4287         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4288         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4289         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4290         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4291         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4292         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4293         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4294         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4295         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4296         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4297         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4298         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4299         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4300         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4301         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4302         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4303         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4304         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4305         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4306         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4307         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4308         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4309         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4310         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4311         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4312         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4313         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4314         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4315         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4316         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4317         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4318         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4319         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4320         (gcov-dump.o): Remove.
4321         (default-c.o): Use COMPILE and POSTCOMPILE.
4322         (CFLAGS-gcc.o): New variable.
4323         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4325 2013-09-25  Tom Tromey  <tromey@redhat.com>
4327         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4328         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4329         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4330         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4331         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4332         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4333         (c-family/array-notation-common.o, c-family/stub-objc.o)
4334         (c-family/c-ubsan.o): Remove.
4336 2013-09-25  Tom Tromey  <tromey@redhat.com>
4338         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4340 2013-09-25  Tom Tromey  <tromey@redhat.com>
4342         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4343         to add -DENABLE_SHARED_LIBGCC.
4344         (gcc.o): Don't use subshell.
4346 2013-09-25  Tom Tromey  <tromey@redhat.com>
4348         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4349         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4350         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4351         * configure, config.in: Rebuild.
4353 2013-09-25  Tom Tromey  <tromey@redhat.com>
4355         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4356         (COMPILE, POSTCOMPILE): New variables.
4357         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4358         (DEPFILES): New variable.
4359         Include ".Po" files.
4360         * configure.ac: Add checks for dependency checking.
4361         * configure, aclocal.m4: Regenerate.
4363 2013-09-25  Tom Tromey  <tromey@redhat.com>
4365         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4366         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4368 2013-09-25  Tom Tromey  <tromey@redhat.com>
4370         * Makefile.in (generated_files): Add options.h,
4371         target-hooks-def.h, insn-opinit.h,
4372         common/common-target-hooks-def.h, pass-instances.def,
4373         c-family/c-target-hooks-def.h.
4375 2013-09-25  Jeff Law  <law@redhat.com>
4377         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4378         than foo[foo.length () - 1] to access last member in a vec.
4379         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4381 2013-09-25  Richard Biener  <rguenther@suse.de>
4383         PR middle-end/58521
4384         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4386 2013-09-25  Jan Hubicka  <jh@suse.cz>
4388         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4389         test.
4391 2013-09-25  Marek Polacek  <polacek@redhat.com>
4393         PR sanitizer/58420
4394         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4395         when determining the type name.
4397 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4399         * config/sh/sh.md: Fix formatting.
4401 2013-09-24  Xinliang David Li  <davidxl@google.com>
4403         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4404         max peel iterations parameter.
4405         * param.def: New parameter.
4406         * doc/invoke.texi: Document New parameter.
4408 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4410         * gimple-pretty-print.c: Various whitespace tweaks.
4411         * tree-core.h: Likewise.
4412         * tree-pretty-print.c: Likewise.
4413         * tree-ssa-alias.c: Likewise.
4414         * tree-ssa-copy.c: Likewise.
4415         * tree-ssanames.c: Likewise.
4416         * tree-ssanames.h: Likewise.
4417         * tree-vrp.c: Likewise.
4419 2013-09-24  Alan Modra  <amodra@gmail.com>
4421         PR middle-end/57134
4422         PR middle-end/57586
4423         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4424         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4425         inout operands.
4427 2013-09-24  Richard Biener  <rguenther@suse.de>
4429         PR middle-end/58513
4430         * tree.c (reference_alias_ptr_type): Move ...
4431         * alias.c (reference_alias_ptr_type): ... here and implement
4432         in terms of the new reference_alias_ptr_type_1.
4433         (ref_all_alias_ptr_type_p): New helper.
4434         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4435         use ref_all_alias_ptr_type_p.
4436         (get_deref_alias_set): Add flag_strict_aliasing check here.
4437         (reference_alias_ptr_type_1): New function, split out from ...
4438         (get_alias_set): ... here.
4439         (alias_ptr_types_compatible_p): New function.
4440         * alias.h (reference_alias_ptr_type): Declare.
4441         (alias_ptr_types_compatible_p): Likewise.
4442         * tree.h (reference_alias_ptr_type): Remove.
4443         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4444         to compare MEM_REF alias types.
4446 2013-09-24  Richard Biener  <rguenther@suse.de>
4448         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4450 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4452         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4453         reload helper function arrays into a single array reg_addr.
4454         (reload_fpr_gpr): Likewise.
4455         (reload_gpr_vsx): Likewise.
4456         (reload_vsx_gpr): Likewise.
4457         (struct rs6000_reg_addr): Likewise.
4458         (reg_addr): Likewise.
4459         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4460         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4461         (rs6000_init_hard_regno_mode_ok): Likewise.
4462         (rs6000_secondary_reload_direct_move): Likewise.
4463         (rs6000_secondary_reload): Likewise.
4465         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4466         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4467         power8 changes.  Put wg constraint in alphabetical order.
4469         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4470         for future work to add ISA 2.07 VSX single precision support.
4471         (-mvsx-scalar-double): Change default from -1 to 1, update
4472         documentation comment.
4473         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4474         (-mupper-regs-df): New debug switch to control whether DF values
4475         can go in the traditional Altivec registers.
4476         (-mupper-regs-sf): New debug switch to control whether SF values
4477         can go in the traditional Altivec registers.
4479         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4480         and wy constraints.
4481         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4482         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4483         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4484         (rs6000_debug_legitimate_address_p): Print if we are running
4485         before, during, or after reload.
4486         (rs6000_secondary_reload): Add a comment.
4487         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4489         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4490         constraints.  Update documentation string.
4491         (wd constraint): Likewise.
4492         (wf constraint): Likewise.
4493         (wg constraint): Likewise.
4494         (wn constraint): Likewise.
4495         (ws constraint): Likewise.
4496         (wt constraint): Likewise.
4497         (wx constraint): Likewise.
4498         (wz constraint): Likewise.
4499         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4500         instructions.
4501         (ww constraint): Likewise.
4502         (wy constraint): Likewise.
4503         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4504         in the previous submissions.
4505         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4507 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4509         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4510         (REG_BR_PROB): Say that the probability is stored in an int_list.
4511         * reg-notes.def: Update commentary to mention INT_LIST.
4512         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4513         (INT_LIST): New rtx.
4514         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4515         * rtlanal.c (int_reg_note_p): New function.
4516         (alloc_reg_note): Assert that the note does not have an int argument.
4517         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4518         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4519         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4520         rather than an INSN_LIST.  Handle INT_LIST.
4521         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4522         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4523         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4524         Manipulate them as ints rather than rtxes.
4525         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4526         * regmove.c (copy_src_to_dest): Likewise.
4527         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4529         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4530         into the cases that need it.
4531         * config/arm/arm.c (arm_unwind_emit): Likewise.
4533         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4534         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4535         * loop-doloop.c (add_test, doloop_modify): Likewise.
4536         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4537         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4538         * predict.c (combine_predictions_for_insn): Likewise.
4539         * print-rtl.c (print_rtx): Handle INT_LIST.
4540         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4541         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4542         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4543         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4544         (ix86_split_fp_branch, predict_jump): Likewise.
4545         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4546         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4547         * config/spu/spu.c (ea_load_store_inline): Likewise.
4549         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4550         value of a REG_BR_PROB note.
4551         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4552         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4553         * emit-rtl.c (try_split): Likewise.
4554         * predict.c (br_prob_note_reliable_p): Likewise.
4555         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4556         * reorg.c (mostly_true_jump): Likewise.
4557         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4558         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4559         * config/i386/i386.c (ix86_print_operand): Likewise.
4560         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4561         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4562         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4563         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4564         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4565         * config/sparc/sparc.c (output_cbranch): Likewise.
4566         * config/spu/spu.c (get_branch_target): Likewise.
4567         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4568         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4570 2013-09-23  Jan Hubicka  <jh@suse.cz>
4572         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4573         for ipa-devirt.
4574         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4575         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4576         of external calls
4577         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4578         (gimple_fold_call): Dump inconsistent devirtualizations; add
4579         sanity check for type based devirtualizations.
4580         * ipa-prop.c: Include ipa-utils.h
4581         (ipa_intraprocedural_devirtualization): Add sanity check.
4582         (try_make_edge_direct_virtual_call): Likewise.
4584 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4586         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4587         assignment statements.
4589 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4591         * gimple-pretty-print.c (dump_ssaname_info): New function.
4592         (dump_gimple_phi): Call it.
4593         (pp_gimple_stmt_1): Likewise.
4594         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4595         (range_info_def): Declare.
4596         * tree-pretty-print.c (pp_double_int): New function.
4597         (dump_generic_node): Call it.
4598         * tree-pretty-print.h (pp_double_int): Declare.
4599         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4600         * tree-ssanames.h (range_info_def): New structure.
4601         (value_range_type): Move definition here.
4602         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4603         Declare.
4604         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4605         initialization.
4606         (set_range_info): New function.
4607         (get_range_info): Likewise.
4608         (duplicate_ssa_name_range_info): Likewise.
4609         (duplicate_ssa_name_fn): Check pointer type and call
4610         duplicate_ssa_name_range_info.
4611         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4612         * tree-vrp.c (value_range_type): Remove definition, now in
4613         tree-ssanames.h.
4614         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4615         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4616         (SSA_NAME_RANGE_INFO): New macro.
4618 2013-09-23  Richard Biener  <rguenther@suse.de>
4620         PR tree-optimization/58464
4621         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4622         (phi_trans_add): Change to add conditionally on being not present.
4623         (phi_translate_1): Remove recursion detection here.
4624         (phi_translate): Pre-seed the cache with NULL to catch
4625         recursion here in a more generic way.
4626         (bitmap_find_leader): Adjust comment.
4627         (get_representative_for): Dump value-numbers.
4628         (create_expression_by_pieces): Likewise.
4629         (insert_into_preds_of_block): Likewise.
4631 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4633         PR target/58475
4634         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4635         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4637 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4639         Revert r202780:
4640         2013-09-20  Renlin Li  <renlin.li@arm.com>
4642         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4643         plus_constant.
4644         (aarch64_expand_epilogue): Likewise.
4645         (aarch64_legitimize_reload_address): Likewise.
4647 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4649         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4650         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4652 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4654         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4655         comparison operators when -fno-trapping-math is in effect.
4656         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4657         comparison operators in TFmode and assert that unsupported operators
4658         cannot reach here.
4659         (ia64_print_operand): Likewise.
4661 2013-09-21  Jan Hubicka  <jh@suse.cz>
4663         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4664         (sse_typeless_stores): Enable for core
4665         (sse_load0_by_pxor): Likewise.
4666         (four_jump_limit): Disable for core.
4667         (pad_returns): Likewise.
4668         (avoid_vector_decode): Likewise.
4669         (fuse_cmp_and_branch): Enable for cores.
4670         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4672 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4674         PR middle-end/56791
4675         * config/pa/pa.c (pa_option_override): Disable auto increment and
4676         decrement instructions until reload is completed.
4678         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4679         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4680         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4682 2013-09-20  DJ Delorie  <dj@redhat.com>
4683             Nick Clifton  <nickc@redhat.com>
4685         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4686         (need_to_save): Save bank 0 on interrupts.
4687         (characterize_address): Strip far address wrappers.
4688         (rl78_as_legitimate_address): Likewise.
4689         (transcode_memory_rtx): Likewise.
4690         (rl78_peep_movhi_p): Disable this peephole after devirt.
4691         (rl78_propogate_register_origins): Forget all origins when a
4692         CLOBBER is seen.
4693         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4694         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4695         peephole2's.
4696         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4697         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4698         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4699         relocating.
4700         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4702 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4704         * config/pa/pa.md: In "scc" insn patterns, change output template to
4705         handle const0_rtx in reg_or_0_operand operands.
4707 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4709         PR target/56875
4710         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4711         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4713 2013-09-20  Richard Biener  <rguenther@suse.de>
4715         PR middle-end/58484
4716         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4717         remembering SSA name version and block index.
4718         (new_scev_info_str): Adjust.
4719         (hash_scev_info): Likewise.  Also hash the block index.
4720         (eq_scev_info): Adjust.
4721         (find_var_scev_info): Likewise.
4722         (struct instantiate_cache_entry): Remove.
4723         (struct instantiate_cache_type): Use a htab to map name, block
4724         to chrec.
4725         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4726         (get_instantiated_value_entry): Likewise.
4727         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4728         (instantiate_scev_name): Adjust.
4730 2013-09-20  Jeff Law  <law@redhat.com>
4732         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4734 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4736         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4737         Call aarch64_simd_expand_args to update op[argc].
4739 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4741         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4742         plugin argument.
4744 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4746         * gengtype.c (file_rules): Added rule for *.cc files.
4747         (get_output_file_with_visibility): Give fatal message when no
4748         rules found.
4750 2013-09-20  Renlin Li  <renlin.li@arm.com>
4752         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4753         (aarch64_expand_epilogue): Likewise.
4754         (aarch64_legitimize_reload_address): Likewise.
4756 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4758         PR middle-end/57748
4759         * expr.c (expand_assignment): Remove misalignp code path.
4761 2013-09-20  Marek Polacek  <polacek@redhat.com>
4763         PR sanitizer/58413
4764         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4765         TYPE_PRECISION.  Add asserts.
4767 2013-09-20  Richard Biener  <rguenther@suse.de>
4769         PR tree-optimization/58453
4770         * tree-loop-distribution.c (distribute_loop): Apply the cost
4771         model for -ftree-loop-distribute-patterns, too.
4773 2013-09-20  Richard Biener  <rguenther@suse.de>
4775         PR middle-end/58473
4776         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4777         make type comparison less strict.
4779 2013-09-20  Alan Modra  <amodra@gmail.com>
4781         * configure: Regenerate.
4782         * aclocal.m4: Regenerate.
4784 2013-09-20  Marek Polacek  <polacek@redhat.com>
4786         PR other/58467
4787         * doc/extend.texi: Document that attribute used is meant to be used
4788         on variables with static storage duration.
4790 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4792         PR tree-optimization/58472
4793         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4794         simd_lane_access set inv_p = false.
4795         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4796         the simduid magic VAR_DECL.
4798 2013-09-19  Jan Hubicka  <jh@suse.cz>
4800         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4802 2013-09-17  Jeff Law  <law@redhat.com>
4804         * tree-ssa-dom.c (record_temporary_equivalences): New function
4805         split out of dom_opt_dom_walker::after_dom_children.
4806         (dom_opt_dom_walker::thread_across_edge): Move common code
4807         in here from dom_opt_dom_walker::after_dom_children.
4808         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4810 2013-09-19  Jan Hubicka  <jh@suse.cz>
4812         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4813         (TARGET_GENERIC): Use PROCESOR_GENERIC
4814         (enum processor_type): Unify generic32 and 64.
4815         * i386.md (cpu): Likewise.
4816         * x86-tune.def (use_leave): Enable for generic32.
4817         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4818         * athlon.md: Change generic64 to generic in all occurences.
4819         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4820         (ix86_target_macros_internal): Likewise.
4821         * driver-i386.c (host_detect_local_cpu): Likewise.
4822         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4823         to ..
4824         (generic_memcpy, generic_memset, generic_cost): This one.
4825         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4826         (m_GENERIC32, m_GENERIC64): Remove.
4827         (m_GENERIC): Turn into one flag.
4828         (processor_target): Unify generic tunnings.
4829         (ix86_option_override_internal): Replace generic32/64 by generic.
4830         (ix86_issue_rate): Likewise.
4831         (ix86_adjust_cost): Likewise.
4833 2013-09-19  Jan Hubicka  <jh@suse.cz>
4835         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4836         of speculative flag.
4838 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4840         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4841         GOMP_sections_start, even if !exit_reachable.
4843 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4845         * lra-constraints.c (need_for_all_save_p): Use macro
4846         HARD_REGNO_CALL_PART_CLOBBERED.
4847         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4848         set up pseudo conflict hard regs.
4850 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4852         PR target/58452
4853         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4854         operands.
4856 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4858         PR rtl-optimization/58438
4859         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4860         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4861         from previous subpasses.
4863 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4865         * arm.c (arm_get_frame_offsets): Validate architecture supports
4866         LDRD/STRD before accepting the tuning preference.
4867         (arm_expand_prologue): Likewise.
4868         (arm_expand_epilogue): Likewise.
4870 2013-09-18  Richard Biener  <rguenther@suse.de>
4872         PR tree-optimization/58417
4873         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4874         have chrecs with symbols defined in the loop as operands.
4875         (chrec_fold_multiply): Likewise.
4876         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4877         parameters before folding binary operations.
4878         (struct instantiate_cache_entry_hasher): Remove.
4879         (struct instantiate_cache_type): Use a pointer-map.
4880         (instantiate_cache_type::instantiate_cache_type): New function.
4881         (instantiate_cache_type::get): Likewise.
4882         (instantiate_cache_type::set): Likewise.
4883         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4884         (get_instantiated_value_entry): Likewise.
4885         (global_cache): New global.
4886         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4887         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4888         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4889         (instantiate_scev_name): Adjust.
4890         (instantiate_scev): Construct global instead of local cache.
4891         (resolve_mixers): Likewise.
4893 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4894             Paolo Carlini  <paolo.carlini@oracle.com>
4896         PR c++/58458
4897         * doc/implement-cxx.texi: Fix references to the C++ standards.
4899 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4901         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4902         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4903         simd_lane_access drs, update also DR_ALIGNED_TO.
4905 2013-09-18  Marek Polacek  <polacek@redhat.com>
4907         PR sanitizer/58411
4908         * doc/extend.texi: Document no_sanitize_undefined attribute.
4909         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4910         no_sanitize_undefined attribute.
4912 2013-09-18  Nick Clifton  <nickc@redhat.com>
4914         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4915         (ASM_DECLARE_FUNCTION_NAME): Define.
4917 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4919         * compare-elim.c (find_comparison_dom_walker): New class
4920         (find_comparisons_in_bb): Rename to
4921         find_comparison_dom_walker::before_dom_children
4922         (find_comparisons): Adjust
4923         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4924         adjust.
4925         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4926         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4927         (init_walk_dominator_tree): Remove declaration.
4928         (fini_walk_dominator_tree): Remove declaration.
4929         * fwprop.c (single_def_use_dom_walker): New class
4930         (single_def_use_enter_block): Convert to
4931         single_def_use_dom_walker::before_dom_children.
4932         (single_def_use_leave_block): Convert to
4933         single_def_use_dom_walker::after_dom_children.
4934         (build_single_def_use_links): Adjust.
4935         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4936         class.
4937         (find_candidates_in_block): Convert to
4938         find_candidates_dom_walker::before_dom_children.
4939         (execute_strength_reduction): Adjust.
4940         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4941         (sese_dom_walker): New class.
4942         (sese_dom_walker::sese_dom_walker): New constructor.
4943         (sese_dom_walker::~sese_dom_walker): New destructor.
4944         (build_sese_conditions_before): Convert to
4945         sese_dom_walker::before_dom_children.
4946         (build_sese_conditions_after): Convert to
4947         sese_dom_walker::after_dom_children.
4948         (build_poly_scop): Adjust
4949         * tree-into-ssa.c (rewrite_dom_walker): New class
4950         (rewrite_enter_block): Convert to
4951         rewrite_dom_walker::before_dom_children.
4952         (rewrite_leave_block): Convert to
4953         rewrite_dom_walker::after_dom_children.
4954         (rewrite_update_dom_walker): New class.
4955         (rewrite_update_enter_block): Convert to
4956         rewrite_update_dom_walker::before_dom_children.
4957         (rewrite_update_leave_block): Convert to
4958         rewrite_update_dom_walker::after_dom_children.
4959         (rewrite_blocks, rewrite_into_ssa): Adjust.
4960         (mark_def_dom_walker): New class.
4961         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4962         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4963         (mark_def_sites_blocks): Convert to
4964         mark_def_dom_walker::before_dom_children.
4965         (mark_def_site_blocks): Remove.
4966         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4967         (tree_ssa_dominator_optimize): Adjust.
4968         (dom_thread_across_edge): Convert to method
4969         dom_opt_dom_walker::thread_across_edge.
4970         (dom_opt_enter_block): Convert to member function
4971         dom_opt_dom_walker::before_dom_children.
4972         (dom_opt_leave_block): Convert to member function
4973         dom_opt_dom_walker::after_dom_children.
4974         * tree-ssa-dse.c (dse_dom_walker): New class.
4975         (dse_enter_block): Convert to member function
4976         dse_dom_walker::before_dom_children.
4977         (tree_ssa_dse): Adjust.
4978         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4979         (determine_invariantness_stmt): Convert to method
4980         invariantness_dom_walker::before_dom_children.
4981         (determine_invariantness): Remove
4982         (move_computations_dom_walker): New class.
4983         (move_computations_stmt): Convert to method
4984         move_computations_dom_walker::before_dom_children.
4985         (move_computations, tree_ssa_lim): Adjust.
4986         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4987         (nt_init_block): Make method
4988         notrappping_dom_walker::before_dom_children.
4989         (nt_fini_block): Make
4990         method nontrapping_dom_walker::after_dom_children.
4991         (get_non_trapping): Adjust.
4992         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4993         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4994         (eliminate_leave_block): Make method.
4995         eliminate_dom_walker::after_dom_children.
4996         (eliminate): Adjust
4997         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4998         (strlen_enter_block): Make method
4999         strlen_dom_walker::before_dom_children.
5000         (strlen_leave_block): Make
5001         method strlen_dom_walker::after_dom_children.
5002         (tree_ssa_strlen): Adjust.
5003         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
5004         (tree_ssa_uncprop): Adjust.
5005         (uncprop_leave_block): Make method
5006         uncprop_dom_walker::after_dom_children.
5007         (uncprop_leave_block): Make method
5008         uncprop_dom_walker::before_dom_children.
5010 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
5012         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
5013         branch in same basic block.  Check both src and dest of the move insn.
5015 2013-09-17  Nick Clifton  <nickc@redhat.com>
5017         * config/rl78/rl78-real.md (bf): New pattern.
5018         (bt): New pattern.
5019         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
5020         (rl78_print_operand): Do not put a # before a %B.
5021         * config/rl78/rl78.opt: Tweak doc strings.
5023 2013-09-17  DJ Delorie  <dj@redhat.com>
5025         * config/rl78/constraints.md (Wcv): Allow up to $r31.
5026         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
5027         (rl78_option_override): Likewise, if -mallregs.
5028         (is_virtual_register): Likewise.
5029         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
5030         (REGNO_OK_FOR_BASE_P): Likewise.
5031         * config/rl78/rl78.opt (-mallregs): New.
5033 2013-09-17  Nick Clifton  <nickc@redhat.com>
5035         * config/rl78/rl78.c (need_to_save): Change return type to bool.
5036         For interrupt functions: save all call clobbered registers if the
5037         interrupt handler is not a leaf function.
5038         (rl78_expand_prologue): Always recompute the frame information.
5039         For interrupt functions: only select bank 0 if one of the bank 0
5040         registers is going to be psuhed.
5042 2013-09-17  DJ Delorie  <dj@redhat.com>
5044         * config/rl78/constraints.md: For each W* constraint, rename to C*
5045         and create a W* constraint that checks for an optional ES: prefix
5046         pattern also.
5047         * config/rl78/rl78.md (UNS_ES_ADDR): New.
5048         (es_addr): New.  Used to wrap far addresses.
5049         * config/rl78/rl78-protos.h (rl78_es_addr): New.
5050         (rl78_es_base): New.
5051         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
5052         wrapped far addresses.
5053         (rl78_print_operand_1): Unwrap far addresses before processing.
5054         (rl78_lo16): Wrap far addresses in unspecs.
5055         (rl78_es_addr): New.
5056         (rl78_es_base): New.
5057         (insn_ok_now): Check for not-yet-wrapped far addresses.
5058         (transcode_memory_rtx): Properly re-wrap far addresses.
5060 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5062         * config/sparc/t-rtems: Add leon3 multilibs.
5064 2013-09-17  Cong Hou  <congh@google.com>
5066         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
5067         when checking the dot production pattern. The type of rhs operand
5068         of multiply is now checked correctly.
5070 2013-09-17  Jeff Law  <law@redhat.com>
5072         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5073         edge implied equivalences into successor phis.
5074         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
5075         here from tree-ssa-threadedge.c.
5076         (mark_threaded_blocks): When threading through a joiner, if both
5077         successors of the joiner's clone reach the same block, verify the
5078         PHI arguments are equal.  If not, cancel the jump threading request.
5079         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
5080         tree-ssa-threadupdate.c
5081         (thread_across_edge): Don't check PHI argument equality when
5082         threading through joiner block here.
5084 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5086         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5087         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5088         (warn_uninit, warn_uninitialized_vars,
5089         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5090         Move to tree-ssa-uninit.c.
5091         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5092         (has_undefined_value_p): New.  Pass dependant parts of
5093         ssa_undefined_value_p.
5094         (uninit_undefined_value_p): Use has_undefined_value_p.
5095         (warn_uninit, warn_uninitialized_vars,
5096         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5097         Move from tree-ssa.c.
5098         * tree-ssa.h: Adjust prototypes.
5100 2013-09-17  Jan Hubicka  <jh@suse.cz>
5102         PR middle-end/58332
5103         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5104         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5105         FUNCTION_NOT_OPTIMIZED.
5106         * ipa-inline-analysis.c (compute_inline_parameters): Function
5107         not optimized is not inlinable unless it is alwaysinline.
5108         (inline_analyze_function): Force calls in not optimized
5109         function not inlinable.
5111 2013-09-17  Jan Hubicka  <jh@suse.cz>
5113         PR middle-end/58329
5114         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5115         to return NULL.
5116         * ipa.c (function_and_variable_visibility): Likewise.
5117         * ipa-profile.c (ipa_profile): Likewise.
5119 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5121         PR ipa/58398
5122         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5123         attribute, and don't inline the resolver in this case.
5125 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5127         * coverage.c (get_coverage_counts): Add missing newline.
5129 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5131         PR tree-optimization/58088
5132         * fold-const.c (mask_with_trailing_zeros): New function.
5133         (fold_binary_loc): Make sure we don't recurse infinitely
5134         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5135         Use mask_with_trailing_zeros where appropriate.
5137 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5139         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5140         of 'prev' var to get better distance estimation.
5142 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5144         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5145         * tree-inline.c (is_parameter_of): New predicate.
5146         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5147         a parameter has been remapped.
5148         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5149         (optimize_inline_calls): Initialize transform_parameter.
5150         (copy_gimple_seq_and_replace_locals): Likewise.
5151         (tree_function_versioning): Likewise.
5152         (maybe_inline_call_in_expr): Likewise.
5154 2013-09-17  Nick Clifton  <nickc@redhat.com>
5156         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5157         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5158         interrupt handlers.
5159         (is_attr_func): New function.
5160         (msp430_is_interrupt_func): New function.
5161         (is_naked_func): New function.
5162         (is_reentrant_func): New function.
5163         (is_critical_func): New function.
5164         (msp430_start_function): Add annotations for function attributes.
5165         (msp430_attr): New function.
5166         (msp430_attribute_table): New.
5167         (msp430_function_section): New function.
5168         (TARGET_ASM_FUNCTION_SECTION): Define.
5169         (msp430_builtin): New enum.
5170         (msp430_init_builtins): New function.
5171         (msp430_builtin_devl): New function.
5172         (msp430_expand_builtin): New function.
5173         (TARGET_INIT_BUILTINS): Define.
5174         (TARGET_EXPAND_BUILTINS): Define.
5175         (TARGET_BUILTIN_DECL): Define.
5176         (msp430_expand_prologue): Add support for naked, interrupt,
5177         critical and reentrant functions.
5178         (msp430_expand_epilogue): Likewise.
5179         (msp430_print_operand): Handle 'O' character.
5180         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5181         NO_TRAMPOLINES.
5182         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5183         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5184         (pushm): Use a 'n' rather than an 'i' constraint.
5185         (msp_return): Add generation of the interrupt return instruction.
5186         (disable_interrupts): New pattern.
5187         (enable_interrupts): New pattern.
5188         (push_intr_state): New pattern.
5189         (pop_intr_state): New pattern.
5190         (bic_SR): New pattern.
5191         (bis_SR): New pattern.
5192         * doc/extend.texi: Document MSP430 function attributes and builtin
5193         functions.
5195 2013-09-17  Richard Biener  <rguenther@suse.de>
5197         PR tree-optimization/58432
5198         * tree-loop-distribution.c (tree_loop_distribution): Also
5199         scan PHIs for outside loop uses and seed a partition from them.
5201 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5203         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5204         (restructure_reference): Call backtrace_base_for_ref.
5206 2013-09-17  Alan Modra  <amodra@gmail.com>
5208         PR target/57589
5209         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5210         patch.
5212 2013-09-16  DJ Delorie  <dj@redhat.com>
5214         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5215         vregs location for RL78/G10.
5216         (rl78_expand_prologue): Avoid SEL on G10.
5217         (rl78_expand_epilogue): Likewise.
5218         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5219         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5220         __RL78_G10__ when appropriate.
5221         (ASM_SPEC): Pass -mg10 along to the assembler.
5222         * config/rl78/rl78.md (sel_rb): Disable for G10.
5223         * config/rl78/rl78.opt: Add -mg10 option.
5224         * config/rl78/t-rl78: Add -mg10 multilib.
5226 2013-09-16  Xinliang David Li  <davidxl@google.com>
5228         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5229         * omp-low.c (omp_max_vf): Ditto.
5230         (expand_omp_simd): Ditto.
5231         * tree-vectorizer.c (vectorize_loops): Ditto.
5232         (gate_vect_slp): Ditto.
5233         (gate_increase_alignment): Ditto.
5234         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5235         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5236         (gate_tree_vectorize): Name change.
5237         (tree_vectorize): Ditto.
5238         (pass_vectorize::gate): Call new function.
5239         (pass_vectorize::execute): Ditto.
5240         * opts.c: O3 default setting change.
5241         (finish_options): Check new flag.
5242         * doc/invoke.texi: Document new flags.
5243         * common.opt: New flags.
5245 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5247         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5248         (Condition Code): Don't reference it.
5249         * doc/tm.texi: Regenerate.
5251 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5253         PR middle-end/58418
5254         * lra-constraints.c (undo_optional_reloads): Consider all optional
5255         reload even if it did not get a hard reg.
5257 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5259         * dumpfile.c (dump_loc): Remove newline emission.
5260         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5261         emission to dump_printf_loc calls where missing.
5262         (vect_mark_for_runtime_alias_test): Ditto.
5263         (vect_analyze_data_ref_dependence): Ditto.
5264         (vect_analyze_data_ref_dependences): Ditto.
5265         (vect_slp_analyze_data_ref_dependence): Ditto.
5266         (vect_slp_analyze_data_ref_dependences): Ditto.
5267         (vect_compute_data_ref_alignment): Ditto.
5268         (vect_update_misalignment_for_peel): Ditto.
5269         (vect_verify_datarefs_alignment): Ditto.
5270         (vector_alignment_reachable_p): Ditto.
5271         (vect_get_data_access_cost): Ditto.
5272         (vect_enhance_data_refs_alignment): Ditto.
5273         (vect_find_same_alignment_drs): Ditto.
5274         (vect_analyze_data_refs_alignment): Ditto.
5275         (vect_analyze_group_access): Ditto.
5276         (vect_analyze_data_ref_access): Ditto.
5277         (vect_analyze_data_ref_accesses): Ditto.
5278         (vect_prune_runtime_alias_test_list): Ditto.
5279         (vect_analyze_data_refs): Ditto.
5280         (vect_create_addr_base_for_vector_ref): Ditto.
5281         (vect_create_data_ref_ptr): Ditto.
5282         (vect_grouped_store_supported): Ditto.
5283         (vect_grouped_load_supported): Ditto.
5284         * value-prof.c (check_counter): Ditto.
5285         (check_ic_target): Ditto.
5286         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5287         (vect_recog_widen_mult_pattern): Ditto.
5288         (vect_recog_widen_sum_pattern): Ditto.
5289         (vect_recog_over_widening_pattern): Ditto.
5290         (vect_recog_widen_shift_pattern): Ditto.
5291         (vect_recog_rotate_pattern): Ditto.
5292         (vect_recog_vector_vector_shift_pattern): Ditto.
5293         (vect_recog_divmod_pattern): Ditto.
5294         (vect_recog_mixed_size_cond_pattern): Ditto.
5295         (vect_recog_bool_pattern): Ditto.
5296         (vect_pattern_recog_1): Ditto.
5297         (vect_pattern_recog): Ditto.
5298         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5299         (vect_is_simple_iv_evolution): Ditto.
5300         (vect_analyze_scalar_cycles_1): Ditto.
5301         (vect_get_loop_niters): Ditto.
5302         (vect_analyze_loop_1): Ditto.
5303         (vect_analyze_loop_form): Ditto.
5304         (vect_analyze_loop_operations): Ditto.
5305         (vect_analyze_loop_2): Ditto.
5306         (vect_analyze_loop): Ditto.
5307         (report_vect_op): Ditto.
5308         (vect_is_slp_reduction): Ditto.
5309         (vect_is_simple_reduction_1): Ditto.
5310         (vect_get_known_peeling_cost): Ditto.
5311         (vect_estimate_min_profitable_iters): Ditto.
5312         (vect_model_reduction_cost): Ditto.
5313         (vect_model_induction_cost): Ditto.
5314         (get_initial_def_for_induction): Ditto.
5315         (vect_create_epilog_for_reduction): Ditto.
5316         (vectorizable_reduction): Ditto.
5317         (vectorizable_induction): Ditto.
5318         (vectorizable_live_operation): Ditto.
5319         (vect_loop_kill_debug_uses): Ditto.
5320         (vect_transform_loop): Ditto.
5321         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5322         (vect_stmt_relevant_p): Ditto.
5323         (process_use): Ditto.
5324         (vect_mark_stmts_to_be_vectorized): Ditto.
5325         (vect_model_simple_cost): Ditto.
5326         (vect_model_promotion_demotion_cost): Ditto.
5327         (vect_model_store_cost): Ditto.
5328         (vect_get_store_cost): Ditto.
5329         (vect_model_load_cost): Ditto.
5330         (vect_get_load_cost): Ditto.
5331         (vect_init_vector_1): Ditto.
5332         (vect_get_vec_def_for_operand): Ditto.
5333         (vect_finish_stmt_generation): Ditto.
5334         (vectorizable_call): Ditto.
5335         (vectorizable_conversion): Ditto.
5336         (vectorizable_assignment): Ditto.
5337         (vectorizable_shift): Ditto.
5338         (vectorizable_operation): Ditto.
5339         (vectorizable_store): Ditto.
5340         (vectorizable_load): Ditto.
5341         (vectorizable_condition): Ditto.
5342         (vect_analyze_stmt): Ditto.
5343         (vect_transform_stmt): Ditto.
5344         (vect_is_simple_use): Ditto.
5345         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5346         (vect_can_advance_ivs_p): Ditto.
5347         (vect_update_ivs_after_vectorizer): Ditto.
5348         (vect_do_peeling_for_loop_bound): Ditto.
5349         (vect_gen_niters_for_prolog_loop): Ditto.
5350         (vect_update_inits_of_drs): Ditto.
5351         (vect_create_cond_for_alias_checks): Ditto.
5352         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5353         (vect_build_slp_tree_1): Ditto.
5354         (vect_supported_load_permutation_p): Ditto.
5355         (vect_analyze_slp_instance): Ditto.
5356         (vect_analyze_slp): Ditto.
5357         (vect_make_slp_decision): Ditto.
5358         (vect_detect_hybrid_slp): Ditto.
5359         (vect_bb_vectorization_profitable_p): Ditto.
5360         (vect_slp_analyze_bb_1): Ditto.
5361         (vect_update_slp_costs_according_to_vf): Ditto.
5362         (vect_get_mask_element): Ditto.
5363         (vect_transform_slp_perm_load): Ditto.
5364         (vect_schedule_slp_instance): Ditto.
5365         (vect_schedule_slp): Ditto.
5366         (vect_slp_transform_bb): Ditto.
5367         * profile.c (read_profile_edge_counts): Ditto.
5368         (compute_branch_probabilities): Ditto.
5369         * coverage.c (get_coverage_counts): Ditto.
5371 2013-09-16  Diego Novillo  <dnovillo@google.com>
5373         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5375 2013-09-16  Jan Hubicka  <jh@suse.cz>
5377         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5378         abstract functions; for static functions check the presence of body.
5380 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5382         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5383         * config/aarch64/aarch64-simd.md
5384         (aarch64_mla_elt<mode>): New.
5385         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5386         (aarch64_mls_elt<mode>): Likewise.
5387         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5388         (aarch64_fma4_elt<mode>): Likewise.
5389         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5390         (aarch64_fma4_elt_to_128v2df): Likewise.
5391         (aarch64_fma4_elt_to_64df): Likewise.
5392         (fnma<mode>4): Likewise.
5393         (aarch64_fnma4_elt<mode>): Likewise.
5394         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5395         (aarch64_fnma4_elt_to_128v2df): Likewise.
5396         (aarch64_fnma4_elt_to_64df): Likewise.
5397         * config/aarch64/iterators.md (VDQSF): New.
5398         * config/aarch64/arm_neon.h
5399         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5400         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5402 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5404         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5405         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5406         (aarch64_mul3_elt_to_128df): Likewise.
5407         (aarch64_mul3_elt_to_64v2df): Likewise.
5408         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5409         (VMUL): New.
5410         (VMUL_CHANGE_NLANES) Likewise.
5411         (h_con): Likewise.
5412         (f): Likewise.
5413         * config/aarch64/arm_neon.h
5414         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5416 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5418         * config/aarch64/arm_neon.h
5419         (vcvtx_high_f32_f64): Fix parameters.
5421 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5422             Uros Bizjak  <ubizjak@gmail.com>
5424         * config/alpha.c: Include tree-ssanames.h.
5426 2013-09-16  Richard Biener  <rguenther@suse.de>
5428         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5429         (dot_rdg_1): Handle control_dd.
5430         (create_edge_for_control_dependence): New function.
5431         (create_rdg_edges): Add control dependences if asked for.
5432         (build_rdg): Likewise.
5433         (generate_loops_for_partition): If there are not necessary
5434         control stmts remove all their dependencies.
5435         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5436         (distribute_loop): Pass on control dependences.
5437         (tree_loop_distribution): Compute control dependences and remove
5438         restriction on number of loop nodes.
5440 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5442         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5443         for internal calls.
5445 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5447         * cse.c (try_const_anchors): Punt on CC modes.
5449 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5451         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5453 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5455         PR target/58382
5456         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5457         calls to word_mode.
5459 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5461         PR target/48094
5462         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5463         seen.
5464         (darwin_objc1_section): Likewise.
5465         (darwin_file_end): Emit Image Info section when required.
5467 2013-09-14  Jan Hubicka  <jh@suse.cz>
5469         * tree-into-ssa.c (gate_into_ssa): New.
5470         (pass_data_build_ssa): Use it.
5471         * cgraph.h (expand_thunk): Update prototype.
5472         * cgraphunit.c (analyze_function): Expand thunks early.
5473         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5474         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5475         set lowered flag; do not add new function.
5476         (assemble_thunks_and_aliases): Update.
5477         * tree-ssa.c (gate_init_datastructures): New gate.
5478         (pass_data_init_datastructures): Use it.
5480 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5482         PR target/58269
5483         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5484         xmm register set described in the psABI.
5486 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5488         * dwarf2out.c (should_emit_struct_debug): Add check
5489         for type_decl variable is not NULL.
5491 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5493         * config.gcc: Use new winnt-c.c target hooks
5494         * config/t-winnt: New file
5495         * config/winnt-c.c: New file
5496         * doc/tm.texi.in: Document new hook
5497         * doc/tm.texi: Regenerated
5499 2013-09-13  Jan Hubicka  <jh@suse.cz>
5501         PR middle-end/58094
5502         * ipa-inline.c (check_callers): New function.
5503         (check_caller_edge): Remove.
5504         (want_inline_function_to_all_callers_p): Also permit alises that are
5505         called dirrectly.
5506         (inline_to_all_callers): Terminate the walk when devirtualization
5507         introduce new calls.
5509 2013-09-13  Jan Hubicka  <jh@suse.cz>
5511         * ipa-inline-analysis.c (struct growth_data): Add node.
5512         (do_estimate_growth_1): Fix detection of recursion.
5514 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5516         PR tree-optimization/58392
5517         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5518         to avoid shadowing of outer loop variable.  If
5519         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5520         replace_by_duplicate_decl simduid of loops that have it set and
5521         set dest_cfun->has_simduid_loops and/or
5522         dest_cfun->has_force_vect_loops.
5523         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5524         instead of maybe_lookup_decl.
5525         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5526         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5527         call.  Copy over force_vect and copy and remap simduid.  Set
5528         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5529         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5530         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5531         caller.  Don't set cfun->has_simduid_loops and/or
5532         cfun->has_force_vect_loops here.
5533         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5534         caller.
5535         (expand_call_inline, tree_function_versioning): Adjust copy_body
5536         callers.
5538 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5540         PR bootstrap/58388
5541         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5542         the assert if the edge was a speculative one.
5544 2013-09-13  Richard Biener  <rguenther@suse.de>
5546         * tree-data-ref.h (known_dependences_p): Move here ...
5547         * tree-loop-distribution.c (known_dependences_p): ... from here.
5548         (dump_rdg_component, debug_rdg_component): Remove.
5549         (dump_rdg): Adjust.
5550         (generate_loops_for_partition): Use gimple_uid instead of
5551         relying on matching stmt visit order.
5552         (rdg_build_partitions): Take starting stmt vector.
5553         (ldist_gen): Merge into ...
5554         (distribute_loop): ... this function.  Do not compute starting
5555         vertices vector.
5556         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5558 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5560         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5561         Set type attribute correctly. Set predicable_short_it attribute.
5562         (cmpsi_shiftsi): Remove %? from output template.
5564 2013-09-13  Richard Biener  <rguenther@suse.de>
5566         * tree-loop-distribution.c (struct rdg_component,
5567         rdg_defs_used_in_other_loops_p, free_rdg_components,
5568         rdg_build_components): Remove.
5569         (stmts_from_loop): Do not record virtual PHIs.
5570         (generate_loops_for_partition): Skip virtual PHIs.
5571         (build_rdg_partition_for_component): Rename to ...
5572         (build_rdg_partition_for_vertex): ... this and adjust.
5573         (rdg_build_partitions): Take a vector of starting vertices
5574         instead of components.  Remove unnecessary leftover handling.
5575         (ldist_gen): Do not build components or record other stores.
5576         (distribute_loop): Do not distribute loops containing stmts
5577         with side-effects.
5579 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5581         PR target/58314
5582         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5584 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5586         * config.gcc: Separate cases for mingw and cygwin targets,
5587         and add 64-bit cygwin target case.
5589         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5590         dll-export inline-functions.
5591         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5594 2013-09-13  Jeff Law  <law@redhat.com>
5596         PR middle-end/58387
5597         Revert:
5598         2013-09-06  Jeff Law  <law@redhat.com>
5600         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5601         edge implied equivalences into successor phis.
5603 2013-09-12  DJ Delorie  <dj@redhat.com>
5605         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5606         separators.
5608 2013-09-12  Brooks Moses  <bmoses@google.com>
5610         PR driver/42955
5611         * Makefile.in: Do not install driver binaries in $(target)/bin.
5613 2013-09-12  DJ Delorie  <dj@redhat.com>
5615         * config/rl78/rl78.opt (mrelax): New.
5616         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5617         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5619         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5620         between SP and FP.
5621         (rl78_expand_epilogue): Likewise.
5623 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5625         PR middle-end/58335
5626         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5627         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5628         pointer elimination with using remove_reg_equal_offset_note.
5630 2013-09-12  DJ Delorie  <dj@redhat.com>
5632         * config/msp430/: New port.
5633         * config.gcc (msp430): Added.
5634         * doc/invoke.texi: Document MSP430 options.
5635         * doc/install.texi: Document msp430-elf
5636         * doc/md.texi: Document msp430-elf
5637         * doc/contrib.texi: Document msp430-elf
5639         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5640         PSImode.
5642 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5644         PR ipa/58389
5645         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5646         reference descriptor is NULL.
5647         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5648         edge to NULL.
5650 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5652         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5653         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5654         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5655         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5656         make_temp_ssa_name): move to tree-ssanames.h
5657         * tree-ssa-alias.h: Move prototype.
5658         * tree-ssa.h: Include tree-ssanames.h.
5659         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5660         * tree-ssanames.h: New.  Move items from tree-flow*.h
5661         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5663 2013-09-12  Richard Biener  <rguenther@suse.de>
5665         PR tree-optimization/58404
5666         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5667         propagate non-invariant addresses into dereferences wrapped
5668         in component references.
5670 2013-09-12  Richard Biener  <rguenther@suse.de>
5672         PR tree-optimization/58402
5673         * passes.def: Move pass_late_warn_uninitialized later.
5675 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5677         * tree-ssa.h: New.  Move content from tree-flow.h and
5678         tree-flow-inline.h.
5679         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5680         Move prototypes belonging to tree-ssa.c.
5681         * tree-flow-inline.h (redirect_edge_var_map_def,
5682         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5683         tree-ssa.h.
5684         * gimple.h: Adjust prototypes.
5685         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5686         to...
5687         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5688         * tree.h: Move prototype to tree-ssa.h.
5689         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5690         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5691         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5692         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5693         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5694         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5695         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5696         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5697         graphite-blocking.c, graphite-clast-to-gimple.c,
5698         graphite-dependences.c, graphite-interchange.c,
5699         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5700         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5701         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5702         ipa-reference.c, ipa-split.c, ipa-utils.c,
5703         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5704         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5705         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5706         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5707         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5708         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5709         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5710         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5711         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5712         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5713         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5714         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5715         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5716         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5717         value-prof.c, var-tracking.c,
5718         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5720 2013-09-12  Richard Biener  <rguenther@suse.de>
5722         PR tree-optimization/58396
5723         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5724         (build_rdg): Take a loop-nest parameter, fix memleaks.
5725         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5727 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5729         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5730         for SLM.
5732 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5734         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5735         The FMA instruction names should have a 'v' prefix.
5737 2013-09-12  Richard Biener  <rguenther@suse.de>
5739         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5740         (dot_rdg): Use popen instead of system in optional code.
5741         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5742         (already_processed_vertex_p): Adjust.
5743         (has_anti_or_output_dependence, predecessor_has_mem_write,
5744         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5745         rdg_flag_uses): Remove.
5746         (rdg_flag_vertex): Simplify.
5747         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5748         remove recursion.
5749         (build_rdg_partition_for_component): Process the first vertex
5750         of a component only.
5751         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5753 2013-09-12  Alan Modra  <amodra@gmail.com>
5755         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5757 2013-09-11  DJ Delorie  <dj@redhat.com>
5758             Nick Clifton  <nickc@redhat.com>
5760         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5761         (rl78_stack_based_mem): New.
5762         * config/rl78/constraints.md (Iv08): New.
5763         (Iv16): New.
5764         (Iv24): New.
5765         (Is09): New.
5766         (Is17): New.
5767         (Is25): New.
5768         (ISsi): New.
5769         (IShi): New.
5770         (ISqi): New.
5771         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5772         (movhi): Likewise.
5773         (movsi): Change from expand to insn-and-split.
5774         (ashrsi3): Clobber AX.
5775         (lshrsi3): New.
5776         (ashlsi3): New.
5777         (cbranchsi4): New.
5778         * config/rl78/rl78.md (CC_REG): Fix.
5779         (addsi3): Allow memory and immediate operands.
5780         (addsi3_internal): Split into...
5781         (addsi3_internal_virt): ...new, and ...
5782         (addsi3_internal_real): ...new.
5783         (subsi): New.
5784         (subsi3_internal_virt): New.
5785         (subsi3_internal_real): New.
5786         (mulsi3): Add memory operand.
5787         (mulsi3_rl78): Likewise.
5788         (mulsi3_g13): Likewise.
5789         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5790         (cbranchqi4_real): Add more constraint options.
5791         (cbranchhi4_real): Expand pattern.
5792         (cbranchhi4_real_signed): New.
5793         (cbranchhi4_real_inverted): New.
5794         (cbranchsi4_real_lt): New.
5795         (cbranchsi4_real_ge): New.
5796         (cbranchsi4_real_signed): New.
5797         (cbranchsi4_real): New.
5798         (peephole2): New.
5799         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5800         constant shifts.
5801         (lshrsi3_virt): Likewise.
5802         (ashlsi3_virt): Likewise.
5803         (cbranchqi4_virt_signed): New.
5804         (cbranchhi4_virt_signed): New.
5805         (cbranchsi4_virt): New.
5806         * config/rl78/rl78.c: Whitespace fixes throughout.
5807         (move_elim_pass): New.
5808         (pass_data_rl78_move_elim): New.
5809         (pass_rl78_move_elim): New.
5810         (make_pass_rl78_move_elim): New.
5811         (rl78_devirt_info): Run devirt earlier.
5812         (rl78_move_elim_info): New.
5813         (rl78_asm_file_start): Register it.
5814         (rl78_split_movsi): New.
5815         (rl78_as_legitimate_address): Allow virtual base registers when
5816         appropriate.
5817         (rl78_addr_space_convert): Remove spurious debug stuff.
5818         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5819         (rl78_print_operand): More cases for not printing '#'.
5820         (rl78_expand_compare): Remove most of the logic.
5821         (content_memory): New.
5822         (clear_content_memory): New.
5823         (get_content_index): New.
5824         (get_content_name): New.
5825         (display_content_memory): New.
5826         (update_content): New.
5827         (record_content): New.
5828         (already_contains): New.
5829         (insn_ok_now): Re-recog insns with virtual registers.
5830         (add_postponed_content_update): New.
5831         (process_postponed_content_update): New.
5832         (gen_and_emit_move): New.
5833         (transcode_memory_rtx): Record new location content.
5834         Use gen_and_emit_move.
5835         (force_into_acc): New.
5836         (move_to_acc): Use gen_and_emit_move.
5837         (move_from_acc): Likewise.
5838         (move_acc_to_reg): Likewise.
5839         (move_to_x): Likewise.
5840         (move_to_hl): Likewise.
5841         (move_to_de): Likewise.
5842         (rl78_alloc_physical_registers_op1): Record location content.
5843         (has_constraint): New.
5844         (rl78_alloc_physical_registers_op2): Record location content.
5845         Optimize use of HL.
5846         (rl78_alloc_physical_registers_ro1): Likewise.
5847         (rl78_alloc_physical_registers_cmp): Likewise.
5848         (rl78_alloc_physical_registers_umul): Likewise.
5849         (rl78_alloc_address_registers_macax): New.
5850         (rl78_alloc_physical_registers): Initialize and set location
5851         content memory as needed.
5852         (rl78_reorg): Make sure split2 is called.
5853         (rl78_rtx_costs): New.
5855 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5857         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5858         for (not (neg ...)) and (neg (not ...)) cases.
5860 2013-09-11  Richard Biener  <rguenther@suse.de>
5862         PR middle-end/58377
5863         * passes.def: Split critical edges before late uninit warning passes.
5864         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5866 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5868         PR tree-optimization/58385
5869         * fold-const.c (build_range_check): If both low and high are NULL,
5870         use omit_one_operand_loc to preserve exp side-effects.
5872 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5874         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5876 2013-09-11  Richard Biener  <rguenther@suse.de>
5878         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5879         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5880         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5881         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5882         stmts_from_loop, known_dependences_p, build_empty_rdg,
5883         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5884         * tree-loop-distribution.c: ... here.
5885         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5886         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5887         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5888         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5889         * tree-loop-distribution.c: ... here.
5890         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5891         (struct partition_s): Add loops member.
5892         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5893         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5894         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5896 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5897             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5898             Sergey Lega  <sergey.s.lega@intel.com>
5899             Anna Tikhonova  <anna.tikhonova@intel.com>
5900             Ilya Tocar  <ilya.tocar@intel.com>
5901             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5902             Ilya Verbin  <ilya.verbin@intel.com>
5903             Kirill Yukhin  <kirill.yukhin@intel.com>
5904             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5906         * config/i386/constraints.md (k): New.
5907         (Yk): Ditto.
5908         * config/i386/i386.c (const regclass_map): Add new mask registers.
5909         (dbx_register_map): Ditto.
5910         (dbx64_register_map): Ditto.
5911         (svr4_dbx_register_map): Ditto.
5912         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5913         disabled.
5914         (ix86_preferred_reload_class): Disable constants for mask registers.
5915         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5916         (ix86_hard_regno_mode_ok): Support new mask registers.
5917         (x86_order_regs_for_local_alloc): Ditto.
5918         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5919         (FIXED_REGISTERS): Add new mask registers.
5920         (CALL_USED_REGISTERS): Ditto.
5921         (REG_ALLOC_ORDER): Ditto.
5922         (VALID_MASK_REG_MODE): New.
5923         (FIRST_MASK_REG): Ditto.
5924         (LAST_MASK_REG): Ditto.
5925         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5926         (MAYBE_MASK_CLASS_P): New.
5927         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5928         (REG_CLASS_CONTENTS): Ditto.
5929         (MASK_REGNO_P): New.
5930         (ANY_MASK_REG_P): Ditto.
5931         (HI_REGISTER_NAMES): Add new mask registers.
5932         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5933         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5934         mask registers.
5935         (attribute "type"): Add mskmov, msklog.
5936         (attribute "length_immediate"): Support them.
5937         (attribute "memory"): Ditto.
5938         (attribute "prefix_0f"): Ditto.
5939         (*movhi_internal): Support new mask registers.
5940         (*movqi_internal): Ditto.
5941         (define_split): Split out clobber pattern is a logic
5942         insn on mask registers.
5943         (*k<logic><mode>): New.
5944         (*andhi_1): Extend to support mask regs.
5945         (*andqi_1): Extend to support mask regs.
5946         (kandn<mode>): New.
5947         (define_split): Split and-not to and and not if operands
5948         are not mask regs.
5949         (*<code><mode>_1): Separate HI mode to new pattern...
5950         (*<code>hi_1): This.
5951         (*<code>qi_1): Extend to support mask regs.
5952         (kxnor<mode>): New.
5953         (kortestzhi): Ditto.
5954         (kortestchi): Ditto.
5955         (kunpckhi): Ditto.
5956         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5957         (*one_cmplhi2_1): ...Here, now with mask registers support.
5958         (*one_cmplqi2_1): Support new mask registers.
5959         (HI/QImode arithmetics splitter): Don't split if mask registers
5960         are used.
5961         (HI/QImode not splitter): Ditto.
5962         * config/i386/predicated.md (mask_reg_operand): New.
5963         (general_reg_operand): Ditto.
5965 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5967         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5968         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5970 2013-09-10  Jeff Law  <law@redhat.com>
5972         PR tree-optimization/58380
5973         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5974         of threading through a buried loop header.
5976         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5977         return value for single successor case.
5979 2013-09-10  Jan Hubicka  <jh@suse.cz>
5981         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5983 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5985         PR target/58361
5986         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5987         support conditional execution.
5988         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5990 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5992         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5993         constraint pass.
5994         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5995         Check destination too to check move insn.
5996         (undo_optional_reloads): Add check that the original peudo did not
5997         changed its allocation and the optional reload was inherited on last
5998         inheritance pass.  Break loop after deciding to keep optional reload.
5999         (lra_undo_inheritance): Add check that inherited pseudo still in
6000         memory.
6002 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
6004         * config/aarch64/aarch64.md (generic_sched): New.
6005         * config/aarch64/aarch64-generic.md (load): Make conditional
6006         on generic_sched attribute.
6007         (nonload): Likewise.
6009 2013-09-10  Jan Hubicka  <jh@suse.cz>
6011         * lto-cgraph.c: Include ipa-utils.h.
6012         (compute_ltrans_boundary): Also add possible targets into the boundary.
6014 2013-09-10  Jan Hubicka  <jh@suse.cz>
6016         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
6017         VAR_DECL of vtable rather than full expression.
6019 2013-09-10  Jan Hubicka  <jh@suse.cz>
6020             Paolo Carlini  <paolo.carlini@oracle.com>
6022         * cgraphunit.c (analyze_functions): Save input_location, set it
6023         to UNKNOWN_LOCATION and restore it at the end.
6025 2013-09-10  Martin Jambor  <mjambor@suse.cz>
6027         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
6028         represented by a thunk.
6030 2013-09-10  Jeff Law  <law@redhat.com>
6032         PR tree-optimization/58343
6033         * tree-ssa-threadupdate.c (thread_block): Identify and disable
6034         jump threading requests through loop headers buried in the middle
6035         of a jump threading path.
6037         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
6038         in return value/type.
6040 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
6042         PR rtl-optimization/58365
6043         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
6044         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
6045         it differs.
6047 2013-09-10  Richard Biener  <rguenther@suse.de>
6049         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
6050         * tree-data-ref.c (create_rdg_vertices): Collect all data
6051         references, signal failure to the caller, use data-ref API.
6052         (build_rdg): Compute data references only once.  Maintain lifetime
6053         of data references and data dependences from within RDG.
6054         (free_rdg): Free dependence relations.
6055         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
6056         inventing extra dependences.
6057         (distribute_loop): Update for RDG API changes.
6059 2013-09-10  Kai Tietz  <ktietz@redhat.com>
6061         * doc/invoke.texi (fms-extensions): Document changed
6062         behavior for ms-abi targets.
6063         * config/i386/i386.c (ix86_option_override_internal):
6064         Set default value of option -fms-extension for ms-abi targets.
6066 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6068         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
6070 2013-09-10  Alan Modra  <amodra@gmail.com>
6072         PR target/58330
6073         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
6075 2013-09-10  Alan Modra  <amodra@gmail.com>
6077         * config/rs6000/predicates.md (add_cint_operand): New.
6078         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
6079         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
6080         using add_cint_operand.
6081         (largetoc_high_plus_aix): Likewise.
6083 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6085         PR tree-optimization/58364
6086         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6087         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6088         the current range can't be an unconditional true or false.
6090 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6092         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6094 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6096         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6098 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6100         PR c++/43452
6101         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6103 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6105         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6106         NO_REGS for immediate that can't be moved directly into FP_REGS.
6108 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6110         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6111         comparison with negated operand.
6112         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6113         RTL form.
6115 2013-09-09  Richard Biener  <rguenther@suse.de>
6117         PR middle-end/58326
6118         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6119         of a subloop record all its block as affecting loop-closed SSA form.
6121 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6123         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6124         of an rtx/bitpos pair.
6125         (store_fixed_bit_field): Update accordingly.
6127 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6129         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6130         GEN_INT.
6131         * builtins.c (expand_errno_check): Likewise.
6132         * dwarf2cfi.c (init_return_column_size): Likewise.
6133         * except.c (sjlj_mark_call_sites): Likewise.
6134         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6135         * lra-constraints.c (emit_inc): Likewise.
6136         * ree.c (combine_set_extension): Likewise.
6137         * regmove.c (fixup_match_2): Likewise.
6138         * reload1.c (inc_for_reload): Likewise.
6140 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6142         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6143         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6144         Use gen_int_mode with the mode of the associated simplify_* call.
6145         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6146         * expmed.c (expand_shift_1): Likewise.
6147         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6148         * loop-iv.c (iv_number_of_iterations): Likewise.
6149         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6150         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6152 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6154         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6155         of the associated expand_* call.
6156         (asan_emit_stack_protection): Likewise.
6157         * builtins.c (round_trampoline_addr): Likewise.
6158         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6159         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6160         (emit_store_flag): Likewise.
6161         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6162         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6163         Likewise.
6164         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6165         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6166         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6167         * modulo-sched.c (generate_prolog_epilog): Likewise.
6168         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6169         (expand_ctz, expand_ffs, expand_unop): Likewise.
6171 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6173         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6174         of the associated gen_rtx_* call.
6175         * caller-save.c (init_caller_save): Likewise.
6176         * combine.c (find_split_point, make_extraction): Likewise.
6177         (make_compound_operation): Likewise.
6178         * dwarf2out.c (mem_loc_descriptor): Likewise.
6179         * explow.c (plus_constant, probe_stack_range): Likewise.
6180         * expmed.c (expand_mult_const): Likewise.
6181         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6182         * reload1.c (init_reload): Likewise.
6183         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6184         * var-tracking.c (adjust_mems): Likewise.
6185         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6186         rather than gen_rtx_fmt_ee.
6188 2013-09-09  Jan Hubicka  <jh@suse.cz>
6190         PR middle-end/58294
6191         * value-prof.c (gimple_ic): Copy also abnormal edges.
6193 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6195         * asan.c (asan_shadow_cst): Use gen_int_mode.
6197 2013-09-08  Jan Hubicka  <jh@suse.cz>
6199         * ipa-profile.c: Add toplevel comment.
6200         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6201         (contains_hot_call_p): New function.
6202         (ipa_propagate_frequency): Set frequencies based on counts when
6203         profile is read.
6204         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6205         profile; do not tamper with profile after inlining if it is read.
6207 2013-09-08  Jan Hubicka  <jh@suse.cz>
6209         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6210         speculative edges.
6212 2013-09-08  Jan Hubicka  <jh@suse.cz>
6214         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6215         summary generation.
6217 2013-09-08  Jeff Law  <law@redhat.com>
6219         PR bootstrap/58340
6220         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6221         of 'found'.
6223 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6225         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6227 2013-09-08  Jan Hubicka  <jh@suse.cz>
6229         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6231 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6233         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6234         for non-debug insns.
6235         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6236         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6237         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6238         accordingly.
6240 2013-09-08  Jan Hubicka  <jh@suse.cz>
6242         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6243         targets and devirtualize to BUILT_IN_UNREACHABLE.
6244         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6245         * ipa.c (walk_polymorphic_call_targets): New function.
6246         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6247         functions; use the new timevar.
6248         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6249         was removed from the program.
6250         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6251         not consider it in the walk when its vtable is dead.
6252         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6253         record_binfo.
6254         (devirt_variable_node_removal_hook): New function.
6255         (possible_polymorphic_call_targets): Also register
6256         devirt_variable_node_removal_hook.
6257         (ipa_devirt): Do not do non-speculative devirtualization.
6258         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6260 2013-09-08  Jan Hubicka  <jh@suse.cz>
6262         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6263         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6264         varpool_remove_variable_insertion_hook): Declare.
6265         * varpool.c (varpool_node_hook_list): New structure.
6266         (first_varpool_node_removal_hook,
6267         first_varpool_variable_insertion_hook): New variables.
6268         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6269         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6270         varpool_remove_variable_insertion_hook,
6271         varpool_call_variable_insertion_hooks): New functions.
6272         (varpool_remove_node): Use it.
6274 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6276         PR c++/54941
6277         * diagnostic.c (diagnostic_build_prefix): When s.file is
6278         "<built-in>" don't output line and column numbers.
6280 2013-09-06  Jan Hubicka  <jh@suse.cz>
6282         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6283         * lto-streamer-out.c: Stream thunks, too.
6284         * lto-streamer-in.c (input_function): Pop cfun here
6285         (lto_read_body): Instead of here.
6287 2013-09-06  Caroline Tice  <cmtice@google.com>
6289         * doc/install.texi: Add documentation for the --enable-vtable-verify
6290         and the --disable-libvtv configure options.
6292 2013-09-06  Jeff Law  <law@redhat.com>
6294         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6295         edge implied equivalences into successor phis.
6297 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6299         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6301 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6303         * resource.c (mark_target_live_regs): Compute resources taking
6304         into account if a call is predicated or not.
6306 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6308         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6309         the compiler in the function names.
6311 2013-09-06  Jan Hubicka  <jh@suse.cz>
6313         PR middle-end/58094
6314         * ipa-inline.c (has_caller_p): New function.
6315         (want_inline_function_to_all_callers_p): Use it.
6316         (sum_callers, inline_to_all_callers): Break out from ...
6317         (ipa_inline): ... here.
6319 2013-09-06  Jan Hubicka  <jh@suse.cz>
6321         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6322         only when AVX is enabled.
6324 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6326         * config/aarch64/aarch64.md
6327         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6328         is fpsimd_<load/store>2.
6329         (load_pair<mode>): Likewise.
6330         (store_pair<mode>): Likewise.
6332 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6334         * config/arm/types.md (type): Add "mrs" type.
6335         * config/aarch64/aarch64.md
6336         (aarch64_load_tp_hard): Make type "mrs".
6337         * config/arm/arm.md
6338         (load_tp_hard): Make type "mrs".
6339         * config/arm/cortex-a15.md: Update with new attributes.
6340         * config/arm/cortex-a5.md: Update with new attributes.
6341         * config/arm/cortex-a53.md: Update with new attributes.
6342         * config/arm/cortex-a7.md: Update with new attributes.
6343         * config/arm/cortex-a8.md: Update with new attributes.
6344         * config/arm/cortex-a9.md: Update with new attributes.
6345         * config/arm/cortex-m4.md: Update with new attributes.
6346         * config/arm/cortex-r4.md: Update with new attributes.
6347         * config/arm/fa526.md: Update with new attributes.
6348         * config/arm/fa606te.md: Update with new attributes.
6349         * config/arm/fa626te.md: Update with new attributes.
6350         * config/arm/fa726te.md: Update with new attributes.
6352 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6354         * config/aarch64/aarch64.md
6355         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6356         (*movtf_aarch64): Likewise.
6357         * config/arm/arm.md
6358         (thumb1_movdi_insn): Use "multiple" for type where more than one
6359         instruction is used for a move.
6360         (*arm32_movhf): Likewise.
6361         (*thumb_movdf_insn): Likewise.
6363 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6365         * config/arm/types.md (type): Rename fcpys to fmov.
6366         * config/arm/vfp.md
6367         (*arm_movsi_vfp): Rename type fcpys as fmov.
6368         (*thumb2_movsi_vfp): Likewise
6369         (*movhf_vfp_neon): Likewise
6370         (*movhf_vfp): Likewise
6371         (*movsf_vfp): Likewise
6372         (*thumb2_movsf_vfp): Likewise
6373         (*movsfcc_vfp): Likewise
6374         (*thumb2_movsfcc_vfp): Likewise
6375         * config/aarch64/aarch64-simd.md
6376         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6377         * config/aarch64/aarch64.md
6378         (*movsi_aarch64): Replace type mov_reg with fmovs.
6379         (*movdi_aarch64): Likewise
6380         (*movsf_aarch64): Likewise
6381         (*movdf_aarch64): Likewise
6382         * config/arm/arm.c
6383         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6384         * config/arm/iwmmxt.md
6385         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6386         * config/arm/arm1020e.md: Update with new attributes.
6387         * config/arm/cortex-a15-neon.md: Update with new attributes.
6388         * config/arm/cortex-a5.md: Update with new attributes.
6389         * config/arm/cortex-a53.md: Update with new attributes.
6390         * config/arm/cortex-a7.md: Update with new attributes.
6391         * config/arm/cortex-a8-neon.md: Update with new attributes.
6392         * config/arm/cortex-a9.md: Update with new attributes.
6393         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6394         * config/arm/cortex-r4f.md: Update with new attributes.
6395         * config/arm/marvell-pj4.md: Update with new attributes.
6396         * config/arm/vfp11.md: Update with new attributes.
6398 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6400         * config/aarch64/aarch64.md
6401         (*madd<mode>): Fix type attribute.
6402         (*maddsi_uxtw): Likewise.
6403         (*msub<mode>): Likewise.
6404         (*msubsi_uxtw): Likewise.
6405         (<su_optab>maddsidi4): Likewise.
6406         (<su_optab>msubsidi4): Likewise.
6408 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6410         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6411         * config/arm/arm.md (core_cycles): Remove fdiv.
6412         * config/arm/vfp.md:
6413         (*sqrtsf2_vfp): Update for attribute changes.
6414         (*sqrtdf2_vfp): Likewise.
6415         * config/aarch64/aarch64.md:
6416         (sqrt<mode>2): Update for attribute changes.
6417         * config/arm/arm1020e.md: Update with new attributes.
6418         * config/arm/cortex-a15-neon.md: Update with new attributes.
6419         * config/arm/cortex-a5.md: Update with new attributes.
6420         * config/arm/cortex-a53.md: Update with new attributes.
6421         * config/arm/cortex-a7.md: Update with new attributes.
6422         * config/arm/cortex-a8-neon.md: Update with new attributes.
6423         * config/arm/cortex-a9.md: Update with new attributes.
6424         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6425         * config/arm/cortex-r4f.md: Update with new attributes.
6426         * config/arm/marvell-pj4.md: Update with new attributes.
6427         * config/arm/vfp11.md: Update with new attributes.
6429 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6431         * config/arm/types.md
6432         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6433         * config/aarch64/aarch64.md
6434         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6435         new attributes.
6436         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6437         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6438         (float<GPI:mode><GPF:mode>2): Likewise.
6439         * config/arm/vfp.md
6440         (*truncsisf2_vfp): Update with new attributes.
6441         (*truncsidf2_vfp): Likewise.
6442         (fixuns_truncsfsi2): Likewise.
6443         (fixuns_truncdfsi2): Likewise.
6444         (*floatsisf2_vfp): Likewise.
6445         (*floatsidf2_vfp): Likewise.
6446         (floatunssisf2): Likewise.
6447         (floatunssidf2): Likewise.
6448         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6449         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6450         * config/arm/arm1020e.md: Update with new attributes.
6451         * config/arm/cortex-a15-neon.md: Update with new attributes.
6452         * config/arm/cortex-a5.md: Update with new attributes.
6453         * config/arm/cortex-a53.md: Update with new attributes.
6454         * config/arm/cortex-a7.md: Update with new attributes.
6455         * config/arm/cortex-a8-neon.md: Update with new attributes.
6456         * config/arm/cortex-a9.md: Update with new attributes.
6457         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6458         * config/arm/cortex-r4f.md: Update with new attributes.
6459         * config/arm/marvell-pj4.md: Update with new attributes.
6460         * config/arm/vfp11.md: Update with new attributes.
6462 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6464         * config/aarch64/arm_neon.h
6465         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6466         (vqtbx<1,2,3,4><q>_s8): Likewise.
6468 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6470         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6471         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6472         (add<mode>3): Add type attribute.
6473         (add<mode>3): Likewise.
6474         (usadd<mode>3): Likewise.
6475         (ssadd<mode>3): Likewise.
6476         (sub<mode>3): Likewise.
6477         (sub<mode>3): Likewise.
6478         (ussub<mode>3): Likewise.
6479         (sssub<mode>3): Likewise.
6480         (ssmulsa3): Likewise.
6481         (usmulusa3): Likewise.
6482         (arm_usatsihi): Likewise.
6483         * config/arm/vfp.md
6484         (*movdi_vfp): Add types for all instructions.
6485         (*movdi_vfp_cortexa8): Likewise.
6486         (*movhf_vfp_neon): Likewise.
6487         (*movhf_vfp): Likewise.
6488         (*movdf_vfp): Likewise.
6489         (*thumb2_movdf_vfp): Likewise.
6490         (*thumb2_movdfcc_vfp): Likewise.
6491         * config/arm/arm.md: Add type attribute to all insn patterns.
6492         (*thumb1_adddi3): Add type attribute.
6493         (*arm_adddi3): Likewise.
6494         (*adddi_sesidi_di): Likewise.
6495         (*adddi_zesidi_di): Likewise.
6496         (*thumb1_addsi3): Likewise.
6497         (addsi3_compare0): Likewise.
6498         (*addsi3_compare0_scratch): Likewise.
6499         (*compare_negsi_si): Likewise.
6500         (cmpsi2_addneg): Likewise.
6501         (*addsi3_carryin_<optab>): Likewise.
6502         (*addsi3_carryin_alt2_<optab>): Likewise.
6503         (*addsi3_carryin_clobercc_<optab>): Likewise.
6504         (*subsi3_carryin): Likewise.
6505         (*subsi3_carryin_const): Likewise.
6506         (*subsi3_carryin_compare): Likewise.
6507         (*subsi3_carryin_compare_const): Likewise.
6508         (*arm_subdi3): Likewise.
6509         (*thumb_subdi3): Likewise.
6510         (*subdi_di_zesidi): Likewise.
6511         (*subdi_di_sesidi): Likewise.
6512         (*subdi_zesidi_di): Likewise.
6513         (*subdi_sesidi_di): Likewise.
6514         (*subdi_zesidi_ze): Likewise.
6515         (thumb1_subsi3_insn): Likewise.
6516         (*arm_subsi3_insn): Likewise.
6517         (*anddi3_insn): Likewise.
6518         (*anddi_zesidi_di): Likewise.
6519         (*anddi_sesdi_di): Likewise.
6520         (*ne_zeroextracts): Likewise.
6521         (*ne_zeroextracts): Likewise.
6522         (*ite_ne_zeroextr): Likewise.
6523         (*ite_ne_zeroextr): Likewise.
6524         (*anddi_notdi_di): Likewise.
6525         (*anddi_notzesidi): Likewise.
6526         (*anddi_notsesidi): Likewise.
6527         (andsi_notsi_si): Likewise.
6528         (thumb1_bicsi3): Likewise.
6529         (*iordi3_insn): Likewise.
6530         (*iordi_zesidi_di): Likewise.
6531         (*iordi_sesidi_di): Likewise.
6532         (*thumb1_iorsi3_insn): Likewise.
6533         (*xordi3_insn): Likewise.
6534         (*xordi_zesidi_di): Likewise.
6535         (*xordi_sesidi_di): Likewise.
6536         (*arm_xorsi3): Likewise.
6537         (*andsi_iorsi3_no): Likewise.
6538         (*smax_0): Likewise.
6539         (*smax_m1): Likewise.
6540         (*arm_smax_insn): Likewise.
6541         (*smin_0): Likewise.
6542         (*arm_smin_insn): Likewise.
6543         (*arm_umaxsi3): Likewise.
6544         (*arm_uminsi3): Likewise.
6545         (*minmax_arithsi): Likewise.
6546         (*minmax_arithsi_): Likewise.
6547         (*satsi_<SAT:code>): Likewise.
6548         (arm_ashldi3_1bit): Likewise.
6549         (arm_ashrdi3_1bit): Likewise.
6550         (arm_lshrdi3_1bit): Likewise.
6551         (*arm_negdi2): Likewise.
6552         (*thumb1_negdi2): Likewise.
6553         (*arm_negsi2): Likewise.
6554         (*thumb1_negsi2): Likewise.
6555         (*negdi_extendsid): Likewise.
6556         (*negdi_zero_extend): Likewise.
6557         (*arm_abssi2): Likewise.
6558         (*thumb1_abssi2): Likewise.
6559         (*arm_neg_abssi2): Likewise.
6560         (*thumb1_neg_abss): Likewise.
6561         (one_cmpldi2): Likewise.
6562         (extend<mode>di2): Likewise.
6563         (*compareqi_eq0): Likewise.
6564         (*arm_extendhisi2addsi): Likewise.
6565         (*arm_movdi): Likewise.
6566         (*thumb1_movdi_insn): Likewise.
6567         (*arm_movt): Likewise.
6568         (*thumb1_movsi_insn): Likewise.
6569         (pic_add_dot_plus_four): Likewise.
6570         (pic_add_dot_plus_eight): Likewise.
6571         (tls_load_dot_plus_eight): Likewise.
6572         (*thumb1_movhi_insn): Likewise.
6573         (*thumb1_movsf_insn): Likewise.
6574         (*movdf_soft_insn): Likewise.
6575         (*thumb_movdf_insn): Likewise.
6576         (cbranchsi4_insn): Likewise.
6577         (cbranchsi4_scratch): Likewise.
6578         (*negated_cbranchsi4): Likewise.
6579         (*tbit_cbranch): Likewise.
6580         (*tlobits_cbranch): Likewise.
6581         (*tstsi3_cbranch): Likewise.
6582         (*cbranchne_decr1): Likewise.
6583         (*addsi3_cbranch): Likewise.
6584         (*addsi3_cbranch_scratch): Likewise.
6585         (*arm_cmpdi_insn): Likewise.
6586         (*arm_cmpdi_unsig): Likewise.
6587         (*arm_cmpdi_zero): Likewise.
6588         (*thumb_cmpdi_zero): Likewise.
6589         (*deleted_compare): Likewise.
6590         (*mov_scc): Likewise.
6591         (*mov_negscc): Likewise.
6592         (*mov_notscc): Likewise.
6593         (*cstoresi_eq0_thumb1_insn): Likewise.
6594         (cstoresi_nltu_thumb1): Likewise.
6595         (cstoresi_ltu_thu): Likewise.
6596         (thumb1_addsi3_addgeu): Likewise.
6597         (*arm_jump): Likewise.
6598         (*thumb_jump): Likewise.
6599         (*check_arch2): Likewise.
6600         (arm_casesi_internal): Likewise.
6601         (thumb1_casesi_dispatch): Likewise.
6602         (*arm_indirect_jump): Likewise.
6603         (*thumb1_indirect_jump): Likewise.
6604         (nop): Likewise.
6605         (*and_scc): Likewise.
6606         (*ior_scc): Likewise.
6607         (*compare_scc): Likewise.
6608         (*cond_move): Likewise.
6609         (*cond_arith): Likewise.
6610         (*cond_sub): Likewise.
6611         (*cmp_ite0): Likewise.
6612         (*cmp_ite1): Likewise.
6613         (*cmp_and): Likewise.
6614         (*cmp_ior): Likewise.
6615         (*ior_scc_scc): Likewise.
6616         (*ior_scc_scc_cmp): Likewise.
6617         (*and_scc_scc): Likewise.
6618         (*and_scc_scc_cmp): Likewise.
6619         (*and_scc_scc_nod): Likewise.
6620         (*negscc): Likewise.
6621         (movcond_addsi): Likewise.
6622         (movcond): Likewise.
6623         (*ifcompare_plus_move): Likewise.
6624         (*if_plus_move): Likewise.
6625         (*ifcompare_move_plus): Likewise.
6626         (*if_move_plus): Likewise.
6627         (*ifcompare_arith_arith): Likewise.
6628         (*if_arith_arith): Likewise.
6629         (*ifcompare_arith_move): Likewise.
6630         (*if_arith_move): Likewise.
6631         (*ifcompare_move_arith): Likewise.
6632         (*if_move_arith): Likewise.
6633         (*ifcompare_move_not): Likewise.
6634         (*if_move_not): Likewise.
6635         (*ifcompare_not_move): Likewise.
6636         (*if_not_move): Likewise.
6637         (*ifcompare_shift_move): Likewise.
6638         (*if_shift_move): Likewise.
6639         (*ifcompare_move_shift): Likewise.
6640         (*if_move_shift): Likewise.
6641         (*ifcompare_shift_shift): Likewise.
6642         (*ifcompare_not_arith): Likewise.
6643         (*ifcompare_arith_not): Likewise.
6644         (*if_arith_not): Likewise.
6645         (*ifcompare_neg_move): Likewise.
6646         (*if_neg_move): Likewise.
6647         (*ifcompare_move_neg): Likewise.
6648         (*if_move_neg): Likewise.
6649         (prologue_thumb1_interwork): Likewise.
6650         (*cond_move_not): Likewise.
6651         (*sign_extract_onebit): Likewise.
6652         (*not_signextract_onebit): Likewise.
6653         (stack_tie): Likewise.
6654         (align_4): Likewise.
6655         (align_8): Likewise.
6656         (consttable_end): Likewise.
6657         (consttable_1): Likewise.
6658         (consttable_2): Likewise.
6659         (consttable_4): Likewise.
6660         (consttable_8): Likewise.
6661         (consttable_16): Likewise.
6662         (*thumb1_tablejump): Likewise.
6663         (prefetch): Likewise.
6664         (force_register_use): Likewise.
6665         (thumb_eh_return): Likewise.
6666         (load_tp_hard): Likewise.
6667         (load_tp_soft): Likewise.
6668         (tlscall): Likewise.
6669         (*arm_movtas_ze): Likewise.
6670         (*arm_rev): Likewise.
6671         (*arm_revsh): Likewise.
6672         (*arm_rev16): Likewise.
6673         * config/arm/thumb2.md
6674         (*thumb2_smaxsi3): Likewise.
6675         (*thumb2_sminsi3): Likewise.
6676         (*thumb32_umaxsi3): Likewise.
6677         (*thumb2_uminsi3): Likewise.
6678         (*thumb2_negdi2): Likewise.
6679         (*thumb2_abssi2): Likewise.
6680         (*thumb2_neg_abss): Likewise.
6681         (*thumb2_movsi_insn): Likewise.
6682         (tls_load_dot_plus_four): Likewise.
6683         (*thumb2_movhi_insn): Likewise.
6684         (*thumb2_mov_scc): Likewise.
6685         (*thumb2_mov_negs): Likewise.
6686         (*thumb2_mov_negs): Likewise.
6687         (*thumb2_mov_nots): Likewise.
6688         (*thumb2_mov_nots): Likewise.
6689         (*thumb2_movsicc_): Likewise.
6690         (*thumb2_movsfcc_soft_insn): Likewise.
6691         (*thumb2_indirect_jump): Likewise.
6692         (*thumb2_and_scc): Likewise.
6693         (*thumb2_ior_scc): Likewise.
6694         (*thumb2_ior_scc_strict_it): Likewise.
6695         (*thumb2_cond_move): Likewise.
6696         (*thumb2_cond_arith): Likewise.
6697         (*thumb2_cond_ari): Likewise.
6698         (*thumb2_cond_sub): Likewise.
6699         (*thumb2_negscc): Likewise.
6700         (*thumb2_movcond): Likewise.
6701         (thumb2_casesi_internal): Likewise.
6702         (thumb2_casesi_internal_pic): Likewise.
6703         (*thumb2_alusi3_short): Likewise.
6704         (*thumb2_mov<mode>_shortim): Likewise.
6705         (*thumb2_addsi_short): Likewise.
6706         (*thumb2_subsi_short): Likewise.
6707         (thumb2_addsi3_compare0): Likewise.
6708         (*thumb2_cbz): Likewise.
6709         (*thumb2_cbnz): Likewise.
6710         (*thumb2_one_cmplsi2_short): Likewise.
6711         (*thumb2_negsi2_short): Likewise.
6712         (*orsi_notsi_si): Likewise.
6713         * config/arm/arm1020e.md: Update with new attributes.
6714         * config/arm/arm1026ejs.md: Update with new attributes.
6715         * config/arm/arm1136jfs.md: Update with new attributes.
6716         * config/arm/arm926ejs.md: Update with new attributes.
6717         * config/arm/cortex-a15.md: Update with new attributes.
6718         * config/arm/cortex-a5.md: Update with new attributes.
6719         * config/arm/cortex-a53.md: Update with new attributes.
6720         * config/arm/cortex-a7.md: Update with new attributes.
6721         * config/arm/cortex-a8.md: Update with new attributes.
6722         * config/arm/cortex-a9.md: Update with new attributes.
6723         * config/arm/cortex-m4.md: Update with new attributes.
6724         * config/arm/cortex-r4.md: Update with new attributes.
6725         * config/arm/fa526.md: Update with new attributes.
6726         * config/arm/fa606te.md: Update with new attributes.
6727         * config/arm/fa626te.md: Update with new attributes.
6728         * config/arm/fa726te.md: Update with new attributes.
6730 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6732         * config/aarch64/aarch64-simd.md
6733         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6734         <vwx> iterator to ensure correct register choice.
6735         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6736         (aarch64_sqdmull_n<mode>): Likewise.
6737         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6738         * config/aarch64/arm_neon.h
6739         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6740         (vml<as><q>_n_<su>16): Likewise.
6741         (vml<as>l_high_lane<q>_<su>16): Likewise.
6742         (vml<as>l_high_n_<su>16): Likewise.
6743         (vml<as>l_lane<q>_<su>16): Likewise.
6744         (vml<as>l_n_<su>16): Likewise.
6745         (vmul<q>_lane<q>_<su>16): Likewise.
6746         (vmul<q>_n_<su>16): Likewise.
6747         (vmull_lane<q>_<su>16): Likewise.
6748         (vmull_n_<su>16): Likewise.
6749         (vmull_high_lane<q>_<su>16): Likewise.
6750         (vmull_high_n_<su>16): Likewise.
6751         (vqrdmulh<q>_n_s16): Likewise.
6753 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6755         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6756         have the correct lane parameter.
6758 2013-09-06  Richard Biener <rguenther@suse.de>
6760         * cfganal.c (control_dependences::~control_dependences):
6761         Properly free all of the vector.
6763 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6765         PR target/58269
6766         * config/i386/i386.c (ix86_conditional_register_usage):
6767         Proper initialize extended SSE registers.
6769 2013-09-06  Jan Hubicka  <jh@suse.cz>
6771         PR tree-optimization/58311
6772         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6774 2013-09-06  Jan Hubicka  <jh@suse.cz>
6776         * Makefile.in (tree-sra.o): Update dependencies.
6777         * tree-sra.c: Include ipa-utils.h
6778         (scan_function): Use recursive_call_p.
6779         (has_caller_p): New function.
6780         (cgraph_for_node_and_aliases): Count also callers of aliases.
6782 2013-09-06  Jan Hubicka  <jh@suse.cz>
6784         PR middle-end/58094
6785         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6786         * tree-tailcall.c: Include ipa-utils.h.
6787         (find_tail_calls): Use it.
6788         * ipa-pure-const.c (check_call): Likewise.
6789         * ipa-utils.c (recursive_call_p): New function.
6790         * ipa-utils.h (recursive_call_p): Dclare.
6791         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6792         (symtab_semantically_equivalent_p): New function.
6793         * Makefile.in (tree-tailcall.o): Update dependencies.
6795 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6797         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6798         non-inlinable part.
6800 2013-09-06  Richard Biener  <rguenther@suse.de>
6802         * lto-streamer.h (lto_global_var_decls): Remove.
6803         * Makefile.in (OBJS): Remove lto-symtab.o.
6804         (lto-symtab.o): Remove.
6805         (GTFILES): Remove lto-symtab.c
6806         * lto-symtab.c: Move to lto/
6808 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6810         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6811         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6812         (UNSPEC_FPINT_RINT): New constant definitions.
6813         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6814         definition with 2 attributes.
6815         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6816         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6817         definitions.
6819 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6821         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6822         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6823         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6824         attribute to "z196_alone".
6825         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6826         "zEC12_simple".
6828 2013-09-06  Richard Biener  <rguenther@suse.de>
6830         * basic-block.h (class control_dependences): New.
6831         * tree-ssa-dce.c (control_dependence_map): Remove.
6832         (cd): New global.
6833         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6834         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6835         find_pdom, find_control_dependence, find_all_control_dependences):
6836         Move to cfganal.c.
6837         (mark_control_dependent_edges_necessary,
6838         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6839         tree_dce_done, perform_tree_ssa_dce): Adjust.
6840         * cfganal.c (set_control_dependence_map_bit,
6841         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6842         find_all_control_dependences): Move from tree-ssa-dce.c and
6843         implement as methods of control_dependences class.
6844         (control_dependences::control_dependences): New.
6845         (control_dependences::~control_dependences): Likewise.
6846         (control_dependences::get_edges_dependent_on): Likewise.
6847         (control_dependences::get_edge): Likewise.
6849 2013-09-04  Jan Hubicka  <jh@suse.cz>
6851         * tree.c (types_same_for_odr): Drop overactive check.
6852         * ipa-devirt.c (hash_type_name): Likewise.
6854 2013-09-04  Jan Hubicka  <jh@suse.cz>
6856         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6857         (analyze_functions): ... here.
6859 2013-09-04  Jan Hubicka  <jh@suse.cz>
6861         PR middle-end/58201
6862         * cgraphunit.c (analyze_functions): Clear AUX fields
6863         after processing; initialize assembler name has.
6865 2013-09-05  Jeff Law  <law@redhat.com>
6867         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6868         from thread_around_empty_block.  Record threading path into PATH.
6869         Recurse if threading through the initial block is successful.
6870         (thread_across_edge): Corresponding changes to slightly simplify.
6872 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6874         * config/aarch64/aarch64.md
6875         (type): Remove frecpe, frecps, frecpx.
6876         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6877         fix to be a TARGET_SIMD instruction.
6878         (aarch64_frecps): Remove.
6879         * config/aarch64/aarch64-simd.md
6880         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6881         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6883 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6884             Sofiane Naci  <sofiane.naci@arm.com>
6886         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6887         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6888         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6889         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6890         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6891         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6892         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6893         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6894         into "clz, "rbit".  Rename "shift" to "shift_imm".
6895         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6896         changes.  Update for attribute changes all occurrences of arlo_* and
6897         shift* types.
6898         * config/arm/arm-fixed.md: Update for attribute changes
6899         all occurrences of arlo_* types.
6900         * config/arm/thumb2.md: Update for attribute changes all occurrences
6901         of arlo_* types.
6902         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6903         (cortexa7_older_only): Likewise.
6904         (cortexa7_younger):  Likewise.
6905         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6906         (1020alu_shift_op): Likewise.
6907         (1020alu_shift_reg_op): Likewise.
6908         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6909         (alu_shift_op): Likewise.
6910         (alu_shift_reg_op): Likewise.
6911         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6912         (11_alu_shift_op): Likewise.
6913         (11_alu_shift_reg_op): Likewise.
6914         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6915         (9_alu_shift_reg_op): Likewise.
6916         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6917         attribute changes.
6918         (cortex_a15_alu_shift): Likewise.
6919         (cortex_a15_alu_shift_reg): Likewise.
6920         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6921         attribute changes.
6922         (cortex_a5_alu_shift): Likewise.
6923         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6924         attribute changes.
6925         (cortex_a53_alu_shift): Likewise.
6926         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6927         attribute changes.
6928         (cortex_a7_alu_reg): Likewise.
6929         (cortex_a7_alu_shift): Likewise.
6930         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6931         attribute changes.
6932         (cortex_a8_alu_shift): Likewise.
6933         (cortex_a8_alu_shift_reg): Likewise.
6934         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6935         (cortex_a9_dp_shift): Likewise.
6936         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6937         attribute changes.
6938         * config/arm/cortex-r4.md
6939         (cortex_r4_alu): Update for attribute changes.
6940         (cortex_r4_mov): Likewise.
6941         (cortex_r4_alu_shift_reg): Likewise.
6942         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6943         (526_alu_shift_op): Likewise.
6944         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6945         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6946         (626te_alu_shift_op): Likewise.
6947         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6948         (726te_alu_shift_op): Likewise.
6949         (726te_alu_shift_reg_op): Likewise.
6950         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6951         (mp626_alu_shift_op): Likewise.
6952         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6953         (pj4_alu_conds): Likewise.
6954         (pj4_shift): Likewise.
6955         (pj4_shift_conds): Likewise.
6956         (pj4_alu_shift): Likewise.
6957         (pj4_alu_shift_conds): Likewise.
6958         * config/aarch64/aarch64.md: Update for attribute change
6959         all occurrences of arlo_* and shift* types.
6961 2013-09-05  Mike Stump  <mikestump@comcast.net>
6963         * tree.h: Move documentation for tree_function_decl to tree-core.h
6964         with the declaration.
6966 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6968         PR target/58139
6969         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6970         looking for widest mode.
6972 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6974         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6976 2013-09-05  Richard Biener  <rguenther@suse.de>
6978         PR tree-optimization/58137
6979         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6980         Do not create vectors of pointers.
6981         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6982         types for the components of the vector initializer.
6983         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6984         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6986 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6988         * ipa-prop.c (remove_described_reference): Accept missing references,
6989         return false if that hppens, otherwise return true.
6990         (cgraph_node_for_jfunc): New function.
6991         (try_decrement_rdesc_refcount): Likewise.
6992         (try_make_edge_direct_simple_call): Use them.
6993         (ipa_edge_removal_hook): Remove references from rdescs.
6994         (ipa_edge_duplication_hook): Clone rdescs and their references
6995         when the new edge has the same caller as the old one.
6996         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6997         reference before removing any edges.
6999 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
7001         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
7002         initial store.
7003         * thumb2.md (thumb2_storewb_parisi): New pattern.
7005 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
7007         * config/aarch64/aarch64-option-extensions.def: Add
7008         AARCH64_OPT_EXTENSION of 'crc'.
7009         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
7010         (AARCH64_ISA_CRC): Ditto.
7011         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
7012         description of the CRC extension.
7014 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7016         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
7017         * config/rs6000/linux.h: Ditto.
7018         * alpha/linux.h: Ditto.
7019         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
7020         no_c99_libc_has_function.
7021         * config/c6x/uclinux-elf.h: Ditto.
7022         * config/lm32/uclinux-elf.h: Ditto.
7023         * config/m68k/uclinux.h: Ditto.
7024         * config/moxie/uclinux.h: Ditto.
7025         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
7026         (crisv32-*-linux*, cris-*-linux*): Ditto.
7027         * config/bfin/bfin.c: Include "tm_p.h".
7029 2013-09-05  Richard Biener  <rguenther@suse.de>
7031         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
7032         check for a definition without a basic-block.
7034 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7035             Sofiane Naci <sofiane.naci@arm.com>
7037         * config/aarch64/aarch64.md
7038         (*movti_aarch64): Rename r_2_f and f_2_r.
7039         (*movsf_aarch64): Likewise.
7040         (*movdf_aarch64): Likewise.
7041         (*movtf_aarch64): Likewise.
7042         (aarch64_movdi_<mode>low): Likewise.
7043         (aarch64_movdi_<mode>high): Likewise.
7044         (aarch64_mov<mode>high_di): Likewise.
7045         (aarch64_mov<mode>low_di): Likewise.
7046         (aarch64_movtilow_tilow): Likewise.
7047         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
7048         values to config/arm/types.md
7049         (attribute "conds"): Update for attribute change.
7050         (anddi3_insn): Likewise.
7051         (iordi3_insn): Likewise.
7052         (xordi3_insn): Likewise.
7053         (one_cmpldi2): Likewise.
7054         * config/arm/types.md (type): Add Neon types.
7055         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
7056         use "type" attribute.
7057         (movmisalign<mode>_neon_store): Likewise.
7058         (movmisalign<mode>_neon_load): Likewise.
7059         (vec_set<mode>_internal): Likewise.
7060         (vec_setv2di_internal): Likewise.
7061         (vec_extract<mode>): Likewise.
7062         (vec_extractv2di): Likewise.
7063         (add<mode>3_neon): Likewise.
7064         (adddi3_neon): Likewise.
7065         (sub<mode>3_neon): Likewise.
7066         (subdi3_neon): Likewise.
7067         (mul<mode>3_neon): Likewise.
7068         (mul<mode>3add<mode>_neon): Likewise.
7069         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
7070         (fma<VCVTF:mode>4)): Likewise.
7071         (fma<VCVTF:mode>4_intrinsic): Likewise.
7072         (fmsub<VCVTF:mode>4)): Likewise.
7073         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
7074         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
7075         (ior<mode>3): Likewise.
7076         (and<mode>3): Likewise.
7077         (anddi3_neon): Likewise.
7078         (orn<mode>3_neon): Likewise.
7079         (orndi3_neon): Likewise.
7080         (bic<mode>3_neon): Likewise.
7081         (bicdi3_neon): Likewise.
7082         (xor<mode>3): Likewise.
7083         (one_cmpl<mode>2): Likewise.
7084         (abs<mode>2): Likewise.
7085         (neg<mode>2): Likewise.
7086         (umin<mode>3_neon): Likewise.
7087         (umax<mode>3_neon): Likewise.
7088         (smin<mode>3_neon): Likewise.
7089         (smax<mode>3_neon): Likewise.
7090         (vashl<mode>3): Likewise.
7091         (vashr<mode>3_imm): Likewise.
7092         (vlshr<mode>3_imm): Likewise.
7093         (ashl<mode>3_signed): Likewise.
7094         (ashl<mode>3_unsigned): Likewise.
7095         (neon_load_count): Likewise.
7096         (ashldi3_neon_noclobber): Likewise.
7097         (signed_shift_di3_neon): Likewise.
7098         (unsigned_shift_di3_neon): Likewise.
7099         (ashrdi3_neon_imm_noclobber): Likewise.
7100         (lshrdi3_neon_imm_noclobber): Likewise.
7101         (widen_ssum<mode>3): Likewise.
7102         (widen_usum<mode>3): Likewise.
7103         (quad_halves_<code>v4si): Likewise.
7104         (quad_halves_<code>v4sf): Likewise.
7105         (quad_halves_<code>v8hi): Likewise.
7106         (quad_halves_<code>v16qi): Likewise.
7107         (reduc_splus_v2di): Likewise.
7108         (neon_vpadd_internal<mode>): Likewise.
7109         (neon_vpsmin<mode>): Likewise.
7110         (neon_vpsmax<mode>): Likewise.
7111         (neon_vpumin<mode>): Likewise.
7112         (neon_vpumax<mode>): Likewise.
7113         (ss_add<mode>_neon): Likewise.
7114         (us_add<mode>_neon): Likewise.
7115         (ss_sub<mode>_neon): Likewise.
7116         (us_sub<mode>_neon): Likewise.
7117         (neon_vadd<mode>_unspec): Likewise.
7118         (neon_vaddl<mode>): Likewise.
7119         (neon_vaddw<mode>): Likewise.
7120         (neon_vhadd<mode>): Likewise.
7121         (neon_vqadd<mode>): Likewise.
7122         (neon_vaddhn<mode>): Likewise.
7123         (neon_vmul<mode>): Likewise.
7124         (neon_vmla<mode>): Likewise.
7125         (neon_vmlal<mode>): Likewise.
7126         (neon_vmls<mode>): Likewise.
7127         (neon_vmlsl<mode>): Likewise.
7128         (neon_vqdmulh<mode>): Likewise.
7129         (neon_vqdmlal<mode>): Likewise.
7130         (neon_vqdmlsl<mode>): Likewise.
7131         (neon_vmull<mode>): Likewise.
7132         (neon_vqdmull<mode>): Likewise.
7133         (neon_vsub<mode>_unspec): Likewise.
7134         (neon_vsubl<mode>): Likewise.
7135         (neon_vsubw<mode>): Likewise.
7136         (neon_vqsub<mode>): Likewise.
7137         (neon_vhsub<mode>): Likewise.
7138         (neon_vsubhn<mode>): Likewise.
7139         (neon_vceq<mode>): Likewise.
7140         (neon_vcge<mode>): Likewise.
7141         (neon_vcgeu<mode>): Likewise.
7142         (neon_vcgt<mode>): Likewise.
7143         (neon_vcgtu<mode>): Likewise.
7144         (neon_vcle<mode>): Likewise.
7145         (neon_vclt<mode>): Likewise.
7146         (neon_vcage<mode>): Likewise.
7147         (neon_vcagt<mode>): Likewise.
7148         (neon_vtst<mode>): Likewise.
7149         (neon_vabd<mode>): Likewise.
7150         (neon_vabdl<mode>): Likewise.
7151         (neon_vaba<mode>): Likewise.
7152         (neon_vabal<mode>): Likewise.
7153         (neon_vmax<mode>): Likewise.
7154         (neon_vmin<mode>): Likewise.
7155         (neon_vpaddl<mode>): Likewise.
7156         (neon_vpadal<mode>): Likewise.
7157         (neon_vpmax<mode>): Likewise.
7158         (neon_vpmin<mode>): Likewise.
7159         (neon_vrecps<mode>): Likewise.
7160         (neon_vrsqrts<mode>): Likewise.
7161         (neon_vqabs<mode>): Likewise.
7162         (neon_vqneg<mode>): Likewise.
7163         (neon_vcls<mode>): Likewise.
7164         (clz<mode>2): Likewise.
7165         (popcount<mode>2): Likewise.
7166         (neon_vrecpe): Likewise.
7167         (neon_vrsqrte): Likewise.
7168         (neon_vget_lane<mode>_sext_internal): Likewise.
7169         (neon_vget_lane<mode>_zext_internal): Likewise.
7170         (neon_vdup_n<mode>): Likewise.
7171         (neon_vdup_nv2di): Likewise.
7172         (neon_vdpu_lane<mode>_internal): Likewise.
7173         (neon_vswp<mode>): Likewise.
7174         (float<mode><V_cvtto>2): Likewise.
7175         (floatuns<mode><V_cvtto>2): Likewise.
7176         (fix_trunc<mode><V_cvtto>)2): Likewise
7177         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7178         (neon_vcvt<mode>): Likewise.
7179         (neon_vcvtv4sfv4hf): Likewise.
7180         (neon_vcvtv4hfv4sf): Likewise.
7181         (neon_vcvt_n<mode>): Likewise.
7182         (neon_vmovn<mode>): Likewise.
7183         (neon_vqmovn<mode>): Likewise.
7184         (neon_vqmovun<mode>): Likewise.
7185         (neon_vmovl<mode>): Likewise.
7186         (neon_vmul_lane<mode>): Likewise.
7187         (neon_vmull_lane<mode>): Likewise.
7188         (neon_vqdmull_lane<mode>): Likewise.
7189         (neon_vqdmulh_lane<mode>): Likewise.
7190         (neon_vmla_lane<mode>): Likewise.
7191         (neon_vmlal_lane<mode>): Likewise.
7192         (neon_vqdmlal_lane<mode>): Likewise.
7193         (neon_vmls_lane<mode>): Likewise.
7194         (neon_vmlsl_lane<mode>): Likewise.
7195         (neon_vqdmlsl_lane<mode>): Likewise.
7196         (neon_vext<mode>): Likewise.
7197         (neon_vrev64<mode>): Likewise.
7198         (neon_vrev32<mode>): Likewise.
7199         (neon_vrev16<mode>): Likewise.
7200         (neon_vbsl<mode>_internal): Likewise.
7201         (neon_vshl<mode>): Likewise.
7202         (neon_vqshl<mode>): Likewise.
7203         (neon_vshr_n<mode>): Likewise.
7204         (neon_vshrn_n<mode>): Likewise.
7205         (neon_vqshrn_n<mode>): Likewise.
7206         (neon_vqshrun_n<mode>): Likewise.
7207         (neon_vshl_n<mode>): Likewise.
7208         (neon_vqshl_n<mode>): Likewise.
7209         (neon_vqshlu_n<mode>): Likewise.
7210         (neon_vshll_n<mode>): Likewise.
7211         (neon_vsra_n<mode>): Likewise.
7212         (neon_vsri_n<mode>): Likewise.
7213         (neon_vsli_n<mode>): Likewise.
7214         (neon_vtbl1v8qi): Likewise.
7215         (neon_vtbl2v8qi): Likewise.
7216         (neon_vtbl3v8qi): Likewise.
7217         (neon_vtbl4v8qi): Likewise.
7218         (neon_vtbx1v8qi): Likewise.
7219         (neon_vtbx2v8qi): Likewise.
7220         (neon_vtbx3v8qi): Likewise.
7221         (neon_vtbx4v8qi): Likewise.
7222         (neon_vtrn<mode>_internal): Likewise.
7223         (neon_vzip<mode>_internal): Likewise.
7224         (neon_vuzp<mode>_internal): Likewise.
7225         (neon_vld1<mode>): Likewise.
7226         (neon_vld1_lane<mode>): Likewise.
7227         (neon_vld1_dup<mode>): Likewise.
7228         (neon_vld1_dupv2di): Likewise.
7229         (neon_vst1<mode>): Likewise.
7230         (neon_vst1_lane<mode>): Likewise.
7231         (neon_vld2<mode>): Likewise.
7232         (neon_vld2_lane<mode>): Likewise.
7233         (neon_vld2_dup<mode>): Likewise.
7234         (neon_vst2<mode>): Likewise.
7235         (neon_vst2_lane<mode>): Likewise.
7236         (neon_vld3<mode>): Likewise.
7237         (neon_vld3qa<mode>): Likewise.
7238         (neon_vld3qb<mode>): Likewise.
7239         (neon_vld3_lane<mode>): Likewise.
7240         (neon_vld3_dup<mode>): Likewise.
7241         (neon_vst3<mode>): Likewise.
7242         (neon_vst3qa<mode>): Likewise.
7243         (neon_vst3qb<mode>): Likewise.
7244         (neon_vst3_lane<mode>): Likewise.
7245         (neon_vld4<mode>): Likewise.
7246         (neon_vld4qa<mode>): Likewise.
7247         (neon_vld4qb<mode>): Likewise.
7248         (neon_vld4_lane<mode>): Likewise.
7249         (neon_vld4_dup<mode>): Likewise.
7250         (neon_vst4<mode>): Likewise.
7251         (neon_vst4qa<mode>): Likewise.
7252         (neon_vst4qb<mode>): Likewise.
7253         (neon_vst4_lane<mode>): Likewise.
7254         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7255         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7256         (neon_vec_<US>mult_lo_<mode>): Likewise.
7257         (neon_vec_<US>mult_hi_<mode>): Likewise.
7258         (neon_vec_<US>shiftl_<mode>): Likewise.
7259         (neon_unpack<US>_<mode>): Likewise.
7260         (neon_vec_<US>mult_<mode>): Likewise.
7261         (vec_pack_trunc_<mode>): Likewise.
7262         (neon_vec_pack_trunk_<mode>): Likewise.
7263         (neon_vabd<mode>_2): Likewise.
7264         (neon_vabd<mode>_3): Likewise.
7265         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7266         (thumb2_movsi_vfp): Likewise.
7267         (movdi_vfp): Likewise.
7268         (movdi_vfp_cortexa8): Likewise.
7269         (movhf_vfp_neon): Likewise.
7270         (movhf_vfp): Likewiwse.
7271         (movsf_vfp): Likewiwse.
7272         (thumb2_movsf_vfp): Likewiwse.
7273         (movdf_vfp): Likewise.
7274         (thumb2_movdf_vfp): Likewise.
7275         (movsfcc_vfp): Likewise.
7276         (thumb2_movsfcc_vfp): Likewise.
7277         (movdfcc_vfp): Likewise.
7278         (thumb2_movdfcc_vfp): Likewise.
7279         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7280         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7281         (v10_v2c): Likewise.
7282         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7283         attribute change.
7284         (cortex_a15_neon_int_2): Likewise.
7285         (cortex_a15_neon_int_3): Likewise.
7286         (cortex_a15_neon_int_4): Likewise.
7287         (cortex_a15_neon_int_5): Likewise.
7288         (cortex_a15_neon_vqneg_vqabs): Likewise.
7289         (cortex_a15_neon_vmov): Likewise.
7290         (cortex_a15_neon_vaba): Likewise.
7291         (cortex_a15_neon_vaba_qqq): Likewise.
7292         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7293         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7294         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7295         scalar_64_32_long_scalar): Likewise.
7296         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7297         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7298         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7299         lotype_qdd_64_32_long): Likewise.
7300         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7301         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7302         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7303         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7304         (cortex_a15_neon_shift_1): Likewise.
7305         (cortex_a15_neon_shift_2): Likewise.
7306         (cortex_a15_neon_shift_3): Likewise.
7307         (cortex_a15_neon_vshl_ddd): Likewise.
7308         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7309         (cortex_a15_neon_vsra_vrsra): Likewise.
7310         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7311         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7312         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7313         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7314         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7315         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7316         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7317         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7318         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7319         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7320         (cortex_a15_neon_bp_simple): Likewise.
7321         (cortex_a15_neon_bp_2cycle): Likewise.
7322         (cortex_a15_neon_bp_3cycle): Likewise.
7323         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7324         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7325         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7326         (cortex_a15_neon_vld2_4_regs): Likewise.
7327         (cortex_a15_neon_vld3_vld4): Likewise.
7328         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7329         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7330         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7331         (cortex_a15_neon_vst3_vst4): Likewise.
7332         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7333         (cortex_a15_neon_vld3_vld4_lane" 10
7334         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7335         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7336         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7337         (cortex_a15_neon_ldm_2): Likewise.0
7338         (cortex_a15_neon_stm_2): Likewise.
7339         (cortex_a15_neon_mcr): Likewise.
7340         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7341         (cortex_a15_neon_mrc): Likewise.
7342         (cortex_a15_neon_mrrc): Likewise.
7343         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7344         change.
7345         (cortex_a15_alu_shift): Likewise.
7346         (cortex_a15_alu_shift_reg): Likewise.
7347         (cortex_a15_mult32): Likewise.
7348         (cortex_a15_mult64): Likewise.
7349         (cortex_a15_block): Likewise.
7350         (cortex_a15_branch): Likewise.
7351         (cortex_a15_load1): Likewise.
7352         (cortex_a15_load3): Likewise.
7353         (cortex_a15_store1): Likewise.
7354         (cortex_a15_store3): Likewise.
7355         (cortex_a15_call): Likewise.
7356         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7357         (cortex_a5_f2r): Likewise.
7358         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7359         change.
7360         (cortex_a53_f2r): Likewise.
7361         * config/arm/cortex-a7.md
7362         (cortex_a7_branch): Update for attribute change.
7363         (cortex_a7_call): Likewise.
7364         (cortex_a7_alu_imm): Likewise.
7365         (cortex_a7_alu_reg): Likewise.
7366         (cortex_a7_alu_shift): Likewise.
7367         (cortex_a7_mul): Likewise.
7368         (cortex_a7_load1): Likewise.
7369         (cortex_a7_store1): Likewise.
7370         (cortex_a7_load2): Likewise.
7371         (cortex_a7_store2): Likewise.
7372         (cortex_a7_load3): Likewise.
7373         (cortex_a7_store3): Likewise.
7374         (cortex_a7_load4): Likewise.
7375         (cortex_a7_store4): Likewise.
7376         (cortex_a7_fpalu): Likewise.
7377         (cortex_a7_fconst): Likewise.
7378         (cortex_a7_fpmuls): Likewise.
7379         (cortex_a7_neon_mul): Likewise.
7380         (cortex_a7_fpmacs): Likewise.
7381         (cortex_a7_neon_mla: Likewise.
7382         (cortex_a7_fpmuld: Likewise.
7383         (cortex_a7_fpmacd: Likewise.
7384         (cortex_a7_fpfmad: Likewise.
7385         (cortex_a7_fdivs: Likewise.
7386         (cortex_a7_fdivd: Likewise.
7387         (cortex_a7_r2f: Likewise.
7388         (cortex_a7_f2r: Likewise.
7389         (cortex_a7_f_flags: Likewise.
7390         (cortex_a7_f_loads: Likewise.
7391         (cortex_a7_f_loadd: Likewise.
7392         (cortex_a7_f_stores: Likewise.
7393         (cortex_a7_f_stored: Likewise.
7394         (cortex_a7_neon): Likewise.
7395         * config/arm/cortex-a8-neon.md
7396         (cortex_a8_neon_mrc): Update for attribute change.
7397         (cortex_a8_neon_mrrc): Likewise.
7398         (cortex_a8_neon_int_1): Likewise.
7399         (cortex_a8_neon_int_2): Likewise.
7400         (cortex_a8_neon_int_3): Likewise.
7401         (cortex_a8_neon_int_4): Likewise.
7402         (cortex_a8_neon_int_5): Likewise.
7403         (cortex_a8_neon_vqneg_vqabs): Likewise.
7404         (cortex_a8_neon_vmov): Likewise.
7405         (cortex_a8_neon_vaba): Likewise.
7406         (cortex_a8_neon_vaba_qqq): Likewise.
7407         (cortex_a8_neon_vsma): Likewise.
7408         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7409         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7410         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7411         long_scalar): Likewise.
7412         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7413         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7414         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7415         long_scalar_qdd_64_32_long): Likewise.
7416         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7417         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7418         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7419         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7420         (cortex_a8_neon_shift_1): Likewise.
7421         (cortex_a8_neon_shift_2): Likewise.
7422         (cortex_a8_neon_shift_3): Likewise.
7423         (cortex_a8_neon_vshl_ddd): Likewise.
7424         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7425         (cortex_a8_neon_vsra_vrsra): Likewise.
7426         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7427         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7428         (cortex_a8_neon_fp_vsum): Likewise.
7429         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7430         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7431         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7432         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7433         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7434         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7435         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7436         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7437         (cortex_a8_neon_bp_simple): Likewise.
7438         (cortex_a8_neon_bp_2cycle): Likewise.
7439         (cortex_a8_neon_bp_3cycle): Likewise.
7440         (cortex_a8_neon_ldr): Likewise.
7441         (cortex_a8_neon_str): Likewise.
7442         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7443         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7444         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7445         (cortex_a8_neon_vld2_4_regs): Likewise.
7446         (cortex_a8_neon_vld3_vld4): Likewise.
7447         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7448         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7449         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7450         (cortex_a8_neon_vst3_vst4): Likewise.
7451         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7452         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7453         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7454         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7455         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7456         (cortex_a8_neon_mcr): Likewise.
7457         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7458         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7459         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7460         change.
7461         (ca9_neon_mrrc): Likewise.
7462         (cortex_a9_neon_int_1): Likewise.
7463         (cortex_a9_neon_int_2): Likewise.
7464         (cortex_a9_neon_int_3): Likewise.
7465         (cortex_a9_neon_int_4): Likewise.
7466         (cortex_a9_neon_int_5): Likewise.
7467         (cortex_a9_neon_vqneg_vqabs): Likewise.
7468         (cortex_a9_neon_vmov): Likewise.
7469         (cortex_a9_neon_vaba): Likewise.
7470         (cortex_a9_neon_vaba_qqq): Likewise.
7471         (cortex_a9_neon_vsma): Likewise.
7472         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7473         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7474         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7475         long_scalar): Likewise.
7476         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7477         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7478         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7479         long_scalar_qdd_64_32_long): Likewise.
7480         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7481         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7482         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7483         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7484         (cortex_a9_neon_shift_1): Likewise.
7485         (cortex_a9_neon_shift_2): Likewise.
7486         (cortex_a9_neon_shift_3): Likewise.
7487         (cortex_a9_neon_vshl_ddd): Likewise.
7488         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7489         (cortex_a9_neon_vsra_vrsra): Likewise.
7490         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7491         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7492         (cortex_a9_neon_fp_vsum): Likewise.
7493         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7494         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7495         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7496         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7497         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7498         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7499         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7500         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7501         (cortex_a9_neon_bp_simple): Likewise.
7502         (cortex_a9_neon_bp_2cycle): Likewise.
7503         (cortex_a9_neon_bp_3cycle): Likewise.
7504         (cortex_a9_neon_ldr): Likewise.
7505         (cortex_a9_neon_str): Likewise.
7506         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7507         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7508         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7509         (cortex_a9_neon_vld2_4_regs): Likewise.
7510         (cortex_a9_neon_vld3_vld4): Likewise.
7511         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7512         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7513         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7514         (cortex_a9_neon_vst3_vst4): Likewise.
7515         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7516         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7517         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7518         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7519         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7520         (cortex_a9_neon_mcr): Likewise.
7521         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7522         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7523         (cortex_a9_fps): Likewise.
7524         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7525         change.
7526         (cortex_m4_fmuls): Likewise.
7527         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7528         change.
7529         (cortex_r4_mrc): Likewise.
7530         * config/arm/iterators.md: Update comment referring to neon_type.
7531         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7532         (iwmmxt_movsi_insn): Likewise.
7533         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7534         attribute change.
7535         (pj4_core_to_vfp): Likewise.
7536         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7537         attribute change.
7538         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7539         (vfp_fstore): Likewise.
7540         * doc/md.texi: Change references to neon_type to refer to type.
7542 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7544         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7546 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7548         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7549         lglobl if not weak.
7551 2013-09-04  Easwaran Raman  <eraman@google.com>
7553         PR middle-end/57370
7554         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7555         (build_and_add_sum): Use it.
7556         (appears_later_in_bb): Simplify code.
7558 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7560         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7562 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7564         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7566 2013-09-04  Jan Hubicka  <jh@suse.cz>
7568         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7569         * ipa-devirt.c: Include diganostic.h
7570         (odr_type_d): Add types and types_set.
7571         (hash_type_name): Work for types with vtables during LTO.
7572         (odr_hasher::remove): Fix comment; destroy types_set.
7573         (add_type_duplicate): New function,
7574         (get_odr_type): Use it.
7575         (dump_type_inheritance_graph): Dump type duplicates.
7576         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7577         graph.
7578         * tree.c (types_same_for_odr): Give exact answers on types with
7579         virtual tables.
7581 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7583         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7584         explaining their differences.
7586 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7588         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7590 2013-09-03  Jeff Law  <law@redhat.com>
7592         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7593         when not threading through a joiner block.  Pass joiner/no joiner
7594         state to register_jump_thread.
7595         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7596         state from argument rather than implying on path length.
7597         Dump the entire jump thread path into debugging dump.
7598         * tree-flow.h (register_jump_thread): Update prototype.
7600 2013-08-29  Xinliang David Li  <davidxl@google.com>
7602         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7603         Remove a trivial gcc_assert.
7605 2013-08-29  Xinliang David Li  <davidxl@google.com>
7607         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7608         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7609         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7610         Delay base decl alignment adjustment.
7611         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7612         * tree-vectorizer.h: New data structure.
7613         (set_dr_misalignment): New function.
7614         (dr_misalignment): Ditto.
7615         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7616         (vectorizable_load): Ditto.
7617         (ensure_base_align): New function.
7618         (vectorize_loops): Add dbg_cnt support.
7619         (execute_vect_slp): Ditto.
7620         * dbgcnt.def: New debug counter.
7621         * Makefile: New dependency.
7623 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7625         Revert:
7627         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7629         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7631 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7633         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7634         function descriptor.
7636 2013-09-03  Richard Biener  <rguenther@suse.de>
7638         * tree-affine.c (add_elt_to_tree): Fix association issue,
7639         avoid useless converts and make sure to always return a
7640         properly typed result.
7642 2013-09-03  Richard Biener  <rguenther@suse.de>
7644         PR middle-end/57656
7645         * fold-const.c (negate_expr_p): Fix division case.
7646         (negate_expr): Likewise.
7648 2013-09-03  Richard Biener  <rguenther@suse.de>
7650         PR lto/58285
7651         * tree-streamer-out.c: Include tm.h.
7652         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7654 2013-09-03  Jan Hubicka  <jh@suse.cz>
7656         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7658 2013-09-03  Alan Modra  <amodra@gmail.com>
7660         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7661         multilibs to big-endian target and vice versa.
7662         * config/rs6000/t-linux64: Use := assignment on all vars.
7663         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7664         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7665         * config/rs6000/t-linux64le: New file.
7666         * config/rs6000/t-linux64bele: New file.
7667         * config/rs6000/t-linux64lebe: New file.
7669 2013-09-02  Jan Hubicka  <jh@suse.cz>
7671         * ipa-inline-transform.c (inline_transform): Do not
7672         optimize_inline_calls when not optimizing.
7674 2013-09-02  Jan Hubicka  <jh@suse.cz>
7676         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7677         duplicated nodes for assembler names.
7678         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7679         hard registers from assembler name hash.
7681 2013-09-02  Jan Hubicka  <jh@suse.cz>
7683         * ipa-split.c (execute_split_functions): Split externally visible
7684         functions called once.
7686 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7688         PR ipa/58106
7689         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7690         linked list.  When finding the correct duplicate, also consider also
7691         the caller in additon to its inlined_to node.
7693 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7695         * config/aarch64/aarch64-simd-builtins.def
7696         (dup_lane_scalar): Remove.
7697         * config/aarch64/aarch64-simd.md
7698         (aarch64_simd_dup): Add 'w->w' alternative.
7699         (aarch64_dup_lane<mode>): Allow for VALL.
7700         (aarch64_dup_lane_scalar<mode>): Remove.
7701         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7702         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7703         (aarch64_get_lane_unsigned<mode>): Likewise.
7704         (aarch64_get_lane<mode>): Likewise.
7705         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7706         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7707         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7708         (VCON): Change container of V2SF.
7709         (vswap_width_name): Likewise.
7710         * config/aarch64/arm_neon.h
7711         (__aarch64_vdup_lane_any): New.
7712         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7713         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7714         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7716 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7718         PR middle-end/56382
7719         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7720         the source or the destination is a single hard register.
7722 2013-09-02  Richard Biener  <rguenther@suse.de>
7724         PR middle-end/57511
7725         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7726         non-linear SCEVs.
7728 2013-09-02  Richard Biener  <rguenther@suse.de>
7730         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7731         arithmetic to sizetype.
7733 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7735         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7736         Find auto-increment use both before and after candidate.
7738 2013-09-02  Marek Polacek  <polacek@redhat.com>
7740         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7742 2013-09-01  Jan Hubicka  <jh@suse.cz>
7744         * Makefile.in: Add ipa-profile.o
7745         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7746         * cgraph.c (struct cgraph_propagate_frequency_data,
7747         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7748         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7749         * cgraph.h (cgraph_propagate_frequency): Remove.
7750         * ipa-inline-analysis.c: Include ipa-utils.h;
7751         drop duplicated cfgloop.h.
7752         (inline_update_callee_summaries): Update.
7753         * ipa-profile.c: New file.
7754         * ipa-utils.h (ipa_propagate_frequency): Declare.
7755         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7756         data-streamer.h, value-prof.h.
7757         (symtab_remove_unreachable_nodes): Update profile.
7758         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7759         account_time_size, cmp_counts, dump_histogram,
7760         ipa_profile_generate_summary, ipa_profile_write_summary,
7761         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7762         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7763         Move to ipa-profile.c.
7765 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7767         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7769 2013-09-01  Jan Hubicka  <jh@suse.cz>
7771         * common.opt (fdevirtualize-speculatively): New function.
7772         * invoke.texi (fdevirtualize-speculatively): Document.
7773         * ipa-devirt.c: Include ipa-inline.h
7774         (likely_target_p): New function.
7775         (ipa_devirt): New function.
7776         (gate_ipa_devirt): New function.
7777         (pass_data_ipa_devirt): New static var.
7778         (pass_ipa_devirt): Likewise.
7779         (make_pass_ipa_devirt): New function.
7780         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7781         (common_handle_option): Disable devirtualization when
7782         value range profiling is available.
7783         * passes.def (pass_ipa_devirt): Add.
7784         * timever.def (TV_IPA_DEVIRT): New timevar.
7785         * tree-pass.h (make_pass_ipa_devirt):
7787 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7789         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7790         include sanitize(undefined).
7792 2013-08-31  Diego Novillo  <dnovillo@google.com>
7794         * Makefile.in (TREE_CORE_H): Define.
7795         (TREE_H): Use.
7796         (GTFILES): Add tree-core.h.
7797         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7798         size the array.
7799         * tree-core.h: New file.
7800         Move all data structures, enum, typedefs, global
7801         declarations and constants from ...
7802         * tree.h: ... here.
7804 2013-08-31  Jan Hubicka  <jh@suse.cz>
7806         * bulitins.c (expand_builtin): Do not early exit for gcov
7807         instrumented functions.
7809 2013-08-31  Marek Polacek  <polacek@redhat.com>
7811         * ubsan.c: Include tm_p.h.
7813 2013-08-31  Jan Hubicka  <jh@suse.cz>
7815         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7816         warning.
7818         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7819         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7820         * tree-cfg.c (verify_gimple_label): ... here.
7821         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7822         (ipa_merge_profiles): New function.
7823         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7824         (lto_input_function_body): Likewise.
7825         * ipa-utils.h (ipa_merge_profiles): Declare.
7826         * lto-streamer.h (lto_input_function_body): Update prototype.
7827         (emit_label_in_global_context_p): Remove.
7828         * lto-symtab.c: Include ipa-utils.h
7829         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7831 2013-08-31  Jan Hubicka  <jh@suse.cz>
7833         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7835 2013-08-31  Jan Hubicka  <jh@suse.cz>
7837         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7839 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7841         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7842         "cmp" RTX before signed_comparison_operator check to account
7843         for "code" changes.
7845 2013-08-30  Jan Hubicka  <jh@suse.cz>
7847         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7848         (detect_type_change_1): Rename to ...
7849         (detect_type_change): ... this one; early return on non-polymorphic
7850         types.
7851         (detect_type_change_ssa): Add comp_type parameter; update
7852         use of detect_type_change.
7853         (compute_complex_assign_jump_func): Add param_type parameter;
7854         update use of detect_type_change_ssa.
7855         (compute_complex_ancestor_jump_func): Likewise.
7856         (ipa_get_callee_param_type): New function.
7857         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7858         update calls to the jump function computation functions.
7860 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7861             Steven Bosscher  <steven@gcc.gnu.org>
7863         * cfgrtl.c (fixup_new_cold_bb): New routine.
7864         (commit_edge_insertions): Invoke fixup_partitions.
7865         (find_partition_fixes): New routine.
7866         (fixup_partitions): Ditto.
7867         (verify_hot_cold_block_grouping): Update comments.
7868         (rtl_verify_edges): Invoke find_partition_fixes.
7869         (rtl_verify_bb_pointers): Update comments.
7870         (rtl_verify_bb_layout): Ditto.
7871         * basic-block.h (probably_never_executed_edge_p): Declare.
7872         (fixup_partitions): Ditto.
7873         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7874         * bb-reorder.c (sanitize_hot_paths): New function.
7875         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7876         sanitize_hot_paths.
7877         * predict.c (probably_never_executed_edge_p): New routine.
7878         * cfg.c (check_bb_profile): Add partition insanity warnings.
7880 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7882         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7884 2013-08-30  Marek Polacek  <polacek@redhat.com>
7886         * Makefile.in (ubsan.o): Add.
7887         (c-family/c-ubsan.o): Add.
7888         (builtins.o): Add ubsan.h dependency.
7889         * ubsan.h: New file.
7890         * ubsan.c: New file.
7891         * common.opt: Add -fsanitize=undefined option.
7892         (flag_sanitize): Add variable.
7893         (fsanitize=): Add option.  Add Driver.
7894         (fsanitize=thread): Remove option.
7895         (fsanitize=address): Likewise.
7896         (static-libubsan): New option.
7897         * doc/invoke.texi: Document the new flag and -static-libubsan.
7898         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7899         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7900         * builtin-attrs.def (ATTR_COLD): Define.
7901         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7902         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7903         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7904         * flag-types.h (sanitize_code): New enum.
7905         * opts.c (common_handle_option): Parse command line arguments
7906         of -fsanitize=.  Add -fsanitize=unreachable option.
7907         * varasm.c (get_variable_section): Adjust.
7908         (assemble_noswitch_variable): Likewise.
7909         (assemble_variable): Likewise.
7910         (output_constant_def_contents): Likewise.
7911         (categorize_decl_for_section): Likewise.
7912         (place_block_symbol): Likewise.
7913         (output_object_block): Likewise.
7914         * builtins.def: Likewise.
7915         * toplev.c (compile_file): Likewise.
7916         (process_options): Likewise.
7917         * cppbuiltin.c: Likewise.
7918         * tsan.c (tsan_pass): Likewise.
7919         (tsan_gate): Likewise.
7920         (tsan_gate_O0): Likewise.
7921         * cfgexpand.c (partition_stack_vars): Likewise.
7922         (expand_stack_vars): Likewise.
7923         (defer_stack_allocation): Likewise.
7924         (expand_used_vars): Likewise.
7925         * cfgcleanup.c (old_insns_match_p): Likewise.
7926         * asan.c (asan_finish_file): Likewise.
7927         (asan_instrument): Likewise.
7928         (gate_asan): Likewise.
7929         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7930         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7931         (asan_global_struct): Use pointer_sized_int_node instead
7932         calling build_nonstandard_integer_type.
7933         (initialize_sanitizer_builtins): Likewise.
7934         (asan_finish_file): Likewise.
7935         * gcc.c: Document %{%:function(args):X}.
7936         (static_spec_functions): Add sanitize.
7937         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7938         store funcval != NULL there.
7939         (do_spec_1): Adjust handle_spec_function caller.
7940         (handle_braces): Allow %:function(args) as condition.
7941         (sanitize_spec_function): New function.
7942         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7943         (LIBUBSAN_SPEC): Likewise.
7944         (LIBUBSAN_EARLY_SPEC): Likewise.
7945         (SANITIZER_SPEC): Handle libubsan.
7946         (SANITIZER_EARLY_SPEC): Likewise.
7947         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7948         instead of fsanitize=address.
7949         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7950         instead of fsanitize=address*.
7951         * builtins.c: Include ubsan.h.
7952         (fold_builtin_0): Instrument __builtin_unreachable.
7953         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7954         instead of flag_asan.
7955         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7956         (pointer_sized_int_node): Define.
7957         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7959 2013-08-30  Mike Stump  <mikestump@comcast.net>
7961         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7962         with RE patterns.
7964 2013-08-29  Jan Hubicka  <jh@suse.cz>
7966         * cgraph.c (cgraph_function_body_availability): Handle weakref
7967         correctly.
7968         * passes.def: Remove pass_fixup_cfg.
7969         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7970         track when we need to remove functions.
7971         (gate_ipa_inline): Execute inlining always; add comment why.
7972         (pass_data_ipa_inline): Remove TODO_remove_functions.
7973         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7974         do not produce summaries.
7975         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7976         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7977         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7978         constant pool and vtable.
7980 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7982         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7983         New arm_neon.h's internal macros to specify 64-bit constants.
7984         Avoid using stdint.h's macros.
7986 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7988         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7989         before applying REG_P.
7991 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7993         PR tree-optimization/58277
7994         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7995         after seeing too many stmts with vdef in between dombb and current
7996         bb, invalidate everything.
7998 2013-08-30  Richard Biener  <rguenther@suse.de>
8000         * fold-const.c (fold_single_bit_test): Fix overflow test.
8002 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
8004         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
8005         and which can live in a register, always retrieve the value on entry.
8006         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
8007         passed by invisible reference specially.
8008         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
8009         (vt_add_function_parameter): Correctly deal with a parameter passed by
8010         invisible reference.
8012 2013-08-30  Jan Hubicka  <jh@suse.cz>
8014         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
8016 2013-08-30  Richard Biener  <rguenther@suse.de>
8018         PR tree-optimization/58228
8019         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
8020         allow invariant loads in nested loop vectorization.
8022 2013-08-30  Richard Biener  <rguenther@suse.de>
8024         PR tree-optimization/58223
8025         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
8026         (has_anti_or_output_dependence): ... this and adjust to also
8027         look for output dependences.
8028         (mark_nodes_having_upstream_mem_writes): Adjust.
8029         (rdg_flag_uses): Likewise.
8031 2013-08-30  Richard Biener  <rguenther@suse.de>
8033         PR tree-optimization/58010
8034         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
8035         assert that we have a loop-closed PHI.
8037 2013-08-29  Jan Hubicka  <jh@suse.cz>
8039         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
8040         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
8041         * lto-section-in.c (lto_free_function_in_decl_state): New function.
8042         (lto_free_function_in_decl_state_for_node): New function.
8044 2013-08-29  Xinliang David Li  <davidxl@google.com>
8046         * loop-unroll.c (report_unroll_peel): Minor message change.
8047         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
8048         Emit alignment peeling message with default -fopt-info.
8049         (vect_loop_versioning): Emit loop version info message.
8050         * tree-vectorizer.c (vectorize_loops): Minor message change.
8051         (execute_vect_slp): Ditto.
8053 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
8055         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
8056         of the clone from the DECL_NAME of the original function.
8058 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
8060         * passes.c (register_pass): Add overload.
8061         * tree-pass.h (register_pass): Forward declare it.  Add comment.
8063 2013-08-29  Jan Hubicka  <jh@suse.cz>
8065         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
8066         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
8067         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
8068         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8069         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
8070         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
8071         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8072         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
8074 2013-08-29  Teresa Johnson  <tejohnson@google.com>
8076         * dumpfile.c (dump_loc): Output column number.
8077         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
8078         * doc/invoke.texi: Document optall -fopt-info flag.
8079         * profile.c (read_profile_edge_counts): Use new dump framework.
8080         (compute_branch_probabilities): Ditto.
8081         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8082         when pass not in any opt group.
8083         * pass_manager.h (pass_manager::get_pass_profile): New method.
8084         * value-prof.c (check_counter): Use new dump framework.
8085         (check_ic_target): Ditto.
8086         * coverage.c (get_coverage_counts): Ditto.
8087         (coverage_init): Setup new dump framework.
8089 2013-08-29  Richard Biener  <rguenther@suse.de>
8091         PR tree-optimization/58246
8092         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8093         handle the dominance check inside a basic-block.
8095 2013-08-29  Richard Biener  <rguenther@suse.de>
8097         PR middle-end/57287
8098         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8099         of default defs that appear in abnormal PHI nodes.
8101 2013-08-29  Richard Biener  <rguenther@suse.de>
8103         PR tree-optimization/57685
8104         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8105         single-use operands to avoid exponential complexity.
8107 2013-08-28  Dehao Chen  <dehao@google.com>
8109         * ipa-inline.c (edge_badness): Fix integer underflow.
8111 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8113         * gtm-builtins.def (_ITM_free): Declare leaf.
8115 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8117         PR target/58067
8118         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8119         (*tls_local_dynamic_base_64_largepic): Likewise.
8120         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8121         Remove predicate from call operand.
8122         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8123         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8125 2013-08-28  Jeff Law  <law@redhat.com>
8127         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8128         checks for the number of predecessors and successors allowed.
8129         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8130         which require copying a joiner block if there is a request which
8131         is a subpath that requires no joiner block copying.
8133 2013-08-28  Jan Hubicka  <jh@suse.cz>
8135         * lto-streamer-out.c (DFS_write_tree_body): Drop
8136         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8137         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8138         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8139         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8140         Do not read DECL_ERROR_ISSUED.
8141         (unpack_ts_decl_with_vis_value_fields): Do not read
8142         DECL_DEFER_OUTPUT.
8143         (lto_input_ts_binfo_tree_pointers): Do not read
8144         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8145         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8146         write DECL_ERROR_ISSUED..
8147         (pack_ts_decl_with_vis_value_fields): Do not write
8148         DECL_DEFER_OUTPUT.
8149         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8150         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8151         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8152         * tree.h (tree_decl_common): Update comment.
8153         (DECL_ERROR_ISSUED): Remove.
8155 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8157         PR middle-end/58257
8158         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8160 2013-08-28  Jan Hubicka  <jh@suse.cz>
8162         * builtins.def (free): Declare leaf.
8164 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8166         * gdbhooks.py: New.
8167         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8168         * configure: Regenerate.
8170 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8172         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8173         (ipa_ancestor_jf_data): Likewise.
8174         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8175         (ipa_get_jf_pass_through_type_preserved): New function.
8176         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8177         (ipa_get_jf_ancestor_type_preserved): New function.
8178         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8179         (ipa_get_jf_ancestor_result): Likewise.
8180         (propagate_vals_accross_pass_through): Use
8181         ipa_get_jf_pass_through_result to do all the value mappings.
8182         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8183         type_preserved flag.
8184         (ipa_set_jf_cst_copy): New function.
8185         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8186         (ipa_set_jf_arith_pass_through): Likewise.
8187         (ipa_set_ancestor_jf): Likewise.
8188         (compute_complex_assign_jump_func): Set type_preserved instead of
8189         punting.
8190         (ipa_compute_jump_functions_for_edge): Likewise.
8191         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8192         (update_jump_functions_after_inlining): Update type_preserved.
8193         Explicitely create jump functions when combining one with pass_through.
8194         (ipa_write_jump_function): Stream the type_preserved flags.
8195         (ipa_read_jump_function): Likewise.
8197 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8198             Aldy Hernandez  <aldyh@redhat.com>
8200         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8201         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8202         * function.h (struct function): Add has_force_vect_loops and
8203         has_simduid_loops.
8204         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8205         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8206         handle it.
8207         * gimple.def: Update CLAUSES comments.
8208         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8209         (gimple_build_omp_for): Add argument to prototype.
8210         (gimple_omp_for_kind): New.
8211         (gimple_omp_for_set_kind): New.
8212         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8213         GOVD_DATA_SHARE_CLASS.
8214         (enum omp_region_type): Add ORT_SIMD.
8215         (gimple_add_tmp_var): Handle ORT_SIMD.
8216         (gimplify_var_or_parm_decl): Same.
8217         (is_gimple_stmt): Same.
8218         (omp_firstprivatize_variable): Same.
8219         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8220         (omp_notice_variable): Handle ORT_SIMD.
8221         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8222         (omp_check_private): Handle ORT_SIMD.
8223         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8224         OMP_CLAUSE_SAFELEN.
8225         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8226         Handle OMP_CLAUSE_LASTPRIVATE.
8227         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8228         OMP_CLAUSE_SAFELEN.
8229         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8230         (gimplify_expr): Handle OMP_SIMD.
8231         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8232         (expand_GOMP_SIMD_VF): New.
8233         (expand_GOMP_SIMD_LAST_LANE): New.
8234         * internal-fn.def (GOMP_SIMD_LANE): New.
8235         (GOMP_SIMD_VF): New.
8236         (GOMP_SIMD_LAST_LANE): New.
8237         * omp-low.c: Include target.h.
8238         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8239         OMP_CLAUSE_SAFELEN.
8240         (check_omp_nesting_restrictions): Same.
8241         (omp_max_vf): New.
8242         (lower_rec_simd_input_clauses): New.
8243         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8244         OMP_CLAUSE_LINEAR.
8245         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8246         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8247         (expand_omp_build_assign): New.
8248         (expand_omp_for_init_counts): New.
8249         (expand_omp_for_init_vars): New.
8250         (extract_omp_for_update_vars): New.
8251         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8252         and rewrite accordingly.
8253         (expand_omp_simd): New.
8254         (expand_omp_for): Use expand_omp_simd.
8255         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8256         (lower_omp_for): Do not lower the body.
8257         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8258         in their own loops.
8259         * tree-flow.h (find_omp_clause): Remove prototype.
8260         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8261         forcing vectorization of the loop, or if flag_tree_vectorize.
8262         (gate_tree_if_conversion): Similarly.
8263         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8264         gimple_build_omp_for.
8265         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8266         * tree-parloops (create_parallel_loop): Pass kind argument to
8267         gimple_build_omp_for.
8268         * tree-pretty-print.c (dump_omp_clause): Add cases for
8269         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8270         OMP_CLAUSE__SIMDUID_.
8271         (dump_generic_node): Handle OMP_SIMD.
8272         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8273         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8274         unroll OMP_SIMD loops here.
8275         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8276         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8277         loop->safelen.
8278         (vect_analyze_data_refs): Handle simd loops.
8279         * tree-vect-loop.c (vectorizable_live_operation): Handle
8280         IFN_GOMP_SIMD*.
8281         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8282         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8283         (vectorizable_load): Same.
8284         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8285         (struct simduid_to_vf): New.
8286         (simduid_to_vf::hash): New.
8287         (simduid_to-vf::equal): New.
8288         (struct simd_array_to_simduid): New.
8289         (simd_array_to_simduid::hash): New.
8290         (simd_array_to_simduid::equal): New.
8291         (adjust_simduid_builtins): New.
8292         (struct note_simd_array_uses_struct): New.
8293         (note_simd_array_uses_cb): New.
8294         (note_simd_array_uses): New.
8295         (vectorize_loops): Handle simd hints and adjust simd builtins
8296         accordingly.
8297         * tree-vectorizer.h (struct _stmt_vec_info): Add
8298         simd_lane_access_p field.
8299         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8300         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8301         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8302         (omp_clause_code_name): Same.
8303         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8304         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8305         * tree.def (OMP_SIMD): New entry.
8306         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8307         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8308         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8309         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8310         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8311         (OMP_CLAUSE_LINEAR_STEP): New.
8312         (OMP_CLAUSE_SAFELEN_EXPR): New.
8313         (OMP_CLAUSE__SIMDUID__DECL): New.
8314         (find_omp_clause): New prototype.
8316 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8318         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8319         Haswell processor detection.
8321 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8323         PR target/57927
8324         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8325         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8326         AVX2 capable processors.
8328 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8330         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8331         of vget_low_* with implementations in terms of other intrinsics.
8333 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8335         PR middle-end/57219
8336         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8337         values to -1, 0 and 1.
8339 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8341         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8342         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8343         (<optab><mode>3_insn): Remove.
8344         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8345         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8346         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8347         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8348         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8349         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8350         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8351         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8352         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8353         (ror<mode>3_insn): Likewise.
8354         * config/aarch64/predicates.md (aarch64_simd_register): New.
8356 2013-08-27  Richard Biener  <rguenther@suse.de>
8358         PR tree-optimization/57521
8359         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8360         one edge is non-critical.
8361         (find_phi_replacement_condition): Make sure to use a non-critical
8362         edge.  Cleanup and remove old bug workarounds.
8363         (bb_postdominates_preds): Remove.
8364         (if_convertible_loop_p_1): Do not compute post-dominators.
8365         (combine_blocks): Do not free post-dominators.
8366         (main_tree_if_conversion): Likewise.
8367         (pass_data_if_conversion): Add TODO_verify_ssa.
8369 2013-08-27  DJ Delorie  <dj@redhat.com>
8371         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8373 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8375         * function.c (assign_parm_find_data_types): Set passed_mode and
8376         nominal_mode to the TYPE_MODE of nominal_type for the built
8377         pointer type in case of the struct-pass-by-reference.
8379 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8381         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8382         (UINT16_TYPE): Change default to "unsigned int".
8384         * config/avr/avr.opt (mfract-convert-truncate): New option.
8385         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8386         is set, round negative fractional integers according to n1169
8387         when converting to integer types.
8389 2013-08-26  Jan Hubicka  <jh@suse.cz>
8391         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8392         methods can not be called indirectly when their address is not taken.
8394 2013-08-26  Jan Hubicka  <jh@suse.cz>
8396         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8397         ctor_for_folding.
8399 2013-08-26  Jan Hubicka  <jh@suse.cz>
8401         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8402         can be unshared.
8404 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8406         * reload.c (find_valid_class): Allow classes that do not include
8407         FIRST_PSEUDO_REGISTER - 1.
8409 2013-08-26  Jan Hubicka  <jh@suse.cz>
8411         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8412         fix edge count/frequency when speculation failed; fix type check
8413         for the direct call.
8415 2013-08-26  Jan Hubicka  <jh@suse.cz>
8417         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8419 2013-08-26  Jan Hubicka  <jh@suse.cz>
8421         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8422         to be changed by edge redirection.
8424 2013-08-26  Jan Hubicka  <jh@suse.cz>
8426         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8427         formating; add sanity check.
8428         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8429         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8430         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8431         (resolve_noninline_speculation): Update callee keys, too.
8433 2013-08-26  Jan Hubicka  <jh@suse.cz>
8435         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8436         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8438 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8440         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8441         into proper place.
8443 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8445         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8446         (x86_64_elf_select_section): Ditto.
8447         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8448         arguments.
8449         (ix86_pass_by_reference): Ditto.
8450         (output_set_got): Ditto.
8451         (ix86_unary_operator_ok): Ditto.
8452         (ix86_expand_builtin): Ditto.
8454 2013-08-23  Jan Hubicka  <jh@suse.cz>
8456         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8458 2013-08-23  Jan Hubicka  <jh@suse.cz>
8460         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8461         (tree_decl_with_vis): Add FINAL field.
8463 2013-08-23  Jeff Law  <law@redhat.com>
8465         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8466         the debugging dump when the expression is fully redundant.
8468 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8470         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8471         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8472         * pretty-print.c (pp_formatted_text_data): Likewise.
8473         (pp_write_text_to_stream): Likewise.
8474         (pp_write_text_as_dot_label_to_stream): Likewise.
8475         (pp_append_r): Likewise.
8476         (pp_format): Likewise.
8477         (pp_flush): Likewise.
8478         (pp_clear_output_area): Likewise.
8479         (pp_append_text): Likewise.
8480         (pp_formatted_text): Likewise.
8481         (pp_remaining_character_count_for_line): Likewise.
8482         (pp_newline): Likewise.
8483         (pp_character): Likewise.
8484         (output_buffer::~output_buffer): Define.
8485         (pretty_printer::~pretty_printer): Destruct output buffer.
8486         * pretty-print.h (output_buffer::~output_buffer): Declare.
8487         (pretty_printer::~pretty_printer): Declare virtual.
8489 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8491         PR other/57324
8492         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8493         HOST_WIDE_INT_M1U): New macros.
8494         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8495         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8496         unsigned -1 for lshift.
8497         * cse.c (cse_insn): Likewise.
8498         * double-int.c (rshift_double, lshift_double): Likewise.
8499         * builtins.c (fold_builtin_bitop): Likewise.
8500         * combine.c (force_to_mode): Likewise.
8501         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8502         * simplify-rtx.c (simplify_const_unary_operation,
8503         simplify_const_binary_operation): Likewise.
8504         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8505         check_va_list_escapes): Likewise.
8506         * rtlanal.c (nonzero_bits1): Likewise.
8507         * expmed.c (expand_smod_pow2): Likewise.
8508         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8510 2013-08-23  Jan Hubicka  <jh@suse.cz>
8512         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8513         as having address taken.
8515 2013-08-23  Jan Hubicka  <jh@suse.cz>
8517         * ipa-utils.h (method_class_type): Declare.
8518         * ipa-devirt.c (method_class_type): Export.
8520         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8521         do not walk base classes of anonymous types.
8523 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8525         PR rtl-optimization/58220
8526         PR regression/58221
8527         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8528         handle SEQUENCE insns properly.
8530 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8532         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8533         definition.
8534         (pp_newline_and_indent): Likewise.
8535         (pp_separate_with): Likewise.
8536         * pretty-print.c (pp_newline_and_flush): Define.
8537         (pp_newline_and_indent): Likewise.
8538         (pp_separate_with): Likewise.
8540 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8542         PR target/58218
8543         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8544         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8546 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8548         * config/i386/predicates.md (ext_sse_reg_operand): New.
8549         * config/i386/i386.md (*movti_internal): Use
8550         predicate to determine if EVEX is needed.
8551         (*movsi_internal): Ditto.
8552         (*movdf_internal): Ditto.
8553         (*movsf_internal): Ditto.
8554         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8556 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8558         PR tree-optimization/58209
8559         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8560         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8561         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8562         emit POINTER_PLUS_EXPR.
8563         (create_tailcall_accumulator): For pointer result type accumulate in
8564         sizetype type.
8566 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8568         * configure.ac: Add backslashes missing from the last change.
8569         * configure: Regenerate.
8571 2013-08-22  Jan Hubicka  <jh@susue.cz>
8573         * ipa.c (function_and_variable_visibility): First remember function
8574         was global and then make it local.
8576 2013-08-22  Julian Brown  <julian@codesourcery.com>
8578         * configure.ac: Add aarch64 to list of arches which use "nop" in
8579         debug_line test.
8580         * configure: Regenerate.
8582 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8584         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8585         gnu_libc_has_function.
8586         * config/s390/tpf.h: Likewise.
8588 2013-08-22  Jan Hubicka  <jh@susue.cz>
8590         * timevar.c (validate_phases): Add cast.
8592 2013-08-22  Jan Hubicka  <jh@susue.cz>
8594         * timevar.c (validate_phases): Use size_t for memory.
8595         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8597 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8599         * pretty-print.h (output_buffer::output_buffer): Declare.
8600         (pretty_printer::pretty_printer): Likewise.
8601         (pp_construct): Remove.
8602         * pretty-print.c (output_buffer::output_buffer): Define.
8603         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8604         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8605         (print_gimple_expr): Likewise.
8606         (print_gimple_seq): Likewise.
8607         (gimple_dump_bb): Likewise.
8608         * sched-vis.c (dump_value_slim): Likewise.
8609         (dump_insn_slim): Likewise.
8610         (dump_rtl_slim): Likewise.
8611         (str_pattern_slim): Likewise.
8612         * tree-mudflap.c (mf_varname_tree): Likewise.
8613         * graph.c (print_graph_cfg): Likewise.
8614         (start_graph_dump): Likewise.
8615         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8616         placement-new.
8617         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8618         pretty printer initialization.
8619         * coretypes.h (diagnostic_context): Remove superflous type alias
8620         declaration.
8621         (pretty_printer): Likewise.  Declare directly as a class.
8622         (pretty_print_info): Remove declaration as class.
8623         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8624         and pp_clear_output_area.
8625         (asan_add_global): Likewise.
8627 2013-08-22  Jan Hubicka  <jh@suse.cz>
8629         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8630         * ipa-utils.h (update_type_inheritance_graph): Declare.
8631         (possible_polymorphic_call_target_p): Declare.
8632         (possible_polymorphic_call_target_p): New.
8633         * ipa-devirt.c: Update toplevel comments.
8634         (cached_polymorphic_call_targets): Move up.
8635         (odr_type_d): Move ID down.
8636         (polymorphic_type_binfo_p): Update comment.
8637         (odr_hasher::remove): Likewise;
8638         (get_odr_type): Set anonymous_namespace.
8639         (dump_odr_type): Dump it.
8640         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8641         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8642         (record_binfo): Add comment.
8643         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8644         built.
8645         (devirt_node_removal_hook): Do not iCE when cache is freed.
8646         (possible_polymorphic_call_target_p): New predicate.
8647         (update_type_inheritance_graph): New function.
8649 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8650             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8651             Sergey Lega  <sergey.s.lega@intel.com>
8652             Anna Tikhonova  <anna.tikhonova@intel.com>
8653             Ilya Tocar  <ilya.tocar@intel.com>
8654             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8655             Ilya Verbin  <ilya.verbin@intel.com>
8656             Kirill Yukhin  <kirill.yukhin@intel.com>
8657             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8659         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8660         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8661         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8662         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8663         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8664         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8665         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8666         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8667         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8668         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8669         OPT_mavx512pf, OPT_mavx512er cases.
8670         * config/i386/constraints.md (v): New constraint.
8671         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8672         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8673         (bit_AVX512CD): New.
8674         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8675         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8676         * config/i386/i386-c.c (ix86_target_macros_internal):
8677         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8678         __AVX512PF__.
8679         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8680         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8681         * config/i386/i386.c (regclass_map, dbx_register_map)
8682         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8683         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8684         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8685         -mavx512pf options.
8686         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8687         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8688         -mavx512pf options.  Fix formatting.
8689         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8690         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8691         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8692         -mavx512cd, -mavx512pf options.
8693         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8694         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8695         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8696         ALL_SSE_REGS.
8697         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8698         (ix86_set_reg_reg_cost): Ditto.
8699         (x86_order_regs_for_local_alloc): Ditto.
8700         (MAX_VECT_LEN): Extend to 64-byte.
8701         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8702         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8703         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8704         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8705         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8706         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8707         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8708         (SSE_REG_MODE_P): Support new modes.
8709         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8710         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8711         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8712         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8713         (REG_CLASS_CONTENTS): Add new registers.
8714         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8715         (EXT_REX_SSE_REGNO_P): New.
8716         (HI_REGISTER_NAMES): Add new registers.
8717         * config/i386/i386.md: Define constants for new registers.
8718         (mode): Add new 512-bit modes.
8719         (prefix): Support evex prefix.
8720         (isa): Support avx512f, noavx512f, fma_avx512f.
8721         (ssemodesuffix): Add new 512-bit modes.
8722         (movxi): New.
8723         (*movxi_internal_avx512f): Ditto.
8724         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8725         Support MODE_XI.
8726         (*movsi_internal): Likewise.
8727         (*movdf_internal): Likewise.
8728         (*movsf_internal): Likewise.
8729         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8730         (<code><mode>3): Likewise.
8731         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8732         New.
8733         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8734         with the new constraint "v".
8735         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8736         modes.
8737         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8738         with the new constraint "v".
8739         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8740         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8741         (*<plusminus_insn><mode>3): Likewise.
8742         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8743         (*mul<mode>3): Likewise.
8744         (<sse>_vmmul<mode>3): Likewise.
8745         (<sse>_div<mode>3): Likewise.
8746         (<sse>_vmdiv<mode>3): Likewise.
8747         (<sse>_sqrt<mode>2): Likewise.
8748         (<sse>_vmsqrt<mode>2): Likewise.
8749         (*<code><mode>3_finite): Likewise.
8750         (*<code><mode>3) <smaxmin>: Likewise.
8751         (<sse>_vm<code><mode>3): Likewise.
8752         (*<code><mode>3) <any_logic>: Likewise.
8753         (*fma_fmadd_<mode>): Likewise.
8754         (*fma_fmsub_<mode>): Likewise.
8755         (*fma_fnmadd_<mode>): Likewise.
8756         (*fma_fnmsub_<mode>): Likewise.
8757         (*fma_fmaddsub_<mode>): Likewise.
8758         (*fma_fmsubadd_<mode>): Likewise.
8759         (*fmai_fmadd_<mode>): Likewise.
8760         (*fmai_fmsub_<mode>): Likewise.
8761         (*fmai_fnmadd_<mode>): Likewise.
8762         (*fmai_fnmsub_<mode>): Likewise.
8763         (sse_cvtsi2ss): Likewise.
8764         (sse_cvtsi2ssq): Likewise.
8765         (sse_cvtss2si): Likewise.
8766         (sse_cvtss2si_2): Likewise.
8767         (sse_cvtss2siq): Likewise.
8768         (sse_cvtss2siq_2): Likewise.
8769         (sse_cvttss2si): Likewise.
8770         (sse_cvtss2siq_2): Likewise.
8771         (float<sseintvecmodelower><mode>2): Likewise.
8772         (sse2_cvtsd2si_2): Likewise.
8773         (sse2_cvtsd2siq_2): Likewise.
8774         (*<plusminus_insn><mode>3): Likewise.
8775         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8776         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8777         (ashr<mode>3): Likewise.
8778         (<shift_insn><mode>3): Likewise.
8779         (avx2_<code><mode>3): Likewise.
8780         (*avx2_<code><mode>3): Likewise.
8781         (*andnot<mode>3): Likewise.
8782         (*<code><mode>3) <any_logic>: Likewise.
8783         (abs<mode>2): Likewise.
8784         (avx2_permvar<mode>): Likewise.
8785         (avx2_perm<mode>_1): Likewise.
8786         (*avx_vpermilp<mode>): Likewise.
8787         (avx_vpermilvar<mode>3): Likewise.
8788         (avx2_ashrv<mode>): Likewise.
8789         (avx2_<shift_insn>v<mode>): Likewise.
8790         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8791         -mavx512cd.
8792         * doc/rtl.texi: Document XImode.
8794 2013-08-21  Jeff Law  <law@redhat.com>
8796         * tree-flow.h (register_jump_thread): Pass vector of edges
8797         instead of each important edge.
8798         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8799         thread path into a vector and pass that to register_jump_thread.
8800         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8801         passed in edge vector to the current 3-edge form.
8803         Revert:
8804         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8806         * dce.c (fini_dce): Call df_analyze again just in case
8807         delete_unmarked_insns removed anything.
8809 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8811         * reload.h (struct reg_equivs): Rename to ..
8812         (struct reg_equivs_s): .. this.
8814 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8816         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8818 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8820         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8822 2013-08-21  Jeff Law  <law@redhat.com>
8824         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8825         simplify assignments too.  If the RHS collapses to a singleton
8826         range, then return the value for the range.
8828 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8830         * config/i386/sse.md (V16): Rename to...
8831         (VMOVE): this.
8832         (mov<mode>): Update iterator name.
8833         (*mov<mode>_internal): Ditto.
8834         (push<mode>1): Ditto.
8835         (movmisalign<mode>): Ditto.
8837 2013-08-20  Jan Hubicka  <jh@suse.cz>
8839         PR bootstrap/58186
8840         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8841         entry for direct edges.
8842         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8844 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8846         Revert my last two changes, r201865 and r201864:
8848         Revert r201865:
8849         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8851         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8852         instances can own GC refs.
8854         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8855         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8856         (gcc::context::gt_pch_nx): Likewise.
8857         (gcc::context::gt_pch_nx):  Likewise.
8858         * ggc.h (gt_ggc_mx <T>): New.
8859         (gt_pch_nx_with_op <T>): New.
8860         (gt_pch_nx <T>): New.
8861         * passes.c (opt_pass::gt_ggc_mx): New.
8862         (opt_pass::gt_pch_nx): New.
8863         (opt_pass::gt_pch_nx_with_op): New.
8864         (pass_manager::gt_ggc_mx): New.
8865         (pass_manager::gt_pch_nx): New.
8866         (pass_manager::gt_pch_nx_with_op): New.
8867         (pass_manager::operator new): Use
8868         ggc_internal_cleared_alloc_stat rather than xcalloc.
8869         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8870         (pass_manager::gt_ggc_mx): New.
8871         (pass_manager::gt_pch_nx): New.
8872         (pass_manager::gt_pch_nx_with_op): New.
8873         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8874         (opt_pass::operator new): New.
8875         (opt_pass::gt_ggc_mx): New.
8876         (opt_pass::gt_pch_nx): New.
8877         (opt_pass::gt_pch_nx_with_op): New.
8879         Revert r201864:
8880         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8882         * Makefile.in (GTFILES): Add context.h.
8883         * context.c (gcc::context::operator new): New.
8884         (gcc::context::gt_ggc_mx): New.
8885         (gcc::context::gt_pch_nx): New.
8886         (gcc::context::gt_pch_nx): New.
8887         * context.h (gcc::context): Add GTY((user)) marking.
8888         (gcc::context::operator new): New.
8889         (gcc::context::gt_ggc_mx): New.
8890         (gcc::context::gt_pch_nx): New.
8891         (gcc::context::gt_pch_nx): New.
8892         (g): Add GTY marking.
8893         (gt_ggc_mx (gcc::context *)): New.
8894         (gt_pch_nx (gcc::context *)): New.
8895         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8896         void *cookie)): New.
8897         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8899 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8901         * dce.c (fini_dce): Call df_analyze again just in case
8902         delete_unmarked_insns removed anything.
8904 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8906         PR rtl-optimizations/57451
8907         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8908         from crossing split section boundaries.
8910 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8912         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8913         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8914         with MULTLIB_DEFAULTS.
8916 2013-08-20  Nick Clifton  <nickc@redhat.com>
8918         * target.def (narrow_volatile_bitfield): Note that the default
8919         value is false, not !TARGET_STRICT_ALIGN.
8920         * doc/tm.texi: Regenerate.
8922 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8924         Fix LIB_SPEC for systems without libpthread.
8926         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8927         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8928         for Android.
8929         * config/i386/linux-common.h: Likewise.
8930         * config/mips/linux-common.h: Likewise.
8932 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8934         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8935         checks.
8937 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8939         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8940         instances can own GC refs.
8942         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8943         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8944         (gcc::context::gt_pch_nx): Likewise.
8945         (gcc::context::gt_pch_nx):  Likewise.
8946         * ggc.h (gt_ggc_mx <T>): New.
8947         (gt_pch_nx_with_op <T>): New.
8948         (gt_pch_nx <T>): New.
8949         * passes.c (opt_pass::gt_ggc_mx): New.
8950         (opt_pass::gt_pch_nx): New.
8951         (opt_pass::gt_pch_nx_with_op): New.
8952         (pass_manager::gt_ggc_mx): New.
8953         (pass_manager::gt_pch_nx): New.
8954         (pass_manager::gt_pch_nx_with_op): New.
8955         (pass_manager::operator new): Use
8956         ggc_internal_cleared_alloc_stat rather than xcalloc.
8957         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8958         (pass_manager::gt_ggc_mx): New.
8959         (pass_manager::gt_pch_nx): New.
8960         (pass_manager::gt_pch_nx_with_op): New.
8961         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8962         (opt_pass::operator new): New.
8963         (opt_pass::gt_ggc_mx): New.
8964         (opt_pass::gt_pch_nx): New.
8965         (opt_pass::gt_pch_nx_with_op): New.
8967 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8969         * Makefile.in (GTFILES): Add context.h.
8970         * context.c (gcc::context::operator new): New.
8971         (gcc::context::gt_ggc_mx): New.
8972         (gcc::context::gt_pch_nx): New.
8973         (gcc::context::gt_pch_nx): New.
8974         * context.h (gcc::context): Add GTY((user)) marking.
8975         (gcc::context::operator new): New.
8976         (gcc::context::gt_ggc_mx): New.
8977         (gcc::context::gt_pch_nx): New.
8978         (gcc::context::gt_pch_nx): New.
8979         (g): Add GTY marking.
8980         (gt_ggc_mx (gcc::context *)): New.
8981         (gt_pch_nx (gcc::context *)): New.
8982         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8983         void *cookie)): New.
8984         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8986 2013-08-20  Alan Modra  <amodra@gmail.com>
8988         PR target/57865
8989         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8990         (rs6000_emit_epilogue): Likewise.
8992 2013-08-19  Dehao Chen  <dehao@google.com>
8994         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8996 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8997             Jakub Jelinek  <jakub@redhat.com>
8999         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
9000         (BUILT_IN_FABSD64): Likewise.
9001         (BUILT_IN_FABSD128): Likewise.
9002         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
9003         (fold_builtin_1): Likewise.
9004         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
9005         destination and source operands.
9006         (*abstd2_fpr): Likewise.
9007         (*nabstd2_fpr): Likewise.
9009 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
9011         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
9012         JUMP_P and INSN_P.
9014 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
9016         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
9017         incomplete.
9019 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9021         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
9022         * builtins.c (default_libc_has_function): New.
9023         (gnu_libc_has_function): Ditto.
9024         (no_c99_libc_has_function): Ditto.
9025         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
9026         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
9027         (fold_builtin_sincos): Likewise.
9028         (fold_builtin_cexp): Likewise.
9029         * builtins.def (DEF_C94_BUILTIN): Likewise.
9030         (DEF_C99_BUILTIN): Likewise.
9031         (DEF_C99_C90RES_BUILTIN): Likewise.
9032         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
9033         definitions to using this define.
9034         * config/darwin-protos.h (darwin_libc_has_function): New.
9035         * config/darwin.c (darwin_libc_has_function): Ditto.
9036         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
9037         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
9038         * config/darwin.h: Ditto.
9039         * config/elfos.h: Ditto.
9040         * config/freebsd.h: Ditto.
9041         * config/i386/cygming.h: Ditto.
9042         * config/i386/djgpp.h: Ditto.
9043         * config/i386/i386-interix.h: Ditto.
9044         * config/microblaze/microblaze.h: Ditto.
9045         * config/mmix/mmix.h: Ditto.
9046         * config/gnu-user.h: Ditto.
9047         * config/ia64/hpux.h: Ditto.
9048         * config/pa/pa-hpux.h: Ditto.
9049         * config/pdp11/pdp11.h: Ditto.
9050         * config/picochip/picochip.h: Ditto.
9051         * config/linux.h: Ditto.
9052         * config/netbsd.h: Ditto.
9053         * config/openbsd.h: Ditto.
9054         * config/rs6000/aix43.h: Ditto.
9055         * config/rs6000/aix51.h: Ditto.
9056         * config/rs6000/aix52.h: Ditto.
9057         * config/rs6000/aix53.h: Ditto.
9058         * config/rs6000/aix61.h: Ditto.
9059         * config/rs6000/darwin.h: Ditto.
9060         * config/rs6000/linux.h: Ditto.
9061         * config/rs6000/linux64.h: Ditto.
9062         * config/s390/tpf.h: Ditto.
9063         * config/sol2-10.h: Ditto.
9064         * config/sol2.h: Ditto.
9065         * config/vms/vms.h: Ditto.
9066         * config/vxworks.h: Ditto.
9067         * config/linux-android.c (linux_android_libc_has_function):
9068         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
9069         * config/linux-protos.h (linux_android_libc_has_function):
9070         New declaration.
9071         * config/i386/i386.c (ix86_libc_has_function): New.
9072         * config/i386/i386-protos.h
9073         (ix86_libc_has_function): New declaration.
9074         * config/i386/i386.md
9075         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
9076         ("isinf<mode>2): Likewise.
9077         * convert.c (convert_to_integer): Using new target hook
9078         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
9079         TARGET_C99_FUNCTIONS.
9080         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9081         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9082         * coretypes.h (function_class): New enum for different
9083         classes of functions.
9084         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9085         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9086         (TARGET_HAS_SINCOS): Likewise.
9087         (TARGET_LIBC_HAS_FUNCTION): New.
9088         * doc/tm.texi: Regenerated.
9089         * targhooks.h (default_libc_has_function): New declaration.
9090         (no_c99_libc_has_function): Ditto.
9091         (gnu_libc_has_function): Ditto.
9092         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9093         and TARGET_HAS_SINCOS.
9095 2013-08-18  Jan Hubicka  <jh@suse.cz>
9097         * Makeifle-in (ipa-devirt.o): New.
9098         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9099         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9100         (analyze_functions): Look into possible targets of polymorphic call.
9101         * dumpfile.c (dump_files): Add type-inheritance dump.
9102         * dumpfile.h (TDI_inheritance): New.
9103         * ipa-devirt.c: New file.
9104         * ipa-utils.h (odr_type_d): Forward declare.
9105         (odr_type): New type.
9106         (build_type_inheritance_graph): Declare.
9107         (possible_polymorphic_call_targets): Declare and introduce inline
9108         variant when only edge is pased.
9109         (dump_possible_polymorphic_call_targets): Likewise.
9110         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9111         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9112         (types_same_for_odr): ... here.
9113         * tree.h (type_in_anonymous_namespace_p): Declare.
9115 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9117         PR tree-optimization/58006
9118         * tree-parloops.c (take_address_of): Don't ICE if get_name
9119         returns NULL.
9120         (eliminate_local_variables_stmt): Remove clobber stmts.
9122 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9124         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9125         error message is issued for an alias to undefined symbol.
9127 2013-08-18  Jan Hubicka  <jh@suse.cz>
9129         * cgraph.c (cgraph_create_indirect_edge): Discover
9130         polymorphic calls and record basic info into indirect_info.
9131         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9132         devirtualization, ignore objc function calls.
9133         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9134         call with no parm index info.
9135         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9136         * tree.c (virtual_method_call_p): New function.
9137         * tree.h (virtual_method_call_p): Declare.
9139 2013-08-16  Jan Hubicka  <jh@suse.cz>
9141         PR middle-end/58179
9142         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9144 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9146         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9147         attribute.
9149 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9151         * gengtype.c (type_for_name): Add special-case support for
9152         locating types within the "gcc::" namespace.
9153         (open_base_files): Emit a "using namespace gcc" directive.
9155 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9157         PR target/58160
9158         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9159         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9161         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9162         array instead of each individual operand as a separate argument.
9163         (emit_fusion_gpr_load): Likewise.
9164         (expand_fusion_gpr_load): Add new function declaration.
9166         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9167         signature to have the operands passed as an array, instead of as
9168         separate arguments.  Allow ZERO_EXTEND to be in the memory
9169         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9170         depend on the register live/dead flags when peepholes are run.
9171         (expand_fusion_gpr_load): New function to be called from the
9172         peephole2 pass, to change the register that addis sets to be the
9173         target register.
9174         (emit_fusion_gpr_load): Change the calling signature to have the
9175         operands passed as an array, instead of as separate arguments.
9176         Allow ZERO_EXTEND to be in the memory address, and also
9177         SIGN_EXTEND if -mpower8-fusion-sign.
9179         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9180         unspec enumeration.
9181         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9182         adjust the register addis loads up in the peephole2 pass.  Do not
9183         depend on the register live/dead state when the peephole pass is done.
9185 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9187         * gengtype.c (create_user_defined_type): Ensure that the kind
9188         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9189         declaration is seen before the GTY((user)) marking.
9191 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9193         PR target/58105
9194         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9196 2013-08-16  Jan Hubicka  <jh@suse.cz>
9198         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9199         arugment expected_type.
9200         (gimple_fold_call): Use it.
9201         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9202         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9203         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9204         (try_make_edge_direct_virtual_call): Likewise.
9205         * tree.c (obj_type_ref_class): New.
9206         * tree.h (obj_type_ref_class): Use it.
9208 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9210         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9211         (rtl_slim_pp): Likewise.
9212         (init_rtl_slim_pretty_print): Likewise.
9213         (dump_value_slim):  Don't call it.  Use local pretty printer.
9214         (dump_insn_slim): Likewise.
9215         (dump_rtl_slim): Likewise.
9216         (str_pattern_slim): Likewise.
9217         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9218         Simplify.
9220 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9222         PR tree-optimization/58164
9223         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9224         walk gimple_goto_dest of GIMPLE_GOTO.
9226         PR tree-optimization/58165
9227         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9228         bi_call must be the last stmt in a bb, don't split_block, instead
9229         use fallthru edge from it and give up if there is none.
9230         Release conds vector when returning early.
9232 2013-08-14  Xinliang David Li  <davidxl@google.com>
9234         * config/i386/i386.c (ix86_option_override_internal):
9235         Remove unused variable and field.
9237 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9239         PR target/57949
9240         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9241         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9242         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9243         and Linux, correct BLKmode alignment when 128-bit alignment is
9244         required and compatibility flag is not set.
9245         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9246         for zero-size arguments when compatibility flag is not set.
9248 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9250         PR tree-optimization/58145
9251         * tree-sra.c (build_ref_for_offset): If prev_base has
9252         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9254 2013-08-14  Xinliang David Li  <davidxl@google.com>
9256         * config/i386/i386.c (ix86_option_override_internal):
9257         Fix uninitialized variable error.
9259 2013-08-14  Xinliang David Li  <davidxl@google.com>
9261         * config/i386/i386.opt: Define two new options.
9262         * config/i386/x86-tune.def: Add arch selector field in macros.
9263         * config/i386/i386.h: Adjust macro definition.
9264         * config/i386/i386.c (ix86_option_override_internal):
9265         Refactor the code.
9266         (parse_mtune_ctrl_str): New function.
9267         (set_ix86_tune_features): New function.
9268         (ix86_function_specific_restore): Call the new helper function.
9270 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9272         PR rtl-optimization/57662
9273         * sel-sched.c (code_motion_process_successors): When the current insn
9274         is removed after the recursive traversal, break from the loop.
9275         Add comments and debug printouts.
9277 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9278             Alexandre Oliva  <aoliva@redhat.com>
9280         PR target/58067
9281         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9282         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9283         there also UNSPEC_PLTOFF.
9285 2013-08-14  Marek Polacek  <polacek@redhat.com>
9287         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9288         NUM_CONDITIONS bit positions.
9290 2013-08-13  Cary Coutant  <ccoutant@google.com>
9292         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9293         (attr_checksum): Hash vector contents instead of pointer.
9294         (attr_checksum_ordered): Likewise.
9296 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9298         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9299         when Pmode != word_mode.  Add length_address attribute.
9300         (sse3_monitor_<mode>): Merge from sse3_monitor and
9301         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9302         Pmode != word_mode.  Update insn length attribute.
9303         * config/i386/i386.c (ix86_option_override_internal): Update
9304         ix86_gen_monitor selection for merged sse3_monitor insn.
9306 2013-08-13  Julian Brown  <julian@codesourcery.com>
9308         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9309         perform invalid legitimization on greater-than-word-size modes for
9310         TARGET_E500_DOUBLE.
9312 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9314         * ira.c (setup_class_translate_array): Use aclass instead of cl
9315         for classes not fully covered by allocno classes.
9317 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9319         PR tree-optimization/57661
9320         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9321         * tree-inline.c (tree_function_versioning): Initialize it.
9322         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9323         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9324         that is not being copied.
9326         PR sanitizer/56417
9327         * asan.c (instrument_strlen_call): Fix typo in comment.
9328         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9330 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9332         * config/mips/mips.md (prefetch): Use lw instead of ld on
9333         loongson in 32bit mode.
9335 2013-08-13  Nick Clifton  <nickc@redhat.com>
9337         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9339 2013-08-13  Jan Hubicka  <jh@suse.cz>
9341         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9342         introduced edge; fix typo in sanity check.
9343         (cgraph_resolve_speculation): Export; improve diagnostic.
9344         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9345         speculation at type mismatch.
9346         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9347         (cgraph_resolve_speculation): Declare.
9348         (symtab_can_be_discarded): New function.
9349         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9350         * ipa-inline-transform.c (speculation_removed): New global var.
9351         (clone_inlined_nodes): See if speculation can be removed.
9352         (inline_call): If speculations was removed, we growths may not match.
9353         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9354         (speculation_useful_p): New function.
9355         (resolve_noninline_speculation): New function.
9356         (inline_small_functions): Resolve useless speculations.
9357         * ipa-inline.h (speculation_useful_p): Declare
9358         * ipa.c (can_replace_by_local_alias): Simplify.
9359         (ipa_profile): Produce speculative calls in non-lto, too;
9360         add simple cost model; produce local aliases.
9362 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9364         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9365         PASS_MANAGER_H.
9367 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9369         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9371 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9373         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9374         iterator instead of X87MODEF.
9376 2013-08-12  Perez Read  <netfirewall@gmail.com>
9378         PR target/58132
9379         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9380         operand 0 for intel asm alternative.
9381         (*movabs<mode>_2): Ditto for operand 1.
9383 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9385         * config/aarch64/arm_none.h
9386         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9388 2013-08-12  Nick Clifton  <nickc@redhat.com>
9390         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9392 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9394         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9395         expand for QI/HImode operand to produce more effictive code for
9396         unsigned char(short) --> float(double) conversion.
9398 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9400         * doc/invoke.texi: Mention that -ftls-model does not force the final
9401         model.
9403 2013-08-12  Marek Polacek  <polacek@redhat.com>
9404             Marc Glisse  <marc.glisse@inria.fr>
9406         PR tree-optimization/57980
9407         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9408         when creating -1 constant.
9410 2013-08-10  Jan Hubicka  <jh@suse.cz>
9412         Workaround binutils PR14342.
9413         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9414         (gimple_init_edge_profiler): Likewise.
9415         (gimple_gen_ic_func_profiler): Likewise.
9417 2013-08-09  Jan Hubicka  <jh@suse.cz>
9419         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9421 2013-08-09  Xinliang David Li  <davidxl@google.com>
9423         * config/i386/stringop.def: New file.
9424         * config/i386/stringop.opt: New file.
9425         * config/i386/i386-opts.h: Include stringopt.def.
9426         * config/i386/i386.opt: Include stringopt.opt.
9427         * config/i386/i386.c (ix86_option_override_internal):
9428         Override default size based stringop inline strategies with options.
9429         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9430         New function.
9432 2013-08-09  Jan Hubicka  <jh@suse.cz>
9434         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9436 2013-08-09  Jan Hubicka  <jh@suse.cz>
9438         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9439         CGRAPH_FREQ_MAX.
9440         (dump_cgraph_node): Dump profile-id.
9441         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9442         and common_target_probability.
9443         * lto-cgraph.c (lto_output_edge): Stream common targets.
9444         (lto_output_node): Stream profile ids.
9445         (input_node): Stream profile ids.
9446         (input_edge): Stream common targets.
9447         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9448         * ipa.c: Include value-prof.h
9449         (ipa_profile_generate_summary): Turn indirect call statement histograms
9450         into common targets.
9451         (ipa_profile): Turn common targets into speculative edges.
9453 2013-08-09  Jan Hubicka  <jh@suse.cz>
9455         * cgraph.h (cgraph_node): Add profile_id.
9456         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9457         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9458         (del_node_map): Update.
9459         (find_func_by_funcdef_no): Replace by ...
9460         (find_func_by_profile_id): ... this one.
9461         (gimple_ic_transform): Do not remove useful histograms when
9462         speculation is not done; dump info when indirect call removal
9463         can happen at LTO.
9464         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9465         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9466         (__gcov_indirect_call_profiler_v2): .. this one.
9467         * profile.h (init_node_map): Update.
9468         * coverage.c (coverage_compute_profile_id): New function.
9469         * coverage.h (coverage_compute_profile_id): Declare.
9470         * tree-profile.c (init_ic_make_global_vars): Make
9471         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9472         (gimple_init_edge_profiler): Update prototype of
9473         __gcov_indirect_call_profiler.
9474         (gimple_gen_ic_func_profiler): Simplify.
9475         (tree_profiling): Use init_node_map
9477 2013-08-09  Jan Hubicka  <jh@suse.cz>
9479         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9480         non-speculative refs.
9481         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9482         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9483         (cgraph_set_call_stmt): Likewise.
9484         (cgraph_create_edge_1): Fix release checking compilatoin;
9485         clear lto_stmt_uid.
9486         (cgraph_free_edge): Free indirect info.
9487         (cgraph_turn_edge_to_speculative): New function.
9488         (cgraph_speculative_call_info): New function.
9489         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9490         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9491         (dump_cgraph_node): Dump speculation.
9492         (verify_edge_count_and_frequency): Accept speculative edges.
9493         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9494         (verify_cgraph_node): Handle speculation.
9495         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9496         (cgraph_set_call_stmt): Update prototype.
9497         (cgraph_make_edge_direct): Update prototype.
9498         (cgraph_speculative_call_info): Declare.
9499         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9500         to change; update call of ipa_find_references.
9501         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9502         lto_stmt_uid and speculative flags.
9503         (ipa_dump_references): Dump speculation.
9504         (ipa_clone_references): Clone speculative flag.
9505         (ipa_clone_referring): Likewise.
9506         (ipa_clone_ref): New function.
9507         (ipa_find_reference): Look into lto_stmt_uids
9508         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9509         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9510         (ipa_find_reference): Update declaration.
9511         (ipa_clone_ref): Declare.
9512         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9513         stream speculative flag.
9514         (lto_output_ref): Stream statements uids and speculation.
9515         (input_ref): Likewise.
9516         (input_edge): Stream speuclation.
9517         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9518         (cgraph_set_call_stmt_including_clones): Handle speculation.
9519         * ipa-inline.c (heap_edge_removal_hook): New function.
9520         (inline_small_functions): Register it.
9521         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9522         also initialize refs.
9523         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9524         edge to change.
9525         (try_make_edge_direct_simple_call): Likewise.
9526         (try_make_edge_direct_simple_call): Likewise.
9527         (update_indirect_edges_after_inlining): Likewise.
9528         (remove_described_reference): Look proper lto_stmt_uid.
9529         (propagate_controlled_uses): Likewise.
9530         (propagate_controlled_uses): Liekwise.
9531         * tree-inline.c (copy_bb): Copy speculative edges.
9532         (redirect_all_calls): New function.
9533         (copy_cfg_body): Do redirection after loop info is updated.
9534         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9536 2013-08-09  Jan Hubicka  <jh@suse.cz>
9538         * lto-streamer-out.c (output_function): Renumber PHIs.
9539         * lto-streamer-in.c (input_function): Likewise.
9541 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9543         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9544         (get_lane_unsigned): Likewise.
9545         (dup_lane_scalar): Likewise.
9546         (get_lane): enable for VALL.
9547         * config/aarch64/aarch64-simd.md
9548         (aarch64_dup_lane_scalar<mode>): Remove.
9549         (aarch64_get_lane_signed<mode>): Likewise.
9550         (aarch64_get_lane_unsigned<mode>): Likewise.
9551         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9552         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9553         (aarch64_get_lane<mode>): Enable for all vector modes.
9554         (aarch64_get_lanedi): Remove misleading constraints.
9555         * config/aarch64/arm_neon.h
9556         (__aarch64_vget_lane_any): Define.
9557         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9558         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9559         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9560         * config/aarch64/iterators.md (VDQQH): New.
9561         (VDQQHS): Likewise.
9562         (vwcore): Likewise.
9564 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9566         * configure.ac: Add GAS check for LEON instructions on SPARC.
9567         * configure: Regenerate.
9568         * config.in: Likewise.
9569         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9570         sparc*-*-* block.
9571         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9572         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9573         for LEON or LEON3.
9574         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9575         (AS_LEON_FLAG): New macro.
9576         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9577         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9578         Deal with LEON and LEON3 for the memory model.
9579         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9580         (atomic_compare_and_swap<mode>_1): Likewise.
9581         (*atomic_compare_and_swap<mode>_1): Likewise.
9583 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9585         * config/arm/neon.md (vcond): Fix floating-point vector
9586         comparisons against 0.
9588 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9590         * lra-constraints.c (emit_spill_move): Remove assert.
9591         (process_alt_operands): Add more debugging
9592         output.  Increase reject for spilling into memory.  Decrease
9593         reject for reloading scratch.
9594         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9596 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9598         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9599         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9600         micromips incompatible.  Add nan2008.
9601         (MULTILIB_DIRNAMES): Add nan2008.
9602         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9603         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9604         and micromips incompatible.  Add nan2008.
9605         (MULTILIB_DIRNAMES): Add nan2008.
9606         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9608 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9610         PR rtl-optimization/58079
9611         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9612         simplify_comparison has widened a comparison with an integer.
9614 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9616         * config/arm/neon.md (movmisalign<mode>): Disable when we
9617         don't allow unaligned accesses.
9618         (*movmisalign<mode>_neon_store): Likewise.
9619         (*movmisalign<mode>_neon_load): Likewise.
9620         (*movmisalign<mode>_neon_store): Likewise.
9621         (*movmisalign<mode>_neon_load): Likewise.
9623 2013-08-08  Jan Hubicka  <jh@suse.cz>
9625         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9626         (make_pass_rebuild_cgraph_edges): Also clear references.
9627         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9628         * ipa-inline-transform.c (inline_transform): Remove all references
9629         after inlining.
9630         * cgraphunit.c (expand_function): Remove all references after
9631         expansion.
9632         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9633         (ipa_find_reference): Rewrite to iterator.
9634         (remove_stmt_references): Likewise.
9635         (ipa_clear_stmts_in_references): New function.
9636         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9637         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9638         clear references.
9639         * ipa-split.c (split_function): Remove references in split function.
9641 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9643         PR target/57431
9644         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9645         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9647 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9649         PR target/56979
9650         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9651         suggested mode for the assignment isn't compatible with the
9652         registers required.
9654 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9656         PR target/58065
9657         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9659 2013-08-07  Xinliang David Li  <davidxl@google.com>
9661         * config/i386/i386.opt: New option -mtune-ctrl=.
9662         * config/i386/x86-tune.def: New file.
9663         * config/i386/i386.h: include x86-tune.def.
9664         * config/i386/i386.c (ix86_option_override_internal):
9665         Parsing -mtune-ctrl= option and set tune features.
9667 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9669         PR other/12081
9670         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9671         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9672         to gen_2arg_fn_t.
9674 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9676         * rtl.h (update_alignments): Declare.
9677         * final.c (grow_label_align): New function extracted from...
9678         (shorten_branches): ...here.  Call it.
9679         (update_alignments): New function.
9680         * reorg.c (sibling_labels): New variable.
9681         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9682         the new label along with it onto the sibling_labels vector.
9683         (fill_simple_delay_slots): Adjust call to get_label_before.
9684         (fill_slots_from_thread): Likewise.
9685         (relax_delay_slots): Likewise.
9686         (make_return_insns): Likewise.
9687         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9689 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9691         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9692         document its semantics.
9693         (diagnostic_report_diagnostic): Adjust accordingly.
9695 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9697         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9698         (sparc_option_override): ...and port to new C++ pass API.
9699         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9701 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9703         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9705 2013-08-06  Caroline Tice  <cmtice@google.com>
9707         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9708         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9709         * tree-pass.h: Add pass_vtable_verify.
9710         * varasm.c (assemble_variable): Add code to properly set the comdat
9711         section and name for the .vtable_map_vars section.
9712         (assemble_vtyv_preinit_initializer): New function.
9713         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9714         LINK_ONCE flag.
9715         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9716         * vtable-verify.c: New file.
9717         * vtable-verify.h: New file.
9718         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9719         initialiation levels.
9720         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9721         * passes.def: Insert pass_vtable_verify.
9722         * aclocal.m4: Reorder includes.
9723         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9724         -fvtv-counts options.
9725         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9726         as appropriate, if -fvtable-verify=... is used.
9727         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9728         -fvtable-verify=... is used.
9729         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9730         (vtable-verify.o): Add new build rule.
9731         (GTFILES): Add vtable-verify.c to list.
9732         * common.opt (fvtable-verify=): New flag.
9733         (vtv_priority): Values for fvtable-verify= flag.
9734         (fvtv-counts): New flag.
9735         (fvtv-debug): New flag.
9736         * tree.h (save_vtable_map_decl): New extern function decl.
9738 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9740         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9741         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9742         (pass_data_rl78_devirt): ...new pass_data instance and...
9743         (make_pass_rl78_devirt): ...new function.
9744         (rl78_asm_file_start): Port pass registration to new C++ API.
9746 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9748         * coretypes.h (rtl_opt_pass): Add.
9749         (gcc::context): Add.
9750         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9751         (epiphany_init): Port to new C++ pass API.
9752         (epiphany_optimize_mode_switching): Likewise.
9753         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9754         (pass_manager::get_pass_mode_switching): New.
9755         (pass_manager::get_pass_peephole2): New.
9756         * mode-switching.c (pass_mode_switching): Add clone method.
9757         * recog.c (pass_peephole2): Add clone method.
9758         (pass_split_all_insns): Add clone method.
9760 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9762         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9763         (mips_option_override): ...here, porting to new C++ API for passes.
9765 2013-08-06  Jan Hubicka  <jh@suse.cz>
9767         * cgraph.c (cgraph_get_body): New function based on lto.c
9768         implementation.
9769         * cgraph.h (cgraph_get_body): Declare.
9770         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9771         LTO paths.
9772         * cgraphunit.c (expand_function): Get body prior expanding.
9773         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9774         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9775         really need.
9776         * passes.c (do_per_function_toporder): Get body.
9777         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9778         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9780 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9782         PR fortran/57987
9783         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9784         is not re-finalized.  Rename second parameter to no_collect.
9786 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9788         PR middle-end/58041
9789         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9790         MEM_REF has proper alignment information.
9792 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9794         PR other/12081
9795         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9796         class insn_gen_fn.
9797         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9798         rtx (*) (rtx, ...) with insn_gen_fn.
9799         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9800         insn_gen_fn::stored_funcptr.  Add initializer braces.
9802 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9804         Rewrite how instances of passes are cloned to remove assumptions
9805         about their sizes (thus allowing pass subclasses to have
9806         additional data fields, albeit non-GC-managed ones at this point).
9808         * passes.c (make_pass_instance): Now that passes have clone
9809         methods, rewrite this function to eliminate XNEW and memcpy
9810         calls that used hardcoded sizes.  Since this function no longer
9811         creates pass instances, rename it to...
9812         (add_pass_instance): ...this.  Document the old way that passes were
9813         numbered and flagged, and rework this function to continue using it.
9814         (next_pass_1): Add an initial_pass argument for use by
9815         add_pass_instance.
9816         (position_pass): When adding multiple instances of a pass, use
9817         the pass's clone method, rather than relying on the XNEW/memcpy
9818         within the former make_pass_instance (now add_pass_instance).
9819         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9820         the initial instance of the current pass within the pass manager.
9822 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9824         This is the automated part of the conversion of passes from C
9825         structs to C++ classes.
9827         Patch autogenerated by refactor_passes.py from
9828         https://github.com/davidmalcolm/gcc-refactoring-scripts
9829         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9831         * asan.c (pass_asan): Convert from a global struct to a subclass of
9832         gimple_opt_pass along with...
9833         (pass_data_asan): ...new pass_data instance and...
9834         (make_pass_asan): ...new function.
9835         (pass_asan_O0): Convert from a global struct to a subclass of
9836         gimple_opt_pass along with...
9837         (pass_data_asan_O0): ...new pass_data instance and...
9838         (make_pass_asan_O0): ...new function.
9839         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9840         subclass of rtl_opt_pass along with...
9841         (pass_data_inc_dec): ...new pass_data instance and...
9842         (make_pass_inc_dec): ...new function.
9843         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9844         a subclass of rtl_opt_pass along with...
9845         (pass_data_reorder_blocks): ...new pass_data instance and...
9846         (make_pass_reorder_blocks): ...new function.
9847         (pass_duplicate_computed_gotos): Convert from a global struct to a
9848         subclass of rtl_opt_pass along with...
9849         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9850         (make_pass_duplicate_computed_gotos): ...new function.
9851         (pass_partition_blocks): Convert from a global struct to a subclass of
9852         rtl_opt_pass along with...
9853         (pass_data_partition_blocks): ...new pass_data instance and...
9854         (make_pass_partition_blocks): ...new function.
9855         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9856         struct to a subclass of rtl_opt_pass along with...
9857         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9858         and...
9859         (make_pass_branch_target_load_optimize1): ...new function.
9860         (pass_branch_target_load_optimize2): Convert from a global struct to a
9861         subclass of rtl_opt_pass along with...
9862         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9863         and...
9864         (make_pass_branch_target_load_optimize2): ...new function.
9865         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9866         of rtl_opt_pass along with...
9867         (pass_data_jump): ...new pass_data instance and...
9868         (make_pass_jump): ...new function.
9869         (pass_jump2): Convert from a global struct to a subclass of
9870         rtl_opt_pass along with...
9871         (pass_data_jump2): ...new pass_data instance and...
9872         (make_pass_jump2): ...new function.
9873         * cfgexpand.c (pass_expand): Convert from a global struct to a
9874         subclass of rtl_opt_pass along with...
9875         (pass_data_expand): ...new pass_data instance and...
9876         (make_pass_expand): ...new function.
9877         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9878         of rtl_opt_pass along with...
9879         (pass_data_free_cfg): ...new pass_data instance and...
9880         (make_pass_free_cfg): ...new function.
9881         (pass_into_cfg_layout_mode): Convert from a global struct to a
9882         subclass of rtl_opt_pass along with...
9883         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9884         (make_pass_into_cfg_layout_mode): ...new function.
9885         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9886         subclass of rtl_opt_pass along with...
9887         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9888         (make_pass_outof_cfg_layout_mode): ...new function.
9889         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9890         struct to a subclass of gimple_opt_pass along with...
9891         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9892         (make_pass_build_cgraph_edges): ...new function.
9893         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9894         subclass of gimple_opt_pass along with...
9895         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9896         (make_pass_rebuild_cgraph_edges): ...new function.
9897         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9898         subclass of gimple_opt_pass along with...
9899         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9900         and...
9901         (make_pass_remove_cgraph_callee_edges): ...new function.
9902         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9903         struct to a subclass of rtl_opt_pass along with...
9904         (pass_data_stack_adjustments): ...new pass_data instance and...
9905         (make_pass_stack_adjustments): ...new function.
9906         * combine.c (pass_combine): Convert from a global struct to a subclass
9907         of rtl_opt_pass along with...
9908         (pass_data_combine): ...new pass_data instance and...
9909         (make_pass_combine): ...new function.
9910         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9911         global struct to a subclass of rtl_opt_pass along with...
9912         (pass_data_compare_elim_after_reload): ...new pass_data instance
9913         and...
9914         (make_pass_compare_elim_after_reload): ...new function.
9915         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9916         of rtl_opt_pass along with...
9917         (pass_data_rtl_cprop): ...new pass_data instance and...
9918         (make_pass_rtl_cprop): ...new function.
9919         * cse.c (pass_cse): Convert from a global struct to a subclass of
9920         rtl_opt_pass along with...
9921         (pass_data_cse): ...new pass_data instance and...
9922         (make_pass_cse): ...new function.
9923         (pass_cse2): Convert from a global struct to a subclass of
9924         rtl_opt_pass along with...
9925         (pass_data_cse2): ...new pass_data instance and...
9926         (make_pass_cse2): ...new function.
9927         (pass_cse_after_global_opts): Convert from a global struct to a
9928         subclass of rtl_opt_pass along with...
9929         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9930         (make_pass_cse_after_global_opts): ...new function.
9931         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9932         of rtl_opt_pass along with...
9933         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9934         (make_pass_ud_rtl_dce): ...new function.
9935         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9936         rtl_opt_pass along with...
9937         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9938         (make_pass_fast_rtl_dce): ...new function.
9939         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9940         a subclass of rtl_opt_pass along with...
9941         (pass_data_df_initialize_opt): ...new pass_data instance and...
9942         (make_pass_df_initialize_opt): ...new function.
9943         (pass_df_initialize_no_opt): Convert from a global struct to a
9944         subclass of rtl_opt_pass along with...
9945         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9946         (make_pass_df_initialize_no_opt): ...new function.
9947         (pass_df_finish): Convert from a global struct to a subclass of
9948         rtl_opt_pass along with...
9949         (pass_data_df_finish): ...new pass_data instance and...
9950         (make_pass_df_finish): ...new function.
9951         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9952         rtl_opt_pass along with...
9953         (pass_data_rtl_dse1): ...new pass_data instance and...
9954         (make_pass_rtl_dse1): ...new function.
9955         (pass_rtl_dse2): Convert from a global struct to a subclass of
9956         rtl_opt_pass along with...
9957         (pass_data_rtl_dse2): ...new pass_data instance and...
9958         (make_pass_rtl_dse2): ...new function.
9959         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9960         subclass of rtl_opt_pass along with...
9961         (pass_data_dwarf2_frame): ...new pass_data instance and...
9962         (make_pass_dwarf2_frame): ...new function.
9963         * except.c (pass_set_nothrow_function_flags): Convert from a global
9964         struct to a subclass of rtl_opt_pass along with...
9965         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9966         and...
9967         (make_pass_set_nothrow_function_flags): ...new function.
9968         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9969         subclass of rtl_opt_pass along with...
9970         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9971         and...
9972         (make_pass_convert_to_eh_region_ranges): ...new function.
9973         * final.c (pass_compute_alignments): Convert from a global struct to a
9974         subclass of rtl_opt_pass along with...
9975         (pass_data_compute_alignments): ...new pass_data instance and...
9976         (make_pass_compute_alignments): ...new function.
9977         (pass_final): Convert from a global struct to a subclass of
9978         rtl_opt_pass along with...
9979         (pass_data_final): ...new pass_data instance and...
9980         (make_pass_final): ...new function.
9981         (pass_shorten_branches): Convert from a global struct to a subclass of
9982         rtl_opt_pass along with...
9983         (pass_data_shorten_branches): ...new pass_data instance and...
9984         (make_pass_shorten_branches): ...new function.
9985         (pass_clean_state): Convert from a global struct to a subclass of
9986         rtl_opt_pass along with...
9987         (pass_data_clean_state): ...new pass_data instance and...
9988         (make_pass_clean_state): ...new function.
9989         * function.c (pass_instantiate_virtual_regs): Convert from a global
9990         struct to a subclass of rtl_opt_pass along with...
9991         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9992         (make_pass_instantiate_virtual_regs): ...new function.
9993         (pass_leaf_regs): Convert from a global struct to a subclass of
9994         rtl_opt_pass along with...
9995         (pass_data_leaf_regs): ...new pass_data instance and...
9996         (make_pass_leaf_regs): ...new function.
9997         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9998         subclass of rtl_opt_pass along with...
9999         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
10000         and...
10001         (make_pass_thread_prologue_and_epilogue): ...new function.
10002         (pass_match_asm_constraints): Convert from a global struct to a
10003         subclass of rtl_opt_pass along with...
10004         (pass_data_match_asm_constraints): ...new pass_data instance and...
10005         (make_pass_match_asm_constraints): ...new function.
10006         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
10007         subclass of rtl_opt_pass along with...
10008         (pass_data_rtl_fwprop): ...new pass_data instance and...
10009         (make_pass_rtl_fwprop): ...new function.
10010         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
10011         rtl_opt_pass along with...
10012         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
10013         (make_pass_rtl_fwprop_addr): ...new function.
10014         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
10015         rtl_opt_pass along with...
10016         (pass_data_rtl_pre): ...new pass_data instance and...
10017         (make_pass_rtl_pre): ...new function.
10018         (pass_rtl_hoist): Convert from a global struct to a subclass of
10019         rtl_opt_pass along with...
10020         (pass_data_rtl_hoist): ...new pass_data instance and...
10021         (make_pass_rtl_hoist): ...new function.
10022         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
10023         subclass of gimple_opt_pass along with...
10024         (pass_data_lower_cf): ...new pass_data instance and...
10025         (make_pass_lower_cf): ...new function.
10026         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
10027         from a global struct to a subclass of gimple_opt_pass along with...
10028         (pass_data_strength_reduction): ...new pass_data instance and...
10029         (make_pass_strength_reduction): ...new function.
10030         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
10031         of rtl_opt_pass along with...
10032         (pass_data_rtl_ifcvt): ...new pass_data instance and...
10033         (make_pass_rtl_ifcvt): ...new function.
10034         (pass_if_after_combine): Convert from a global struct to a subclass of
10035         rtl_opt_pass along with...
10036         (pass_data_if_after_combine): ...new pass_data instance and...
10037         (make_pass_if_after_combine): ...new function.
10038         (pass_if_after_reload): Convert from a global struct to a subclass of
10039         rtl_opt_pass along with...
10040         (pass_data_if_after_reload): ...new pass_data instance and...
10041         (make_pass_if_after_reload): ...new function.
10042         * init-regs.c (pass_initialize_regs): Convert from a global struct to
10043         a subclass of rtl_opt_pass along with...
10044         (pass_data_initialize_regs): ...new pass_data instance and...
10045         (make_pass_initialize_regs): ...new function.
10046         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
10047         of ipa_opt_pass_d along with...
10048         (pass_data_ipa_cp): ...new pass_data instance and...
10049         (make_pass_ipa_cp): ...new function.
10050         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
10051         global struct to a subclass of gimple_opt_pass along with...
10052         (pass_data_inline_parameters): ...new pass_data instance and...
10053         (make_pass_inline_parameters): ...new function.
10054         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
10055         subclass of gimple_opt_pass along with...
10056         (pass_data_early_inline): ...new pass_data instance and...
10057         (make_pass_early_inline): ...new function.
10058         (pass_ipa_inline): Convert from a global struct to a subclass of
10059         ipa_opt_pass_d along with...
10060         (pass_data_ipa_inline): ...new pass_data instance and...
10061         (make_pass_ipa_inline): ...new function.
10062         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
10063         struct to a subclass of gimple_opt_pass along with...
10064         (pass_data_local_pure_const): ...new pass_data instance and...
10065         (make_pass_local_pure_const): ...new function.
10066         (pass_ipa_pure_const): Convert from a global struct to a subclass of
10067         ipa_opt_pass_d along with...
10068         (pass_data_ipa_pure_const): ...new pass_data instance and...
10069         (make_pass_ipa_pure_const): ...new function.
10070         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
10071         to a subclass of ipa_opt_pass_d along with...
10072         (pass_data_ipa_reference): ...new pass_data instance and...
10073         (make_pass_ipa_reference): ...new function.
10074         * ipa-split.c (pass_split_functions): Convert from a global struct to
10075         a subclass of gimple_opt_pass along with...
10076         (pass_data_split_functions): ...new pass_data instance and...
10077         (make_pass_split_functions): ...new function.
10078         (pass_feedback_split_functions): Convert from a global struct to a
10079         subclass of gimple_opt_pass along with...
10080         (pass_data_feedback_split_functions): ...new pass_data instance and...
10081         (make_pass_feedback_split_functions): ...new function.
10082         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10083         global struct to a subclass of simple_ipa_opt_pass along with...
10084         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10085         instance and...
10086         (make_pass_ipa_function_and_variable_visibility): ...new function.
10087         (pass_ipa_free_inline_summary): Convert from a global struct to a
10088         subclass of simple_ipa_opt_pass along with...
10089         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10090         (make_pass_ipa_free_inline_summary): ...new function.
10091         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10092         subclass of ipa_opt_pass_d along with...
10093         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10094         and...
10095         (make_pass_ipa_whole_program_visibility): ...new function.
10096         (pass_ipa_profile): Convert from a global struct to a subclass of
10097         ipa_opt_pass_d along with...
10098         (pass_data_ipa_profile): ...new pass_data instance and...
10099         (make_pass_ipa_profile): ...new function.
10100         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10101         ipa_opt_pass_d along with...
10102         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10103         (make_pass_ipa_cdtor_merge): ...new function.
10104         * ira.c (pass_ira): Convert from a global struct to a subclass of
10105         rtl_opt_pass along with...
10106         (pass_data_ira): ...new pass_data instance and...
10107         (make_pass_ira): ...new function.
10108         (pass_reload): Convert from a global struct to a subclass of
10109         rtl_opt_pass along with...
10110         (pass_data_reload): ...new pass_data instance and...
10111         (make_pass_reload): ...new function.
10112         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10113         subclass of rtl_opt_pass along with...
10114         (pass_data_cleanup_barriers): ...new pass_data instance and...
10115         (make_pass_cleanup_barriers): ...new function.
10116         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10117         of rtl_opt_pass along with...
10118         (pass_data_loop2): ...new pass_data instance and...
10119         (make_pass_loop2): ...new function.
10120         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10121         rtl_opt_pass along with...
10122         (pass_data_rtl_loop_init): ...new pass_data instance and...
10123         (make_pass_rtl_loop_init): ...new function.
10124         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10125         rtl_opt_pass along with...
10126         (pass_data_rtl_loop_done): ...new pass_data instance and...
10127         (make_pass_rtl_loop_done): ...new function.
10128         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10129         subclass of rtl_opt_pass along with...
10130         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10131         (make_pass_rtl_move_loop_invariants): ...new function.
10132         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10133         rtl_opt_pass along with...
10134         (pass_data_rtl_unswitch): ...new pass_data instance and...
10135         (make_pass_rtl_unswitch): ...new function.
10136         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10137         subclass of rtl_opt_pass along with...
10138         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10139         and...
10140         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10141         (pass_rtl_doloop): Convert from a global struct to a subclass of
10142         rtl_opt_pass along with...
10143         (pass_data_rtl_doloop): ...new pass_data instance and...
10144         (make_pass_rtl_doloop): ...new function.
10145         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10146         a subclass of rtl_opt_pass along with...
10147         (pass_data_lower_subreg): ...new pass_data instance and...
10148         (make_pass_lower_subreg): ...new function.
10149         (pass_lower_subreg2): Convert from a global struct to a subclass of
10150         rtl_opt_pass along with...
10151         (pass_data_lower_subreg2): ...new pass_data instance and...
10152         (make_pass_lower_subreg2): ...new function.
10153         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10154         struct to a subclass of ipa_opt_pass_d along with...
10155         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10156         (make_pass_ipa_lto_gimple_out): ...new function.
10157         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10158         of ipa_opt_pass_d along with...
10159         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10160         (make_pass_ipa_lto_finish_out): ...new function.
10161         * mode-switching.c (pass_mode_switching): Convert from a global struct
10162         to a subclass of rtl_opt_pass along with...
10163         (pass_data_mode_switching): ...new pass_data instance and...
10164         (make_pass_mode_switching): ...new function.
10165         * modulo-sched.c (pass_sms): Convert from a global struct to a
10166         subclass of rtl_opt_pass along with...
10167         (pass_data_sms): ...new pass_data instance and...
10168         (make_pass_sms): ...new function.
10169         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10170         subclass of gimple_opt_pass along with...
10171         (pass_data_expand_omp): ...new pass_data instance and...
10172         (make_pass_expand_omp): ...new function.
10173         (pass_lower_omp): Convert from a global struct to a subclass of
10174         gimple_opt_pass along with...
10175         (pass_data_lower_omp): ...new pass_data instance and...
10176         (make_pass_lower_omp): ...new function.
10177         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10178         of gimple_opt_pass along with...
10179         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10180         (make_pass_diagnose_omp_blocks): ...new function.
10181         * passes.c (pass_early_local_passes): Convert from a global struct to
10182         a subclass of simple_ipa_opt_pass along with...
10183         (pass_data_early_local_passes): ...new pass_data instance and...
10184         (make_pass_early_local_passes): ...new function.
10185         (pass_all_early_optimizations): Convert from a global struct to a
10186         subclass of gimple_opt_pass along with...
10187         (pass_data_all_early_optimizations): ...new pass_data instance and...
10188         (make_pass_all_early_optimizations): ...new function.
10189         (pass_all_optimizations): Convert from a global struct to a subclass
10190         of gimple_opt_pass along with...
10191         (pass_data_all_optimizations): ...new pass_data instance and...
10192         (make_pass_all_optimizations): ...new function.
10193         (pass_all_optimizations_g): Convert from a global struct to a subclass
10194         of gimple_opt_pass along with...
10195         (pass_data_all_optimizations_g): ...new pass_data instance and...
10196         (make_pass_all_optimizations_g): ...new function.
10197         (pass_rest_of_compilation): Convert from a global struct to a subclass
10198         of rtl_opt_pass along with...
10199         (pass_data_rest_of_compilation): ...new pass_data instance and...
10200         (make_pass_rest_of_compilation): ...new function.
10201         (pass_postreload): Convert from a global struct to a subclass of
10202         rtl_opt_pass along with...
10203         (pass_data_postreload): ...new pass_data instance and...
10204         (make_pass_postreload): ...new function.
10205         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10206         subclass of rtl_opt_pass along with...
10207         (pass_data_gcse2): ...new pass_data instance and...
10208         (make_pass_gcse2): ...new function.
10209         * postreload.c (pass_postreload_cse): Convert from a global struct to
10210         a subclass of rtl_opt_pass along with...
10211         (pass_data_postreload_cse): ...new pass_data instance and...
10212         (make_pass_postreload_cse): ...new function.
10213         * predict.c (pass_profile): Convert from a global struct to a subclass
10214         of gimple_opt_pass along with...
10215         (pass_data_profile): ...new pass_data instance and...
10216         (make_pass_profile): ...new function.
10217         (pass_strip_predict_hints): Convert from a global struct to a subclass
10218         of gimple_opt_pass along with...
10219         (pass_data_strip_predict_hints): ...new pass_data instance and...
10220         (make_pass_strip_predict_hints): ...new function.
10221         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10222         of rtl_opt_pass along with...
10223         (pass_data_peephole2): ...new pass_data instance and...
10224         (make_pass_peephole2): ...new function.
10225         (pass_split_all_insns): Convert from a global struct to a subclass of
10226         rtl_opt_pass along with...
10227         (pass_data_split_all_insns): ...new pass_data instance and...
10228         (make_pass_split_all_insns): ...new function.
10229         (pass_split_after_reload): Convert from a global struct to a subclass
10230         of rtl_opt_pass along with...
10231         (pass_data_split_after_reload): ...new pass_data instance and...
10232         (make_pass_split_after_reload): ...new function.
10233         (pass_split_before_regstack): Convert from a global struct to a
10234         subclass of rtl_opt_pass along with...
10235         (pass_data_split_before_regstack): ...new pass_data instance and...
10236         (make_pass_split_before_regstack): ...new function.
10237         (pass_split_before_sched2): Convert from a global struct to a subclass
10238         of rtl_opt_pass along with...
10239         (pass_data_split_before_sched2): ...new pass_data instance and...
10240         (make_pass_split_before_sched2): ...new function.
10241         (pass_split_for_shorten_branches): Convert from a global struct to a
10242         subclass of rtl_opt_pass along with...
10243         (pass_data_split_for_shorten_branches): ...new pass_data instance
10244         and...
10245         (make_pass_split_for_shorten_branches): ...new function.
10246         * ree.c (pass_ree): Convert from a global struct to a subclass of
10247         rtl_opt_pass along with...
10248         (pass_data_ree): ...new pass_data instance and...
10249         (make_pass_ree): ...new function.
10250         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10251         subclass of rtl_opt_pass along with...
10252         (pass_data_stack_regs): ...new pass_data instance and...
10253         (make_pass_stack_regs): ...new function.
10254         (pass_stack_regs_run): Convert from a global struct to a subclass of
10255         rtl_opt_pass along with...
10256         (pass_data_stack_regs_run): ...new pass_data instance and...
10257         (make_pass_stack_regs_run): ...new function.
10258         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10259         subclass of rtl_opt_pass along with...
10260         (pass_data_cprop_hardreg): ...new pass_data instance and...
10261         (make_pass_cprop_hardreg): ...new function.
10262         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10263         subclass of rtl_opt_pass along with...
10264         (pass_data_reginfo_init): ...new pass_data instance and...
10265         (make_pass_reginfo_init): ...new function.
10266         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10267         of rtl_opt_pass along with...
10268         (pass_data_regmove): ...new pass_data instance and...
10269         (make_pass_regmove): ...new function.
10270         * regrename.c (pass_regrename): Convert from a global struct to a
10271         subclass of rtl_opt_pass along with...
10272         (pass_data_regrename): ...new pass_data instance and...
10273         (make_pass_regrename): ...new function.
10274         * reorg.c (pass_delay_slots): Convert from a global struct to a
10275         subclass of rtl_opt_pass along with...
10276         (pass_data_delay_slots): ...new pass_data instance and...
10277         (make_pass_delay_slots): ...new function.
10278         (pass_machine_reorg): Convert from a global struct to a subclass of
10279         rtl_opt_pass along with...
10280         (pass_data_machine_reorg): ...new pass_data instance and...
10281         (make_pass_machine_reorg): ...new function.
10282         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10283         of rtl_opt_pass along with...
10284         (pass_data_sched): ...new pass_data instance and...
10285         (make_pass_sched): ...new function.
10286         (pass_sched2): Convert from a global struct to a subclass of
10287         rtl_opt_pass along with...
10288         (pass_data_sched2): ...new pass_data instance and...
10289         (make_pass_sched2): ...new function.
10290         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10291         to a subclass of rtl_opt_pass along with...
10292         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10293         (make_pass_stack_ptr_mod): ...new function.
10294         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10295         to a subclass of rtl_opt_pass along with...
10296         (pass_data_rtl_store_motion): ...new pass_data instance and...
10297         (make_pass_rtl_store_motion): ...new function.
10298         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10299         of gimple_opt_pass along with...
10300         (pass_data_tracer): ...new pass_data instance and...
10301         (make_pass_tracer): ...new function.
10302         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10303         to a subclass of gimple_opt_pass along with...
10304         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10305         (make_pass_diagnose_tm_blocks): ...new function.
10306         (pass_lower_tm): Convert from a global struct to a subclass of
10307         gimple_opt_pass along with...
10308         (pass_data_lower_tm): ...new pass_data instance and...
10309         (make_pass_lower_tm): ...new function.
10310         (pass_tm_init): Convert from a global struct to a subclass of
10311         gimple_opt_pass along with...
10312         (pass_data_tm_init): ...new pass_data instance and...
10313         (make_pass_tm_init): ...new function.
10314         (pass_tm_mark): Convert from a global struct to a subclass of
10315         gimple_opt_pass along with...
10316         (pass_data_tm_mark): ...new pass_data instance and...
10317         (make_pass_tm_mark): ...new function.
10318         (pass_tm_edges): Convert from a global struct to a subclass of
10319         gimple_opt_pass along with...
10320         (pass_data_tm_edges): ...new pass_data instance and...
10321         (make_pass_tm_edges): ...new function.
10322         (pass_tm_memopt): Convert from a global struct to a subclass of
10323         gimple_opt_pass along with...
10324         (pass_data_tm_memopt): ...new pass_data instance and...
10325         (make_pass_tm_memopt): ...new function.
10326         (pass_ipa_tm): Convert from a global struct to a subclass of
10327         simple_ipa_opt_pass along with...
10328         (pass_data_ipa_tm): ...new pass_data instance and...
10329         (make_pass_ipa_tm): ...new function.
10330         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10331         subclass of gimple_opt_pass along with...
10332         (pass_data_call_cdce): ...new pass_data instance and...
10333         (make_pass_call_cdce): ...new function.
10334         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10335         subclass of gimple_opt_pass along with...
10336         (pass_data_build_cfg): ...new pass_data instance and...
10337         (make_pass_build_cfg): ...new function.
10338         (pass_split_crit_edges): Convert from a global struct to a subclass of
10339         gimple_opt_pass along with...
10340         (pass_data_split_crit_edges): ...new pass_data instance and...
10341         (make_pass_split_crit_edges): ...new function.
10342         (pass_warn_function_return): Convert from a global struct to a
10343         subclass of gimple_opt_pass along with...
10344         (pass_data_warn_function_return): ...new pass_data instance and...
10345         (make_pass_warn_function_return): ...new function.
10346         (pass_warn_function_noreturn): Convert from a global struct to a
10347         subclass of gimple_opt_pass along with...
10348         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10349         (make_pass_warn_function_noreturn): ...new function.
10350         (pass_warn_unused_result): Convert from a global struct to a subclass
10351         of gimple_opt_pass along with...
10352         (pass_data_warn_unused_result): ...new pass_data instance and...
10353         (make_pass_warn_unused_result): ...new function.
10354         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10355         a subclass of gimple_opt_pass along with...
10356         (pass_data_merge_phi): ...new pass_data instance and...
10357         (make_pass_merge_phi): ...new function.
10358         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10359         a subclass of gimple_opt_pass along with...
10360         (pass_data_lower_complex): ...new pass_data instance and...
10361         (make_pass_lower_complex): ...new function.
10362         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10363         gimple_opt_pass along with...
10364         (pass_data_lower_complex_O0): ...new pass_data instance and...
10365         (make_pass_lower_complex_O0): ...new function.
10366         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10367         subclass of gimple_opt_pass along with...
10368         (pass_data_lower_eh): ...new pass_data instance and...
10369         (make_pass_lower_eh): ...new function.
10370         (pass_refactor_eh): Convert from a global struct to a subclass of
10371         gimple_opt_pass along with...
10372         (pass_data_refactor_eh): ...new pass_data instance and...
10373         (make_pass_refactor_eh): ...new function.
10374         (pass_lower_resx): Convert from a global struct to a subclass of
10375         gimple_opt_pass along with...
10376         (pass_data_lower_resx): ...new pass_data instance and...
10377         (make_pass_lower_resx): ...new function.
10378         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10379         of gimple_opt_pass along with...
10380         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10381         (make_pass_lower_eh_dispatch): ...new function.
10382         (pass_cleanup_eh): Convert from a global struct to a subclass of
10383         gimple_opt_pass along with...
10384         (pass_data_cleanup_eh): ...new pass_data instance and...
10385         (make_pass_cleanup_eh): ...new function.
10386         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10387         to a subclass of simple_ipa_opt_pass along with...
10388         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10389         (make_pass_ipa_lower_emutls): ...new function.
10390         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10391         a subclass of gimple_opt_pass along with...
10392         (pass_data_if_conversion): ...new pass_data instance and...
10393         (make_pass_if_conversion): ...new function.
10394         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10395         subclass of gimple_opt_pass along with...
10396         (pass_data_build_ssa): ...new pass_data instance and...
10397         (make_pass_build_ssa): ...new function.
10398         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10399         global struct to a subclass of gimple_opt_pass along with...
10400         (pass_data_loop_distribution): ...new pass_data instance and...
10401         (make_pass_loop_distribution): ...new function.
10402         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10403         subclass of gimple_opt_pass along with...
10404         (pass_data_mudflap_1): ...new pass_data instance and...
10405         (make_pass_mudflap_1): ...new function.
10406         (pass_mudflap_2): Convert from a global struct to a subclass of
10407         gimple_opt_pass along with...
10408         (pass_data_mudflap_2): ...new pass_data instance and...
10409         (make_pass_mudflap_2): ...new function.
10410         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10411         subclass of gimple_opt_pass along with...
10412         (pass_data_mudflap_1): ...new pass_data instance and...
10413         (make_pass_mudflap_1): ...new function.
10414         (pass_mudflap_2): Convert from a global struct to a subclass of
10415         gimple_opt_pass along with...
10416         (pass_data_mudflap_2): ...new pass_data instance and...
10417         (make_pass_mudflap_2): ...new function.
10418         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10419         gimple_opt_pass along with...
10420         (pass_data_nrv): ...new pass_data instance and...
10421         (make_pass_nrv): ...new function.
10422         (pass_return_slot): Convert from a global struct to a subclass of
10423         gimple_opt_pass along with...
10424         (pass_data_return_slot): ...new pass_data instance and...
10425         (make_pass_return_slot): ...new function.
10426         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10427         to a subclass of gimple_opt_pass along with...
10428         (pass_data_object_sizes): ...new pass_data instance and...
10429         (make_pass_object_sizes): ...new function.
10430         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10431         global struct to a subclass of gimple_opt_pass along with...
10432         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10433         and...
10434         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10435         (pass_fixup_cfg): Convert from a global struct to a subclass of
10436         gimple_opt_pass along with...
10437         (pass_data_fixup_cfg): ...new pass_data instance and...
10438         (make_pass_fixup_cfg): ...new function.
10439         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10440         (make_pass_mudflap_1): ...new function.
10441         (pass_mudflap_2): Replace declaration with that of...
10442         (make_pass_mudflap_2): ...new function.
10443         (pass_asan): Replace declaration with that of...
10444         (make_pass_asan): ...new function.
10445         (pass_asan_O0): Replace declaration with that of...
10446         (make_pass_asan_O0): ...new function.
10447         (pass_tsan): Replace declaration with that of...
10448         (make_pass_tsan): ...new function.
10449         (pass_tsan_O0): Replace declaration with that of...
10450         (make_pass_tsan_O0): ...new function.
10451         (pass_lower_cf): Replace declaration with that of...
10452         (make_pass_lower_cf): ...new function.
10453         (pass_refactor_eh): Replace declaration with that of...
10454         (make_pass_refactor_eh): ...new function.
10455         (pass_lower_eh): Replace declaration with that of...
10456         (make_pass_lower_eh): ...new function.
10457         (pass_lower_eh_dispatch): Replace declaration with that of...
10458         (make_pass_lower_eh_dispatch): ...new function.
10459         (pass_lower_resx): Replace declaration with that of...
10460         (make_pass_lower_resx): ...new function.
10461         (pass_build_cfg): Replace declaration with that of...
10462         (make_pass_build_cfg): ...new function.
10463         (pass_early_tree_profile): Replace declaration with that of...
10464         (make_pass_early_tree_profile): ...new function.
10465         (pass_cleanup_eh): Replace declaration with that of...
10466         (make_pass_cleanup_eh): ...new function.
10467         (pass_sra): Replace declaration with that of...
10468         (make_pass_sra): ...new function.
10469         (pass_sra_early): Replace declaration with that of...
10470         (make_pass_sra_early): ...new function.
10471         (pass_early_ipa_sra): Replace declaration with that of...
10472         (make_pass_early_ipa_sra): ...new function.
10473         (pass_tail_recursion): Replace declaration with that of...
10474         (make_pass_tail_recursion): ...new function.
10475         (pass_tail_calls): Replace declaration with that of...
10476         (make_pass_tail_calls): ...new function.
10477         (pass_tree_loop): Replace declaration with that of...
10478         (make_pass_tree_loop): ...new function.
10479         (pass_tree_loop_init): Replace declaration with that of...
10480         (make_pass_tree_loop_init): ...new function.
10481         (pass_lim): Replace declaration with that of...
10482         (make_pass_lim): ...new function.
10483         (pass_tree_unswitch): Replace declaration with that of...
10484         (make_pass_tree_unswitch): ...new function.
10485         (pass_predcom): Replace declaration with that of...
10486         (make_pass_predcom): ...new function.
10487         (pass_iv_canon): Replace declaration with that of...
10488         (make_pass_iv_canon): ...new function.
10489         (pass_scev_cprop): Replace declaration with that of...
10490         (make_pass_scev_cprop): ...new function.
10491         (pass_empty_loop): Replace declaration with that of...
10492         (make_pass_empty_loop): ...new function.
10493         (pass_record_bounds): Replace declaration with that of...
10494         (make_pass_record_bounds): ...new function.
10495         (pass_graphite): Replace declaration with that of...
10496         (make_pass_graphite): ...new function.
10497         (pass_graphite_transforms): Replace declaration with that of...
10498         (make_pass_graphite_transforms): ...new function.
10499         (pass_if_conversion): Replace declaration with that of...
10500         (make_pass_if_conversion): ...new function.
10501         (pass_loop_distribution): Replace declaration with that of...
10502         (make_pass_loop_distribution): ...new function.
10503         (pass_vectorize): Replace declaration with that of...
10504         (make_pass_vectorize): ...new function.
10505         (pass_slp_vectorize): Replace declaration with that of...
10506         (make_pass_slp_vectorize): ...new function.
10507         (pass_complete_unroll): Replace declaration with that of...
10508         (make_pass_complete_unroll): ...new function.
10509         (pass_complete_unrolli): Replace declaration with that of...
10510         (make_pass_complete_unrolli): ...new function.
10511         (pass_parallelize_loops): Replace declaration with that of...
10512         (make_pass_parallelize_loops): ...new function.
10513         (pass_loop_prefetch): Replace declaration with that of...
10514         (make_pass_loop_prefetch): ...new function.
10515         (pass_iv_optimize): Replace declaration with that of...
10516         (make_pass_iv_optimize): ...new function.
10517         (pass_tree_loop_done): Replace declaration with that of...
10518         (make_pass_tree_loop_done): ...new function.
10519         (pass_ch): Replace declaration with that of...
10520         (make_pass_ch): ...new function.
10521         (pass_ccp): Replace declaration with that of...
10522         (make_pass_ccp): ...new function.
10523         (pass_phi_only_cprop): Replace declaration with that of...
10524         (make_pass_phi_only_cprop): ...new function.
10525         (pass_build_ssa): Replace declaration with that of...
10526         (make_pass_build_ssa): ...new function.
10527         (pass_build_alias): Replace declaration with that of...
10528         (make_pass_build_alias): ...new function.
10529         (pass_build_ealias): Replace declaration with that of...
10530         (make_pass_build_ealias): ...new function.
10531         (pass_dominator): Replace declaration with that of...
10532         (make_pass_dominator): ...new function.
10533         (pass_dce): Replace declaration with that of...
10534         (make_pass_dce): ...new function.
10535         (pass_dce_loop): Replace declaration with that of...
10536         (make_pass_dce_loop): ...new function.
10537         (pass_cd_dce): Replace declaration with that of...
10538         (make_pass_cd_dce): ...new function.
10539         (pass_call_cdce): Replace declaration with that of...
10540         (make_pass_call_cdce): ...new function.
10541         (pass_merge_phi): Replace declaration with that of...
10542         (make_pass_merge_phi): ...new function.
10543         (pass_split_crit_edges): Replace declaration with that of...
10544         (make_pass_split_crit_edges): ...new function.
10545         (pass_pre): Replace declaration with that of...
10546         (make_pass_pre): ...new function.
10547         (pass_profile): Replace declaration with that of...
10548         (make_pass_profile): ...new function.
10549         (pass_strip_predict_hints): Replace declaration with that of...
10550         (make_pass_strip_predict_hints): ...new function.
10551         (pass_lower_complex_O0): Replace declaration with that of...
10552         (make_pass_lower_complex_O0): ...new function.
10553         (pass_lower_complex): Replace declaration with that of...
10554         (make_pass_lower_complex): ...new function.
10555         (pass_lower_vector): Replace declaration with that of...
10556         (make_pass_lower_vector): ...new function.
10557         (pass_lower_vector_ssa): Replace declaration with that of...
10558         (make_pass_lower_vector_ssa): ...new function.
10559         (pass_lower_omp): Replace declaration with that of...
10560         (make_pass_lower_omp): ...new function.
10561         (pass_diagnose_omp_blocks): Replace declaration with that of...
10562         (make_pass_diagnose_omp_blocks): ...new function.
10563         (pass_expand_omp): Replace declaration with that of...
10564         (make_pass_expand_omp): ...new function.
10565         (pass_expand_omp_ssa): Replace declaration with that of...
10566         (make_pass_expand_omp_ssa): ...new function.
10567         (pass_object_sizes): Replace declaration with that of...
10568         (make_pass_object_sizes): ...new function.
10569         (pass_strlen): Replace declaration with that of...
10570         (make_pass_strlen): ...new function.
10571         (pass_fold_builtins): Replace declaration with that of...
10572         (make_pass_fold_builtins): ...new function.
10573         (pass_stdarg): Replace declaration with that of...
10574         (make_pass_stdarg): ...new function.
10575         (pass_early_warn_uninitialized): Replace declaration with that of...
10576         (make_pass_early_warn_uninitialized): ...new function.
10577         (pass_late_warn_uninitialized): Replace declaration with that of...
10578         (make_pass_late_warn_uninitialized): ...new function.
10579         (pass_cse_reciprocals): Replace declaration with that of...
10580         (make_pass_cse_reciprocals): ...new function.
10581         (pass_cse_sincos): Replace declaration with that of...
10582         (make_pass_cse_sincos): ...new function.
10583         (pass_optimize_bswap): Replace declaration with that of...
10584         (make_pass_optimize_bswap): ...new function.
10585         (pass_optimize_widening_mul): Replace declaration with that of...
10586         (make_pass_optimize_widening_mul): ...new function.
10587         (pass_warn_function_return): Replace declaration with that of...
10588         (make_pass_warn_function_return): ...new function.
10589         (pass_warn_function_noreturn): Replace declaration with that of...
10590         (make_pass_warn_function_noreturn): ...new function.
10591         (pass_cselim): Replace declaration with that of...
10592         (make_pass_cselim): ...new function.
10593         (pass_phiopt): Replace declaration with that of...
10594         (make_pass_phiopt): ...new function.
10595         (pass_forwprop): Replace declaration with that of...
10596         (make_pass_forwprop): ...new function.
10597         (pass_phiprop): Replace declaration with that of...
10598         (make_pass_phiprop): ...new function.
10599         (pass_tree_ifcombine): Replace declaration with that of...
10600         (make_pass_tree_ifcombine): ...new function.
10601         (pass_dse): Replace declaration with that of...
10602         (make_pass_dse): ...new function.
10603         (pass_nrv): Replace declaration with that of...
10604         (make_pass_nrv): ...new function.
10605         (pass_rename_ssa_copies): Replace declaration with that of...
10606         (make_pass_rename_ssa_copies): ...new function.
10607         (pass_sink_code): Replace declaration with that of...
10608         (make_pass_sink_code): ...new function.
10609         (pass_fre): Replace declaration with that of...
10610         (make_pass_fre): ...new function.
10611         (pass_check_data_deps): Replace declaration with that of...
10612         (make_pass_check_data_deps): ...new function.
10613         (pass_copy_prop): Replace declaration with that of...
10614         (make_pass_copy_prop): ...new function.
10615         (pass_vrp): Replace declaration with that of...
10616         (make_pass_vrp): ...new function.
10617         (pass_uncprop): Replace declaration with that of...
10618         (make_pass_uncprop): ...new function.
10619         (pass_return_slot): Replace declaration with that of...
10620         (make_pass_return_slot): ...new function.
10621         (pass_reassoc): Replace declaration with that of...
10622         (make_pass_reassoc): ...new function.
10623         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10624         (make_pass_rebuild_cgraph_edges): ...new function.
10625         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10626         (make_pass_remove_cgraph_callee_edges): ...new function.
10627         (pass_build_cgraph_edges): Replace declaration with that of...
10628         (make_pass_build_cgraph_edges): ...new function.
10629         (pass_local_pure_const): Replace declaration with that of...
10630         (make_pass_local_pure_const): ...new function.
10631         (pass_tracer): Replace declaration with that of...
10632         (make_pass_tracer): ...new function.
10633         (pass_warn_unused_result): Replace declaration with that of...
10634         (make_pass_warn_unused_result): ...new function.
10635         (pass_diagnose_tm_blocks): Replace declaration with that of...
10636         (make_pass_diagnose_tm_blocks): ...new function.
10637         (pass_lower_tm): Replace declaration with that of...
10638         (make_pass_lower_tm): ...new function.
10639         (pass_tm_init): Replace declaration with that of...
10640         (make_pass_tm_init): ...new function.
10641         (pass_tm_mark): Replace declaration with that of...
10642         (make_pass_tm_mark): ...new function.
10643         (pass_tm_memopt): Replace declaration with that of...
10644         (make_pass_tm_memopt): ...new function.
10645         (pass_tm_edges): Replace declaration with that of...
10646         (make_pass_tm_edges): ...new function.
10647         (pass_split_functions): Replace declaration with that of...
10648         (make_pass_split_functions): ...new function.
10649         (pass_feedback_split_functions): Replace declaration with that of...
10650         (make_pass_feedback_split_functions): ...new function.
10651         (pass_strength_reduction): Replace declaration with that of...
10652         (make_pass_strength_reduction): ...new function.
10653         (pass_ipa_lower_emutls): Replace declaration with that of...
10654         (make_pass_ipa_lower_emutls): ...new function.
10655         (pass_ipa_function_and_variable_visibility): Replace declaration with
10656         that of...
10657         (make_pass_ipa_function_and_variable_visibility): ...new function.
10658         (pass_ipa_tree_profile): Replace declaration with that of...
10659         (make_pass_ipa_tree_profile): ...new function.
10660         (pass_early_local_passes): Replace declaration with that of...
10661         (make_pass_early_local_passes): ...new function.
10662         (pass_ipa_whole_program_visibility): Replace declaration with that
10663         of...
10664         (make_pass_ipa_whole_program_visibility): ...new function.
10665         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10666         (make_pass_ipa_lto_gimple_out): ...new function.
10667         (pass_ipa_increase_alignment): Replace declaration with that of...
10668         (make_pass_ipa_increase_alignment): ...new function.
10669         (pass_ipa_inline): Replace declaration with that of...
10670         (make_pass_ipa_inline): ...new function.
10671         (pass_ipa_free_lang_data): Replace declaration with that of...
10672         (make_pass_ipa_free_lang_data): ...new function.
10673         (pass_ipa_free_inline_summary): Replace declaration with that of...
10674         (make_pass_ipa_free_inline_summary): ...new function.
10675         (pass_ipa_cp): Replace declaration with that of...
10676         (make_pass_ipa_cp): ...new function.
10677         (pass_ipa_reference): Replace declaration with that of...
10678         (make_pass_ipa_reference): ...new function.
10679         (pass_ipa_pure_const): Replace declaration with that of...
10680         (make_pass_ipa_pure_const): ...new function.
10681         (pass_ipa_pta): Replace declaration with that of...
10682         (make_pass_ipa_pta): ...new function.
10683         (pass_ipa_lto_finish_out): Replace declaration with that of...
10684         (make_pass_ipa_lto_finish_out): ...new function.
10685         (pass_ipa_tm): Replace declaration with that of...
10686         (make_pass_ipa_tm): ...new function.
10687         (pass_ipa_profile): Replace declaration with that of...
10688         (make_pass_ipa_profile): ...new function.
10689         (pass_ipa_cdtor_merge): Replace declaration with that of...
10690         (make_pass_ipa_cdtor_merge): ...new function.
10691         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10692         of...
10693         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10694         (pass_init_datastructures): Replace declaration with that of...
10695         (make_pass_init_datastructures): ...new function.
10696         (pass_fixup_cfg): Replace declaration with that of...
10697         (make_pass_fixup_cfg): ...new function.
10698         (pass_expand): Replace declaration with that of...
10699         (make_pass_expand): ...new function.
10700         (pass_instantiate_virtual_regs): Replace declaration with that of...
10701         (make_pass_instantiate_virtual_regs): ...new function.
10702         (pass_rtl_fwprop): Replace declaration with that of...
10703         (make_pass_rtl_fwprop): ...new function.
10704         (pass_rtl_fwprop_addr): Replace declaration with that of...
10705         (make_pass_rtl_fwprop_addr): ...new function.
10706         (pass_jump): Replace declaration with that of...
10707         (make_pass_jump): ...new function.
10708         (pass_jump2): Replace declaration with that of...
10709         (make_pass_jump2): ...new function.
10710         (pass_lower_subreg): Replace declaration with that of...
10711         (make_pass_lower_subreg): ...new function.
10712         (pass_cse): Replace declaration with that of...
10713         (make_pass_cse): ...new function.
10714         (pass_fast_rtl_dce): Replace declaration with that of...
10715         (make_pass_fast_rtl_dce): ...new function.
10716         (pass_ud_rtl_dce): Replace declaration with that of...
10717         (make_pass_ud_rtl_dce): ...new function.
10718         (pass_rtl_dce): Replace declaration with that of...
10719         (make_pass_rtl_dce): ...new function.
10720         (pass_rtl_dse1): Replace declaration with that of...
10721         (make_pass_rtl_dse1): ...new function.
10722         (pass_rtl_dse2): Replace declaration with that of...
10723         (make_pass_rtl_dse2): ...new function.
10724         (pass_rtl_dse3): Replace declaration with that of...
10725         (make_pass_rtl_dse3): ...new function.
10726         (pass_rtl_cprop): Replace declaration with that of...
10727         (make_pass_rtl_cprop): ...new function.
10728         (pass_rtl_pre): Replace declaration with that of...
10729         (make_pass_rtl_pre): ...new function.
10730         (pass_rtl_hoist): Replace declaration with that of...
10731         (make_pass_rtl_hoist): ...new function.
10732         (pass_rtl_store_motion): Replace declaration with that of...
10733         (make_pass_rtl_store_motion): ...new function.
10734         (pass_cse_after_global_opts): Replace declaration with that of...
10735         (make_pass_cse_after_global_opts): ...new function.
10736         (pass_rtl_ifcvt): Replace declaration with that of...
10737         (make_pass_rtl_ifcvt): ...new function.
10738         (pass_into_cfg_layout_mode): Replace declaration with that of...
10739         (make_pass_into_cfg_layout_mode): ...new function.
10740         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10741         (make_pass_outof_cfg_layout_mode): ...new function.
10742         (pass_loop2): Replace declaration with that of...
10743         (make_pass_loop2): ...new function.
10744         (pass_rtl_loop_init): Replace declaration with that of...
10745         (make_pass_rtl_loop_init): ...new function.
10746         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10747         (make_pass_rtl_move_loop_invariants): ...new function.
10748         (pass_rtl_unswitch): Replace declaration with that of...
10749         (make_pass_rtl_unswitch): ...new function.
10750         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10751         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10752         (pass_rtl_doloop): Replace declaration with that of...
10753         (make_pass_rtl_doloop): ...new function.
10754         (pass_rtl_loop_done): Replace declaration with that of...
10755         (make_pass_rtl_loop_done): ...new function.
10756         (pass_web): Replace declaration with that of...
10757         (make_pass_web): ...new function.
10758         (pass_cse2): Replace declaration with that of...
10759         (make_pass_cse2): ...new function.
10760         (pass_df_initialize_opt): Replace declaration with that of...
10761         (make_pass_df_initialize_opt): ...new function.
10762         (pass_df_initialize_no_opt): Replace declaration with that of...
10763         (make_pass_df_initialize_no_opt): ...new function.
10764         (pass_reginfo_init): Replace declaration with that of...
10765         (make_pass_reginfo_init): ...new function.
10766         (pass_inc_dec): Replace declaration with that of...
10767         (make_pass_inc_dec): ...new function.
10768         (pass_stack_ptr_mod): Replace declaration with that of...
10769         (make_pass_stack_ptr_mod): ...new function.
10770         (pass_initialize_regs): Replace declaration with that of...
10771         (make_pass_initialize_regs): ...new function.
10772         (pass_combine): Replace declaration with that of...
10773         (make_pass_combine): ...new function.
10774         (pass_if_after_combine): Replace declaration with that of...
10775         (make_pass_if_after_combine): ...new function.
10776         (pass_ree): Replace declaration with that of...
10777         (make_pass_ree): ...new function.
10778         (pass_partition_blocks): Replace declaration with that of...
10779         (make_pass_partition_blocks): ...new function.
10780         (pass_match_asm_constraints): Replace declaration with that of...
10781         (make_pass_match_asm_constraints): ...new function.
10782         (pass_regmove): Replace declaration with that of...
10783         (make_pass_regmove): ...new function.
10784         (pass_split_all_insns): Replace declaration with that of...
10785         (make_pass_split_all_insns): ...new function.
10786         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10787         (make_pass_fast_rtl_byte_dce): ...new function.
10788         (pass_lower_subreg2): Replace declaration with that of...
10789         (make_pass_lower_subreg2): ...new function.
10790         (pass_mode_switching): Replace declaration with that of...
10791         (make_pass_mode_switching): ...new function.
10792         (pass_sms): Replace declaration with that of...
10793         (make_pass_sms): ...new function.
10794         (pass_sched): Replace declaration with that of...
10795         (make_pass_sched): ...new function.
10796         (pass_ira): Replace declaration with that of...
10797         (make_pass_ira): ...new function.
10798         (pass_reload): Replace declaration with that of...
10799         (make_pass_reload): ...new function.
10800         (pass_clean_state): Replace declaration with that of...
10801         (make_pass_clean_state): ...new function.
10802         (pass_branch_prob): Replace declaration with that of...
10803         (make_pass_branch_prob): ...new function.
10804         (pass_value_profile_transformations): Replace declaration with that
10805         of...
10806         (make_pass_value_profile_transformations): ...new function.
10807         (pass_postreload_cse): Replace declaration with that of...
10808         (make_pass_postreload_cse): ...new function.
10809         (pass_gcse2): Replace declaration with that of...
10810         (make_pass_gcse2): ...new function.
10811         (pass_split_after_reload): Replace declaration with that of...
10812         (make_pass_split_after_reload): ...new function.
10813         (pass_branch_target_load_optimize1): Replace declaration with that
10814         of...
10815         (make_pass_branch_target_load_optimize1): ...new function.
10816         (pass_thread_prologue_and_epilogue): Replace declaration with that
10817         of...
10818         (make_pass_thread_prologue_and_epilogue): ...new function.
10819         (pass_stack_adjustments): Replace declaration with that of...
10820         (make_pass_stack_adjustments): ...new function.
10821         (pass_peephole2): Replace declaration with that of...
10822         (make_pass_peephole2): ...new function.
10823         (pass_if_after_reload): Replace declaration with that of...
10824         (make_pass_if_after_reload): ...new function.
10825         (pass_regrename): Replace declaration with that of...
10826         (make_pass_regrename): ...new function.
10827         (pass_cprop_hardreg): Replace declaration with that of...
10828         (make_pass_cprop_hardreg): ...new function.
10829         (pass_reorder_blocks): Replace declaration with that of...
10830         (make_pass_reorder_blocks): ...new function.
10831         (pass_branch_target_load_optimize2): Replace declaration with that
10832         of...
10833         (make_pass_branch_target_load_optimize2): ...new function.
10834         (pass_leaf_regs): Replace declaration with that of...
10835         (make_pass_leaf_regs): ...new function.
10836         (pass_split_before_sched2): Replace declaration with that of...
10837         (make_pass_split_before_sched2): ...new function.
10838         (pass_compare_elim_after_reload): Replace declaration with that of...
10839         (make_pass_compare_elim_after_reload): ...new function.
10840         (pass_sched2): Replace declaration with that of...
10841         (make_pass_sched2): ...new function.
10842         (pass_stack_regs): Replace declaration with that of...
10843         (make_pass_stack_regs): ...new function.
10844         (pass_stack_regs_run): Replace declaration with that of...
10845         (make_pass_stack_regs_run): ...new function.
10846         (pass_df_finish): Replace declaration with that of...
10847         (make_pass_df_finish): ...new function.
10848         (pass_compute_alignments): Replace declaration with that of...
10849         (make_pass_compute_alignments): ...new function.
10850         (pass_duplicate_computed_gotos): Replace declaration with that of...
10851         (make_pass_duplicate_computed_gotos): ...new function.
10852         (pass_variable_tracking): Replace declaration with that of...
10853         (make_pass_variable_tracking): ...new function.
10854         (pass_free_cfg): Replace declaration with that of...
10855         (make_pass_free_cfg): ...new function.
10856         (pass_machine_reorg): Replace declaration with that of...
10857         (make_pass_machine_reorg): ...new function.
10858         (pass_cleanup_barriers): Replace declaration with that of...
10859         (make_pass_cleanup_barriers): ...new function.
10860         (pass_delay_slots): Replace declaration with that of...
10861         (make_pass_delay_slots): ...new function.
10862         (pass_split_for_shorten_branches): Replace declaration with that of...
10863         (make_pass_split_for_shorten_branches): ...new function.
10864         (pass_split_before_regstack): Replace declaration with that of...
10865         (make_pass_split_before_regstack): ...new function.
10866         (pass_convert_to_eh_region_ranges): Replace declaration with that
10867         of...
10868         (make_pass_convert_to_eh_region_ranges): ...new function.
10869         (pass_shorten_branches): Replace declaration with that of...
10870         (make_pass_shorten_branches): ...new function.
10871         (pass_set_nothrow_function_flags): Replace declaration with that of...
10872         (make_pass_set_nothrow_function_flags): ...new function.
10873         (pass_dwarf2_frame): Replace declaration with that of...
10874         (make_pass_dwarf2_frame): ...new function.
10875         (pass_final): Replace declaration with that of...
10876         (make_pass_final): ...new function.
10877         (pass_rtl_seqabstr): Replace declaration with that of...
10878         (make_pass_rtl_seqabstr): ...new function.
10879         (pass_release_ssa_names): Replace declaration with that of...
10880         (make_pass_release_ssa_names): ...new function.
10881         (pass_early_inline): Replace declaration with that of...
10882         (make_pass_early_inline): ...new function.
10883         (pass_inline_parameters): Replace declaration with that of...
10884         (make_pass_inline_parameters): ...new function.
10885         (pass_update_address_taken): Replace declaration with that of...
10886         (make_pass_update_address_taken): ...new function.
10887         (pass_convert_switch): Replace declaration with that of...
10888         (make_pass_convert_switch): ...new function.
10889         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10890         to a subclass of simple_ipa_opt_pass along with...
10891         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10892         (make_pass_ipa_tree_profile): ...new function.
10893         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10894         subclass of gimple_opt_pass along with...
10895         (pass_data_sra_early): ...new pass_data instance and...
10896         (make_pass_sra_early): ...new function.
10897         (pass_sra): Convert from a global struct to a subclass of
10898         gimple_opt_pass along with...
10899         (pass_data_sra): ...new pass_data instance and...
10900         (make_pass_sra): ...new function.
10901         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10902         gimple_opt_pass along with...
10903         (pass_data_early_ipa_sra): ...new pass_data instance and...
10904         (make_pass_early_ipa_sra): ...new function.
10905         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10906         subclass of gimple_opt_pass along with...
10907         (pass_data_ccp): ...new pass_data instance and...
10908         (make_pass_ccp): ...new function.
10909         (pass_fold_builtins): Convert from a global struct to a subclass of
10910         gimple_opt_pass along with...
10911         (pass_data_fold_builtins): ...new pass_data instance and...
10912         (make_pass_fold_builtins): ...new function.
10913         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10914         subclass of gimple_opt_pass along with...
10915         (pass_data_copy_prop): ...new pass_data instance and...
10916         (make_pass_copy_prop): ...new function.
10917         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10918         global struct to a subclass of gimple_opt_pass along with...
10919         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10920         (make_pass_rename_ssa_copies): ...new function.
10921         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10922         subclass of gimple_opt_pass along with...
10923         (pass_data_dce): ...new pass_data instance and...
10924         (make_pass_dce): ...new function.
10925         (pass_dce_loop): Convert from a global struct to a subclass of
10926         gimple_opt_pass along with...
10927         (pass_data_dce_loop): ...new pass_data instance and...
10928         (make_pass_dce_loop): ...new function.
10929         (pass_cd_dce): Convert from a global struct to a subclass of
10930         gimple_opt_pass along with...
10931         (pass_data_cd_dce): ...new pass_data instance and...
10932         (make_pass_cd_dce): ...new function.
10933         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10934         subclass of gimple_opt_pass along with...
10935         (pass_data_dominator): ...new pass_data instance and...
10936         (make_pass_dominator): ...new function.
10937         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10938         gimple_opt_pass along with...
10939         (pass_data_phi_only_cprop): ...new pass_data instance and...
10940         (make_pass_phi_only_cprop): ...new function.
10941         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10942         subclass of gimple_opt_pass along with...
10943         (pass_data_dse): ...new pass_data instance and...
10944         (make_pass_dse): ...new function.
10945         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10946         a subclass of gimple_opt_pass along with...
10947         (pass_data_forwprop): ...new pass_data instance and...
10948         (make_pass_forwprop): ...new function.
10949         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10950         struct to a subclass of gimple_opt_pass along with...
10951         (pass_data_tree_ifcombine): ...new pass_data instance and...
10952         (make_pass_tree_ifcombine): ...new function.
10953         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10954         subclass of gimple_opt_pass along with...
10955         (pass_data_ch): ...new pass_data instance and...
10956         (make_pass_ch): ...new function.
10957         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10958         subclass of gimple_opt_pass along with...
10959         (pass_data_tree_loop): ...new pass_data instance and...
10960         (make_pass_tree_loop): ...new function.
10961         (pass_tree_loop_init): Convert from a global struct to a subclass of
10962         gimple_opt_pass along with...
10963         (pass_data_tree_loop_init): ...new pass_data instance and...
10964         (make_pass_tree_loop_init): ...new function.
10965         (pass_lim): Convert from a global struct to a subclass of
10966         gimple_opt_pass along with...
10967         (pass_data_lim): ...new pass_data instance and...
10968         (make_pass_lim): ...new function.
10969         (pass_tree_unswitch): Convert from a global struct to a subclass of
10970         gimple_opt_pass along with...
10971         (pass_data_tree_unswitch): ...new pass_data instance and...
10972         (make_pass_tree_unswitch): ...new function.
10973         (pass_predcom): Convert from a global struct to a subclass of
10974         gimple_opt_pass along with...
10975         (pass_data_predcom): ...new pass_data instance and...
10976         (make_pass_predcom): ...new function.
10977         (pass_vectorize): Convert from a global struct to a subclass of
10978         gimple_opt_pass along with...
10979         (pass_data_vectorize): ...new pass_data instance and...
10980         (make_pass_vectorize): ...new function.
10981         (pass_graphite): Convert from a global struct to a subclass of
10982         gimple_opt_pass along with...
10983         (pass_data_graphite): ...new pass_data instance and...
10984         (make_pass_graphite): ...new function.
10985         (pass_graphite_transforms): Convert from a global struct to a subclass
10986         of gimple_opt_pass along with...
10987         (pass_data_graphite_transforms): ...new pass_data instance and...
10988         (make_pass_graphite_transforms): ...new function.
10989         (pass_check_data_deps): Convert from a global struct to a subclass of
10990         gimple_opt_pass along with...
10991         (pass_data_check_data_deps): ...new pass_data instance and...
10992         (make_pass_check_data_deps): ...new function.
10993         (pass_iv_canon): Convert from a global struct to a subclass of
10994         gimple_opt_pass along with...
10995         (pass_data_iv_canon): ...new pass_data instance and...
10996         (make_pass_iv_canon): ...new function.
10997         (pass_scev_cprop): Convert from a global struct to a subclass of
10998         gimple_opt_pass along with...
10999         (pass_data_scev_cprop): ...new pass_data instance and...
11000         (make_pass_scev_cprop): ...new function.
11001         (pass_record_bounds): Convert from a global struct to a subclass of
11002         gimple_opt_pass along with...
11003         (pass_data_record_bounds): ...new pass_data instance and...
11004         (make_pass_record_bounds): ...new function.
11005         (pass_complete_unroll): Convert from a global struct to a subclass of
11006         gimple_opt_pass along with...
11007         (pass_data_complete_unroll): ...new pass_data instance and...
11008         (make_pass_complete_unroll): ...new function.
11009         (pass_complete_unrolli): Convert from a global struct to a subclass of
11010         gimple_opt_pass along with...
11011         (pass_data_complete_unrolli): ...new pass_data instance and...
11012         (make_pass_complete_unrolli): ...new function.
11013         (pass_parallelize_loops): Convert from a global struct to a subclass
11014         of gimple_opt_pass along with...
11015         (pass_data_parallelize_loops): ...new pass_data instance and...
11016         (make_pass_parallelize_loops): ...new function.
11017         (pass_loop_prefetch): Convert from a global struct to a subclass of
11018         gimple_opt_pass along with...
11019         (pass_data_loop_prefetch): ...new pass_data instance and...
11020         (make_pass_loop_prefetch): ...new function.
11021         (pass_iv_optimize): Convert from a global struct to a subclass of
11022         gimple_opt_pass along with...
11023         (pass_data_iv_optimize): ...new pass_data instance and...
11024         (make_pass_iv_optimize): ...new function.
11025         (pass_tree_loop_done): Convert from a global struct to a subclass of
11026         gimple_opt_pass along with...
11027         (pass_data_tree_loop_done): ...new pass_data instance and...
11028         (make_pass_tree_loop_done): ...new function.
11029         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
11030         struct to a subclass of gimple_opt_pass along with...
11031         (pass_data_cse_reciprocals): ...new pass_data instance and...
11032         (make_pass_cse_reciprocals): ...new function.
11033         (pass_cse_sincos): Convert from a global struct to a subclass of
11034         gimple_opt_pass along with...
11035         (pass_data_cse_sincos): ...new pass_data instance and...
11036         (make_pass_cse_sincos): ...new function.
11037         (pass_optimize_bswap): Convert from a global struct to a subclass of
11038         gimple_opt_pass along with...
11039         (pass_data_optimize_bswap): ...new pass_data instance and...
11040         (make_pass_optimize_bswap): ...new function.
11041         (pass_optimize_widening_mul): Convert from a global struct to a
11042         subclass of gimple_opt_pass along with...
11043         (pass_data_optimize_widening_mul): ...new pass_data instance and...
11044         (make_pass_optimize_widening_mul): ...new function.
11045         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
11046         subclass of gimple_opt_pass along with...
11047         (pass_data_phiopt): ...new pass_data instance and...
11048         (make_pass_phiopt): ...new function.
11049         (pass_cselim): Convert from a global struct to a subclass of
11050         gimple_opt_pass along with...
11051         (pass_data_cselim): ...new pass_data instance and...
11052         (make_pass_cselim): ...new function.
11053         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
11054         subclass of gimple_opt_pass along with...
11055         (pass_data_phiprop): ...new pass_data instance and...
11056         (make_pass_phiprop): ...new function.
11057         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
11058         subclass of gimple_opt_pass along with...
11059         (pass_data_pre): ...new pass_data instance and...
11060         (make_pass_pre): ...new function.
11061         (pass_fre): Convert from a global struct to a subclass of
11062         gimple_opt_pass along with...
11063         (pass_data_fre): ...new pass_data instance and...
11064         (make_pass_fre): ...new function.
11065         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
11066         subclass of gimple_opt_pass along with...
11067         (pass_data_reassoc): ...new pass_data instance and...
11068         (make_pass_reassoc): ...new function.
11069         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
11070         subclass of gimple_opt_pass along with...
11071         (pass_data_sink_code): ...new pass_data instance and...
11072         (make_pass_sink_code): ...new function.
11073         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
11074         subclass of gimple_opt_pass along with...
11075         (pass_data_strlen): ...new pass_data instance and...
11076         (make_pass_strlen): ...new function.
11077         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
11078         struct to a subclass of gimple_opt_pass along with...
11079         (pass_data_build_alias): ...new pass_data instance and...
11080         (make_pass_build_alias): ...new function.
11081         (pass_build_ealias): Convert from a global struct to a subclass of
11082         gimple_opt_pass along with...
11083         (pass_data_build_ealias): ...new pass_data instance and...
11084         (make_pass_build_ealias): ...new function.
11085         (pass_ipa_pta): Convert from a global struct to a subclass of
11086         simple_ipa_opt_pass along with...
11087         (pass_data_ipa_pta): ...new pass_data instance and...
11088         (make_pass_ipa_pta): ...new function.
11089         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11090         subclass of gimple_opt_pass along with...
11091         (pass_data_uncprop): ...new pass_data instance and...
11092         (make_pass_uncprop): ...new function.
11093         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11094         global struct to a subclass of gimple_opt_pass along with...
11095         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11096         (make_pass_late_warn_uninitialized): ...new function.
11097         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11098         to a subclass of gimple_opt_pass along with...
11099         (pass_data_init_datastructures): ...new pass_data instance and...
11100         (make_pass_init_datastructures): ...new function.
11101         (pass_early_warn_uninitialized): Convert from a global struct to a
11102         subclass of gimple_opt_pass along with...
11103         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11104         (make_pass_early_warn_uninitialized): ...new function.
11105         (pass_update_address_taken): Convert from a global struct to a
11106         subclass of gimple_opt_pass along with...
11107         (pass_data_update_address_taken): ...new pass_data instance and...
11108         (make_pass_update_address_taken): ...new function.
11109         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11110         struct to a subclass of gimple_opt_pass along with...
11111         (pass_data_release_ssa_names): ...new pass_data instance and...
11112         (make_pass_release_ssa_names): ...new function.
11113         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11114         subclass of gimple_opt_pass along with...
11115         (pass_data_stdarg): ...new pass_data instance and...
11116         (make_pass_stdarg): ...new function.
11117         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11118         global struct to a subclass of gimple_opt_pass along with...
11119         (pass_data_convert_switch): ...new pass_data instance and...
11120         (make_pass_convert_switch): ...new function.
11121         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11122         to a subclass of gimple_opt_pass along with...
11123         (pass_data_tail_recursion): ...new pass_data instance and...
11124         (make_pass_tail_recursion): ...new function.
11125         (pass_tail_calls): Convert from a global struct to a subclass of
11126         gimple_opt_pass along with...
11127         (pass_data_tail_calls): ...new pass_data instance and...
11128         (make_pass_tail_calls): ...new function.
11129         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11130         struct to a subclass of gimple_opt_pass along with...
11131         (pass_data_lower_vector): ...new pass_data instance and...
11132         (make_pass_lower_vector): ...new function.
11133         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11134         gimple_opt_pass along with...
11135         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11136         (make_pass_lower_vector_ssa): ...new function.
11137         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11138         to a subclass of gimple_opt_pass along with...
11139         (pass_data_slp_vectorize): ...new pass_data instance and...
11140         (make_pass_slp_vectorize): ...new function.
11141         (pass_ipa_increase_alignment): Convert from a global struct to a
11142         subclass of simple_ipa_opt_pass along with...
11143         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11144         (make_pass_ipa_increase_alignment): ...new function.
11145         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11146         gimple_opt_pass along with...
11147         (pass_data_vrp): ...new pass_data instance and...
11148         (make_pass_vrp): ...new function.
11149         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11150         subclass of simple_ipa_opt_pass along with...
11151         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11152         (make_pass_ipa_free_lang_data): ...new function.
11153         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11154         gimple_opt_pass along with...
11155         (pass_data_tsan): ...new pass_data instance and...
11156         (make_pass_tsan): ...new function.
11157         (pass_tsan_O0): Convert from a global struct to a subclass of
11158         gimple_opt_pass along with...
11159         (pass_data_tsan_O0): ...new pass_data instance and...
11160         (make_pass_tsan_O0): ...new function.
11161         * var-tracking.c (pass_variable_tracking): Convert from a global
11162         struct to a subclass of rtl_opt_pass along with...
11163         (pass_data_variable_tracking): ...new pass_data instance and...
11164         (make_pass_variable_tracking): ...new function.
11165         * web.c (pass_web): Convert from a global struct to a subclass of
11166         rtl_opt_pass along with...
11167         (pass_data_web): ...new pass_data instance and...
11168         (make_pass_web): ...new function.
11169         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11170         declaration with that of...
11171         (make_pass_mode_switch_use): ...new function.
11172         (pass_resolve_sw_modes): Replace declaration with that of...
11173         (make_pass_resolve_sw_modes): ...new function.
11174         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11175         from a global struct to a subclass of rtl_opt_pass along with...
11176         (pass_data_mode_switch_use): ...new pass_data instance and...
11177         (make_pass_mode_switch_use): ...new function.
11178         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11179         from a global struct to a subclass of rtl_opt_pass along with...
11180         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11181         (make_pass_resolve_sw_modes): ...new function.
11182         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11183         struct to a subclass of rtl_opt_pass along with...
11184         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11185         (make_pass_insert_vzeroupper): ...new function.
11186         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11187         global struct to a subclass of rtl_opt_pass along with...
11188         (pass_data_work_around_errata): ...new pass_data instance and...
11189         (make_pass_work_around_errata): ...new function.
11190         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11191         struct to a subclass of rtl_opt_pass along with...
11192         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11193         (make_pass_mips_machine_reorg2): ...new function.
11195 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11197         * passes.c (pass_manager::operator new): New.
11199 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11201         Handwritten part of conversion of passes to C++ classes.
11203         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11204         (toplev.o): Add dep on PASS_MANAGER_H.
11205         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11206         of early local pases to reflect this moving from a global to a
11207         member of gcc::pass_manager.
11208         (cgraph_add_new_function): Likewise.
11209         * lto-cgraph.c (lto_output_node): Update for conversion of
11210         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11211         * passes.c (opt_pass::clone): New.
11212         (opt_pass::gate): New.
11213         (opt_pass::execute): New.
11214         (opt_pass::opt_pass): New.
11215         (pass_manager::execute_early_local_passes): New.
11216         (pass_manager::execute_pass_mode_switching): new.
11217         (finish_optimization_passes): Convert to...
11218         (pass_manager::finish_optimization_passes): ...this.
11219         (finish_optimization_passes): Update for conversion of passes to
11220         C++ classes.
11221         (register_dump_files_1): Use has_gate since we cannot portably
11222         check a vtable entry against NULL.
11223         (dump_one_pass): Likewise.
11224         (ipa_write_summaries_2): Likewise.
11225         (ipa_write_optimization_summaries_1): Likewise.
11226         (ipa_read_summaries_1): Likewise.
11227         (ipa_read_optimization_summaries_1): Likewise.
11228         (execute_ipa_stmt_fixups): Likewise.
11229         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11230         pass-creation functions rather than wiring up globals, and
11231         storing the results in fields of pass_manager generated using
11232         pass-instances.def.
11233         (pass_manager::dump_profile_report): Update for conversion of
11234         passes to C++ classes.
11235         (pass_manager::execute_ipa_summary_passes): Likewise.
11236         (execute_one_ipa_transform_pass): Likewise.
11237         (execute_one_pass): Use has_gate and has_execute since we cannot
11238         portably check a vtable entry against NULL.
11239         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11240         (pass_manager): Use pass-instances.def to add fields for the
11241         various pass instances.
11242         * toplev.c (finalize): Update for move of
11243         finish_optimization_passes to a method of gcc::pass_manager.
11244         * toplev.h (finish_optimization_passes): Move to method of class
11245         pass_manager.
11246         * tree-pass.h (struct pass_data): New.
11247         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11248         (opt_pass::gate): Convert to virtual function.
11249         (opt_pass::~opt_pass): New.
11250         (opt_pass::clone): New.
11251         (opt_pass::execute): Convert to virtual function.
11252         (opt_pass::opt_pass): New.
11253         (opt_pass::ctxt_): new.
11254         (gimple_opt_pass): Convert to subclass of opt_pass.
11255         (gimple_opt_pass::gimple_opt_pass): New.
11256         (rtl_opt_pass): Convert to subclass of opt_pass.
11257         (rtl_opt_pass::rtl_opt_pass): New.
11258         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11259         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11260         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11261         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11262         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11263         invocation of pass_mode_switching to reflect this moving from a
11264         global to a member of gcc::pass_manager.
11265         (ix86_option_override): Rework how pass_insert_vzeroupper is
11266         added to the pass_manager to reflect autogenerated changes.
11267         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11269 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11271         PR rtl-optimization/57708
11272         * recog.c (peep2_find_free_register): Validate all regs in a
11273         multi-reg mode.
11275 2013-08-05  Jan Hubicka  <jh@suse.cz>
11277         PR lto/57602
11278         * cgraph.c (verify_cgraph_node): Accept local flags from other
11279         partitions.
11280         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11281         (function_and_variable_visibility): Likewise.
11282         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11284 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11286         * graph.c (init_graph_slim_pretty_print): Remove.
11287         (print_graph_cfg): Do not call it.  Use local pretty printer.
11288         (start_graph_dump): Likewise.
11290 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11292         * gimple-pretty-print.c (buffer): Remove.
11293         (initialized): Likewise.
11294         (maybe_init_pretty_print): Likewise.
11295         (print_gimple_stmt): Do not call it.  Use non-static local
11296         pretty_printer variable.
11297         (print_gimple_expr): Likewise.
11298         (print_gimple_seq): Likewise.
11299         (gimple_dump_bb): Likewise.
11301 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11303         * asan.c (asan_pp): Remove.
11304         (asan_pp_initialized): Likewise.
11305         (asan_pp_initialize): Likewise.
11306         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11307         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11308         (asan_add_global): Likewise.
11310 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11312         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11313         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11314         * pretty-print.c (pp_formatted_text_data): Likewise.
11315         (pp_indent): Rename from pp_base_indent.
11316         (pp_format): Rename from pp_base_format.
11317         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11318         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11319         (pp_flush): Rename from pp_base_flush.
11320         (pp_set_line_maximum_length): Rename from
11321         pp_base_set_line_maximum_length.
11322         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11323         (pp_set_prefix): Rename from pp_base_set_prefix.
11324         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11325         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11326         (pp_append_text): Rename from pp_base_append_text.
11327         (pp_formatted_text): Rename from pp_base_formatted_text.
11328         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11329         (pp_remaining_character_count_for_line): Rename from
11330         pp_base_remaining_character_count_for_line.
11331         (pp_newline): Rename from pp_base_newline.
11332         (pp_character): Rename from pp_base_character.
11333         (pp_string): Rename from pp_base_string.
11334         (pp_maybe_space): Rename from pp_base_maybe_space.
11335         * asan.c (asan_pp_string): Adjust.
11336         (asan_emit_stack_protection): Likewise.
11337         (asan_add_global): Likewise.
11338         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11339         * tree-mudflap.c (mf_varname_tree): Likewise.
11340         * tree-pretty-print.c (pp_tree_identifier): Rename from
11341         pp_base_tree_identifier.
11342         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11343         Declare as function.
11345 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11347         * pretty-print.h (pp_bar_bar): New.
11348         (pp_ampersand_ampersand): Likewise.
11349         (pp_less_equal): Likewise.
11350         (pp_greater_equal): Likewise.
11351         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11352         printer functions instead of pp_string or operators and punctuators.
11353         (dump_gimple_call): Likewise.
11354         (dump_gimple_omp_for): Likewise.
11355         (dump_gimple_transaction): Likewise.
11356         (dump_gimple_phi): Likewise.
11357         (pp_gimple_stmt_1): Likewise.
11358         * sched-vis.c (print_insn): Likewise.
11359         * tree-mudflap.c (mf_varname_tree): Likewise.
11360         * tree-pretty-print.c (dump_block_node): Likewise.
11361         (dump_generic_node): Likewise.
11363 2013-08-02  Jan Hubicka  <jh@suse.cz>
11365         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11366         boundaries.
11367         * lto-streamer-out.c (tree_is_indexable): Results decls and
11368         parm decls are not indexable.
11369         (DFS_write_tree_body): Do not follow args and results.
11370         (hash_tree): Likewise.
11371         (output_functions): Rearrange so struct function is needed
11372         only when real body is output; be able to also ouptut abstract
11373         functions; output DECL_ARGUMENTS and DECL_RESULT.
11374         (lto_output): When not in WPA, ale store abstract functions.
11375         (write_symbol): Do not care about RESULT_DECL.
11376         (output_symbol_p): Handle correctly sbtract decls.
11377         * lto-streamer-in.c (input_function): Rearrange so struct
11378         function can be NULL at entry; allow streaming of
11379         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11380         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11381         sanity check during LTO.
11382         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11383         RESULT_DECl and DECL_ARGUMENTS.
11384         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11385         Likewise.
11387 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11389         * pretty-print.h (pp_underscore): New.
11390         (pp_comma): Tidy.
11391         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11392         printer functions instead of pp_character.
11393         (dump_binary_rhs): Likewise.
11394         (dump_ternary_rhs): Likewise.
11395         (dump_gimple_call_args): Likewise.
11396         (pp_points_to_solution): Likewise.
11397         (dump_gimple_call): Likewise.
11398         (dump_gimple_switch): Likewise.
11399         (dump_gimple_cond): Likewise.
11400         (dump_gimple_bind): Likewise.
11401         (dump_gimple_try): Likewise.
11402         (dump_gimple_omp_for): Likewise.
11403         (dump_gimple_omp_continue): Likewise.
11404         (dump_gimple_omp_single): Likewise.
11405         (dump_gimple_omp_sections): Likewise.
11406         (dump_gimple_omp_block): Likewise.
11407         (dump_gimple_omp_critical): Likewise.
11408         (dump_gimple_transaction): Likewise.
11409         (dump_gimple_asm): Likewise.
11410         (dump_gimple_phi): Likewise.
11411         (dump_gimple_omp_parallel): Likewise.
11412         (dump_gimple_omp_task): Likewise.
11413         (dump_gimple_omp_atomic_load): Likewise.
11414         (dump_gimple_omp_atomic_store): Likewise.
11415         (dump_gimple_mem_ops): Likewise.
11416         (pp_gimple_stmt_1): Likewise.
11417         (pp_cfg_jump): Likewise.
11418         (dump_implicit_edges): Likewise.
11419         (gimple_dump_bb_for_graph): Likewise.
11420         * graph.c (draw_cfg_node): Likewise.
11421         * langhooks.c (lhd_print_error_function): Likewise.
11422         * sched-vis.c (print_exp): Likewise.
11423         (print_value): Likewise.
11424         (print_pattern): Likewise.
11425         (print_insn): Likewise.
11426         (rtl_dump_bb_for_graph): Likewise.
11427         * tree-pretty-print.c (dump_function_declaration): Likewise.
11428         (dump_array_domain): Likewise.
11429         (dump_omp_clause): Likewise.
11430         (dump_location): Likewise.
11431         (dump_generic_node): Likewise.
11432         (print_struct_decl): Likewise.
11433         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11435 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11437         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11438         candidate table when replacing a candidate statement.
11439         (replace_rhs_if_not_dup): Likewise.
11440         (replace_one_candidate): Likewise.
11442 2013-08-02  Jan Hubicka  <jh@suse.cz>
11443             Martin Liska  <marxin.liska@gmail.com>
11445         * cgraphunit.c (add_new_function): Fix logic when adding from
11446         late IPA pass.
11447         (assemble_thunk): Rename to ...
11448         (expand_thunk); .. this one; export; get it working with
11449         general functions; make produced gimple valid.
11450         * cgraph.h (expand_thunk): Declare.
11452 2013-08-02  Jan Hubicka  <jh@suse.cz>
11454         * ipa-cp.c (gather_context_independent_values): Use
11455         ipa_get_param_move_cost.
11456         (get_replacement_map): Remove PARAM; move parameter folding
11457         into tree-inline.c
11458         (create_specialized_node): Update.
11459         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11460         assert that we have gimple body; update move_cost.
11461         (count_formal_params): Assert that we have gimple body.
11462         (ipa_dump_param): New function.
11463         (ipa_alloc_node_params): Break out from ...
11464         (ipa_initialize_node_params): ... here.
11465         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11466         (ipa_write_node_info): Stream move costs.
11467         (ipa_read_node_info): Read move costs.
11468         (ipa_update_after_lto_read): Do not recompute node params.
11469         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11470         (ipa_get_param): Check we are not in WPA.
11471         (ipa_get_param_move_cost): New.
11472         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11473         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11474         parm numbers to be present.
11476 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11478         PR rtl-optimization/58048
11479         * lra-constraints.c (process_alt_operands): Don't check asm
11480         operand on register.
11482 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11484         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11485         the implied StoreLoad barrier for atomic operations if before.
11487 2013-08-02  Jan Hubicka  <jh@suse.cz>
11488             Martin Liska  <marxin.liska@gmail.com>
11490         * cgraph.c (cgraph_function_body_availability): Do not check
11491         cgraph flags.
11492         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11493         symtab_node_availability): Declare.
11494         * ipa.c (can_replace_by_local_alias): New.
11495         (function_and_variable_visibility): Use it.
11496         * symtab.c (symtab_for_node_and_aliases,
11497         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11499 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11501         PR rtl-optimization/57963
11502         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11503         (lra_constraints): Use them.
11505 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11507         * config/arm/types.md (define_attr "type"): Add "load_acq"
11508         and "store_rel".
11509         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11510         changes.
11511         (cortex_a53_store1): Likewise.
11513 2013-08-01  Jan Hubicka  <jh@suse.cz>
11515         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11516         partitions are not needed.
11518 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11520         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11521         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11522         MAYBE_NON_Q_CLASS_P where appropriate.
11524 2013-08-01  Jan Hubicka  <jh@suse.cz>
11526         * cgraph.h (release_function_body): Declare.
11527         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11528         of unused delcarations.
11530 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11532         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11533         RTL form when subtracting a constant.
11535 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11537         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11538         Generate canonical plus rtx with negated immediate instead of minus
11539         where appropriate.
11540         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11542 2013-08-01  Jan Hubicka  <jh@suse.cz>
11544         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11545         (cgraph_release_function_body): Likewise.
11546         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11547         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11548         to used_as_abstract_origin.
11549         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11550         symbols used as abstract origins.
11551         * cgraphunit.c (analyze_functions): Update.
11552         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11553         used_as_abstract_origin.
11554         * tree-inline.c (tree_function_versioning): Update
11555         used_as_abstract_origin; be ready for DECL_RESULT and
11556         DECL_ARGUMENTS to be NULL.
11558         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11559         for abstract functions.
11560         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11561         real symbols.
11563 2013-08-01  Jan Hubicka  <jh@suse.cz>
11565         * profile.c (compute_value_histograms): Fix thinko.
11567 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11569         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11570         aarch-common-protos.h to extra_headers.
11571         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11572         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11573         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11575 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11577         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11578         Include "../arm/types.md".  Define "type" attribute for all patterns.
11579         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11580         attribute changes.
11582 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11584         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11585         to support power8 load fusion.
11586         (fusion_gpr_mem_load): Likewise.
11588         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11590         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11591         declarations for power8 load fusion.
11592         (emit_fusion_gpr_load): Likewise.
11594         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11595         tuning for power8, turn on fusion mode by default.  Turn on sign
11596         extending fusion mode if normal fusion mode is on, and we are at
11597         -O2 or -O3.
11598         (fusion_gpr_load_p): New function, return true if we can fuse an
11599         addis instruction with a dependent load to a GPR.
11600         (emit_fusion_gpr_load): Emit the instructions for power8 load
11601         fusion to GPRs.
11603         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11604         (VSX load fusion peepholes): New peepholes to fuse together an
11605         addi instruction with a VSX load instruction.
11607         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11608         peepholes to fuse an addis instruction with a load to a GPR base
11609         register.  If we are supporting sign extending fusions, convert
11610         sign extending loads to zero extending loads and add an explicit
11611         sign extension.
11613 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11615         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11616         aarch-common-protos.h to extra_headers.
11617         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11618         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11619         (arm_early_store_addr_dep): Likewise.
11620         (arm_no_early_alu_shift_dep): Likewise.
11621         (arm_no_early_alu_shift_value_dep): Likewise.
11622         (arm_no_early_mul_dep): Likewise.
11623         (arm_no_early_store_addr_dep): Likewise.
11624         (arm_mac_accumulator_is_mul_result): Likewise.
11625         (arm_mac_accumulator_is_result): Likewise.
11626         * config/arm/aarch-common.c: ... here.  New file.
11627         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11628         here to ...
11629         (arm_early_store_addr_dep): Likewise.
11630         (arm_no_early_alu_shift_dep): Likewise.
11631         (arm_no_early_alu_shift_value_dep): Likewise.
11632         (arm_no_early_mul_dep): Likewise.
11633         (arm_no_early_store_addr_dep): Likewise.
11634         (arm_mac_accumulator_is_mul_result): Likewise.
11635         (arm_mac_accumulator_is_result): Likewise.
11636         * config/arm/aarch-common-protos.h: ... here.  New file.
11637         * config/arm/t-arm (aarch-common.o): Define.
11639 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11641         * config/arm/arm.md: Include new file "types.md".
11642         (define_attr "type"): Move from here to ...
11643         (define_attr "mul32"): Likewise.
11644         (define_attr "mul64"): Likewise.
11645         * config/arm/types.md: ... here.  New file.
11647 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11649         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11650         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11652 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11654         * gen-pass-instances.awk: Fix offset of substr().
11656 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11658         * Makefile.in (pass-instances.def): New.
11659         (passes.o): Replace dependency on passes.def with one on
11660         pass-instances.def
11662         * gen-pass-instances.awk: New.
11664         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11665         rather than passes.def, updating local definition of NEXT_PASS
11666         macro to add an extra NUM parameter (currently unused).
11668 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11670         * Makefile.in (PASS_MANAGER_H): New.
11671         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11672         (passes.o): Likewise.
11673         (statistics.o): Likewise.
11674         (cgraphunit.o): Likewise.
11675         (context.o): Depend on PASS_MANAGER_H.
11677         * pass_manager.h: New.
11679         * cgraphunit.c (cgraph_add_new_function): Update for moves
11680         of globals to fields of pass_manager.
11681         (analyze_function): Likewise.
11682         (expand_function): Likewise.
11683         (ipa_passes): Likewise.
11684         (compile): Likewise.
11686         * context.c (context::context): New.
11687         * context.h  (context::context): New.
11688         (context::get_passes): New.
11689         (context::passes_): New.
11691         * lto-cgraph.c (input_node): Update for moves of globals to
11692         fields of pass_manager.
11694         * passes.c (all_passes): Remove, in favor of a field of the
11695         same name within the new class pass_manager.
11696         (all_small_ipa_passes): Likewise.
11697         (all_lowering_passes): Likewise.
11698         (all_regular_ipa_passes): Likewise.
11699         (all_late_ipa_passes): Likewise.
11700         (all_lto_gen_passes): Likewise.
11701         (passes_by_id): Likewise.
11702         (passes_by_id_size): Likewise.
11703         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11704         the new class pass_manager.
11705         (set_pass_for_id): Convert to...
11706         (pass_manager::set_pass_for_id): ...method.
11707         (get_pass_for_id): Convert to...
11708         (pass_manager::get_pass_for_id): ...method.
11709         (register_one_dump_file): Move body of implementation into...
11710         (pass_manager::register_one_dump_file): ...here.
11711         (register_dump_files_1): Convert to...
11712         (pass_manager::register_dump_files_1): ...method.
11713         (register_dump_files): Convert to...
11714         (pass_manager::register_dump_files): ...method.
11715         (create_pass_tab): Update for moves of globals to fields of
11716         pass_manager.
11717         (dump_passes): Move body of implementation into...
11718         (pass_manager::dump_passes): ...here.
11719         (register_pass): Move body of implementation into...
11720         (pass_manager::register_pass): ...here.
11721         (init_optimization_passes): Convert into...
11722         (pass_manager::pass_manager): ...constructor for new
11723         pass_manager class, and initialize the pass_lists array.
11724         (check_profile_consistency): Update for moves of globals to
11725         fields of pass_manager.
11726         (dump_profile_report): Move body of implementation into...
11727         (pass_manager::dump_profile_report): ...here.
11728         (ipa_write_summaries_1): Update for moves of pass lists from
11729         being globals to fields of pass_manager.
11730         (ipa_write_optimization_summaries): Likewise.
11731         (ipa_read_summaries):  Likewise.
11732         (ipa_read_optimization_summaries): Likewise.
11733         (execute_all_ipa_stmt_fixups): Likewise.
11735         * statistics.c (statistics_fini): Update for moves of globals to
11736         fields of pass_manager.
11738         * toplev.c (general_init): Replace call to
11739         init_optimization_passes with construction of the pass_manager
11740         instance.
11742         * tree-pass.h (all_passes): Remove, in favor of a field of the
11743         same name within the new class pass_manager.
11744         (all_small_ipa_passes): Likewise.
11745         (all_lowering_passes): Likewise.
11746         (all_regular_ipa_passes): Likewise.
11747         (all_lto_gen_passes): Likewise.
11748         (all_late_ipa_passes): Likewise.
11749         (passes_by_id): Likewise.
11750         (passes_by_id_size): Likewise.
11751         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11752         the new class pass_manager.
11753         (get_pass_for_id): Remove.
11755 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11757         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11758         configs.
11760 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11762         * arm.md (mulhi3): New expand pattern.
11764 2013-07-30  Jan Hubicka  <jh@suse.cz>
11765             Martin Liska  <marxin.liska@gmail.com>
11767         * profile.c (compute_value_histograms): Do not ICE when
11768         there is mismatch only on some counters.
11770 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11772         PR rtl-optimization/57637
11773         * function.c (move_insn_for_shrink_wrap): Also check the
11774         GEN set of the LIVE problem for the liveness analysis
11775         if it exists, otherwise give up.
11777 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11779         PR tree-optimization/57993
11780         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11781         replaced statement in the candidate table.
11782         (phi_add_costs): Return infinite cost when the hidden basis does
11783         not dominate all phis on which the candidate is dependent.
11784         (replace_one_candidate): Record replaced statement in the
11785         candidate table.
11787 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11789         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11790         (ashlv2si3): New expander.
11791         (*ashlv2si3_i): New define_insn_and_split.
11792         * predicates.md (float_operation): Allow patterns with three
11793         basic sub-patterns.
11795         PR rtl-optimization/58021
11796         * mode-switching.c (create_pre_exit): Always split off preceding
11797         insns if we are not at the basic block head.
11799 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11801         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11802         (UCLIBC_DYNAMIC_LINKER): New macro.
11803         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11804         `-mnan=2008'.
11805         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11806         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11807         `-mnan=2008'.
11808         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11809         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11810         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11811         for SF and DF modes.  Use ieee_quad_format for TF mode.
11812         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11813         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11814         (mips_option_override): Handle `-mnan=legacy'.
11815         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11816         `-mabs=2008' and `-mnan=2008'.
11817         (OPTION_DEFAULT_SPECS): Add "nan" default.
11818         (ASM_SPEC): Handle `-mnan='.
11819         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11820         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11821         comment accordingly.
11822         (neg<mode>2): Likewise.
11823         * config/mips/mips.opt (mabs, mnan): New options.
11824         * doc/install.texi (Configuration): Document `--with-nan=' option.
11825         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11826         `-mnan=' options.
11827         (MIPS Options): Document them.
11828         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11829         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11830         * configure: Regenerate.
11831         * config.in: Regenerate.
11833 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11835         * config/i386/i386.md (float post-reload splitters): Do not check
11836         for subregs of SSE registers.
11838 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11839             H.J. Lu  <hongjiu.lu@intel.com>
11841         PR target/57954
11842         PR target/57988
11843         * config/i386/i386.md (post-reload splitter
11844         to avoid partial SSE reg dependency stalls): New pattern.
11846 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11848         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11850 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11852         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11853         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11855 2013-07-26  Cary Coutant  <ccoutant@google.com>
11857         * dwarf2out.c (die_checksum_ordered): Don't include template
11858         instantiations in signature.
11859         (is_template_parameter): New function.
11860         (is_template_instantiation): New function.
11861         (generate_skeleton_bottom_up): Don't include template instantiations
11862         in type unit DIE.
11863         (generate_skeleton): Likewise.
11864         (break_out_comdat_types): Move recursive call to break out nested
11865         types earlier.
11866         (prune_unused_types_mark_generic_parms_dies): Call
11867         is_template_parameter.
11869 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11871         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11872         uses vector registers.
11873         * config/aarch64/iterators.md: Add attributes rtn and vas.
11875 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11876             Richard Earnshaw  <richard.earnshaw@arm.com>
11878         * combine.c (simplify_comparison): Re-canonicalize operands
11879         where appropriate.
11880         * config/arm/arm.md (movcond_addsi): New splitter.
11882 2013-07-25  Sterling Augustine  <saugustine@google.com>
11884         * dwarf2out.c (size_of_pubnames): Move code to...
11885         (include_pubname_in_output): ...here.  New.
11886         (want_pubnames): Rearrange.
11887         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11889 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11891         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11893 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11895         PR target/38836
11896         * doc/extend.texi: Remove obsolete builtins. Fix
11897         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11899 2013-07-25  Jan Hubicka  <jh@suse.cz>
11901         * cgraph.c (release_function_body): Break out from ...
11902         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11903         and DECL_ARGUMENTS.
11904         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11905         old_tree in the map.
11906         (create_specialized_node): Update.
11907         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11908         into index.
11909         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11910         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11911         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11912         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11913         DECL_RESULT.
11915 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11917         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11918         addsi3_carryin_alt2_<optab>): Correct output template.
11920 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11922         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11923         Adjust for arm_restrict_it.
11924         Remove trailing whitespace.
11926 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11928         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11929         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11931         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11933 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11935         PR rtl-optimization/57960
11936         * lra-constraints.c (process_alt_operands): Use the right mode
11937         when checking strict_low.
11939 2013-07-25  Jan Hubicka  <jh@suse.cz>
11941         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11942         * cgraph.c (cgraph_remove_node): Do not release function body
11943         when in cgraph streaming.
11944         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11945         in other partitions are not considered reachable; fix handling of
11946         clones.
11948 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11950         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11952 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11954         PR target/19599
11955         PR target/57731
11956         PR target/57837
11957         * config/arm/arm.md ("*sibcall_insn): Replace use of
11958         Ss with US. Adjust output for v5 and v4t.
11959         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11961         * config/arm/constraints.md ("Ss"): Rename to US.
11963 2013-07-25  Terry Guo  <terry.guo@arm.com>
11965         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11966         shift_add/shift_sub0/shift_sub1 RTXs.
11968 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11969             Anton Blanchard  <anton@au1.ibm.com>
11971         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11972         (altivec_vpks<VI_char>ss): Likewise.
11973         (altivec_vpks<VI_char>us): Likewise.
11974         (altivec_vpku<VI_char>us): Likewise.
11975         (altivec_vpku<VI_char>um): Likewise.
11977 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11979         Introduce context class.
11981         * Makefile.in (CONTEXT_H): New.
11982         (OBJS): Add context.o.
11983         (toplev.o): Add CONTEXT_H to dependencies.
11984         (context.o): New.
11986         * toplev.c (general_init): Create the singleton gcc::context instance.
11988         * context.c: New.
11990         * context.h: New.
11992 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11994         PR rtl-optimization/57968
11995         * mode-switching.c (create_pre_exit): Allow instructions that
11996         don't set a return register to need a non-exit mode.
11998 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11999             Anton Blanchard  <anton@au1.ibm.com>
12001         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
12002         operands to vperm for little endian.
12003         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
12004         of lvsl to create the control mask for a vperm for little endian.
12006 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12007             Anton Blanchard  <anton@au1.ibm.com>
12009         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
12010         two operands for little-endian.
12012 2013-07-23  Steve Ellcey  <sellcey@mips.com>
12014         * config/mips/mips.c (mips_case_values_threshold): New.
12015         (TARGET_CASE_VALUES_THRESHOLD): Define.
12017 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12018             Anton Blanchard  <anton@au1.ibm.com>
12020         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
12021         selection of field for vector splat in little endian mode.
12023 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
12025         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
12026         expanders to rs6000.md.
12027         (ior<mode>3): Likewise.
12028         (and<mode>3): Likewise.
12029         (one_cmpl<mode>2): Likewise.
12030         (nor<mode>3): Likewise.
12031         (andc<mode>3): Likewise.
12032         (eqv<mode>3): Likewise.
12033         (nand<mode>3): Likewise.
12034         (orc<mode>3): Likewise.
12036         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
12037         declaration.
12039         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
12040         to split multi-word logical operations.
12041         (rs6000_split_logical_di): Likewise.
12042         (rs6000_split_logical): Likewise.
12044         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
12045         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
12046         and allow TImode operations in 32-bit.
12047         (vsx_and<mode>3_64bit): Likewise.
12048         (vsx_ior<mode>3_32bit): Likewise.
12049         (vsx_ior<mode>3_64bit): Likewise.
12050         (vsx_xor<mode>3_32bit): Likewise.
12051         (vsx_xor<mode>3_64bit): Likewise.
12052         (vsx_one_cmpl<mode>2_32bit): Likewise.
12053         (vsx_one_cmpl<mode>2_64bit): Likewise.
12054         (vsx_nor<mode>3_32bit): Likewise.
12055         (vsx_nor<mode>3_64bit): Likewise.
12056         (vsx_andc<mode>3_32bit): Likewise.
12057         (vsx_andc<mode>3_64bit): Likewise.
12058         (vsx_eqv<mode>3_32bit): Likewise.
12059         (vsx_eqv<mode>3_64bit): Likewise.
12060         (vsx_nand<mode>3_32bit): Likewise.
12061         (vsx_nand<mode>3_64bit): Likewise.
12062         (vsx_orc<mode>3_32bit): Likewise.
12063         (vsx_orc<mode>3_64bit): Likewise.
12065         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
12066         logical types in GPRs.
12068         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
12069         logical insns to rs6000.md, and allow TImode operations in
12070         32-bit.
12071         (altivec_ior<mode>3): Likewise.
12072         (altivec_xor<mode>3): Likewise.
12073         (altivec_one_cmpl<mode>2): Likewise.
12074         (altivec_nor<mode>3): Likewise.
12075         (altivec_andc<mode>3): Likewise.
12077         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
12078         attributes for moving the 128-bit logical operations into
12079         rs6000.md.
12080         (BOOL_REGS_OUTPUT): Likewise.
12081         (BOOL_REGS_OP1): Likewise.
12082         (BOOL_REGS_OP2): Likewise.
12083         (BOOL_REGS_UNARY): Likewise.
12084         (BOOL_REGS_AND_CR0): Likewise.
12085         (one_cmpl<mode>2): Add support for DI logical operations on
12086         32-bit, splitting the operations to 32-bit.
12087         (anddi3): Likewise.
12088         (iordi3): Likewise.
12089         (xordi3): Likewise.
12090         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12091         changes to combine the 32/64-bit code, allow logical operations on
12092         TI mode in 32-bit, and to use similar match_operator patterns like
12093         scalar mode uses.  Combine the Altivec and VSX code for logical
12094         operations, and move it here.
12095         (ior<mode>3, 128-bit types): Likewise.
12096         (xor<mode>3, 128-bit types): Likewise.
12097         (one_cmpl<mode>3, 128-bit types): Likewise.
12098         (nor<mode>3, 128-bit types): Likewise.
12099         (andc<mode>3, 128-bit types): Likewise.
12100         (eqv<mode>3, 128-bit types): Likewise.
12101         (nand<mode>3, 128-bit types): Likewise.
12102         (orc<mode>3, 128-bit types): Likewise.
12103         (and<mode>3_internal): Likewise.
12104         (bool<mode>3_internal): Likewise.
12105         (boolc<mode>3_internal1): Likewise.
12106         (boolc<mode>3_internal2): Likewise.
12107         (boolcc<mode>3_internal1): Likewise.
12108         (boolcc<mode>3_internal2): Likewise.
12109         (eqv<mode>3_internal1): Likewise.
12110         (eqv<mode>3_internal2): Likewise.
12111         (one_cmpl1<mode>3_internal): Likewise.
12113 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12115         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12116         Rename flag_stack_usage to flag_stack_usage_info.
12118 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12120         * config/microblaze/sync.md: New file.
12121         * config/microblaze/microblaze.md: Include sync.md
12122         * config/microblaze/microblaze.c: Add print_operand 'y'.
12123         * config/microblaze/constraints.md: Add memory_contraint
12124         'Q' which is a single register.
12126 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12128         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12130 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12132         * reload.c (find_reloads): Exit loop once we find this operand
12133         cannot be reloaded somehow for this alternative.
12135         * reload.c (find_reloads): Exit loop once we find a hard register.
12137         * rtlanal.c (computed_jump_p): Exit loop once we find label
12138         reference is used.
12140         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12142         * cfgloopmanip.c (remove_path): Exit loop after setting
12143         irred_invalidated.
12145         * gensupport.c (subst_dup): Avoid loop if code is not
12146         MATCH_DUP nor MATCH_OP_DUP.
12148 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12150         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12152 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12154         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12155         true for SP_REGNUM if mode == ptr_mode.
12156         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12157         with value R0_REGNUM + 31.
12159 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12161         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12162         pad pointer-typed argument downward.
12164 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12166         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12167         and __ILP32__ when the ILP32 model is in use.
12169 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12171         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12172         (aarch64_load_symref_appropriately): In the case of
12173         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12174         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12175         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12176         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12177         if 'mode' doesn't equal to 'ptr_mode'.
12178         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12179         'vcall_offset'; change to call aarch64_emit_move differently depending
12180         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12181         to calculate the upper bound of 'vcall_offset'.
12182         (aarch64_cannot_force_const_mem): Change to also return true if
12183         mode != ptr_mode.
12184         (aarch64_legitimize_reload_address): In the case of large
12185         displacements, add new local variable 'xmode' and an assertion
12186         based on it; change to use 'xmode' to generate the new rtx and
12187         reload.
12188         (aarch64_asm_trampoline_template): Change to generate the template
12189         differently depending on TARGET_ILP32 or not; change to use
12190         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12191         (aarch64_trampoline_size): Removed.
12192         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12193         and replace immediate literals with it.  Change to use 'ptr_mode'
12194         instead of 'DImode' and call convert_memory_address if the mode
12195         of 'fnaddr' doesn't equal to 'ptr_mode'.
12196         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12197         to output symbol.
12198         (aarch64_elf_asm_destructor): Likewise.
12199         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12200         on TARGET_ILP32 instead of aarch64_trampoline_size.
12201         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12202         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12203         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12204         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12205         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12206         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12207         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12208         depending on the value of 'mode'.
12209         (add_losym_<mode>): New.
12210         (ldr_got_small_<mode>): New, based on ldr_got_small.
12211         (ldr_got_small): Remove.
12212         (ldr_got_small_sidi): New.
12213         * config/aarch64/iterators.md (P): New.
12214         (PTR): Change to 'ptr_mode' in the condition.
12216 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12218         * config.gcc (aarch64*-*-*): Support --with-abi.
12219         (aarch64*-*-elf): Support --with-multilib-list.
12220         (aarch64*-*-linux*): Likewise.
12221         (supported_defaults): Add abi to aarch64*-*-*.
12222         * configure.ac: Mention AArch64 for --with-multilib-list.
12223         * configure: Re-generated.
12224         * config/aarch64/biarchilp32.h: New file.
12225         * config/aarch64/biarchlp64.h: New file.
12226         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12227         (ABI_SPEC): Ditto.
12228         (MULTILIB_DEFAULTS): Ditto.
12229         (DRIVER_SELF_SPECS): Ditto.
12230         (ASM_SPEC): Update to also substitute -mabi.
12231         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12232         file whose name depends on -mabi= and -mbig-endian.
12233         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12234         TARGET_ILP32.
12235         (POINTER_SIZE): New define.
12236         (POINTERS_EXTEND_UNSIGNED): Ditto.
12237         (enum aarch64_abi_type): New enumeration tag.
12238         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12239         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12240         (TARGET_ILP32): New define.
12241         * config/aarch64/aarch64.opt (mabi): New.
12242         (aarch64_abi): New.
12243         (ilp32, lp64): New values for -mabi.
12244         * config/aarch64/t-aarch64 (comma): New define.
12245         (MULTILIB_OPTIONS): Ditto.
12246         (MULTILIB_DIRNAMES): Ditto.
12247         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12248         * doc/invoke.texi: Document -mabi for AArch64.
12250 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12252         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12254 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12255             Anton Blanchard  <anton@au1.ibm.com>
12257         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12258         endianness when selecting field to splat.
12260 2013-07-22  Eric Christopher  <echristo@gmail.com>
12262         * dwarf2out.c (die_odr_checksum): New function to use
12263         CHECKSUM_ macros and ULEB128 for DIE tag.
12264         (generate_type_signature): Use.
12266 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12268         * config.gcc (sparc*-*-*): Accept leon3 processor.
12269         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12270         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12271         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12272         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12273         (mfix-ut699): Adjust comment.
12274         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12275         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12276         (CPP_CPU_SPEC): Likewise.
12277         (ASM_CPU_SPEC): Likewise.
12278         * config/sparc/sparc.c (leon3_cost): New constant.
12279         (sparc_option_override): Add leon3 support.
12280         (mem_ref): New function.
12281         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12282         (sparc_do_work_around_errata): Look into the instruction in the delay
12283         slot and adjust accordingly.  Add fix for the data cache nullify issues
12284         of the UT699.  Change insertion position for the NOP.
12285         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12286         (leon3_load): New reservation.
12287         (leon_store): Bump latency to 2.
12288         (grfpu): New automaton.
12289         (grfpu_alu): New unit.
12290         (grfpu_ds): Likewise.
12291         (leon_fp_alu): Adjust.
12292         (leon_fp_mult): Delete.
12293         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12294         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12295         * config/sparc/sparc.md (cpu): Add leon3.
12296         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12297         (swapsi): Likewise.
12298         (atomic_test_and_set): Likewise.
12299         (ldstub): Likewise.
12301 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12303         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12304         default for R5900 targets.
12305         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12306         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12307         * config/mips/mips.c (mips_option_override): Report an error for
12308         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12309         for -march=r5900 -mhard-float.
12311 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12313         * df-problems.c (can_move_insns_across): Exit loop once we
12314         find a non-fixed, non-global register.
12316         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12317         setting can_throw.
12319         * omega.c (omega_eliminate_red): Break after setting red_found.
12320         (omega_problem_has_red_equations): Similarly after setting found.
12321         (omega_query_variable): Similarly after setting coupled.
12323 2013-07-22  Marek Polacek  <polacek@redhat.com>
12325         * gimplify.c: Don't include gimple.h twice.
12327 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12329         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12330         instead of TARGET_THUMB1.
12331         (Pz): New constraint.
12332         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12333         encodings.
12334         (compare_negsi_si): Likewise.
12335         (compare_addsi2_op0): Likewise.
12336         (compare_addsi2_op1): Likewise.
12337         (addsi3_carryin_<optab>): Likewise.
12338         (addsi3_carryin_alt2_<optab>): Likewise.
12339         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12340         for arm_restrict_it.
12341         (subsi3_carryin): Likewise.
12342         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12343         (minmax_arithsi): Disable for arm_restrict_it.
12344         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12345         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12346         (satsi_<SAT:code>_shift): Likewise.
12347         (arm_shiftsi3): Add alternative for 16-bit encoding.
12348         (arm32_movhf): Disable for arm_restrict_it.
12349         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12350         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12352 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12354         * config/arm/arm.md (attribute "insn"): Delete.
12355         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12356         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12357         (not_shiftsi): Update for attribute change.
12358         (not_shiftsi_compare0): Likewise.
12359         (not_shiftsi_compare0_scratch): Likewise.
12360         (arm_one_cmplsi2): Likewise.
12361         (thumb1_one_cmplsi2): Likewise.
12362         (notsi_compare0): Likewise.
12363         (notsi_compare0_scratch): Likewise.
12364         (thumb1_movdi_insn): Likewise.
12365         (arm_movsi_insn): Likewise.
12366         (movhi_insn_arch4): Likewise.
12367         (movhi_bytes): Likewise.
12368         (arm_movqi_insn): Likewise.
12369         (thumb1_movqi_insn): Likewise.
12370         (arm32_movhf): Likewise.
12371         (thumb1_movhf): Likewise.
12372         (arm_movsf_soft_insn): Likewise.
12373         (thumb1_movsf_insn): Likewise.
12374         (thumb_movdf_insn): Likewise.
12375         (movsicc_insn): Likewise.
12376         (movsfcc_soft_insn): Likewise.
12377         (and_scc): Likewise.
12378         (cond_move): Likewise.
12379         (if_move_not): Likewise.
12380         (if_not_move): Likewise.
12381         (if_shift_move): Likewise.
12382         (if_move_shift): Likewise.
12383         (if_shift_shift): Likewise.
12384         (if_not_arith): Likewise.
12385         (if_arith_not): Likewise.
12386         (cond_move_not): Likewise.
12387         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12388         (neon_mov<mode>): Likewise.
12389         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12390         (thumb2_movsi_vfp): Likewise.
12391         (movsf_vfp): Likewise.
12392         (thumb2_movsf_vfp): Likewise.
12393         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12394         change.
12395         (cortexa7_older_only): Likewise.
12396         (cortexa7_younger): Likewise.
12397         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12398         (1020alu_shift_op): Likewise.
12399         (1020alu_shift_reg_op): Likewise.
12400         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12401         (alu_shift_op): Likewise.
12402         (alu_shift_reg_op): Likewise.
12403         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12404         (11_alu_shift_op): Likewise.
12405         (11_alu_shift_reg_op): Likewise.
12406         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12407         (9_alu_shift_reg_op): Likewise.
12408         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12409         change.
12410         (cortex_a15_alu_shift): Likewise.
12411         (cortex_a15_alu_shift_reg): Likewise.
12412         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12413         (cortex_a5_alu_shift): Likewise.
12414         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12415         change.
12416         (cortex_a53_alu_shift): Likewise.
12417         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12418         change.
12419         (cortex_a7_alu_reg): Likewise.
12420         (cortex_a7_alu_shift): Likewise.
12421         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12422         (cortex_a8_alu_shift): Likewise.
12423         (cortex_a8_alu_shift_reg): Likewise.
12424         (cortex_a8_mov): Likewise.
12425         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12426         (cortex_a9_dp_shift): Likewise.
12427         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12428         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12429         (cortex_r4_mov): Likewise.
12430         (cortex_r4_alu_shift): Likewise.
12431         (cortex_r4_alu_shift_reg): Likewise.
12432         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12433         (526_alu_shift_op): Likewise.
12434         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12435         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12436         (626te_alu_shift_op): Likewise.
12437         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12438         (726te_alu_op): Likewise.
12439         (726te_alu_shift_op): Likewise.
12440         (726te_alu_shift_reg_op): Likewise.
12441         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12442         (mp626_alu_shift_op): Likewise.
12443         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12444         (pj4_alu_e1_conds): Likewise.
12445         (pj4_alu): Likewise.
12446         (pj4_alu_conds): Likewise.
12447         (pj4_shift): Likewise.
12448         (pj4_shift_conds): Likewise.
12449         (pj4_alu_shift): Likewise.
12450         (pj4_alu_shift_conds): Likewise.
12452 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12454         * config/arm/predicates.md (shiftable_operator_strict_it):
12455         New predicate.
12456         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12457         Disable cond_exec version for arm_restrict_it.
12458         (thumb2_smaxsi3): Convert to generate cond_exec.
12459         (thumb2_sminsi3): Likewise.
12460         (thumb32_umaxsi3): Likewise.
12461         (thumb2_uminsi3): Likewise.
12462         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12463         (thumb2_neg_abssi2): Likewise.
12464         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12465         (thumb2_movsicc_insn): Adjust alternatives.
12466         (thumb2_mov_negscc): Disable for arm_restrict_it.
12467         (thumb2_mov_negscc_strict_it): New pattern.
12468         (thumb2_mov_notscc_strict_it): New pattern.
12469         (thumb2_mov_notscc): Disable for arm_restrict_it.
12470         (thumb2_ior_scc): Likewise.
12471         (thumb2_ior_scc_strict_it): New pattern.
12472         (thumb2_cond_move): Adjust for arm_restrict_it.
12473         (thumb2_cond_arith): Disable for arm_restrict_it.
12474         (thumb2_cond_arith_strict_it): New pattern.
12475         (thumb2_cond_sub): Adjust for arm_restrict_it.
12476         (thumb2_movcond): Likewise.
12477         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12478         (thumb2_zero_extendhisi2_v6): Likewise.
12479         (thumb2_zero_extendqisi2_v6): Likewise.
12480         (orsi_notsi_si): Likewise.
12481         (orsi_not_shiftsi_si): Likewise.
12483 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12485         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12486         instruction sequence is 1 byte shorter.
12488 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12490         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12491         it is not needed after split.
12493 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12495         PR target/51784
12496         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12497         second label for nonlocal goto receivers. Don't output pic base labels
12498         unless we're producing PIC; mark that action unreachable().
12499         (ix86_save_reg): If the function contains a nonlocal label, save the
12500         PIC base reg.
12501         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12502         * config/darwin.c (emitted_pic_label_num): New GTY.
12503         (update_pic_label_number_if_needed): New.
12504         (machopic_output_function_base_name): Adjust for nonlocal receiver
12505         case.
12506         (machopic_should_output_picbase_label): New.
12507         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12508         (nonlocal_goto_receiver): New insn and split.
12510 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12512         * config/aarch64/aarch64-builtins.c
12513         (aarch64_fold_builtin): Fold abs in all modes.
12514         * config/aarch64/aarch64-simd-builtins.def
12515         (abs): Enable for all modes.
12516         * config/aarch64/arm_neon.h
12517         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12518         (vabs_f64): Add missing intrinsic.
12520 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12522         * config/aarch64/arm_neon.h (vabs_s64): New function
12524 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12526         PR target/57516
12527         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12528         * config/avr/avr.md (adjust_len): Add `round'.
12529         * config/avr/avr-protos.h (avr_out_round): New prototype.
12530         (avr_out_plus): Add `out_label' argument.
12531         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12532         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12533         Handle the case where `insn' is just a pattern.
12534         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12535         (avr_out_round): New function.
12536         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12538 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12540         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12541         Add check for flag_stack_usage to handle -fstack-usage support
12543 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12545         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12546         interaction for new Power8 flags and VSX.
12548 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12550         PR middle-end/57698
12551         * tree-inline.c (expand_call_inline): Emit errors during
12552         early_inlining only if optimization is not turned on.
12554 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12556         * passes.def: New.
12558         * passes.c (init_optimization_passes): Move the construction of
12559         the pass hierarchy into a new passes.def file.
12561         * Makefile.in (passes.o): Add dependency on passes.def.
12563 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12565         * passes.c (init_optimization_passes): Introduce macros for
12566         constructing the tree of passes (INSERT_PASSES_AFTER,
12567         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12568         TERMINATE_PASS_LIST).
12570 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12571             Wei Mi  <wmi@google.com>
12573         PR rtl-optimization/57878
12574         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12575         top.
12576         (reload_pseudo_compare_func): Check nregs first for reload
12577         pseudos.
12579 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12581         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12583 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12585         * read-rtl.c (validate_const_int): Once an invalid character is
12586         seen, quit the loop.
12588         * gengtype.c (write_roots): Similarly once we find the "deletable"
12589         or "if_marked" option.
12591 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12593         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12594         "xtab" and "sat".  Move value "clz" from here to ...
12595         (attriubte "type"): ... here.
12596         (satsi_<SAT:code>): Delete "insn" attribute.
12597         (satsi_<SAT:code>_shift): Likewise.
12598         (arm_zero_extendqisi2addsi): Likewise.
12599         (arm_extendqisi2addsi): Likewise.
12600         (clzsi2): Update for attribute changes.
12601         (rbitsi2): Likewise.
12602         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12603         attribute.
12604         (arm_usatsihi): Likewise.
12605         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12607 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12609         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12610         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12611         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12612         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12613         in alphabetical order.
12614         (attribute "core_cycles"): Update for attribute changes.
12615         (arm_addsi3): Likewise.
12616         (addsi3_compare0): Likewise.
12617         (addsi3_compare0_scratch): Likewise.
12618         (addsi3_compare_op1): Likewise.
12619         (addsi3_compare_op2): Likewise.
12620         (compare_addsi2_op0): Likewise.
12621         (compare_addsi2_op1): Likewise.
12622         (addsi3_carryin_shift_<optab>): Likewise.
12623         (subsi3_carryin_shift): Likewise.
12624         (rsbsi3_carryin_shift): Likewise.
12625         (arm_subsi3_insn): Likewise.
12626         (subsi3_compare0): Likewise.
12627         (subsi3_compare): Likewise.
12628         (arm_andsi3_insn): Likewise.
12629         (thumb1_andsi3_insn): Likewise.
12630         (andsi3_compare0): Likewise.
12631         (andsi3_compare0_scratch): Likewise.
12632         (zeroextractsi_compare0_scratch
12633         (andsi_not_shiftsi_si): Likewise.
12634         (iorsi3_insn): Likewise.
12635         (iorsi3_compare0): Likewise.
12636         (iorsi3_compare0_scratch): Likewise.
12637         (arm_xorsi3): Likewise.
12638         (thumb1_xorsi3_insn): Likewise.
12639         (xorsi3_compare0): Likewise.
12640         (xorsi3_compare0_scratch): Likewise.
12641         (satsi_<SAT:code>_shift): Likewise.
12642         (rrx): Likewise.
12643         (arm_shiftsi3): Likewise.
12644         (shiftsi3_compare0): Likewise.
12645         (not_shiftsi): Likewise.
12646         (not_shiftsi_compare0): Likewise.
12647         (not_shiftsi_compare0_scratch): Likewise.
12648         (arm_one_cmplsi2): Likewise.
12649         (thumb_one_complsi2): Likewise.
12650         (notsi_compare0): Likewise.
12651         (notsi_compare0_scratch): Likewise.
12652         (thumb1_zero_extendhisi2): Likewise.
12653         (arm_zero_extendhisi2): Likewise.
12654         (arm_zero_extendhisi2_v6): Likewise.
12655         (arm_zero_extendhisi2addsi): Likewise.
12656         (thumb1_zero_extendqisi2): Likewise.
12657         (thumb1_zero_extendqisi2_v6): Likewise.
12658         (arm_zero_extendqisi2): Likewise.
12659         (arm_zero_extendqisi2_v6): Likewise.
12660         (arm_zero_extendqisi2addsi): Likewise.
12661         (thumb1_extendhisi2): Likewise.
12662         (arm_extendhisi2): Likewise.
12663         (arm_extendhisi2_v6): Likewise.
12664         (arm_extendqisi): Likewise.
12665         (arm_extendqisi_v6): Likewise.
12666         (arm_extendqisi2addsi): Likewise.
12667         (thumb1_extendqisi2): Likewise.
12668         (thumb1_movdi_insn): Likewise.
12669         (arm_movsi_insn): Likewise.
12670         (movsi_compare0): Likewise.
12671         (movhi_insn_arch4): Likewise.
12672         (movhi_bytes): Likewise.
12673         (arm_movqi_insn): Likewise.
12674         (thumb1_movqi_insn): Likewise.
12675         (arm32_movhf): Likewise.
12676         (thumb1_movhf): Likewise.
12677         (arm_movsf_soft_insn): Likewise.
12678         (thumb1_movsf_insn): Likewise.
12679         (movdf_soft_insn): Likewise.
12680         (thumb_movdf_insn): Likewise.
12681         (arm_cmpsi_insn): Likewise.
12682         (cmpsi_shiftsi): Likewise.
12683         (cmpsi_shiftsi_swp): Likewise.
12684         (arm_cmpsi_negshiftsi_si): Likewise.
12685         (movsicc_insn): Likewise.
12686         (movsfcc_soft_insn): Likewise.
12687         (arith_shiftsi): Likewise.
12688         (arith_shiftsi_compare0
12689         (arith_shiftsi_compare0_scratch
12690         (sub_shiftsi): Likewise.
12691         (sub_shiftsi_compare0
12692         (sub_shiftsi_compare0_scratch
12693         (and_scc): Likewise.
12694         (cond_move): Likewise.
12695         (if_plus_move): Likewise.
12696         (if_move_plus): Likewise.
12697         (if_move_not): Likewise.
12698         (if_not_move): Likewise.
12699         (if_shift_move): Likewise.
12700         (if_move_shift): Likewise.
12701         (if_shift_shift): Likewise.
12702         (if_not_arith): Likewise.
12703         (if_arith_not): Likewise.
12704         (cond_move_not): Likewise.
12705         (thumb1_ashlsi3): Set type attribute.
12706         (thumb1_ashrsi3): Likewise.
12707         (thumb1_lshrsi3): Likewise.
12708         (thumb1_rotrsi3): Likewise.
12709         (shiftsi3_compare0_scratch): Likewise.
12710         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12711         (neon_mov<mode>): Likewise.
12712         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12713         attribute changes.
12714         (thumb2_movsi_insn): Likewise.
12715         (thumb2_cmpsi_neg_shiftsi): Likewise.
12716         (thumb2_extendqisi_v6): Likewise.
12717         (thumb2_zero_extendhisi2_v6): Likewise.
12718         (thumb2_zero_extendqisi2_v6): Likewise.
12719         (thumb2_shiftsi3_short): Likewise.
12720         (thumb2_addsi3_compare0_scratch): Likewise.
12721         (orsi_not_shiftsi_si): Likewise.
12722         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12723         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12724         changes.
12725         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12726         (1020alu_shift_op): Likewise.
12727         (1020alu_shift_reg_op): Likewise.
12728         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12729         (alu_shift_op): Likewise.
12730         (alu_shift_reg_op): Likewise.
12731         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12732         (11_alu_shift_op): Likewise.
12733         (11_alu_shift_reg_op): Likewise.
12734         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12735         (9_alu_shift_reg_op): Likewise.
12736         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12737         changes.
12738         (cortex_a15_alu_shift): Likewise.
12739         (cortex_a15_alu_shift_reg): Likewise.
12740         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12741         changes.
12742         (cortex_a5_alu_shift): Likewise.
12743         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12744         changes.
12745         (cortex_a53_alu_shift): Likewise.
12746         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12747         changes.
12748         (cortex_a7_alu_reg): Likewise.
12749         (cortex_a7_alu_shift): Likewise.
12750         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12751         changes.
12752         (cortex_a8_alu_shift): Likewise.
12753         (cortex_a8_alu_shift_reg): Likewise.
12754         (cortex_a8_mov): Likewise.
12755         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12756         (cortex_a9_dp_shift): Likewise.
12757         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12758         changes.
12759         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12760         changes.
12761         (cortex_r4_mov): Likewise.
12762         (cortex_r4_alu_shift): Likewise.
12763         (cortex_r4_alu_shift_reg): Likewise.
12764         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12765         (526_alu_shift_op): Likewise.
12766         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12767         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12768         (626te_alu_shift_op): Likewise.
12769         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12770         (726te_alu_op): Likewise.
12771         (726te_alu_shift_op): Likewise.
12772         (726te_alu_shift_reg_op): Likewise.
12773         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12774         (mp626_alu_shift_op): Likewise.
12775         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12776         (pj4_alu_e1_conds): Likewise.
12777         (pj4_alu): Likewise.
12778         (pj4_alu_conds): Likewise.
12779         (pj4_shift): Likewise.
12780         (pj4_shift_conds): Likewise.
12781         (pj4_alu_shift): Likewise.
12782         (pj4_alu_shift_conds): Likewise.
12783         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12784         changes.
12785         (cortexa7_older_only): Likewise.
12786         (cortexa7_younger): Likewise.
12788 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12790         * ipa-pure-const.c (generate_summary): Rename to...
12791         (pure_const_generate_summary): ... this.
12793 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12795         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12797 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12799         PR target/57909
12800         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12801         usage in HI mode.
12803 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12805         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12806         enabled without -march=zEC12.
12807         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12808         flags to be set.
12810 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12812         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12813         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12814         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12815         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12816         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12817         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12818         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12819         ISA_HAS_FP4.
12820         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12821         and ISA_HAS_NMADD3_NMSUB3.
12822         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12823         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12824         (nmsub4<mode>, nmsub3<mode>): Likewise.
12825         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12827 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12829         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12830         TARGET_MIPS5400 checking.
12832 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12833             Peter Bergner  <bergner@vnet.ibm.com>
12835         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12836         registers in the comment.
12837         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12838         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12839         rather than FIRST_PSEUDO_REGISTERS.
12841 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12843         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12844         enable extra ISA flags with TARGET_HTM.
12846 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12848         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12849         Fix comment typos.
12851 2013-07-15  Cong Hou  <congh@google.com>
12853         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12854         in compare function for sorting.
12856 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12858         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12859         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12860         * config/rs6000/rs6000.opt: Add -mhtm option.
12861         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12862         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12863         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12864         __HTM__ if the HTM instructions are available.
12865         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12866         htm_spr_reg_operand): New define_predicates.
12867         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12868         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12869         Include htm.md.
12870         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12871         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12872         HTM builtin functions.
12873         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12874         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12875         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12876         (rs6000_builtin_mask_calculate): Likewise.
12877         (rs6000_option_override_internal): Likewise.
12878         (bdesc_htm): Add new HTM builtin support.
12879         (htm_spr_num): New function.
12880         (htm_spr_regno): Likewise.
12881         (rs6000_htm_spr_icode): Likewise.
12882         (htm_expand_builtin): Likewise.
12883         (htm_init_builtins): Likewise.
12884         (rs6000_expand_builtin): Add support for HTM builtin functions.
12885         (rs6000_init_builtins): Likewise.
12886         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12887         option.
12888         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12889         (TARGET_HTM, MASK_HTM): Define macros.
12890         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12891         (FIXED_REGISTERS): Likewise.
12892         (CALL_USED_REGISTERS): Likewise.
12893         (CALL_REALLY_USED_REGISTERS): Likewise.
12894         (REG_ALLOC_ORDER): Likewise.
12895         (enum reg_class): Likewise.
12896         (REG_CLASS_NAMES): Likewise.
12897         (REG_CLASS_CONTENTS): Likewise.
12898         (REGISTER_NAMES): Likewise.
12899         (ADDITIONAL_REGISTER_NAMES): Likewise.
12900         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12901         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12902         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12903         * config/rs6000/htm.md: New file.
12904         * config/rs6000/htmintrin.h: New file.
12905         * config/rs6000/htmxlintrin.h: New file.
12907 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12909         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12910         Define SYMBOL_TINY_GOT, update comment.
12911         * config/aarch64/aarch64.c
12912         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12913         (aarch64_expand_mov_immediate): Likewise.
12914         (aarch64_print_operand): Likewise.
12915         (aarch64_classify_symbol): Likewise.
12916         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12917         (ldr_got_tiny): Define.
12919 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12921         PR tree-optimization/54094
12922         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12923           scheduling dimension for the parallelism check from the polyhedral
12924           information in the AST.
12925         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12926           in 2D + 1 form.
12928 2013-07-13  Jason Merrill  <jason@redhat.com>
12930         * print-tree.c (debug_vec_tree): Use debug_raw.
12931         (debug_raw (vec<tree, va_gc> &)): New.
12932         (debug_raw (vec<tree, va_gc> *)): New.
12933         * tree.h: Declare them.
12935 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12937         * ifcvt.c (ifcvt_after_combine): New static variable.
12938         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12939         for size.
12940         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12941         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12942         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12944 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12946         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12948 2013-07-12  Michael Matz  <matz@suse.de>
12950         PR middle-end/55771
12951         * convert.c (convert_to_real): Reject non-float inner types.
12953 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12955         * config/aarch64/aarch64-protos.h
12956         (aarch64_simd_immediate_valid_for_move): Remove.
12957         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12958         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12959         cases.
12960         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12962 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12964         * config/mips/mips.c (mips_conditional_register_usage): Do not
12965         use t[0-7] registers in MIPS16 mode when optimizing for size.
12967 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12969         * config/i386/i386.c (dispatch_function_versions): Fix array
12970         indexing of function_version_info to match actual_versions.
12972 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12974         * vec.h (struct va_gc): Move release out-of-line.
12975         (va_gc::release): Call ggc_free on released vec.
12977 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12979         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12980         Require GOT register as additional operand in UNSPEC.
12981         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12982         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12983         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12984         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12985         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12986         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12987         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12989 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12991         PR target/57631
12992         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12993         name seen by assembler/linker rather if available.
12995 2013-07-11  Andreas Schwab  <schwab@suse.de>
12997         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12999 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
13001         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
13003 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
13005         * doc/tm.texi.in: Move hook documentation to ....
13006         * target.def: ... here.
13008         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
13009         text on @hook line.
13010         * doc/tm.texi: Regenerate.
13012 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
13014         PR c++/57869
13015         * doc/invoke.texi: Document -Wconditionally-supported.
13017 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13019         PR target/57844
13020         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
13021         of my_fp.
13023 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13025         PR target/57506
13026         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
13027         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
13028         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
13029         Remove duplicate devices.
13030         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
13031         * config/avr/t-multilib: Regenerate.
13032         * config/avr/avr-tables.opt: Regenerate.
13033         * doc/avr-mmcu.texi: Regenerate.
13035 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13037         PR target/56987
13038         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
13040 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
13042         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
13043         the cost of MULT when optimizing for size.
13045 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
13047         * config/cr16/cr16-protos.h: Don't include target.h.
13049 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
13051         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
13052         adjust register size for TDmode and TFmode for VSX registers.
13054 2013-07-08  Kai Tietz  <ktietz@redhat.com>
13056         PR target/56892
13057         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
13058         hook_bool_const_tree_true.
13060 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13062         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
13063         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
13064         * config/s390/s390.md: Define FPR*_REGNUM constants.
13065         Fix FPR2_REGNUM constant (18 -> 17).
13066         ("*trunc<BFP:mode><DFP_ALL:mode>2")
13067         ("*trunc<DFP_ALL:mode><BFP:mode>2")
13068         ("trunc<BFP:mode><DFP_ALL:mode>2")
13069         ("trunc<DFP_ALL:mode><BFP:mode>2")
13070         ("*extend<BFP:mode><DFP_ALL:mode>2")
13071         ("*extend<DFP_ALL:mode><BFP:mode>2")
13072         ("extend<BFP:mode><DFP_ALL:mode>2")
13073         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
13074         FPR4_REGNUM.
13076 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
13078         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
13080 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13082         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13083         and cfun_fpr_bit_p to cfun_fpr_save_p.
13084         (s390_frame_area, s390_register_info, s390_frame_info)
13085         (s390_emit_prologue, s390_emit_epilogue)
13086         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13087         register numbers.
13088         * config/s390/s390.h: Define *_REGNUM macros for floating point
13089         register numbers.
13091 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13093         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13095 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13097         PR rtl-optimization/57786
13098         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13099         and break out of the loop when it is set to false.
13101 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13103         PR target/57819
13104         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13105         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13106         (const_int 63)) 0)).
13107         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13108         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13109         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13111         PR rtl-optimization/57829
13112         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13113         mask bits outside of mode are just sign-extension from mode to HWI.
13115 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13117         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13118         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13119         adjust_address instead of change_address to keep info about alignment.
13120         (emit_strmov): Remove.
13121         (emit_memmov): New function.
13122         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13123         (expand_movmem_epilogue): Likewise and return updated rtx for
13124         destination.
13125         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13126         destination and source.
13127         (decide_alignment): Refactor, handle vector_loop.
13128         (ix86_expand_movmem): Likewise.
13129         (ix86_expand_setmem): Likewise.
13130         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13132 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13134         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13135         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13137 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13139         * config/i386/sse.md (sse_movlhps): Change alternative 3
13140         of operand 2 to "m".
13142 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13144         PR target/57807
13145         * config/i386/sse.md (iptr): New mode attribute.
13146         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13147         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13148         (<sse>_vmmul<mode>3): Ditto.
13149         (<sse>_vmdiv<mode>3): Ditto.
13150         (sse_vmrcpv4sf2): Ditto.
13151         (<sse>_vmsqrt<mode>2): Ditto.
13152         (sse_vmrsqrtv4sf2): Ditto.
13153         (<sse>_vm<code><mode>3): Ditto.
13154         (avx_vmcmp<mode>3): Ditto.
13155         (<sse>_vmmaskcmp<mode>3): Ditto.
13156         (<sse>_comi): Ditto.
13157         (<sse>_ucomi): Ditto.
13158         (*xop_vmfrcz_<mode>): Ditto.
13159         (*fmai_fmadd_<mode>): Ditto.
13160         (*fmai_fmsub_<mode>): Ditto.
13161         (*fmai_fnmadd_<mode>): Ditto.
13162         (*fmai_fnmsub_<mode>): Ditto.
13163         (*fma4i_vmfmadd_<mode>): Ditto.
13164         (*fma4i_vmfmsub_<mode>): Ditto.
13165         (*fma4i_vmfnmadd_<mode>): Ditto.
13166         (*fma4i_vmfnmsub_<mode>): Ditto.
13167         (*xop_vmfrcz_<mode>): Ditto.
13168         (sse_cvtps2pi): Ditto.
13169         (sse_cvttps2pi): Ditto.
13170         (sse_cvtss2si): Ditto.
13171         (sse_cvtss2si_2): Ditto.
13172         (sse_cvtss2siq_2): Ditto.
13173         (sse_cvttss2si): Ditto.
13174         (sse_cvttss2siq): Ditto.
13175         (sse_cvtsd2si): Ditto.
13176         (sse_cvtsd2si_2): Ditto.
13177         (sse_cvtsd2siq_2): Ditto.
13178         (sse_cvttsd2si): Ditto.
13179         (sse_cvttsd2siq): Ditto.
13180         (sse_cvtsd2ss): Ditto.
13181         (sse_cvtss2sd): Ditto.
13182         (avx2_pbroadcast<mode>): Ditto.
13183         (avx2_pbroadcast<mode>_1): Ditto.
13184         (*avx_vperm_broadcast_v4sf): Ditto.
13186         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13187         (sse_movlhps): Ditto.
13188         (sse_storehps): Ditto.
13189         (sse_loadhps): Ditto.
13190         (sse_storelps): Ditto.
13191         (sse_loadlps): Ditto.
13192         (*vec_concatv4sf): Ditto.
13193         (*vec_interleave_highv2df): Ditto.
13194         (*vec_interleave_lowv2df): Ditto.
13195         (*vec_extractv2df_1_sse): Ditto.
13196         (*vec_extractv2df_0_sse): Ditto.
13197         (sse2_storelpd): Ditto.
13198         (sse2_loadlpd): Ditto.
13199         (sse2_movsd): Ditto.
13200         (*vec_concatv4si): Ditto.
13201         (vec_concatv2di): Ditto.
13203         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13204         for Intel asm dialect.
13205         (mmx_punpcklwd): Ditto.
13206         (mmx_punpckldq): Ditto.
13208         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13209         for intel assembler dialect.
13211 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13213         PR target/29776
13214         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13215         for BUILT_IN_C{LZ,LRSB}*.
13216         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13217         * tree-vrp.c (extract_range_basic): Handle
13218         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13219         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13220         fall thru to code calling set_value*.
13221         * builtins.c (expand_builtin): Remove *IMAX cases.
13222         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13223         if width is bigger than 2*HWI.
13225 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13227         PR rtl-optimization/55342
13228         * lra-int.h (lra_subreg_reload_pseudos): New.
13229         * lra.c: Add undoing optional reloads to the block diagram.
13230         (lra_subreg_reload_pseudos): New.
13231         (lra_optional_reload_pseudos): Change comments.
13232         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13233         lra_optional_reload_pseudos after undo transformations.
13234         * lra-assigns.c (pseudo_prefix_title): New.
13235         (lra_setup_reg_renumber): Use it.
13236         (spill_for): Ditto.  Check subreg reload pseudos too.
13237         (assign_by_spills): Consider subreg reload pseudos too.
13238         * lra-constraints.c (simplify_operand_subreg): Use
13239         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13240         (curr_insn_transform): Recognize and do optional reloads.
13241         (undo_optional_reloads): New.
13242         (lra_undo_inheritance): Call undo_optional_reloads.
13244 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13246         * tree-complex.c (expand_complex_operations_1): Fix typo.
13248 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13250         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13251         (tune_params): New member 'const vec_costs'.
13252         * config/aarch64/aarch64.c (generic_vector_cost): New.
13253         (generic_tunings): New member 'generic_vector_cost'.
13254         (aarch64_builtin_vectorization_cost): New.
13255         (aarch64_add_stmt_cost): New.
13256         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13257         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13259 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13261         PR target/57777
13262         * config/i386/predicates.md (vsib_address_operand): Disallow
13263         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13265 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13267         PR middle-end/55030
13268         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13269         expand_builtin_setjmp_receiver.
13270         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13271         with NULL for the label parameter.
13272         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13273         the frame-pointer.  Adjust comments.
13274         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13275         only if LABEL is non-NULL.
13277 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13279         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13280         (ARM_ABI_AAPCS64): Ditto.
13281         (arm_abi): Ditto.
13282         (ARM_DEFAULT_ABI): Ditto.
13284 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13286         * config/aarch64/aarch64-builtins.c
13287         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13288         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13289         (st1): Likewise.
13290         * config/aarch64/aarch64-simd.md
13291         (aarch64_ld1<VALL:mode>): New.
13292         (aarch64_st1<VALL:mode>): Likewise.
13293         * config/aarch64/arm_neon.h
13294         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13296 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13298         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13299         target ISA is AVX.
13300         (ix86_option_override_internal):Turn on all -mavx target flags by
13301         default as they are dependent on AVX anyway.
13303 2013-07-02  Cary Coutant  <ccoutant@google.com>
13305         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13306         deterministic hash.
13307         (loc_checksum_ordered): Likewise.
13308         (hash_loc_operands): Remove inline keyword.
13310 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13312         PR tree-optimization/57741
13313         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13314         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13315         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13316         Allow REAL_CST step_exprs if flag_associative_math.
13317         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13319 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13321         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13323 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13325         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13327 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13329         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13330         encoding.
13331         (iorsi3_insn): Likewise.
13332         (arm_xorsi3): Likewise.
13334 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13336         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13337         to ...
13338         (attribute "type"): ... here, and prefix with "wmmx_".
13339         (attribute "core_cycles"): Update for attribute changes.
13340         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13341         (tbcstv4hi): Likewise.
13342         (tbcstv2si): Likewise.
13343         (iwmmxt_iordi3): Likewise.
13344         (iwmmxt_xordi3): Likewise.
13345         (iwmmxt_anddi3): Likewise.
13346         (iwmmxt_nanddi3): Likewise.
13347         (iwmmxt_arm_movdi): Likewise.
13348         (iwmmxt_movsi_insn): Likewise.
13349         (mov<mode>_internal): Likewise.
13350         (and<mode>3_iwmmxt): Likewise.
13351         (ior<mode>3_iwmmxt): Likewise.
13352         (xor<mode>3_iwmmxt): Likewise.
13353         (add<mode>3_iwmmxt): Likewise.
13354         (ssaddv8qi3): Likewise.
13355         (ssaddv4hi3): Likewise.
13356         (ssaddv2si3): Likewise.
13357         (usaddv8qi3): Likewise.
13358         (usaddv4hi3): Likewise.
13359         (usaddv2si3): Likewise.
13360         (sub<mode>3_iwmmxt): Likewise.
13361         (sssubv8qi3): Likewise.
13362         (sssubv4hi3): Likewise.
13363         (sssubv2si3): Likewise.
13364         (ussubv8qi3): Likewise.
13365         (ussubv4hi3): Likewise.
13366         (ussubv2si3): Likewise.
13367         (mulv4hi3_iwmmxt): Likewise.
13368         (smulv4hi3_highpart): Likewise.
13369         (umulv4hi3_highpart): Likewise.
13370         (iwmmxt_wmacs): Likewise.
13371         (iwmmxt_wmacsz): Likewise.
13372         (iwmmxt_wmacu): Likewise.
13373         (iwmmxt_wmacuz): Likewise.
13374         (iwmmxt_clrdi): Likewise.
13375         (iwmmxt_clrv8qi): Likewise.
13376         (iwmmxt_clr4hi): Likewise.
13377         (iwmmxt_clr2si): Likewise.
13378         (iwmmxt_uavgrndv8qi3): Likewise.
13379         (iwmmxt_uavgrndv4hi3): Likewise.
13380         (iwmmxt_uavgv8qi3): Likewise.
13381         (iwmmxt_uavgv4hi3): Likewise.
13382         (iwmmxt_tinsrb): Likewise.
13383         (iwmmxt_tinsrh): Likewise.
13384         (iwmmxt_tinsrw): Likewise.
13385         (iwmmxt_textrmub): Likewise.
13386         (iwmmxt_textrmsb): Likewise.
13387         (iwmmxt_textrmuh): Likewise.
13388         (iwmmxt_textrmsh): Likewise.
13389         (iwmmxt_textrmw): Likewise.
13390         (iwmxxt_wshufh): Likewise.
13391         (eqv8qi3): Likewise.
13392         (eqv4hi3): Likewise.
13393         (eqv2si3): Likewise.
13394         (gtuv8qi3): Likewise.
13395         (gtuv4hi3): Likewise.
13396         (gtuv2si3): Likewise.
13397         (gtv8qi3): Likewise.
13398         (gtv4hi3): Likewise.
13399         (gtv2si3): Likewise.
13400         (smax<mode>3_iwmmxt): Likewise.
13401         (umax<mode>3_iwmmxt): Likewise.
13402         (smin<mode>3_iwmmxt): Likewise.
13403         (umin<mode>3_iwmmxt): Likewise.
13404         (iwmmxt_wpackhss): Likewise.
13405         (iwmmxt_wpackwss): Likewise.
13406         (iwmmxt_wpackdss): Likewise.
13407         (iwmmxt_wpackhus): Likewise.
13408         (iwmmxt_wpackwus): Likewise.
13409         (iwmmxt_wpackdus): Likewise.
13410         (iwmmxt_wunpckihb): Likewise.
13411         (iwmmxt_wunpckihh): Likewise.
13412         (iwmmxt_wunpckihw): Likewise.
13413         (iwmmxt_wunpckilb): Likewise.
13414         (iwmmxt_wunpckilh): Likewise.
13415         (iwmmxt_wunpckilw): Likewise.
13416         (iwmmxt_wunpckehub): Likewise.
13417         (iwmmxt_wunpckehuh): Likewise.
13418         (iwmmxt_wunpckehuw): Likewise.
13419         (iwmmxt_wunpckehsb): Likewise.
13420         (iwmmxt_wunpckehsh): Likewise.
13421         (iwmmxt_wunpckehsw): Likewise.
13422         (iwmmxt_wunpckelub): Likewise.
13423         (iwmmxt_wunpckeluh): Likewise.
13424         (iwmmxt_wunpckeluw): Likewise.
13425         (iwmmxt_wunpckelsb): Likewise.
13426         (iwmmxt_wunpckelsh): Likewise.
13427         (iwmmxt_wunpckelsw): Likewise.
13428         (ror<mode>3): Likewise.
13429         (ashr<mode>3_iwmmxt): Likewise.
13430         (lshr<mode>3_iwmmxt): Likewise.
13431         (ashl<mode>3_iwmmxt): Likewise.
13432         (ror<mode>3_di): Likewise.
13433         (ashr<mode>3_di): Likewise.
13434         (lshr<mode>3_di): Likewise.
13435         (ashl<mode>3_di): Likewise.
13436         (iwmmxt_wmadds): Likewise.
13437         (iwmmxt_wmaddu): Likewise.
13438         (iwmmxt_tmia): Likewise.
13439         (iwmmxt_tmiaph): Likewise.
13440         (iwmmxt_tmiabb): Likewise.
13441         (iwmmxt_tmiatb): Likewise.
13442         (iwmmxt_tmiabt): Likewise.
13443         (iwmmxt_tmiatt): Likewise.
13444         (iwmmxt_tmovmskb): Likewise.
13445         (iwmmxt_tmovmskh): Likewise.
13446         (iwmmxt_tmovmskw): Likewise.
13447         (iwmmxt_waccb): Likewise.
13448         (iwmmxt_wacch): Likewise.
13449         (iwmmxt_waccw): Likewise.
13450         (iwmmxt_waligni): Likewise.
13451         (iwmmxt_walignr): Likewise.
13452         (iwmmxt_walignr0): Likewise.
13453         (iwmmxt_walignr1): Likewise.
13454         (iwmmxt_walignr2): Likewise.
13455         (iwmmxt_walignr3): Likewise.
13456         (iwmmxt_wsadb): Likewise.
13457         (iwmmxt_wsadh): Likewise.
13458         (iwmmxt_wsadbz): Likewise.
13459         (iwmmxt_wsadhz): Likewise.
13460         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13461         (iwmmxt_wabsdiffb): Likewise.
13462         (iwmmxt_wabsdiffh): Likewise.
13463         (iwmmxt_wabsdiffw): Likewise.
13464         (iwmmxt_waddsubhx): Likewise
13465         (iwmmxt_wsubaddhx): Likewise.
13466         (addc<mode>3): Likewise.
13467         (iwmmxt_avg4): Likewise.
13468         (iwmmxt_avg4r): Likewise.
13469         (iwmmxt_wmaddsx): Likewise.
13470         (iwmmxt_wmaddux): Likewise.
13471         (iwmmxt_wmaddsn): Likewise.
13472         (iwmmxt_wmaddun): Likewise.
13473         (iwmmxt_wmulwsm): Likewise.
13474         (iwmmxt_wmulwum): Likewise.
13475         (iwmmxt_wmulsmr): Likewise.
13476         (iwmmxt_wmulumr): Likewise.
13477         (iwmmxt_wmulwsmr): Likewise.
13478         (iwmmxt_wmulwumr): Likewise.
13479         (iwmmxt_wmulwl): Likewise.
13480         (iwmmxt_wqmulm): Likewise.
13481         (iwmmxt_wqmulwm): Likewise.
13482         (iwmmxt_wqmulmr): Likewise.
13483         (iwmmxt_wqmulwmr): Likewise.
13484         (iwmmxt_waddbhusm): Likewise.
13485         (iwmmxt_waddbhusl): Likewise.
13486         (iwmmxt_wqmiabb): Likewise.
13487         (iwmmxt_wqmiabt): Likewise.
13488         (iwmmxt_wqmiatb): Likewise.
13489         (iwmmxt_wqmiatt): Likewise.
13490         (iwmmxt_wqmiabbn): Likewise.
13491         (iwmmxt_wqmiabtn): Likewise.
13492         (iwmmxt_wqmiatbn): Likewise.
13493         (iwmmxt_wqmiattn): Likewise.
13494         (iwmmxt_wmiabb): Likewise.
13495         (iwmmxt_wmiabt): Likewise.
13496         (iwmmxt_wmiatb): Likewise.
13497         (iwmmxt_wmiatt): Likewise.
13498         (iwmmxt_wmiabbn): Likewise.
13499         (iwmmxt_wmiabtn): Likewise.
13500         (iwmmxt_wmiatbn): Likewise.
13501         (iwmmxt_wmiattn): Likewise.
13502         (iwmmxt_wmiawbb): Likewise.
13503         (iwmmxt_wmiawbt): Likewise.
13504         (iwmmxt_wmiawtb): Likewise.
13505         (iwmmxt_wmiawtt): Likewise.
13506         (iwmmxt_wmiawbbn): Likewise.
13507         (iwmmxt_wmiawbtn): Likewise.
13508         (iwmmxt_wmiawtbn): Likewise.
13509         (iwmmxt_wmiawttn): Likewise.
13510         (iwmmxt_wmerge): Likewise.
13511         (iwmmxt_tandc<mode>3): Likewise.
13512         (iwmmxt_torc<mode>3): Likewise.
13513         (iwmmxt_torvsc<mode>3): Likewise.
13514         (iwmmxt_textrc<mode>3): Likewise.
13515         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13516         (wmmxt_pack): Likewise.
13517         (wmmxt_mult_c1): Likewise.
13518         (wmmxt_mult_c2): Likewise.
13519         (wmmxt_alu_c1): Likewise.
13520         (wmmxt_alu_c2): Likewise.
13521         (wmmxt_alu_c3): Likewise.
13522         (wmmxt_transfer_c1): Likewise.
13523         (wmmxt_transfer_c2): Likewise.
13524         (wmmxt_transfer_c3): Likewise.
13525         (marvell_f_iwmmxt_wstr): Likewise.
13526         (marvell_f_iwmmxt_wldr): Likewise.
13528 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13530         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13532 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13534         Revert:
13535         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13536         * lra-constraints.c (need_for_split_p): Check call used hard regs
13537         living through calls.
13539         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13540         call used regs for call insn.
13542 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13544         PR target/57736
13545         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13546         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13548 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13550         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13552 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13554         * lra-constraints.c (need_for_split_p): Check call used hard regs
13555         living through calls.
13557 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13559         PR target/57744
13560         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13561         to tie with any other modes.  Eliminate Altivec vector mode tests,
13562         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13563         code, to return 0 if testing MODE2 for a condition, if we've
13564         already tested MODE1 for the same condition.
13566 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13568         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13569         layout.
13571 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13573         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13574         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13576 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13578         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13579         Define.
13580         (aarch64_symbolic_constant_p): Remove.
13581         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13582         static.  Fix line length and white space.
13583         (aarch64_symbolic_constant_p): Remove.
13584         * config/aarch64/predicates.md (aarch64_valid_symref):
13585         Use aarch64_classify_symbol_expression.
13587 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13589         * config/arm/constraints.md (Ts): New constraint.
13590         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13591         16-bit encodings.
13592         (compare_scc): Use "Ts" constraint for operand 0.
13593         (ior_scc_scc): Likewise.
13594         (and_scc_scc): Likewise.
13595         (and_scc_scc_nodom): Likewise.
13596         (ior_scc_scc_cmp): Likewise for operand 7.
13597         (and_scc_scc_cmp): Likewise.
13598         * config/arm/thumb2.md (thumb2_movsi_insn):
13599         Add alternatives for 16-bit encodings.
13600         (thumb2_movhi_insn): Likewise.
13601         (thumb2_movsicc_insn): Likewise.
13602         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13603         (thumb2_negscc): Use "Ts" constraint.
13604         Move mvn instruction outside cond_exec block.
13605         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13606         for 16-bit encodings.
13608 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13610         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13611         encoding.
13612         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13613         (mulsi3subsi): Likewise.
13614         (mulsidi3adddi): Likewise.
13615         (mulsidi3_v6): Likewise.
13616         (umulsidi3_v6): Likewise.
13617         (umulsidi3adddi_v6): Likewise.
13618         (smulsi3_highpart_v6): Likewise.
13619         (umulsi3_highpart_v6): Likewise.
13620         (mulhisi3tb): Likewise.
13621         (mulhisi3bt): Likewise.
13622         (mulhisi3tt): Likewise.
13623         (maddhisi4): Likewise.
13624         (maddhisi4tb): Likewise.
13625         (maddhisi4tt): Likewise.
13626         (maddhidi4): Likewise.
13627         (maddhidi4tb): Likewise.
13628         (maddhidi4tt): Likewise.
13629         (zeroextractsi_compare0_scratch): Likewise.
13630         (insv_zero): Likewise.
13631         (insv_t2): Likewise.
13632         (anddi_notzesidi_di): Likewise.
13633         (anddi_notsesidi_di): Likewise.
13634         (andsi_notsi_si): Likewise.
13635         (iordi_zesidi_di): Likewise.
13636         (xordi_zesidi_di): Likewise.
13637         (andsi_iorsi3_notsi): Likewise.
13638         (smax_0): Likewise.
13639         (smax_m1): Likewise.
13640         (smin_0): Likewise.
13641         (not_shiftsi): Likewise.
13642         (unaligned_loadsi): Likewise.
13643         (unaligned_loadhis): Likewise.
13644         (unaligned_loadhiu): Likewise.
13645         (unaligned_storesi): Likewise.
13646         (unaligned_storehi): Likewise.
13647         (extv_reg): Likewise.
13648         (extzv_t2): Likewise.
13649         (divsi3): Likewise.
13650         (udivsi3): Likewise.
13651         (arm_zero_extendhisi2addsi): Likewise.
13652         (arm_zero_extendqisi2addsi): Likewise.
13653         (compareqi_eq0): Likewise.
13654         (arm_extendhisi2_v6): Likewise.
13655         (arm_extendqisi2addsi): Likewise.
13656         (arm_movt): Likewise.
13657         (thumb2_ldrd): Likewise.
13658         (thumb2_ldrd_base): Likewise.
13659         (thumb2_ldrd_base_neg): Likewise.
13660         (thumb2_strd): Likewise.
13661         (thumb2_strd_base): Likewise.
13662         (thumb2_strd_base_neg): Likewise.
13663         (arm_negsi2): Add alternative for 16-bit encoding.
13664         (arm_one_cmplsi2): Likewise.
13666 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13668         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13669         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13670         (movdfcc): Likewise.
13671         * config/arm/vfp.md (*thumb2_movsf_vfp):
13672         Disable predication for arm_restrict_it.
13673         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13674         (*thumb2_movdfcc_vfp): Likewise.
13675         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13676         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13677         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13678         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13679         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13680         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13681         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13682         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13683         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13684         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13685         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13686         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13687         Disable predication for arm_restrict_it.
13689 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13691         * config/i386/bmiintrin.h (_bextr_u32): New.
13692         (_bextr_u64): Ditto.
13694 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13696         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13697         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13698         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13699         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13700         * config/mips/n32-elf.h: ...this new file.
13702 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13704         PR target/57224
13705         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13706         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13708 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13710         * config/mips/mips-tables.opt: Regenerate.
13711         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13712         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13713         * doc/invoke.texi: Add -m14kc.
13715 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13717         PR target/57623
13718         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13719         constraints of operand 1 and 2.
13721         PR target/57623
13722         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13723         to match RTL canonicalization.  Swap predicates and
13724         constraints of operand 1 and 2.
13726 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13728         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13729         Process OP_INOUT regs for splitting too.
13731 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13733         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13734         decl before the loop, initialize to NULL.
13735         (vectorizable_load): Initialize ptr_incr to NULL.
13737 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13739         PR lto/57208
13740         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13741         * ipa-ref.c (ipa_maybe_record_reference): New function.
13742         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13743         * ipa-cp.c (create_specialized_node): Record potential references from
13744         aggvals.
13745         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13747 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13749         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13750         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13751         'mode' to force_reg.
13752         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13753         (aarch64_expand_mov_immediate): Likewise.
13755 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13757         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13758         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13760 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13762         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13763         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13764         (struct machine_function): Add tbegin_p.
13765         (s390_canonicalize_comparison): Fold CC mode compares to
13766         conditional jump if possible.
13767         (s390_emit_jump): Return the emitted jump.
13768         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13769         Handle CCRAWmode compares.
13770         (s390_option_override): Default to -mhtm if available.
13771         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13772         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13773         FPRs instead of df_regs_ever_live_p.
13774         (s390_optimize_nonescaping_tx): New function.
13775         (s390_init_frame_layout): Extend clobbered_regs array to cover
13776         FPRs as well.
13777         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13778         (s390_expand_tbegin): New function.
13779         (enum s390_builtin): New enum definition.
13780         (code_for_builtin): New array definition.
13781         (s390_init_builtins): New function.
13782         (s390_expand_builtin): New function.
13783         (TARGET_INIT_BUILTINS): Define.
13784         (TARGET_EXPAND_BUILTIN): Define.
13785         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13786         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13787         (s390_alc_comparison): Likewise.
13788         * config/s390/s390-modes.def: Add CCRAWmode.
13789         * config/s390/s390.h (processor_flags): Add PF_TX.
13790         (TARGET_CPU_HTM): Define macro.
13791         (TARGET_HTM): Define macro.
13792         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13793         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13794         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13795         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13796         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13797         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13798         ("*cc_to_int"): Move up.
13799         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13800         constants other than 0.
13801         ("*ccraw_to_int"): New insn and splitter definition.
13802         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13803         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13804         ("tx_assist"): New expander.
13805         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13806         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13807         * config/s390/s390.opt: Add -mhtm option.
13808         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13809         * config/s390/htmxlintrin.h: New file.
13810         * config/s390/htmintrin.h: New file.
13811         * config/s390/s390intrin.h: New file.
13812         * doc/extend.texi: Document htm builtins.
13813         * config.gcc: Add the new header files to extra_headers.
13815 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13817         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13818         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13820 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13821             Pat Haugen  <pthaugen@us.ibm.com>
13822             Peter Bergner  <bergner@vnet.ibm.com>
13824         * config/rs6000/power8.md: New.
13825         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13826         setting for power8 entry.
13827         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13828         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13829         test for Power4/Power5 only.
13830         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13831         support.
13832         (force_new_group): Adjust comment.
13833         * config/rs6000/rs6000.md: Include power8.md.
13835 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13837         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13838         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13839         declaration.
13840         (tune_params): Update comment.
13841         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13842         (arm_max_conditional_execute): New function.
13843         (thumb2_final_prescan_insn): Use max_insn_skipped and
13844         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13846 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13848         PR tree-optimization/57705
13849         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13850         SSA_NAME step, provided that it is not defined inside the loop.
13851         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13852         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13854 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13856         PR middle-end/57670
13857         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13858         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13859         calls in the dump.
13860         (ipa_note_param_call): Initialize member_ptr flag.
13861         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13862         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13863         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13864         (ipa_read_indirect_edge_info): Likewise.
13866 2013-06-25  Richard Biener  <rguenther@suse.de>
13868         PR middle-end/56977
13869         * passes.c (init_optimization_passes): Move pass_fold_builtins
13870         and pass_dce earlier with -Og.
13872 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13874         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13875         <BIT_FIELD_REF>: Remove trailing TAB.
13876         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13877         remove blank line.
13879 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13881         PR tree-optimization/57358
13882         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13883         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13884         (ipa_analyze_params_uses): Generate pessimistic info when true.
13886 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13888         PR tree-optimization/57539
13889         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13890         global.inlined_to of the new node to it.  All callers changed.
13891         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13892         inlining_into, pass it to cgraph_clone_node.
13893         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13894         ipa_free_edge_args_substructures.
13895         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13896         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13897         Assert we have found an rdesc in the rdesc list.
13899 2013-06-24  Richard Biener  <rguenther@suse.de>
13901         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13902         (pointer_set_lookup): Declare.
13903         (class pointer_map): New template class implementing a
13904         generic pointer to T map.
13905         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13906         pointer_map<T>::contains, pointer_map<T>::insert,
13907         pointer_map<T>::traverse): New functions.
13908         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13909         (pointer_set_lookup): New function.
13910         (pointer_set_contains): Use pointer_set_lookup.
13911         (pointer_set_insert): Likewise.
13912         (insert_aux): Remove.
13913         (struct pointer_map_t): Embed a pointer_set_t.
13914         (pointer_map_create): Adjust.
13915         (pointer_map_destroy): Likewise.
13916         (pointer_map_contains): Likewise.
13917         (pointer_map_insert): Likewise.
13918         (pointer_map_traverse): Likewise.
13919         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13920         pointer_map<unsigned> instead of a pointer_map_t.
13921         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13922         (streamer_tree_cache_lookup): Likewise.
13923         (streamer_tree_cache_create): Likewise.
13924         (streamer_tree_cache_delete): Likewise.
13925         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13926         pointer_map<unsigned> instead of a pointer_map_t.
13927         (lto_init_tree_ref_encoder): Adjust.
13928         (lto_destroy_tree_ref_encoder): Likewise.
13929         * lto-section-out.c (lto_output_decl_index): Likewise.
13930         (lto_record_function_out_decl_state): Likewise.
13931         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13933 2013-06-24  Richard Biener  <rguenther@suse.de>
13935         PR tree-optimization/57488
13936         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13938 2013-06-24  Alan Modra  <amodra@gmail.com>
13940         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13941         (gen_easy_altivec_constant): Likewise.
13942         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13943         easy_vector_constant_msb): Likewise.
13945 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13947         PR target/57688
13948         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13949         add missing return true.
13951 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13953         PR target/52483
13954         * config/sh/predicates.md (general_extend_operand): Invoke
13955         general_movsrc_operand for memory operands.
13956         (general_movsrc_operand): Allow reg+reg addressing, do not use
13957         general_operand for memory operands.
13959 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13961         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13962         when current target options does not apply.
13963         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13964         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13965         * config/i386/bmiintrin.h: Pass appropriate target
13966         attributes to header.
13967         * config/i386/mmintrin.h: Ditto.
13968         * config/i386/nmmintrin.h: Ditto.
13969         * config/i386/avx2intrin.h: Ditto.
13970         * config/i386/fxsrintrin.h: Ditto.
13971         * config/i386/tbmintrin.h: Ditto.
13972         * config/i386/xsaveintrin.h: Ditto.
13973         * config/i386/f16cintrin.h: Ditto.
13974         * config/i386/xtestintrin.h: Ditto.
13975         * config/i386/xsaveoptintrin.h: Ditto.
13976         * config/i386/bmi2intrin.h: Ditto.
13977         * config/i386/lzcntintrin.h: Ditto.
13978         * config/i386/smmintrin.h: Ditto.
13979         * config/i386/wmmintrin.h: Ditto.
13980         * config/i386/x86intrin.h: Remove all header include guards.
13981         * config/i386/prfchwintrin.h: Ditto.
13982         * config/i386/pmmintrin.h: Ditto.
13983         * config/i386/tmmintrin.h: Ditto.
13984         * config/i386/xmmintrin.h: Ditto.
13985         * config/i386/popcntintrin.h: Ditto.
13986         * config/i386/rdseedintrin.h: Ditto.
13987         * config/i386/ammintrin.h: Ditto.
13988         * config/i386/emmintrin.h: Ditto.
13989         * config/i386/immintrin.h: Remove all header include guards.
13990         * config/i386/fma4intrin.h: Ditto.
13991         * config/i386/lwpintrin.h: Ditto.
13992         * config/i386/xopintrin.h: Ditto.
13993         * config/i386/ia32intrin.h: Ditto.
13994         * config/i386/avxintrin.h: Ditto.
13995         * config/i386/rtmintrin.h: Ditto.
13996         * config/i386/fmaintrin.h: Ditto.
13997         * config/i386/mm3dnow.h: Ditto.
13999 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
14001         * common/config/i386/i386-common.c: Handle LZCNT.
14003 2013-06-22  Andi Kleen  <ak@linux.intel.com>
14005         * doc/extend.texi: Use __atomic_store_n instead of
14006         __atomic_store in HLE example.
14008 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
14010         * config/sh/sh.c: Remove <cstdlib> workaround.
14012 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14014         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
14016 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14018         * doc/extend.texi: Document that __atomic_clear and
14019           __atomic_test_and_set should only be used with bool.
14021 2013-06-20  Jan Hubicka  <jh@suse.cz>
14023         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
14024         types_same_for_odr.
14025         * tree.c (decls_same_for_odr): New function.
14026         (same_for_edr): New function.
14027         (types_same_for_odr): New function.
14028         (get_binfo_at_offset): Use it.
14029         * tree.h (types_same_for_odr): Declare.
14031 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
14032             Jason Merrill  <jason@redhat.com>
14034         * system.h: Include <cstdlib> as well as <stdlib.h>.
14036 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
14038         PR target/57655
14039         * config/i386/i386.c (construct_container): Report error if
14040         long double is used with disabled x87 float returns.
14042 2013-06-20  Jan Hubicka  <jh@suse.cz>
14044         * lto-cgraph.c (input_symtab): Do not set cgraph state.
14046 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
14048         PR rtl-optimization/57425
14049         PR rtl-optimization/57569
14050         * alias.c (write_dependence_p): Remove parameters mem_mode and
14051         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
14052         Changed all callers.
14053         (canon_anti_dependence): Get comments and semantics in sync.
14054         Add parameter mem_canonicalized.  Changed all callers.
14055         * rtl.h (canon_anti_dependence): Update prototype.
14057 2013-06-20  Richard Biener  <rguenther@suse.de>
14059         * data-streamer-in.c (streamer_read_uhwi): Optimize single
14060         byte case, inline streamer_read_uchar and defer section
14061         overrun check.
14063 2013-06-20  Richard Biener  <rguenther@suse.de>
14065         PR tree-optimization/57584
14066         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
14067         SSA names into the expanded expression that take part in
14068         abnormal coalescing.
14070 2013-06-19  Sharad Singhai  <singhai@google.com>
14072         * gcov.c (print_usage): Handle new option.
14073         (process_args): Ditto.
14074         (get_gcov_intermediate_filename): New function.
14075         (output_intermediate_file): New function.
14076         (output_gcov_file): New function
14077         (generate_results): Handle new option.
14078         (release_function): Relase demangled name.
14079         (read_graph_file): Handle demangled name.
14080         (output_lines): Ditto.
14081         * doc/gcov.texi: Document gcov intermediate format.
14083 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14085         PR bootstrap/57604
14086         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14087         (lra_emit_add): Use the functions.  Add comment about Y as an
14088         address segment.
14090 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14092         PR driver/57652
14093         * collect2.c (collect_atexit): New.
14094         (collect_exit): Delete.
14095         (main): Register collect_atexit with atexit.
14096         (collect_wait): Change collect_exit to exit.
14097         (do_wait): Same.
14098         * collect2.h (collect_exit): Delete.
14099         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14101 2013-06-19  Wei Mi  <wmi@google.com>
14103         PR rtl-optimization/57518
14104         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14105         if regno is used in paradoxical subreg.
14106         (update_equiv_regs): Check pdx_subregs[regno] before
14107         set a reg to be equivalent with a mem.
14109 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14111         PR driver/57651
14112         * file-find.h (find_a_file): Add a mode parameter.
14113         * file-find.c (find_a_file): Likewise.
14114         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14115         with X_OK for the executables.
14116         * collect2.c (main): Call find_a_file with X_OK.
14118 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14120         PR target/56942
14121         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14122         Use NEXT_INSN instead of next_real_insn.
14124 2013-06-19  Jan Hubicka  <jh@suse.cz>
14126         * cgraph.h (const_value_known_p): Replace by ...
14127         (ctor_for_folding): .. this one.
14128         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14129         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14130         * expr.c (expand_expr_real_1): Likewise.
14131         (string_constant): Likewise.
14132         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14133         * ipa.c (process_references): Likewise.
14134         (symtab_remove_unreachable_nodes): Likewise.
14135         * ipa-inline-analysis.c (param_change_prob): Likewise.
14136         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14137         (get_base_constructor): Likwise.
14138         * varpool.c (varpool_remove_node): Likewise.
14139         (varpool_remove_initializer): LIkewise.
14140         (dump_varpool_node): LIkwise.
14141         (const_value_known_p): Rewrite to ...
14142         (ctor_for_folding): ... this one.
14144 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14146         PR driver/57651
14147         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14148         PERSONALITY in $PATH derived prefixes.
14150 2013-06-19  Jeff Law  <law@redhat.com>
14152         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14153         in comment.
14155         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14156         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14157         booleans.
14159 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14161         * config/arm/vfp.md: Move VFP instruction classification documentation
14162         to ...
14163         * config/arm/arm.md: ... here.  Update instruction classification
14164         documentation.
14166 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14168         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14169         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14170         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14172 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14174         PR target/57609
14175         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14176         with NEXT_INSN.  Use tablejump_p to check for jump table data
14177         insns.
14179 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14181         PR c++/56544
14182         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14183         that now in C++ the value is correct per the C++ standards.
14185 2013-06-19  Richard Biener  <rguenther@suse.de>
14187         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14188         for global context.
14190 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14192         Revert:
14193         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14195         PR target/57609
14196         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14197         with next_active_insn.
14199 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14201         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14202         functions are inlined during failures to flag an error.
14203         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14204         in early inline pass.
14206 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14208         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14209         in comments.
14211 2013-06-18  Julian Brown  <julian@codesourcery.com>
14213         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14214         Permit virtual register pre-reload if !strict.
14215         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14216         change.
14217         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14218         prototype.
14219         * config/arm/neon.md (movmisalign<mode>): Use
14220         neon_perm_struct_or_reg_operand instead of
14221         neon_struct_or_register_operand.
14222         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14223         neon_permissive_struct_operand instead of neon_struct_operand.
14224         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14225         neon_vector_mem_operand.
14226         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14227         neon_vector_mem_operand.
14228         (neon_permissive_struct_operand): New.
14229         (neon_struct_or_register_operand): Rename to...
14230         (neon_perm_struct_or_reg_operand): This. Adjust call to
14231         neon_vector_mem_operand.
14233 2013-06-18  Richard Biener  <rguenther@suse.de>
14235         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14236         * lto-streamer.h: Include pointer-set.h.
14237         (struct lto_decl_slot): Remove.
14238         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14239         Remove next_index entry.
14240         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14241         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14242         (lto_init_tree_ref_encoder): Adjust.
14243         (lto_destroy_tree_ref_encoder): Likewise.
14244         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14245         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14246         (lto_output_decl_index): Adjust.
14247         (lto_new_out_decl_state): Likewise.
14248         (lto_record_function_out_decl_state): Likewise.
14249         * lto-streamer-out.c (copy_function): Likewise.
14251 2013-06-18  Richard Biener  <rguenther@suse.de>
14253         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14254         * cgraphunit.c: Include cfgloop.h.
14255         (init_lowered_empty_function): Initialize the loop tree.
14256         (assemble_thunk): Insert new BBs into loops.
14258 2013-06-18  Richard Biener  <rguenther@suse.de>
14260         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14261         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14262         the map from cache entry to cache index optional.
14263         (streamer_tree_cache_replace_tree): Adjust accordingly.
14264         (streamer_tree_cache_append): Likewise.
14265         (streamer_tree_cache_delete): Likewise.
14266         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14267         streamer cache map from cache entry to cache index.
14268         * lto-streamer-out.c (create_output_block): Adjust.
14270 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14272         * config/arm/arm.md (attribute "insn"): Move multiplication and
14273         division attributes to...
14274         (attribute "type"): ... here.  Remove mult.
14275         (attribute "mul32"): New attribute.
14276         (attribute "mul64"): Add umaal.
14277         (*arm_mulsi3): Update attributes.
14278         (*arm_mulsi3_v6): Likewise.
14279         (*thumb_mulsi3): Likewise.
14280         (*thumb_mulsi3_v6): Likewise.
14281         (*mulsi3_compare0): Likewise.
14282         (*mulsi3_compare0_v6): Likewise.
14283         (*mulsi_compare0_scratch): Likewise.
14284         (*mulsi_compare0_scratch_v6): Likewise.
14285         (*mulsi3addsi): Likewise.
14286         (*mulsi3addsi_v6): Likewise.
14287         (*mulsi3addsi_compare0): Likewise.
14288         (*mulsi3addsi_compare0_v6): Likewise.
14289         (*mulsi3addsi_compare0_scratch): Likewise.
14290         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14291         (*mulsi3subsi): Likewise.
14292         (*mulsidi3adddi): Likewise.
14293         (*mulsi3addsi_v6): Likewise.
14294         (*mulsidi3adddi_v6): Likewise.
14295         (*mulsidi3_nov6): Likewise.
14296         (*mulsidi3_v6): Likewise.
14297         (*umulsidi3_nov6): Likewise.
14298         (*umulsidi3_v6): Likewise.
14299         (*umulsidi3adddi): Likewise.
14300         (*umulsidi3adddi_v6): Likewise.
14301         (*smulsi3_highpart_nov6): Likewise.
14302         (*smulsi3_highpart_v6): Likewise.
14303         (*umulsi3_highpart_nov6): Likewise.
14304         (*umulsi3_highpart_v6): Likewise.
14305         (mulhisi3): Likewise.
14306         (*mulhisi3tb): Likewise.
14307         (*mulhisi3bt): Likewise.
14308         (*mulhisi3tt): Likewise.
14309         (maddhisi4): Likewise.
14310         (*maddhisi4tb): Likewise.
14311         (*maddhisi4tt): Likewise.
14312         (maddhidi4): Likewise.
14313         (*maddhidi4tb): Likewise.
14314         (*maddhidi4tt): Likewise.
14315         (divsi3): Likewise.
14316         (udivsi3): Likewise.
14317         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14318         (thumb2_mulsi_short_compare0): Likewise.
14319         (thumb2_mulsi_short_compare0_scratch): Likewise.
14320         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14321         (1020mult2): Likewise.
14322         (1020mult3): Likewise.
14323         (1020mult4): Likewise.
14324         (1020mult5): Likewise.
14325         (1020mult6): Likewise.
14326         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14327         change.
14328         (cortex_a15_mult64): Likewise.
14329         (cortex_a15_sdiv): Likewise.
14330         (cortex_a15_udiv): Likewise.
14331         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14332         (mult2): Likewise.
14333         (mult3): Likewise.
14334         (mult4): Likewise.
14335         (mult5): Likewise.
14336         (mult6): Likewise.
14337         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14338         (pj4_ir_div): Likewise.
14339         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14340         (11_mult2): Likewise.
14341         (11_mult3): Likewise.
14342         (11_mult4): Likewise.
14343         (11_mult5): Likewise.
14344         (11_mult6): Likewise.
14345         (11_mult7): Likewise.
14346         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14347         (cortex_a8_mla): Likewise.
14348         (cortex_a8_mull): Likewise.
14349         (cortex_a8_smulwy): Likewise.
14350         (cortex_a8_smlald): Likewise.
14351         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14352         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14353         (cortex_r4_mul_3): Likewise.
14354         (cortex_r4_mla_4): Likewise.
14355         (cortex_r4_mla_3): Likewise.
14356         (cortex_r4_smlald): Likewise.
14357         (cortex_r4_mull): Likewise.
14358         (cortex_r4_sdiv): Likewise.
14359         (cortex_r4_udiv): Likewise.
14360         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14361         (cortex_a7_idiv): Likewise.
14362         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14363         (9_mult2): Likewise.
14364         (9_mult3): Likewise.
14365         (9_mult4): Likewise.
14366         (9_mult5): Likewise.
14367         (9_mult6): Likewise.
14368         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14369         (cortex_a53_sdiv): Likewise.
14370         (cortex_a53_udiv): Likewise.
14371         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14372         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14373         (mp626_mult2): Likewise.
14374         (mp626_mult3): Likewise.
14375         (mp626_mult4): Likewise.
14376         * config/arm/fa526.md (526_mult1): Update attribute change.
14377         (526_mult2): Likewise.
14378         * config/arm/arm-generic.md (mult): Update attribute change.
14379         (mult_ldsched_strongarm): Likewise.
14380         (mult_ldsched): Likewise.
14381         (multi_cycle): Likewise.
14382         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14383         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14384         (606te_mult2): Likewise.
14385         (606te_mult3): Likewise.
14386         (606te_mult4): Likewise.
14387         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14388         (cortex_a9_mac16): Likewise.
14389         (cortex_a9_multiply): Likewise.
14390         (cortex_a9_mac): Likewise.
14391         (cortex_a9_multiply_long): Likewise.
14392         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14393         (626te_mult2): Likewise.
14394         (626te_mult3): Likewise.
14395         (626te_mult4): Likewise.
14397 2013-06-18  Richard Biener  <rguenther@suse.de>
14399         PR lto/57334
14400         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14402 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14404         PR target/57609
14405         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14406         with next_active_insn.
14408 2013-06-18  Alan Modra  <amodra@gmail.com>
14410         * config/rs6000/rs6000.h (enum data_align): New.
14411         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14412         (DATA_ABI_ALIGNMENT): Define.
14413         (CONSTANT_ALIGNMENT): Correct comment.
14414         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14415         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14417 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14419         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14420         ATTRIBUTE_UNUSED marking.
14422 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14424         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14425         alternative and update.
14426         (aarch64_dup_lanedi): Delete.
14427         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14428         * config/aarch64/aarch64-simd-builtins.def: Update.
14430 2013-06-17  Richard Biener  <rguenther@suse.de>
14432         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14433         (lto_input_scc): Declare.
14434         (lto_input_tree_1): Likewise.
14435         (struct lto_stats_d): Add num_tree_bodies_output and
14436         num_pickle_refs_output.
14437         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14438         (lto_read_tree_1): Split out from ...
14439         (lto_read_tree): ... this.
14440         (lto_input_scc): New function.
14441         (lto_input_tree_1): Split out from ...
14442         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14443         (lto_data_in_create): Create the streamer cache without hashes.
14444         * lto-streamer-out.c (create_output_block): Create the streamer
14445         cache with hashes when not doing WPA.
14446         (lto_write_tree_1): Split out from ...
14447         (lto_write_tree): ... this.
14448         (get_symbol_initial_value): New function.
14449         (lto_output_tree_1): Split out from ...
14450         (lto_output_tree): ... this.  Write trees as series of SCCs
14451         using a DFS walk via DFS_write_tree.
14452         (struct sccs, struct scc_entry): New types.
14453         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14454         (DFS_write_tree_body): New function.
14455         (DFS_write_tree): Likewise.
14456         (hash_tree): Likewise.
14457         (scc_entry_compare): Likewise.
14458         (hash_scc): Likewise.
14459         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14460         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14461         TREE_CHAIN as regular reference.
14462         (streamer_read_integer_cst): Remove.
14463         (streamer_get_pickled_tree): Adjust.
14464         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14465         of DECL_EXTERNALs in BLOCK_VARS for now.
14466         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14467         reference.
14468         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14469         Add hash value argument and record that if hashes are recorded
14470         in the cache.
14471         (streamer_tree_cache_insert_1): Adjust.
14472         (streamer_tree_cache_insert): Likewise.
14473         (streamer_tree_cache_insert_at): Rename to ...
14474         (streamer_tree_cache_replace_tree): ... this and adjust.
14475         (streamer_tree_cache_append): Adjust.
14476         (record_common_node): Likewise.
14477         (streamer_tree_cache_create): Add argument whether to
14478         record hash values together with trees.
14479         (streamer_tree_cache_delete): Adjust.
14480         * tree-streamer.h (struct streamer_tree_cache_d): Add
14481         vector of hashes.
14482         (streamer_read_integer_cst): Remove.
14483         (streamer_tree_cache_insert): Adjust.
14484         (streamer_tree_cache_append): Likewise.
14485         (streamer_tree_cache_insert_at): Rename to ...
14486         (streamer_tree_cache_replace_tree): ... this and adjust.
14487         (streamer_tree_cache_create): Add argument whether to record hashes.
14488         (streamer_tree_cache_get): Rename to ...
14489         (streamer_tree_cache_get_tree): ... this.
14490         (streamer_tree_cache_get_hash): New function.
14491         * tree.c (cache_integer_cst): New function.
14492         * tree.h (cache_integer_cst): Declare.
14493         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14494         * lto-symtab.c (lto_varpool_replace_node): Only release
14495         DECL_INITIAL of non-prevailing decls.
14496         * varpool.c (varpool_remove_initializer): Do not release
14497         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14499 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14501         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14502         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14503         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14504         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14505         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14506         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14507         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14508         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14509         instead of TARGET_64BIT.
14510         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14511         Require ISA_HAS_<D>DIV.
14513 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14515         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14516         (mips*-*-linux*): Move default with_llsc setting to where other
14517         defaults are set.
14518         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14519         with_arch block.
14520         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14521         Likewise.  Remove default with_tune setting.  Move default float
14522         setting to its own block.  Handle with_llsc in the same block as above.
14524 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14526         PR rtl-optimization/57425
14527         PR rtl-optimization/57569
14528         * alias.c (write_dependence_p): Add new parameters mem_mode,
14529         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14530         Changed all callers.
14531         (canon_anti_dependence): New function.
14532         * cse.c (check_dependence): Use canon_anti_dependence.
14533         * cselib.c (cselib_invalidate_mem): Likewise.
14534         * rtl.h (canon_anti_dependence): Declare.
14536 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14538         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14539         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14540         ".set mips3" for 64-bit targets.
14542 2013-06-15  Dehao Chen  <dehao@google.com>
14544         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14545         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14546         (gimple_check_call_args): Likewise.
14547         * value-prof.c (check_ic_target): Likewise.
14548         * ipa-inline.c (early_inliner): Likewise.
14549         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14550         * cgraph.c (cgraph_create_edge_1): Likewise.
14551         (cgraph_make_edge_direct): Likewise.
14553 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14555         PR target/57615
14556         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14557         rs6000_output_move_128bit to handle emitting quad memory
14558         operations.  Set attribute length to 8 bytes.
14560 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14562         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14563         New pattern.
14564         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14565         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14566         (aarch64_<su>mlsl<mode>): Likewise.
14568 2013-06-14  Mike Stump  <mikestump@comcast.net>
14570         * Makefile.in (TARGET_H): Add insn-codes.h.
14572 2013-06-14  Alan Modra  <amodra@gmail.com>
14574         PR middle-end/57134
14575         PR middle-end/57586
14576         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14577         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14578         bitfield expansion when EXPAND_MEMORY.
14579         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14581 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14583         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14584         test for clearing quad memory on 32-bit later.
14586 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14588         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14589         (fold_negate_expr): Likewise.
14590         (fold_real_zero_addition_p): Handle vectors.
14591         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14593 2013-06-14  Alan Modra  <amodra@gmail.com>
14595         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14597 2013-06-13  Jan Hubicka  <jh@suse.cz>
14599         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14600         Local comdats are not externally visible.
14601         * symtab.c (dump_symtab_base): Dump externally visible.
14602         (verify_symtab_base): Verify back links in the symtab hash.
14604 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14606         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14607         CONVERT_EXPR as equal nodes.
14609 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14611         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14613 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14615         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14616         Generalize to complex and vector.
14617         * tree.c (build_all_ones_cst): New function.
14618         * tree.h (build_all_ones_cst): Declare it.
14620 2013-06-13  Alan Modra  <amodra@gmail.com>
14622         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14623         * config/rs6000/rs6000.md (signbittf2): New insn.
14624         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14625         (abstf2_internal, cmptf_internal2): Likewise.
14626         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14628 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14629             Pat Haugen  <pthaugen@us.ibm.com>
14630             Peter Bergner  <bergner@vnet.ibm.com>
14632         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14633         power8 byte, half-word, and quad-word atomic instructions.
14634         (emit_store_conditional): Likewise.
14635         (rs6000_expand_atomic_compare_and_swap): Likewise.
14636         (rs6000_expand_atomic_op): Likewise.
14638         * config/rs6000/sync.md (larx): Add new modes for power8.
14639         (stcx): Likewise.
14640         (AINT): New mode iterator to include TImode as well as normal
14641         integer modes on power8.
14642         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14643         that VSX registers are not considered.  Use AINT mode iterator
14644         instead of INT1 to allow inclusion of quad word atomic operations
14645         on power8.
14646         (load_locked<mode>): Likewise.
14647         (store_conditional<mode>): Likewise.
14648         (atomic_compare_and_swap<mode>): Likewise.
14649         (atomic_exchange<mode>): Likewise.
14650         (atomic_nand<mode>): Likewise.
14651         (atomic_fetch_<fetchop_name><mode>): Likewise.
14652         (atomic_nand_fetch<mode>): Likewise.
14653         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14654         each type.
14655         (ATOMIC): On power8, add QImode, HImode modes.
14656         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14657         modes that promote to SImode.
14658         (load_lockedti): Convert TImode arguments to PTImode, so that we
14659         get a guaranteed even/odd register pair.
14660         (load_lockedpti): Likewise.
14661         (store_conditionalti): Likewise.
14662         (store_conditionalpti): Likewise.
14664         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14665         atomic load/store instructions.
14666         (HSI): Likewise.
14668 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14670         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14671         loads.
14672         (insn_count): New attribute, with most cases extracted from...
14673         (length): ...here.  Redefine most cases in terms of insn_count.
14674         (single_insn): Delete.
14675         (can_delay): Use insn_count to check for single instructions.
14676         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14677         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14678         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14679         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14680         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14681         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14682         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14683         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14684         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14685         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14686         rather than "length".
14687         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14688         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14689         Use "insn_count" rather than "length".
14690         * config/mips/mips-dsp.md
14691         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14692         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14693         length attributes.
14695 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14697         PR tree-optimization/57361
14698         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14700 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14702         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14703         to split.
14704         (aarch64_simd_combine<mode>): New instruction expansion.
14705         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14706         function prototype.
14707         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14708         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14710 2013-06-12  Jan Hubicka  <jh@suse.cz>
14712         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14713         decl has when in streaming stage.
14714         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14715         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14717 2013-06-12  Roland Stigge  <stigge@antcom.de>
14719         PR target/57578
14720         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14722 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14724         PR tree-optimization/57537
14725         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14726         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14728 2013-06-12  Richard Biener  <rguenther@suse.de>
14730         * data-streamer.h (streamer_write_char_stream): CSE
14731         obs->current_pointer.
14732         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14733         streamer_write_char_stream manually and optimize the resulting loop.
14734         (streamer_write_hwi_stream): Likewise.
14736 2013-06-12  Jan Hubicka  <jh@suse.cz>
14738         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14739         * cgraph.h (varpool_create_empty_node): Declare.
14740         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14741         duplicated nodes.
14742         * symtab.c (symtab_unregister_node): Be lax about missin entries
14743         in node hash.
14744         (symtab_get_node): Update comment.
14745         * varpool.c (varpool_create_empty_node): Break out from ...
14746         (varpool_node_for_decl): ... here.
14747         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14749 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14751         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14752         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14753         part.  Use straight-line flow at the end.
14754         <COMPONENT_REF>: Remove superfluous else.
14755         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14757 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14759         PR target/56564
14760         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14761         target hook even for !TREE_PUBLIC decls.  If no resolution info
14762         is available, return false for common and external decls.
14764 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14766         * config/rl78/constraints.md (U): New constraint.
14767         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14768         valloc attribute.
14770 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14772         PR target/57589
14773         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14774         to allow returning address to AT_PLATFORM name.
14776 2013-06-11  Jan Hubicka  <jh@suse.cz>
14778         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14779         * cgraph.h (symtab_node_base): Add weakref flag.
14780         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14781         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14782         (output_weakrefs): Use weakref flag.
14783         * fold-const.c (simple_operand_p): Handle WEAK.
14784         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14785         * ipa.c (varpool_externally_visible_p): Drop weakref.
14786         (function_and_variable_visibility): Update comment; fix weakref
14787         sanity checks; do not clear DECL_WEAK on them.
14788         * lto-cgraph.c (lto_output_node): update.
14789         (lto_output_varpool_node): Update.
14790         (input_overwrite_node): Update.
14791         (input_node): Update.
14792         (input_varpool_node): Update.
14793         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14794         (lto_symtab_merge_symbols): Add sanity check.
14795         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14796         * passes.c (rest_of_decl_compilation): Set static flag, too.
14797         * symtab.c (dump_symtab_base): Dump weakref.
14798         (verify_symtab_base): Sanity check weakrefs.
14799         (symtab_make_decl_local): Remove duplicated code.
14800         (symtab_alias_ultimate_target): Simplify.
14801         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14803 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14805         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14806         sequence_vect == NULL.
14808 2013-06-11  DJ Delorie  <dj@redhat.com>
14810         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14811         (rl78_unwind_word_mode): New.
14813 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14815         * final.c (debug_prefix_maps): Make static.
14817 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14819         * function.c (initial_trampoline): Remove stray copy.
14821 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14823         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14825 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14827         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14828         within bounds at the beginning of the function.
14830 2013-06-11  Alan Modra  <amodra@gmail.com>
14832         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14833         reporting.
14834         (get_named_section): Don't NULL !DECL_P decl.
14836 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14838         * doc/invoke.texi (core-avx2): Document.
14839         (slm): Likewise.
14840         (atom): Updated with MOVBE.
14842 2013-06-11  Richard Biener  <rguenther@suse.de>
14844         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14846 2013-06-11  Anton Blanchard  <anton@samba.org>
14848         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14849         correct shift value in little-endian mode.
14851 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14853         PR target/56564
14854         * varasm.c (get_variable_align): Move #endif to the right place.
14856 2013-06-10  Cary Coutant  <ccoutant@google.com>
14858         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14859         for hash so that hash table traversal order is deterministic.
14861 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14862             Pat Haugen  <pthaugen@us.ibm.com>
14863             Peter Bergner  <bergner@vnet.ibm.com>
14865         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14866         of vectors in GPRS if they are direct moves or quad word load or
14867         store moves.
14869         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14870         declaration.
14871         (direct_move_p): Likewise.
14872         (quad_load_store_p): Likewise.
14874         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14875         classes into bins based on the physical register type.
14876         (reg_class_to_reg_type): Likewise.
14877         (IS_STD_REG_TYPE): Likewise.
14878         (IS_FP_VECT_REG_TYPE): Likewise.
14879         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14880         use direct move instructions.
14881         (reload_gpr_vsx): Likewise.
14882         (reload_vsx_gpr): Likewise.
14883         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14884         information that is a simplification of register classes.  Also
14885         precalculate direct move reload helpers.
14886         (direct_move_p): New function to return true if the operation can
14887         be done as a direct move instruciton.
14888         (quad_load_store_p): New function to return true if the operation
14889         is a quad memory operation.
14890         (rs6000_legitimize_address): If quad memory, only allow register
14891         indirect for TImode addresses.
14892         (rs6000_legitimate_address_p): Likewise.
14893         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14894         (rs6000_reload_register_type): Likewise.
14895         (register_to_reg_type): Return register type.
14896         (rs6000_secondary_reload_simple_move): New helper function for
14897         secondary reload and secondary memory needed to identify anything
14898         that is a simple move, and does not need reloading.
14899         (rs6000_secondary_reload_direct_move): New helper function for
14900         secondary reload to identify cases that can be done with several
14901         instructions via the direct move instructions.
14902         (rs6000_secondary_reload_move): New helper function for secondary
14903         reload to identify moves between register types that can be done.
14904         (rs6000_secondary_reload): Add support for quad memory operations
14905         and for direct move.
14906         (rs6000_secondary_memory_needed): Likewise.
14907         (rs6000_debug_secondary_memory_needed): Change argument names.
14908         (rs6000_output_move_128bit): New function to return the move to
14909         use for 128-bit moves, including knowing about the various
14910         limitations of quad memory operations.
14912         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14913         memory operations.  call rs6000_output_move_128bit for the actual
14914         instruciton(s) to generate.
14915         (vsx_movti_64bit): Likewise.
14917         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14918         (UNSPEC_P8V_MTVSRWZ): Likewise.
14919         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14920         (UNSPEC_P8V_MTVSRD): Likewise.
14921         (UNSPEC_P8V_XXPERMDI): Likewise.
14922         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14923         (UNSPEC_FUSION_GPR): Likewise.
14924         (FMOVE128_GPR): New iterator for direct move.
14925         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14926         (f32_sv): Likewise.
14927         (f32_dm): Likewise.
14928         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14929         loads and direct move instructions.
14930         (zero_extendsidi2_lfiwzx): Likewise.
14931         (extendsidi2_lfiwax): Likewise.
14932         (extendsidi2_nocell): Likewise.
14933         (floatsi<mode>2_lfiwax): Likewise.
14934         (lfiwax): Likewise.
14935         (floatunssi<mode>2_lfiwzx): Likewise.
14936         (lfiwzx): Likewise.
14937         (fix_trunc<mode>_stfiwx): Likewise.
14938         (fixuns_trunc<mode>_stfiwx): Likewise.
14939         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14940         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14941         (parity<mode>2_cmpb): Set length/type attr.
14942         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14943         for 'mr.' to fast_compare.
14944         (bpermd_<mode>): Change type attr to popcnt.
14945         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14946         (p8_mtvsrwz_1): Likewise.
14947         (p8_mtvsrwz_2): Likewise.
14948         (reload_fpr_from_gpr<mode>): Likewise.
14949         (p8_mtvsrd_1): Likewise.
14950         (p8_mtvsrd_2): Likewise.
14951         (p8_xxpermdi_<mode>): Likewise.
14952         (reload_vsx_from_gpr<mode>): Likewise.
14953         (reload_vsx_from_gprsf): Likewise.
14954         (p8_mfvsrd_3_<mode>): LIkewise.
14955         (reload_gpr_from_vsx<mode>): Likewise.
14956         (reload_gpr_from_vsxsf): Likewise.
14957         (p8_mfvsrd_4_disf): Likewise.
14958         (multi-word GPR splits): Do not split direct moves or quad memory
14959         operations.
14961 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14963         * tree-into-ssa.c (interesting_blocks): Make static.
14965 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14967         PR target/56564
14968         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14969         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14970         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14971         (get_variable_align): New function.
14972         (get_variable_section, emit_bss, emit_common,
14973         assemble_variable_contents, place_block_symbol): Use
14974         get_variable_align instead of DECL_ALIGN.
14975         (assemble_noswitch_variable): Add align argument, use it
14976         instead of DECL_ALIGN.
14977         (assemble_variable): Adjust caller.  Use get_variable_align
14978         instead of DECL_ALIGN.
14979         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14980         caller.
14981         (DATA_ABI_ALIGNMENT): Define.
14982         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14983         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14984         opt is false, only return the psABI mandated alignment increase.
14985         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14986         (DATA_ABI_ALIGNMENT): ... this.
14987         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14988         (DATA_ABI_ALIGNMENT): ... this.
14989         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14990         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14991         (DATA_ABI_ALIGNMENT): ... this.
14992         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14993         * doc/tm.texi: Regenerated.
14995 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14997         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14998         cmp_code to construct REG_EQUAL note.
15000 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
15002         PR target/57568
15003         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
15004         that operands[2] doesn't overlap with operands[0].
15006 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
15007             Jan Hubicka  <jh@suse.cz>
15009         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
15010         hack to mark symbols as used.
15012 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
15014         PR rtl-optimization/57559
15015         * lra-constraints.c (process_alt_operands): Don't discourage
15016         memory with known offset for offsetable memory constraint.
15017         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
15019 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
15021         * varasm.c (struct oc_local_state): Reorder fields.
15022         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
15023         and adjust accordingly.
15024         (output_constructor): Reorder initialization code and adjust call to
15025         output_constructor_bitfield.
15027 2013-06-07  Jan Hubicka  <jh@suse.cz>
15029         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
15031 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
15033         * tree-object-size.c (unknown): Make const.
15035 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15037         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
15038         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
15039         for last alternative in the cpu_facility attribute.
15041 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15043         PR target/56315
15044         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
15045         (xordi3): Change operand 2 constraint to arm_xordi_operand.
15046         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
15047         * config/arm/constraints.md (Dg): New constraint.
15048         * config/arm/neon.md (xordi3_neon): Remove.
15049         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
15050         * config/arm/predicates.md (arm_xordi_operand): New predicate.
15052 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15054         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
15055         Clean up alternatives.
15057 2013-06-07  Alan Modra  <amodra@gmail.com>
15059         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
15060         va_list_gpr_size.
15062 2013-06-07  Alan Modra  <amodra@gmail.com>
15064         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
15066 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15068         * config/arm/constraints.md (Df): New constraint.
15069         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
15070         Correct length attribute for last two alternatives.
15072 2013-06-07  Alan Modra  <amodra@gmail.com>
15074         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
15075         override user -mfp-in-toc.
15076         (offsettable_ok_by_alignment): Consider just the current access
15077         rather than the whole object, unless BLKmode.  Handle
15078         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
15079         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
15080         for -mcmodel=medium.
15081         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15082         override user -mfp-in-toc or -msum-in-toc.  Default to
15083         -mno-fp-in-toc for -mcmodel=medium.
15085 2013-06-06  DJ Delorie  <dj@redhat.com>
15087         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15088         TARGET_VALID_POINTER_MODE.
15090 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15091             Pat Haugen  <pthaugen@us.ibm.com>
15092             Peter Bergner  <bergner@vnet.ibm.com>
15094         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15095         Document new power8 builtins.
15097         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15098         condition code register, to allow 128-bit logical operations to be
15099         done in the VSX or GPR registers.
15100         (nor<mode>3): Use the canonical form for nor.
15101         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15102         vclz*, and vpopcnt* vector instructions.
15103         (nand<mode>3): Likewise.
15104         (orc<mode>3): Likewise.
15105         (clz<mode>2): LIkewise.
15106         (popcount<mode>2): Likewise.
15108         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15109         that only the GPRs are recognized.
15111         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15112         support for new power8 builtins.
15114         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15115         builtin functions.
15116         (xscvdpspn): Likewise.
15117         (vclz): Likewise.
15118         (vclzb): Likewise.
15119         (vclzh): Likewise.
15120         (vclzw): Likewise.
15121         (vclzd): Likewise.
15122         (vpopcnt): Likewise.
15123         (vpopcntb): Likewise.
15124         (vpopcnth): Likewise.
15125         (vpopcntw): Likewise.
15126         (vpopcntd): Likewise.
15127         (vgbbd): Likewise.
15128         (vmrgew): Likewise.
15129         (vmrgow): Likewise.
15130         (eqv): Likewise.
15131         (eqv_v16qi3): Likewise.
15132         (eqv_v8hi3): Likewise.
15133         (eqv_v4si3): Likewise.
15134         (eqv_v2di3): Likewise.
15135         (eqv_v4sf3): Likewise.
15136         (eqv_v2df3): Likewise.
15137         (nand): Likewise.
15138         (nand_v16qi3): Likewise.
15139         (nand_v8hi3): Likewise.
15140         (nand_v4si3): Likewise.
15141         (nand_v2di3): Likewise.
15142         (nand_v4sf3): Likewise.
15143         (nand_v2df3): Likewise.
15144         (orc): Likewise.
15145         (orc_v16qi3): Likewise.
15146         (orc_v8hi3): Likewise.
15147         (orc_v4si3): Likewise.
15148         (orc_v2di3): Likewise.
15149         (orc_v4sf3): Likewise.
15150         (orc_v2df3): Likewise.
15152         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15153         allow power8 quad mode in 64-bit.
15154         (rs6000_builtin_vectorized_function): Add support to vectorize
15155         ISA 2.07 count leading zeros, population count builtins.
15156         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15157         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15158         (builtin_function_type): Add vgbbd builtin function which takes an
15159         unsigned argument.
15160         (altivec_expand_vec_perm_const): Add support for new power8 merge
15161         instructions.
15163         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15164         that does not include TImdoe for use with 32-bit.
15165         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15166         instructions.
15167         (UNSPEC_VSX_CVDPSPN): Likewise.
15168         (vsx_xscvdpspn): Likewise.
15169         (vsx_xscvspdpn): Likewise.
15170         (vsx_xscvdpspn_scalar): Likewise.
15171         (vsx_xscvspdpn_directmove): Likewise.
15172         (vsx_and<mode>3): Split logical operations into 32-bit and
15173         64-bit. Add support to do logical operations on TImode as well as
15174         VSX vector types.  Allow logical operations to be done in either
15175         VSX registers or in general purpose registers in 64-bit mode.  Add
15176         splitters if GPRs were used. For AND, add clobber of CCmode to
15177         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15178         encoding.
15179         (vsx_and<mode>3_32bit): Likewise.
15180         (vsx_and<mode>3_64bit): Likewise.
15181         (vsx_ior<mode>3): Likewise.
15182         (vsx_ior<mode>3_32bit): Likewise.
15183         (vsx_ior<mode>3_64bit): Likewise.
15184         (vsx_xor<mode>3): Likewise.
15185         (vsx_xor<mode>3_32bit): Likewise.
15186         (vsx_xor<mode>3_64bit): Likewise.
15187         (vsx_one_cmpl<mode>2): Likewise.
15188         (vsx_one_cmpl<mode>2_32bit): Likewise.
15189         (vsx_one_cmpl<mode>2_64bit): Likewise.
15190         (vsx_nor<mode>3): Likewise.
15191         (vsx_nor<mode>3_32bit): Likewise.
15192         (vsx_nor<mode>3_64bit): Likewise.
15193         (vsx_andc<mode>3): Likewise.
15194         (vsx_andc<mode>3_32bit): Likewise.
15195         (vsx_andc<mode>3_64bit): Likewise.
15196         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15197         and xxlorc instructions.
15198         (vsx_eqv<mode>3_64bit): Likewise.
15199         (vsx_nand<mode>3_32bit): Likewise.
15200         (vsx_nand<mode>3_64bit): Likewise.
15201         (vsx_orc<mode>3_32bit): Likewise.
15202         (vsx_orc<mode>3_64bit): Likewise.
15204         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15206         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15207         instruction.
15208         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15209         (p8_vmrgow): Likewise.
15210         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15211         GPRs to be split under VSX.
15212         (p8v_clz<mode>2): Add power8 count leading zero support.
15213         (p8v_popcount<mode>2): Add power8 population count support.
15214         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15215         support.
15217         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15218         instruction.
15220         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15221         builtin functions.
15222         (vec_nand): Likewise.
15223         (vec_vclz): Likewise.
15224         (vec_vclzb): Likewise.
15225         (vec_vclzd): Likewise.
15226         (vec_vclzh): Likewise.
15227         (vec_vclzw): Likewise.
15228         (vec_vgbbd): Likewise.
15229         (vec_vmrgew): Likewise.
15230         (vec_vmrgow): Likewise.
15231         (vec_vpopcnt): Likewise.
15232         (vec_vpopcntb): Likewise.
15233         (vec_vpopcntd): Likewise.
15234         (vec_vpopcnth): Likewise.
15235         (vec_vpopcntw): Likewise.
15237 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15239         PR rtl-optimization/57468
15240         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15241         spilled pseudos.
15243 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15245         PR rtl-optimization/57459
15246         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15247         type when setting live regs.
15249 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15251         * config/s390/s390.opt (mlra): New option.
15252         * config/s390/s390.c (s390_decompose_address): Check displacement
15253         for all registers for LRA.
15254         (s390_secondary_reload): Don't used secondary reloads for LRA.
15255         (s390_lra_p): New function.
15256         (TARGET_LRA_P): Define.
15257         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15258         of attribute cpu_facility to zarch for the last alternative.
15259         (*cmpmem_short): Ditto.
15261 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15263         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15264         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15265         (arm_expand_prologue): Likewise.
15267 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15269         PR c++/53743
15270         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15271         as this is now done by redirect_edge_and_branch_force.
15272         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15273         barriers, and fix interaction with splitting.
15274         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15275         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15276         reflect changes made in the routine.
15277         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15278         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15279         since this is called in cfglayout mode, and replace partition fixup
15280         with assert as that is now done by force_nonfallthru_and_redirect.
15281         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15282         already be marked with region crossing note.
15283         (insert_section_boundary_note): Make non-static, gate on flag
15284         has_bb_partition, rewrite to also check for multiple partitions.
15285         (rest_of_handle_reorder_blocks): Remove call to
15286         insert_section_boundary_note, now done later during free_cfg.
15287         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15288         * bb-reorder.h (insert_section_boundary_note): Declare.
15289         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15290         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15291         invoke insert_section_boundary_note.
15292         (try_redirect_by_replacing_jump): Remove unnecessary
15293         check for region crossing note.
15294         (fixup_partition_crossing): New function.
15295         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15296         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15297         in non-cfglayout mode.
15298         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15299         remove old code that tried to do this. Emit barrier correctly
15300         when we are in cfglayout mode.
15301         (last_bb_in_partition): New function.
15302         (rtl_split_edge): Correctly fixup partition boundaries.
15303         (commit_one_edge_insertion): Remove old code that tried to
15304         fixup region crossing edge since this is now handled in
15305         split_block, and set up insertion point correctly since
15306         block may now end in a jump.
15307         (verify_hot_cold_block_grouping): Guard against checking when not in
15308         linearized RTL mode.
15309         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15310         notes.
15311         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15312         rtl_verify_flow_info, so not called in cfglayout mode.
15313         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15314         (fixup_reorder_chain): Remove old code that attempted to fixup region
15315         crossing note as this is now handled in force_nonfallthru_and_redirect.
15316         (duplicate_insn_chain): Don't duplicate switch section notes.
15317         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15318         note.
15319         * basic-block.h (emit_barrier_after_bb): Declare.
15321 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15323         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15324         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15325         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15327 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15329         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15330         where appropriate.
15331         * config/arm/ldmstm.md: Regenerate.
15333 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15335         * config/arm/sync.md (atomic_loaddi_1):
15336         Disable predication for arm_restrict_it.
15337         (arm_load_exclusive<mode>): Likewise.
15338         (arm_load_exclusivesi): Likewise.
15339         (arm_load_exclusivedi): Likewise.
15340         (arm_load_acquire_exclusive<mode>): Likewise.
15341         (arm_load_acquire_exclusivesi): Likewise.
15342         (arm_load_acquire_exclusivedi): Likewise.
15343         (arm_store_exclusive<mode>): Likewise.
15344         (arm_store_exclusive<mode>): Likewise.
15345         (arm_store_release_exclusivedi): Likewise.
15346         (arm_store_release_exclusive<mode>): Likewise.
15348 2013-06-06  Richard Biener  <rguenther@suse.de>
15350         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15351         after LTO_null.
15352         (lto_tag_is_tree_code_p): Adjust.
15353         (lto_tag_is_gimple_code_p): Likewise.
15354         (lto_gimple_code_to_tag): Likewise.
15355         (lto_tag_to_gimple_code): Likewise.
15356         (lto_tree_code_to_tag): Likewise.
15357         (lto_tag_to_tree_code): Likewise.
15358         * data-streamer.h (streamer_write_hwi_in_range): Use
15359         uhwi streaming to stream the normalized range.
15360         (streamer_read_hwi_in_range): Likewise.
15362 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15364         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15365         (predicable_short_it): Likewise.
15366         (predicated): Likewise.
15367         (enabled): Handle above.
15368         (define_cond_exec): Set predicated attribute to yes.
15370 2013-06-05  Mike Stump  <mikestump@comcast.net>
15372         * gdbinit.in (__FUNCTION__): Add.
15374 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15376         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15377         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15379 2013-06-05  Jan Hubicka  <jh@suse.cz>
15381         * varasm.c (mark_decl_referenced): Revert the removal until targets
15382         are fixed.
15384 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15386         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15387         instead of mark_decl_referenced.
15389 2013-06-05  Jan Hubicka  <jh@suse.cz>
15391         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15392         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15393         and symtab_used_from_object_file_p.
15394         (cgraph_make_node_local_1): Clear forced_by_abi.
15395         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15396         * cgraph.h (symtab_node_base): Add forced_by_abi.
15397         (decide_is_variable_needed): Remove.
15398         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15399         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15400         (decide_is_symbol_needed): ... this one; handle symbols in general;
15401         always analyze virtuals; honnor forced_by_abi.
15402         (cgraph_finalize_function): Update.
15403         (varpool_finalize_decl): Update.
15404         (symbol_defined_and_needed): Remove.
15405         (analyze_functions): Update.
15406         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15407         output_refs, input_overwrite_node): Handle forced_by_abi.
15408         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15409         (address_taken_from_non_vtable_p): ... this one.
15410         (comdat_can_be_unshared_p_1): New function.
15411         (cgraph_comdat_can_be_unshared_p): Rename to ...
15412         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15413         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15414         (function_and_variable_visibility): Clear forced_by_abi as needed.
15415         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15416         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15417         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15418         * varpool.c (decide_is_variable_needed): Remove.
15420 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15422         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15423         (arm_option_override): Override arm_restrict_it where appropriate.
15424         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15425         * config/arm/arm.opt (mrestrict-it): New command-line option.
15426         * doc/invoke.texi: Document -mrestrict-it.
15428 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15430         * tsan.c (tsan_atomic_table): Make const.
15432 2013-06-05  Richard Biener  <rguenther@suse.de>
15434         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15435         index associated with the tree we are supposed to replace.
15436         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15437         TREE_ASM_WRITTEN as zero for everything but SSA names.
15439 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15441         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15443 2013-06-04  Jan Hubicka  <jh@suse.cz>
15445         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15446         (input_node, input_varpool_node): Handle correctly external same
15447         body aliases.
15448         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15449         nodes at ltrans stage.
15451 2013-06-04  Jan Hubicka  <jh@suse.cz>
15453         * ipa-inline.c (update_caller_keys): Fix availability test.
15454         (update_callee_keys): Likewise.
15455         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15456         to follow ELF standard.
15458 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15460         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15461         (mips64r5900el-*-elf*): New configurations.
15462         * config/mips/mips-cpus.def (r5900): New processor.
15463         * config/mips/mips-tables.opt: Regenerate.
15464         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15465         (mips_issue_rate): Handle PROCESSOR_R5900.
15466         (mips_reorg_process_insns): Force reorder mode for the R5900.
15467         * config/mips/mips.h (TARGET_MIPS5900): Define.
15468         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15469         TARGET_MIPS5900.
15470         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15471         TARGET_MIPS5900.
15472         * config/mips/mips.md (processor): Add r5900.
15473         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15475 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15477         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15478         into function to generate MOVI instruction.
15479         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15480         (aarch64_preferred_simd_mode): Turn into wrapper.
15481         (aarch64_output_scalar_simd_mov_immediate): New function.
15482         * config/aarch64/aarch64-protos.h: Add prototype for above.
15484 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15486         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15487         element_char member.
15488         (sizetochar): Return signed char.
15489         (aarch64_simd_valid_immediate): Remove elchar and other
15490         unnecessary variables.
15491         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15492         Calculate element_char as required.
15493         * config/aarch64/aarch64-protos.h: Update and move prototype
15494         for aarch64_output_simd_mov_immediate.
15495         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15496         Update arguments.
15498 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15500         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15501         information completed by aarch64_simd_valid_immediate.
15502         (aarch64_legitimate_constant_p): Update arguments.
15503         (aarch64_simd_valid_immediate): Work with struct rather than many
15504         pointers.
15505         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15506         (aarch64_simd_make_constant): Update arguments.
15507         (aarch64_output_simd_mov_immediate): Work with struct rather than
15508         many pointers.  Output immediate directly rather than as operand.
15509         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15510         Update prototype.
15511         * config/aarch64/constraints.md (Dn): Update arguments.
15513 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15515         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15516         longer static.
15517         (aarch64_simd_immediate_valid_for_move): Remove.
15518         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15519         (aarch64_simd_make_constant): Update call.
15520         (aarch64_output_simd_mov_immediate): Update call.
15521         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15522         Add prototype.
15523         * config/aarch64/constraints.md (Dn): Update call.
15525 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15527         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15528         return type to bool for prototype.
15529         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15530         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15531         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15533 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15535         * config/mips/mips.opt (meva): New.
15536         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15537         (ASM_SPEC): Handle -meva.
15538         * doc/invoke.texi (meva):  Document.
15540 2013-06-04  Alan Modra  <amodra@gmail.com>
15542         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15543         constant output.
15545 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15547         * rtl.def: Add extra fourth optional field to define_cond_exec.
15548         * gensupport.c (process_one_cond_exec): Process attributes from
15549         define_cond_exec.
15550         * doc/md.texi: Document fourth field in define_cond_exec.
15552 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15554         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15555         out the processing order as in store_bit_field_1.
15557 2013-06-04  Jan Hubicka  <jh@suse.cz>
15559         PR middle-end/57500
15560         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15561         non-VAR_DECL node if it does not exist yet.
15563 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15565         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15566         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15567         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15568         target_cpu_default setting.
15570 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15572         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15573         default to -fopt-info=optimized instead of all.
15574         * doc/invoke.texi: Ditto.
15575         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15576         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15577         (execute_vect_slp): Emit BB vectorization success under
15578         MSG_OPTIMIZED_LOCATIONS.
15579         * tree-vect-slp.c (vect_slp_transform_bb): Change
15580         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15581         * tree-vect-loop.c (vect_transform_loop): Ditto.
15583 2013-06-03  Jason Merrill  <jason@redhat.com>
15585         PR c++/57415
15586         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15587         Use TARGET_EXPR for C++.
15589 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15591         PR rtl-optimization/57268
15592         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15593         if DEBUG_INSN_P (insn).
15595         Reapply
15596         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15598         PR rtl-optimization/57268
15599         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15600         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15602 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15604         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15605         (ix86_avoid_lea_for_addr): Likewise.
15606         (exact_dependency_1): Likewise.
15607         (ix86_adjust_cost): Likewise.
15608         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15609         removed.
15610         (do_reorder_for_imul): Fix typo, formatting and
15611         !reload_completed check removed.
15612         (ix86_sched_reorder): Fix typo and formatting.
15613         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15614         list.
15616 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15618         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15620 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15622         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15623         <STRING_CST>: Likewise.
15624         <VECTOR_CST>: Likewise.
15626 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15627             Mikael Morin  <mikael@gcc.gnu.org>
15629         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15630         * config.in: Regenerated.
15631         * configure: Regenerated.
15633 2013-06-01  Jan Hubicka  <jh@suse.cz>
15635         PR middle-end/57366
15636         * cgraphunit.c (compile): When weakref is not supported,
15637         set up transparent aliases before final output pass.
15638         * varasm.c (assemble_alias): Do not try to do it here.
15640 2013-06-01  Jan Hubicka  <jh@suse.cz>
15642         PR middle-end/57467
15643         * passes.c (for_per_function): Skip unanalyzed functions.
15645 2013-06-01  Jan Hubicka  <jh@suse.cz>
15647         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15648         (lto_symtab_merge_symbols_1): ... this one.
15649         (lto_symtab_merge_cgraph_nodes): Rename to ...
15650         (lto_symtab_merge_symbols): ... this one; simplify.
15651         * cgraph.c (same_body_aliases_done): Rename to ...
15652         (cpp_implicit_aliases_done): ... this one.
15653         (cgraph_create_function_alias): Update.
15654         (cgraph_same_body_alias): Update.
15655         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15656         (verify_edge_corresponds_to_fndecl): Simplify.
15657         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15658         (cgraph_node): Remove same_body_alias.
15659         (varpool_node): Remove alias_of and extra_name_alias.
15660         (same_body_aliases_done): Rename to ..
15661         (cpp_implicit_aliases_done): ... this one.
15662         (symtab_alias_ultimate_target): Add default parameter.
15663         (symtab_resolve_alias): New function.
15664         (fixup_same_cpp_alias_visibility): Declare.
15665         (cgraph_function_node): Add default parameter.
15666         (cgraph_node_asm_name): Likewise.
15667         (cgraph_function_or_thunk_node): Add default parameter; do
15668         not ICE when it is NULL.
15669         (varpool_variable_node): Likewise.
15670         * tree-emutls.c (create_emultls_var): Update.
15671         (ipa_lower_emutls): Update.
15672         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15673         (cgraph_reset_node): Reset alias info.
15674         (cgraph_finalize_function): Update.
15675         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15676         (analyze_function): Simplify.
15677         (cgraph_process_same_body_aliases): Simplify.
15678         (analyze_functions): Fixup same body aliases.
15679         (handle_alias_pairs): Simplify.
15680         (assemble_thunk): Update.
15681         (assemble_thunks_and_aliases): Update.
15682         (output_weakrefs): Rewrite.
15683         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15684         (lto_output_varpool_node): Likewise.
15685         (compute_ltrans_boundary): Remve assert.
15686         (get_alias_symbol): New functoin.
15687         (input_node): Rewrite alias handling.
15688         (input_varpool_node): Likewise.
15689         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15690         * ipa.c (process_references): Handle weakrefs correctly.
15691         (symtab_remove_unreachable_nodes): Likewise.
15692         * trans-mem.c (get_cg_data): Update.
15693         (ipa_tm_create_version_alias): Update.
15694         (ipa_tm_execute): Update.
15695         * symtab.c (dump_symtab_base): Dump aliases.
15696         (verify_symtab_base): Verify aliases.
15697         (symtab_node_availability): New function.
15698         (symtab_alias_ultimate_target): Simplify.
15699         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15700         handle all the fixup cases.
15701         (symtab_resolve_alias): New function.
15702         * passes.c (ipa_write_summaries): Handle weakrefs.
15703         * varpool.c (varpool_analyze_node): Simplify.
15704         (assemble_aliases): Update.
15705         (varpool_create_variable_alias): Simplify.
15706         (varpool_extra_name_alias): Simplify.
15707         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15708         (lto_symtab_merge_symbols): ... this one.
15710 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15712         Revert
15713         PR rtl-optimization/57268
15714         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15715         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15717 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15719         Partially reverted:
15720         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15722         PR middle-end/57073
15723         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15724         further up.
15726 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15728         PR rtl-optimization/57268
15729         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15730         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15732 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15734         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15735         unordered comparison operators when -fno-trapping-math is in effect
15736         on the e500.
15737         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15738         and implement unordered comparison operators properly on the e500.
15740 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15742         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15743         for constant scalar integers.
15744         (simplify_relational_operation_1): Likewise.
15746 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15748         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15749         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15750         Fix comment.
15752 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15753             Igor Zamyatin  <igor.zamyatin@intel.com>
15755         Silvermont (SLM) architecture performance tuning.
15756         * config/i386/i386.h (enum ix86_tune_indices): Add
15757         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15758         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15760         * config/i386/i386.c (initial_ix86_tune_features)
15761         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15762         (ix86_lea_outperforms): Handle Silvermont tuning.
15763         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15764         call.
15765         (ix86_use_lea_for_mov): Likewise.
15766         (ix86_avoid_lea_for_addr): Likewise.
15767         (ix86_lea_for_add_ok): Likewise.
15768         (exact_dependency_1): New function.
15769         (exact_store_load_dependency): Likewise.
15770         (ix86_adjust_cost): Handle Silvermont tuning.
15771         (do_reoder_for_imul): Likewise.
15772         (swap_top_of_ready_list): New function.
15773         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15775         * config/i386/i386.md (peepholes that split memory operand in fp
15776         converts): New.
15778 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15780         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15781         Remove un-necessary braces.
15783 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15785         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15786         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15788 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15790         PR middle-end/57073
15791         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15793 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15795         PR target/56315
15796         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15797         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15798         * config/arm/neon.md (iordi3_neon): Remove.
15799         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15800         * config/arm/predicates.md (imm_for_neon_logic_operand):
15801         Move to earlier in the file.
15802         (neon_logic_op2): Likewise.
15803         (arm_iordi_operand_neon): New predicate.
15805 2013-05-31  Richard Biener  <rguenther@suse.de>
15807         PR tree-optimization/57478
15808         PR tree-optimization/57453
15809         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15810         are life as well.
15812 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15814         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15815         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15817 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15818             Thomas Koenig  <tkoenig@gcc.gnu.org>
15820         PR middle-end/57073
15821         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15822         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15824 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15826         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15828 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15830         * target.def (register_usage_leveling_p): New hook.
15831         * targhooks.c (default_register_usage_leveling_p): New.
15832         * targhooks.h (default_register_usage_leveling_p): New prototype.
15833         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15834         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15835         * doc/tm.texi: Update.
15836         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15838 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15840         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15841         (*insv_reg<mode>): New define_insn.
15843 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15845         PR rtl-optimization/57439
15846         * postreload.c (move2add_valid_value_p): Check that we have
15847         a zero subreg_regno_offset when accessing the register in
15848         the requested mode.
15850 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15851             Igor Zamyatin  <igor.zamyatin@intel.com>
15853         Silvermont (SLM) architecture pipeline model, tuning and
15854         insn selection.
15855         * config.gcc: Add slm config options and target.
15857         * config/i386/slm.md: New.
15859         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15861         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15862         PROCESSOR_SLM.
15863         (ix86_target_macros_internal): Likewise.
15865         * config/i386/i386.c (slm_cost): New cost.
15866         (m_SLM): New macro flag.
15867         (initial_ix86_tune_features): Set m_SLM.
15868         (x86_accumulate_outgoing_args): Likewise.
15869         (x86_arch_always_fancy_math_387): Likewise.
15870         (processor_target_table): Add slm cost.
15871         (cpu_names): Add slm cpu name.
15872         (x86_option_override_internal): Set SLM ISA.
15873         (ix86_issue_rate): New case PROCESSOR_SLM.
15874         (ia32_multipass_dfa_lookahead): Likewise.
15875         (fold_builtin_cpu): Add slm.
15877         * config/i386/i386.h (TARGET_SLM): New target macro.
15878         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15879         (processor_type): Add PROCESSOR_SLM.
15881         * config/i386/i386.md (cpu): Add new value "slm".
15882         (slm.md): Include slm.md.
15884 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15885             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15887         * config/arm/arm-protos.h: Add and update function protos.
15888         * config/arm/arm.c (use_simple_return_p): New added.
15889         (thumb2_expand_return): Check simple_return flag.
15890         * config/arm/arm.md: Add simple_return and conditional simple_return.
15891         * config/arm/iterators.md: Add iterator for return and simple_return.
15893 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15895         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15896         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15897         (arm_emit_vfp_multi_reg_pop): Likewise.
15898         (thumb2_emit_ldrd_pop): Likewise.
15899         (arm_expand_epilogue): Add misc REG_CFA notes.
15900         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15902 2013-05-29  Lawrence Crowl  <crowl@google.com>
15904         * config/arm/t-arm: Update for below.
15906         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15907         Change type to hash_table.  Update dependent calls and types.
15909         * config/i386/t-cygming: Update for below.
15911         * config/i386/t-interix: Update for below.
15913         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15914         Change type to hash_table.  Update dependent calls and types.
15915         (i386_find_on_wrapper_list::wrappers): Likewise.
15917         * config/ia64/t-ia64: Update for below.
15919         * config/ia64/ia64.c (bundle_state_table):
15920         Change type to hash_table.  Update dependent calls and types.
15922         * config/mips/mips.c (mips_reorg_process_insns::htab):
15923         Change type to hash_table.  Update dependent calls and types.
15925         * config/sol2.c (solaris_comdat_htab):
15926         Change type to hash_table.  Update dependent calls and types.
15928         * config/t-sol2: Update for above.
15930 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15932         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15933         functions are not yet marked as defined.
15935 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15936             Pat Haugen  <pthaugen@us.ibm.com>
15937             Peter Bergner  <bergner@vnet.ibm.com>
15939         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15940         instructions.
15941         (VEC_A): Likewise.
15942         (VEC_C): Likewise.
15943         (vrotl<mode>3): Likewise.
15944         (vashl<mode>3): Likewise.
15945         (vlshr<mode>3): Likewise.
15946         (vashr<mode>3): Likewise.
15948         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15949         support for power8 V2DI builtins.
15951         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15952         power8 V2DI builtins.
15953         (vupkhsw): Likewise.
15954         (vupklsw): Likewise.
15955         (vaddudm): Likewise.
15956         (vminsd): Likewise.
15957         (vmaxsd): Likewise.
15958         (vminud): Likewise.
15959         (vmaxud): Likewise.
15960         (vpkudum): Likewise.
15961         (vpksdss): Likewise.
15962         (vpkudus): Likewise.
15963         (vpksdus): Likewise.
15964         (vrld): Likewise.
15965         (vsld): Likewise.
15966         (vsrd): Likewise.
15967         (vsrad): Likewise.
15968         (vsubudm): Likewise.
15969         (vcmpequd): Likewise.
15970         (vcmpgtsd): Likewise.
15971         (vcmpgtud): Likewise.
15972         (vcmpequd_p): Likewise.
15973         (vcmpgtsd_p): Likewise.
15974         (vcmpgtud_p): Likewise.
15975         (vupkhsw): Likewise.
15976         (vupklsw): Likewise.
15977         (vaddudm): Likewise.
15978         (vmaxsd): Likewise.
15979         (vmaxud): Likewise.
15980         (vminsd): Likewise.
15981         (vminud): Likewise.
15982         (vpksdss): Likewise.
15983         (vpksdus): Likewise.
15984         (vpkudum): Likewise.
15985         (vpkudus): Likewise.
15986         (vrld): Likewise.
15987         (vsld): Likewise.
15988         (vsrad): Likewise.
15989         (vsrd): Likewise.
15990         (vsubudm): Likewise.
15992         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15993         support for power8 V2DI instructions.
15995         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15996         power8 V2DI instructions.  Combine pack and unpack insns to use an
15997         iterator for each mode.  Check whether a particular mode supports
15998         Altivec instructions instead of just checking TARGET_ALTIVEC.
15999         (UNSPEC_VPKUWUM): Likewise.
16000         (UNSPEC_VPKSHSS): Likewise.
16001         (UNSPEC_VPKSWSS): Likewise.
16002         (UNSPEC_VPKUHUS): Likewise.
16003         (UNSPEC_VPKSHUS): Likewise.
16004         (UNSPEC_VPKUWUS): Likewise.
16005         (UNSPEC_VPKSWUS): Likewise.
16006         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
16007         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
16008         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
16009         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
16010         (UNSPEC_VUPKHSB): Likewise.
16011         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
16012         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
16013         (UNSPEC_VUPKHSH): Likewise.
16014         (UNSPEC_VUPKLSB): Likewise.
16015         (UNSPEC_VUPKLSH): Likewise.
16016         (VI2): Likewise.
16017         (VI_char): Likewise.
16018         (VI_scalar): Likewise.
16019         (VI_unit): Likewise.
16020         (VP): Likewise.
16021         (VP_small): Likewise.
16022         (VP_small_lc): Likewise.
16023         (VU_char): Likewise.
16024         (add<mode>3): Likewise.
16025         (altivec_vaddcuw): Likewise.
16026         (altivec_vaddu<VI_char>s): Likewise.
16027         (altivec_vadds<VI_char>s): Likewise.
16028         (sub<mode>3): Likewise.
16029         (altivec_vsubcuw): Likewise.
16030         (altivec_vsubu<VI_char>s): Likewise.
16031         (altivec_vsubs<VI_char>s): Likewise.
16032         (altivec_vavgs<VI_char>): Likewise.
16033         (altivec_vcmpbfp): Likewise.
16034         (altivec_eq<mode>): Likewise.
16035         (altivec_gt<mode>): Likewise.
16036         (altivec_gtu<mode>): Likewise.
16037         (umax<mode>3): Likewise.
16038         (smax<mode>3): Likewise.
16039         (umin<mode>3): Likewise.
16040         (smin<mode>3): Likewise.
16041         (altivec_vpkuhum): Likewise.
16042         (altivec_vpkuwum): Likewise.
16043         (altivec_vpkshss): Likewise.
16044         (altivec_vpkswss): Likewise.
16045         (altivec_vpkuhus): Likewise.
16046         (altivec_vpkshus): Likewise.
16047         (altivec_vpkuwus): Likewise.
16048         (altivec_vpkswus): Likewise.
16049         (altivec_vpks<VI_char>ss): Likewise.
16050         (altivec_vpks<VI_char>us): Likewise.
16051         (altivec_vpku<VI_char>us): Likewise.
16052         (altivec_vpku<VI_char>um): Likewise.
16053         (altivec_vrl<VI_char>): Likewise.
16054         (altivec_vsl<VI_char>): Likewise.
16055         (altivec_vsr<VI_char>): Likewise.
16056         (altivec_vsra<VI_char>): Likewise.
16057         (altivec_vsldoi_<mode>): Likewise.
16058         (altivec_vupkhsb): Likewise.
16059         (altivec_vupkhs<VU_char>): Likewise.
16060         (altivec_vupkls<VU_char>): Likewise.
16061         (altivec_vupkhsh): Likewise.
16062         (altivec_vupklsb): Likewise.
16063         (altivec_vupklsh): Likewise.
16064         (altivec_vcmpequ<VI_char>_p): Likewise.
16065         (altivec_vcmpgts<VI_char>_p): Likewise.
16066         (altivec_vcmpgtu<VI_char>_p): Likewise.
16067         (abs<mode>2): Likewise.
16068         (vec_unpacks_hi_v16qi): Likewise.
16069         (vec_unpacks_hi_v8hi): Likewise.
16070         (vec_unpacks_lo_v16qi): Likewise.
16071         (vec_unpacks_hi_<VP_small_lc>): Likewise.
16072         (vec_unpacks_lo_v8hi): Likewise.
16073         (vec_unpacks_lo_<VP_small_lc>): Likewise.
16074         (vec_pack_trunc_v8h): Likewise.
16075         (vec_pack_trunc_v4si): Likewise.
16076         (vec_pack_trunc_<mode>): Likewise.
16078         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
16079         V2DI builtins.
16080         (vec_vmaxsd): Likewise.
16081         (vec_vmaxud): Likewise.
16082         (vec_vminsd): Likewise.
16083         (vec_vminud): Likewise.
16084         (vec_vpksdss): Likewise.
16085         (vec_vpksdus): Likewise.
16086         (vec_vpkudum): Likewise.
16087         (vec_vpkudus): Likewise.
16088         (vec_vrld): Likewise.
16089         (vec_vsld): Likewise.
16090         (vec_vsrad): Likewise.
16091         (vec_vsrd): Likewise.
16092         (vec_vsubudm): Likewise.
16093         (vec_vupkhsw): Likewise.
16094         (vec_vupklsw): Likewise.
16096 2013-05-29  Jan Hubicka  <jh@suse.cz>
16098         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16099         flags; reorder rest of fields in more consistent way.
16100         (varpool_node): Remove analyzed, finalized and alias.
16101         (cgraph_ndoe): Likewise.
16102         (symtab_alias_ultimate_target): New function.
16103         (cgraph_function_node): Move offline.
16104         (cgraph_reset_node): Declare.
16105         (cgraph_comdat_can_be_unshared_p): Remove.
16106         (varpool_remove_initializer): Declare.
16107         (varpool_first_defined_variable, varpool_next_defined_variable
16108         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16109         (cgraph_function_with_gimple_body_p): Update.
16110         (varpool_all_refs_explicit_p): Update.
16111         (symtab_alias_target): New function.
16112         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16113         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16114         (cgraph_function_or_thunk_node): Simplify using
16115         symtab_alias_ultimate_target.
16116         (varpool_variable_node): Likewise.
16117         * cgraph.c (cgraph_create_function_alias): Update.
16118         (cgraph_add_thunk): Update.
16119         (cgraph_remove_node): Update.
16120         (dump_cgraph_node): Do not dump removed flags.
16121         (cgraph_function_body_availability): Update.
16122         (cgraph_propagate_frequency): Update.
16123         (verify_cgraph_node): Check sanity of local flag.
16124         (cgraph_function_node): Move here from cgraph.h; revamp for
16125         cgraph_function_or_thunk_node.
16126         * lto-symtab.c (lto_varpool_replace_node): Update.
16127         (lto_symtab_resolve_can_prevail_p): Update.
16128         (lto_symtab_merge_cgraph_nodes): Update.
16129         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16130         propagate_constants_accross_call, devirtualization_time_bonus,
16131         ipcp_propagate_stage): Update.
16132         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16133         * ipa-inline-transform.c (clone_inlined_nodes,
16134         preserve_function_body_p): Update.
16135         * ipa-reference.c (propagate): Update.
16136         (write_node_summary_p): Update.
16137         * toplev.c (wrapup_global_declaration_2): Update.
16138         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16139         (analyze_function) ... this one.
16140         (cgraph_process_new_functions): Update.
16141         (cgraph_reset_node): Export.
16142         (cgraph_finalize_function): Update.
16143         (cgraph_add_new_function): Update.
16144         (process_function_and_variable_attributes): Update.
16145         (varpool_finalize_decl): Update.
16146         (symbol_finalized): Remove.
16147         (symbol_finalized_and_needed): Rename to ...
16148         (symbol_defined_and_needed): ... update.
16149         (cgraph_analyze_functions): Update.
16150         (handle_alias_pairs): Update.
16151         (mark_functions_to_output): Update.
16152         (assemble_thunk): Update.
16153         (output_in_order): Update.
16154         (output_weakrefs): Update.
16155         (finalize_compilation_unit): Update.
16156         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16157         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16158         input_node, input_varpool_node): Update.
16159         * dbxout.c (dbxout_expand_expr): Update.
16160         * cgraphclones.c (cgraph_clone_node): Update.
16161         (cgraph_copy_node_for_versioning): Update.
16162         (cgraph_materialize_clone): Update.
16163         (cgraph_materialize_all_clones): Update.
16164         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16165         propagate_pure_const, propagate_nothrow): Update.
16166         * lto-streamer-out.c (lto_output, write_symbol): Update.
16167         * ipa-utils.c (ipa_reverse_postorder): Update.
16168         * ipa-inline.c (can_inline_edge_p): Update.
16169         (update_caller_keys, ipa_inline): Update.
16170         * dwarf2out.c (reference_to_unused,
16171         premark_types_used_by_global_vars_helper): Update.
16172         * tree-eh.c (tree_could_trap_p): Update.
16173         * ipa-split.c (consider_split, execute_split_functions): Update.
16174         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16175         has_addr_references_p): Update; move ahead in file for better
16176         readability.
16177         (process_references): Simplify.
16178         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16179         bodies are removed.
16180         (cgraph_comdat_can_be_unshared_p): Make static.
16181         (cgraph_externally_visible_p): Update.
16182         (varpool_externally_visible_p): Update.
16183         (function_and_variable_visibility): Update.
16184         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16185         ipa_tm_mark_force_output_node): Update.
16186         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16187         estimate_edge_devirt_benefit, inline_generate_summary,
16188         inline_write_summary): Update.
16189         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16190         * ipa-prop.c (ipa_compute_jump_functions): Update.
16191         (ipa_print_node_params, ipa_prop_read_section,
16192         ipa_update_after_lto_read, read_replacements_section): Update.
16193         * varasm.c (mark_decl_referenced): Update.
16194         (assemble_alias, dump_tm_clone_pairs): Update.
16195         * tree-inline.c (copy_bb): Update.
16196         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16197         Update.
16198         * symtab.c (dump_symtab_base): Print new flags.
16199         (verify_symtab_base): Verify new flags.
16200         (symtab_alias_ultimate_target): New function.
16201         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16202         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16203         Update.
16204         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16205         Update.
16206         * i386.c (ix86_get_function_versions_dispatcher,
16207         ix86_generate_version_dispatcher_body): Update.
16208         (fold_builtin_cpu): Use varpool_add_new_variable.
16209         * varpool.c (varpool_remove_initializer): Break out from ...
16210         (varpool_remove_node): ... this one.
16211         (dump_varpool_node, varpool_node_for_asm,
16212         cgraph_variable_initializer_availability, varpool_analyze_node,
16213         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16214         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16215         Update.
16217 2013-05-29  Jan Hubicka  <jh@suse.cz>
16219         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16221 2013-05-29  Easwaran Raman  <eraman@google.com>
16223         PR tree-optimization/57442
16224         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16225         when control exits the main loop.
16227 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16229         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16230         and RX600.
16231         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16232         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16233         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16234         option.
16236 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16238         PR tree-optimization/57441
16239         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16240         Don't limit size of incr_vec to number of candidates.
16242 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16244         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16245         and mips16 directories.
16246         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16247         (MULTILIB_DIRNAMES): Ditto.
16248         (MULTILIB_EXCEPTIONS): Add new exceptions.
16249         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16250         (MULTILIB_DIRNAMES): Ditto.
16251         (MULTILIB_EXCEPTIONS): Add new exceptions.
16253 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16254             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16256         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16257         SYMBOL_TINY_ABSOLUTE.
16258         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16259         SYMBOL_TINY_ABSOLUTE.
16260         (aarch64_expand_mov_immediate): Likewise.
16261         (aarch64_classify_symbol): Likewise.
16262         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16263         Permit SYMBOL_TINY_ABSOLUTE.
16264         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16266 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16267             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16269         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16270         Refactor if/switch.  Replace gcc_assert with if.
16272 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16274         * config/i386/i386.c (initial_ix86_tune_features): Enable
16275         FP Reassociation for AMD bdver1 and bdver2.
16277 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16279         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16280         and IMAGPART_EXPR do not occur within other handled_components.
16282 2013-05-29  Richard Biener  <rguenther@suse.de>
16284         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16285         access on whether the use is in the BB we currently try to
16286         vectorize.
16287         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16288         vectorize to vect_bb_slp_scalar_cost.
16290 2013-05-29  Richard Biener  <rguenther@suse.de>
16292         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16293         computing scalar cost offsetted by stmts that are kept live
16294         by scalar uses.
16295         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16296         for computation of scalar cost.
16298 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16300         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16302 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16304         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16305         array notation built-in reduction functions.
16306         * doc/passes.texi (Passes): Added documentation about changes done
16307         for Cilk Plus.
16308         * doc/invoke.texi (C Dialect Options): Added documentation about
16309         the -fcilkplus flag.
16310         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16311         (BUILTINS_DEF): Depend on cilkplus.def.
16312         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16313         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16314         * cilkplus.def: New file.
16316 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16318         PR rtl-optimization/57439
16319         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16321 2013-05-28  Easwaran Raman  <eraman@google.com>
16323         PR tree-optimization/57337
16324         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16325         (find_insert_point): Correctly identify the insertion point
16326         when two statements with the same UID is compared.
16328 2013-05-28  Richard Biener  <rguenther@suse.de>
16330         PR tree-optimization/56787
16331         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16332         from the list of data references.
16333         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16334         clobbers.
16335         (vect_analyze_loop_operations): Likewise.
16336         (vect_transform_loop): Remove clobbers.
16338 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16340         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16341         and REALPART_EXPRs have scalar type.
16343 2013-05-28  Richard Biener  <rguenther@suse.de>
16345         PR tree-optimization/57411
16346         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16347         virtual operands.
16348         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16349         virtual operand propagation.
16351 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16353         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16354         destination register for bmasksi_vis.
16355         (vector_init_bshuffle): Likewise.
16356         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16358 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16360         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16361         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16362         mode if the instruction isn't available in the original mode.
16363         * config/sparc/sparc.opt (mfix-ut699): New option.
16364         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16365         (divdf3): Turn into expander.
16366         (divdf3_nofix): New insn.
16367         (divdf3_fix): Likewise.
16368         (divsf3): Disable if -mfix-ut699.
16369         (sqrtdf2): Turn into expander.
16370         (sqrtdf2_nofix): New insn.
16371         (sqrtdf2_fix): Likewise.
16372         (sqrtsf2): Disable if -mfix-ut699.
16374 2013-05-27  Richard Biener  <rguenther@suse.de>
16376         PR middle-end/57412
16377         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16378         block for the new loop.
16380 2013-05-27  Richard Biener  <rguenther@suse.de>
16382         PR tree-optimization/57343
16383         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16384         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16385         (number_of_iterations_cond): Do not build the folded tree.
16387 2013-05-27  Richard Biener  <rguenther@suse.de>
16389         Revert
16390         PR middle-end/57381
16391         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16392         OEP_CONSTANT_ADDRESS_OF retained.
16394         PR tree-optimization/57417
16395         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16396         for unchanged base.
16397         (set_ssa_val_to): Compare addresses using
16398         get_addr_base_and_unit_offset.
16400 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16402         PR rtl-optimization/56833
16403         * postreload.c (move2add_record_mode): New function.
16404         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16405         (move2add_use_add2_insn): Use move2add_record_sym_value.
16406         (move2add_use_add3_insn): Likewise.
16407         (reload_cse_move2add): Use move2add_valid_value_p and
16408         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16409         affected regs by setting reg_mode to VOIDmode.
16410         (move2add_note_store): Don't pretend the inside of a SUBREG is
16411         the actual destination.  Invalidate single/leading registers by
16412         setting reg_mode to VOIDmode.
16413         Use move2add_record_sym_value, move2add_valid_value_p and
16414         move2add_record_mode.
16416 2013-05-27  Richard Biener  <rguenther@suse.de>
16418         PR tree-optimization/57396
16419         * tree-affine.c (double_int_constant_multiple_p): Properly
16420         return false for val == 0 and div != 0.
16422 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16424         * config/mips/mips.h: Use #elif in preprocessor conditions.
16426 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16428         PR target/53916
16429         * config/mips/constraints.md (kl): New constraint.
16430         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16431         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16432         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16433         from LO for MIPS16.
16434         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16436 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16438         PR target/55777
16439         * config/mips/mips.c (mips_can_inline_p): New function.
16440         (TARGET_CAN_INLINE_P): Define.
16442 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16444         * sched-int.h (ds_t, dw_t): Make unsigned int.
16445         Fix documentation that describes how all the ds_t bits are used.
16446         Reserve the last bit for delayed-branch scheduling.
16447         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16448         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16449         (gen_dep_weak_1): Remove prototype.
16450         * sched-deps.c (get_dep_weak_1): Make static.
16451         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16452         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16453         * doc/tm.texi: Regenerate.
16454         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16456 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16458         PR debug/56950
16459         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16461 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16462             Sandra Loosemore  <sandra@codesourcery.com>
16464         * config.gcc (powerpc-*): Allow native for with-cpu.
16466 2013-05-24  Jeff Law  <law@redhat.com>
16468         PR tree-optimization/57124
16469         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16470         conversion feeding a condition if the range has an overflow
16471         if -fstrict-overflow.  Add warnings for when we do make the
16472         transformation.
16474 2013-05-24  Dehao Chen  <dehao@google.com>
16476         * tree-cfg.c (locus_discrim_map): Fix the typo.
16477         (locus_discrim_hasher): Likewise.
16478         (locus_discrim_hasher::hash): Likewise.
16479         (locus_discrim_hasher::equal): Likewise.
16481 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16483         PR tree-optimization/57294
16484         * cgraph.h (ipa_record_stmt_references): Declare.
16485         * cgraphbuild.c (ipa_record_stmt_references): New function.
16486         (build_cgraph_edges): Use ipa_record_stmt_references.
16487         (rebuild_cgraph_edges): Likewise.
16488         (cgraph_rebuild_references): Likewise.
16489         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16490         associated with the old statement and build references from the
16491         newly built statements.
16492         * ipa-ref.c (ipa_remove_stmt_references): New function.
16493         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16495 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16497         * lra-constraints.c (emit_spill_move): Use smaller mode for
16498         mem-mem moves.
16499         (check_and_process_move): Consider mem-reg moves for secondary
16500         too.
16501         (curr_insn_transform): Don't lose insns emitted before for
16502         secondary memory moves.
16503         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16504         reg set up in the current insn.
16506 2013-05-24  Dehao Chen  <dehao@google.com>
16508         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16509         hash function.
16510         (locus_descrim_hasher::equal): Likewise.
16511         (build_gimple_cfg): New discriminator assignment algorithm.
16512         (make_edges): Likewise.
16513         (next_discriminator_for_locus): Likewise.
16514         (same_line_p): Likewise.
16515         (assign_discriminators): Likewise.
16516         (make_cond_expr_edges): Likewise.
16517         (make_gimple_switch_edges): Likewise.
16518         (make_goto_expr_edges): Likewise.
16519         (make_gimple_asm_edges): Likewise.
16521 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16523         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16524         X format specifier to only display bottom 16 bits.
16525         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16526         immediate to match for operand 2, since it will be masked.
16528 2013-05-24  Richard Biener  <rguenther@suse.de>
16530         PR tree-optimization/57287
16531         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16532         all SSA names that occur in abnormal PHIs.
16534 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16536         PR tree-ssa/57385
16537         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16538         that index is not negative.
16540 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16542         PR rtl-optimization/55177
16543         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16544         (simplify_byte_swapping_operation): New.
16545         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16546         (simplify_relational_operation_1): Deal with BSWAP.
16548 2013-05-23  Richard Henderson  <rth@redhat.com>
16550         PR target/56742
16551         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16552         (ix86_reorg): Call it.
16554 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16556         PR target/57379
16557         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16558         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16559         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16561 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16563         PR debug/57351
16564         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16566 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16567             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16569         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16570         * config/aarch64/constraints.md (Usa): Remove.
16571         * doc/md.texi (AArch64 Usa): Remove.
16573 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16574             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16576         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16577         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16578         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16579         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16581 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16583         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16584         instruction (AdvSIMD).
16585         * config/aarch64/aarch64-builtins.c
16586         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16587         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16589 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16591         PR middle-end/57347
16592         * tree.h (contains_bitfld_component_ref_p): Declare.
16593         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16594         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16595         caller.
16596         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16597         not access a bit-field.  Assert all final offsets are byte-aligned.
16599 2013-05-23  Richard Biener  <rguenther@suse.de>
16601         PR tree-optimization/57380
16602         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16603         least one invariant or re-used load.
16604         * passes.c (init_optimization_passes): Move pass_phiprop before
16605         pass_forwprop.
16607 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16609         * config/aarch64/aarch64-simd.md
16610         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16612 2013-05-23  Richard Biener  <rguenther@suse.de>
16614         PR middle-end/57381
16615         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16616         OEP_CONSTANT_ADDRESS_OF retained.
16618 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16620         PR middle-end/57344
16621         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16622         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16624 2013-05-23  Richard Biener  <rguenther@suse.de>
16626         PR rtl-optimization/57341
16627         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16628         instead of true_dependence.
16630 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16632         * bb-reorder.c (branch_threshold): Make const.
16633         (exec_threshold): Ditto.
16635 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16636             Pat Haugen  <pthaugen@us.ibm.com>
16637             Peter Bergner  <bergner@vnet.ibm.com>
16639         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16640         documentation for the power8 crypto builtins.
16642         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16644         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16645         macros for defining power8 builtin functions.
16646         (BU_P8V_AV_2): Likewise.
16647         (BU_P8V_AV_P): Likewise.
16648         (BU_P8V_VSX_1): Likewise.
16649         (BU_P8V_OVERLOAD_1): Likewise.
16650         (BU_P8V_OVERLOAD_2): Likewise.
16651         (BU_CRYPTO_1): Likewise.
16652         (BU_CRYPTO_2): Likewise.
16653         (BU_CRYPTO_3): Likewise.
16654         (BU_CRYPTO_OVERLOAD_1): Likewise.
16655         (BU_CRYPTO_OVERLOAD_2): Likewise.
16656         (XSCVSPDP): Fix typo, point to the correct instruction.
16657         (VCIPHER): Add power8 crypto builtins.
16658         (VCIPHERLAST): Likewise.
16659         (VNCIPHER): Likewise.
16660         (VNCIPHERLAST): Likewise.
16661         (VPMSUMB): Likewise.
16662         (VPMSUMH): Likewise.
16663         (VPMSUMW): Likewise.
16664         (VPERMXOR_V2DI): Likewise.
16665         (VPERMXOR_V4SI: Likewise.
16666         (VPERMXOR_V8HI: Likewise.
16667         (VPERMXOR_V16QI: Likewise.
16668         (VSHASIGMAW): Likewise.
16669         (VSHASIGMAD): Likewise.
16670         (VPMSUM): Likewise.
16671         (VPERMXOR): Likewise.
16672         (VSHASIGMA): Likewise.
16674         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16675         __CRYPTO__ if the crypto instructions are available.
16676         (altivec_overloaded_builtins): Add support for overloaded power8
16677         builtins.
16679         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16680         support for power8 crypto builtins.
16681         (builtin_function_type): Likewise.
16682         (altivec_init_builtins): Add support for builtins that take vector
16683         long long (V2DI) arguments.
16685         * config/rs6000/crypto.md: New file, define power8 crypto
16686         instructions.
16688 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16689             Pat Haugen  <pthaugen@us.ibm.com>
16690             Peter Bergner  <bergner@vnet.ibm.com>
16692         * doc/invoke.texi (Option Summary): Add power8 options.
16693         (RS/6000 and PowerPC Options): Likewise.
16695         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16696         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16697         wm, wn, wr documentation.
16699         * config/rs6000/constraints.md (wm): New constraint for VSX
16700         registers if direct move instructions are enabled.
16701         (wn): New constraint for no registers.
16702         (wq): New constraint for quad word even GPR registers.
16703         (wr): New constraint if 64-bit instructions are enabled.
16704         (wv): New constraint if power8 vector instructions are enabled.
16705         (wQ): New constraint for quad word memory locations.
16707         * config/rs6000/predicates.md (const_0_to_15_operand): New
16708         constraint for 0..15 for crypto instructions.
16709         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16710         as GPR and floating point registers.
16711         (int_reg_operand): New predicate to match only GPR registers.
16712         (base_reg_operand): New predicate to match base registers.
16713         (quad_int_reg_operand): New predicate to match even GPR registers
16714         for quad memory operations.
16715         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16716         operations in both GPR and VSX registers.
16717         (quad_memory_operand): New predicate for quad memory operations.
16718         (reg_or_indexed_operand): New predicate for direct move support.
16720         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16721         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16722         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16723         (POWERPC_MASKS): Add power8 options.
16724         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16725         various options.
16727         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16728         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16730         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16731         (-mpower8-fusion): New power8 options.
16732         (-mpower8-fusion-sign): Likewise.
16733         (-mpower8-vector): Likewise.
16734         (-mcrypto): Likewise.
16735         (-mdirect-move): Likewise.
16736         (-mquad-memory): Likewise.
16738         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16739         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16740         registers.
16741         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16742         (rs6000_debug_vector_unit): Add p8_vector.
16743         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16744         definitions.  Also print fusion state.
16745         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16746         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16747         (rs6000_option_override_internal): Add support for power8.
16748         (rs6000_common_init_builtins): Add debugging for skipped builtins
16749         if -mdebug=builtin.
16750         (rs6000_adjust_cost): Add power8 support.
16751         (rs6000_issue_rate): Likewise.
16752         (insn_must_be_first_in_group): Likewise.
16753         (insn_must_be_last_in_group): Likewise.
16754         (force_new_group): Likewise.
16755         (rs6000_register_move_cost): Likewise.
16756         (rs6000_opt_masks): Likewise.
16758         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16759         power8 capable assembler, default to power7 options.
16760         (TARGET_DIRECT_MOVE): Likewise.
16761         (TARGET_CRYPTO): Likewise.
16762         (TARGET_P8_VECTOR): Likewise.
16763         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16764         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16765         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16766         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16767         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16768         (TARGET_XSCVDPSPN): Likewise.
16769         (TARGET_XSCVSPDPN): Likewsie.
16770         (TARGET_SYNC_HI_QI): Likewise.
16771         (TARGET_SYNC_TI): Likewise.
16772         (MASK_CRYPTO): Likewise.
16773         (MASK_DIRECT_MOVE): Likewise.
16774         (MASK_P8_FUSION): Likewise.
16775         (MASK_P8_VECTOR): Likewise.
16776         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16777         temporary used by some of the direct move instructions to get two FP
16778         temporary registers does not force creation of a stack frame.
16779         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16780         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16781         that any VSX registers are tieable, even if they are also an
16782         Altivec vector mode.
16783         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16784         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16785         (RS6000_BTM_CRYPTO): Likewise.
16786         (RS6000_BTM_COMMON): Likewise.
16788         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16789         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16790         (enum rs6000_vector): Add power8 vector support.
16792 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16794         PR target/19599
16795         PR target/57340
16796         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16797         (any_sibcall_could_use_r3): this and handle indirect calls.
16798         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16800 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16802         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16804 2013-05-22  Richard Biener  <rguenther@suse.de>
16806         PR middle-end/57349
16807         * profile.c (branch_prob): Do not split blocks that are
16808         abnormally receiving from ECF_RETURNS_TWICE functions.
16810 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16812         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16813         address mode.  Move pointer mode initialization to the same place.
16815 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16817         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16818         while it has any effect.
16820 2013-05-21  Easwaran Raman  <eraman@google.com>
16822         PR tree-optimization/57322
16823         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16824         UID of the statement added to the BB to be 1.
16826 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16828         PR tree-optimization/57331
16829         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16830         of conversion from pointer type to integral type with integer.
16832 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16834         PR lto/57289
16835         * ipa-prop.c (ipa_read_node_info): Process param_used and
16836         controlled_uses in the same order as when writing.
16838 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16840         PR plugins/56754
16841         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16843 2013-05-21  Richard Biener  <rguenther@suse.de>
16845         PR tree-optimization/57318
16846         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16847         estimate stmts with side-effects as likely eliminated.
16849 2013-05-21  Richard Biener  <rguenther@suse.de>
16851         PR tree-optimization/57330
16852         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16853         preserve the call stmts fntype.
16855 2013-05-21  Richard Biener  <rguenther@suse.de>
16857         PR tree-optimization/57303
16858         * tree-ssa-sink.c (statement_sink_location): Improve killing
16859         stmt detection and properly handle self-assignments.
16861 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16863         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16864         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16865         registers. Set register size out of the PARALLEL loop.
16867 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16869         PR target/56547
16870         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16871         (*fmasf4, *fmasf4_media): New insns.
16873 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16875         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16876         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16877         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16878         (mips_idiv_insns): Update the comments to say that the returned
16879         instruction counts are in units of BASE_INSN_LENGTH.
16880         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16881         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16882         using 2 rather than 4 as the length of indirect MIPS16 and
16883         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16884         length of a NOP.  Don't divide MIPS16 lengths by 2.
16885         (mips16_split_long_branches): Assume a branch is long if the
16886         length is greater than 4 rather than 8.
16887         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16888         rather than multiplying them by 2.  Multiply instruction counts
16889         by BASE_INSN_LENGTH rather than 4.
16890         (*jump_mips16, tls_get_tp_mips16_<mode>)
16891         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16893 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16895         * config/mips/mips.md (extended_mips16): Remove branch case.
16896         (length): Remove duplicated extended_mips16 test.
16898 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16900         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16902 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16904         * recog.h (Recog_data): Rename to...
16905         (recog_data_d): ...this.
16906         (recog_data): Update accordingly.
16907         * recog.c (recog_data): Likewise.
16908         * reload.c (save_recog_data): Likewise.
16909         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16910         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16912 2013-05-17  Julian Brown  <julian@codesourcery.com>
16914         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16915         found in a REG_EQUAL note, invalidate it.
16917 2013-05-17  Easwaran Raman  <eraman@google.com>
16919         * tree-ssa-reassoc.c (find_insert_point): New function.
16920         (insert_stmt_after): Likewise.
16921         (get_def_stmt): Likewise.
16922         (ensure_ops_are_available): Likewise.
16923         (not_dominated_by): Likewise.
16924         (rewrite_expr_tree): Do not move statements beyond what is
16925         necessary. Remove call to swap_ops_for_binary_stmt...
16926         (reassociate_bb): ... and move it here.
16927         (build_and_add_sum): Assign UIDs for new statements.
16928         (linearize_expr): Likewise.
16929         (do_reassoc): Renumber gimple statement UIDs.
16931 2013-05-17  Jan Hubicka  <jh@suse.cz>
16933         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16934         weakrefs.
16935         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16936         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16937         weakrefs.
16938         (output_weakrefs): Update.
16940 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16941             Martin Jambor  <mjambor@suse.cz>
16943         PR middle-end/57276
16944         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16945         value that corresponds to the given aggval is found in values vector.
16947 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16949         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16950         sse, sse2, sse3, ssse3 and sse4a flags to options.
16952 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16954         * gengtype-state.c: (s_expr_writer): New class, to handle
16955         prettifying of output layout of s-expressions.
16956         (state_writer): New class, to write out gtype.state.
16957         (state_written_type_count): Move this variable into member data of
16958         state_writer.
16959         (s_expr_writer::s_expr_writer): New code: constructor for new class
16960         (state_writer::state_writer(): ditto
16961         (s_expr_writer::write_new_line): New function
16962         (s_expr_writer::write_any_indent): ditto
16963         (s_expr_writer::begin_s_expr): ditto
16964         (s_expr_writer::end_s_expr): ditto
16965         (write_state_fileloc): convert to method of state_writer...
16966         (state_writer:: write_state_fileloc): ...and use methods of
16967         s_expr_writer to write indentation into the gtype.state output file
16968         to visually represent the hierarchical structure of the list
16969         structures
16970         (write_state_fields): ditto, renaming to...
16971         (state_writer::write_state_fields)
16972         (write_state_a_string): ditto, renaming to...
16973         (state_writer::write_state_a_string)
16974         (write_state_string_option): ditto, renaming to...
16975         (state_writer::write_state_string_option)
16976         (write_state_type_option): ditto, renaming to...
16977         (state_writer::write_state_type_option)
16978         (write_state_nested_option): ditto, renaming to...
16979         (state_writer::write_state_nested_option)
16980         (write_state_option): ditto, renaming to...
16981         (state_writer::write_state_option)
16982         (write_state_options): ditto, renaming to...
16983         (state_writer::write_state_options)
16984         (write_state_lang_bitmap): ditto, renaming to...
16985         (state_writer::write_state_lang_bitmap)
16986         (write_state_version): ditto, renaming to...
16987         (state_writer::write_state_version)
16988         (write_state_scalar_type): ditto, renaming to...
16989         (state_writer::write_state_scalar_type)
16990         (write_state_string_type): ditto, renaming to...
16991         (state_writer::write_state_string_type)
16992         (write_state_undefined_type): ditto, renaming to...
16993         (state_writer::write_state_undefined_type)
16994         (write_state_struct_union_type): ditto, renaming to...
16995         (state_writer::write_state_struct_union_type)
16996         (write_state_struct_type): ditto, renaming to...
16997         (state_writer::write_state_struct_type)
16998         (write_state_user_struct_type): ditto, renaming to...
16999         (state_writer::write_state_user_struct_type)
17000         (write_state_lang_struct_type): ditto, renaming to...
17001         (state_writer::write_state_lang_struct_type)
17002         (write_state_param_struct_type): ditto, renaming to...
17003         (state_writer::write_state_param_struct_type)
17004         (write_state_pointer_type): ditto, renaming to...
17005         (state_writer::write_state_pointer_type)
17006         (write_state_array_type): ditto, renaming to...
17007         (state_writer::write_state_array_type)
17008         (write_state_gc_used): ditto, renaming to...
17009         (state_writer::write_state_gc_used)
17010         (write_state_common_type_content): ditto, renaming to...
17011         (state_writer::write_state_common_type_content)
17012         (write_state_type): ditto, renaming to...
17013         (state_writer::write_state_type)
17014         (write_state_pair_list): ditto, renaming to...
17015         (state_writer::write_state_pair_list)
17016         (write_state_pair): ditto, renaming to...
17017         (state_writer::write_state_pair)
17018         (write_state_typedefs): ditto, renaming to...
17019         (state_writer::write_state_typedefs)
17020         (write_state_structures): ditto, renaming to...
17021         (state_writer::write_state_structures)
17022         (write_state_param_structs): ditto, renaming to...
17023         (state_writer::write_state_param_structs)
17024         (write_state_variables): ditto, renaming to...
17025         (state_writer::write_state_variables)
17026         (write_state_srcdir): ditto, renaming to...
17027         (state_writer::write_state_srcdir)
17028         (write_state_files_list): ditto, renaming to...
17029         (state_writer::write_state_files_list)
17030         (write_state_languages): ditto, renaming to...
17031         (state_writer::write_state_languages)
17032         (write_state): create a state_writer instance and use it when
17033         writing out the state file
17035 2013-05-17  Mike Stump  <mikestump@comcast.net>
17037         PR rtl-optimization/57304
17038         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
17039         accessing DF_REF_REAL_LOC.
17041 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17043         PR rtl-optimization/57281
17044         PR rtl-optimization/57300
17045         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
17046         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
17047         what the other splitter did if the registers are dead.
17049 2013-05-17  Richard Biener  <rguenther@suse.de>
17051         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
17052         MEM_REF offsets.
17054 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17056         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
17057         linking.
17059 2013-05-17  Marek Polacek  <polacek@redhat.com>
17061         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
17062         length when doing non-zero store of storing '\0' to '\0'.
17064 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17066         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
17067         vect_external_def oprnd1 with loop_vinfo, try to emit
17068         optional cast, negation and and stmts on the loop preheader
17069         edge instead of into the pattern def seq.
17071         PR tree-optimization/57051
17072         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
17073         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
17075 2013-05-16  Nick Clifton  <nickc@redhat.com>
17077         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
17078         (rl78_is_naked_func): New function.
17079         (rl78_expand_prologue): Skip prologue generation for naked functions.
17080         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17081         * doc/extend.texi (naked): Add RL78 to the list of processors
17082         that supports this attribute.
17084 2013-05-16  Jeff Law  <law@redhat.com>
17086         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17088 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17090         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17091         cache parameters using detect_caches_amd also for CYRIX,
17092         NSC and TM2 signatures.
17094 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17095             Dzianis Kahanovich  <mahatma@eu.by>
17097         PR target/45359
17098         PR target/46396
17099         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17100         VIA/Centaur processors and determine their cache parameters
17101         using detect_caches_amd.
17103 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17105         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17106         (rtl_verify_edges): New function.
17107         (rtl_verify_bb_insns): Ditto.
17108         (rtl_verify_bb_pointers): Ditto.
17109         (rtl_verify_bb_insn_chain): Ditto.
17110         (rtl_verify_fallthru): Ditto.
17111         (rtl_verify_bb_layout): Ditto.
17112         (rtl_verify_flow_info_1): Outline checks into new functions.
17113         (rtl_verify_flow_info): Ditto.
17115 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17117         * cfghooks.c (copy_bbs): Add update_dominance argument.
17118         * cfghooks.h (copy_bbs): Update prototype.
17119         * tree-cfg.c (gimple_duplicate_sese_region):
17120         Add update_dominance argument.
17121         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17122         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17123         gimple_duplicate_sese_region call.
17124         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17125         Update copy_bbs call.
17126         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17127         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17129 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17131         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17132         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17133         vect_recog_rotate_pattern.
17134         (vect_recog_rotate_pattern): New function.
17136 2013-05-16  Jason Merrill  <jason@redhat.com>
17138         * Makefile.in (LLINKER): New variable.
17139         (mostlyclean): Remove link mutex.
17140         * configure.ac: Handle --enable-link-mutex.
17141         * lock-and-run.sh: New script.
17143 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17145         PR target/19599
17146         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17147         for NULL decl.
17149 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17151         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17153 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17155         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17156         * config/arm/arm.c (next_consecutive_mem): New function.
17157         (gen_movmem_ldrd_strd): Likewise.
17158         * config/arm/arm.md (movmemqi): Update condition and code.
17159         (unaligned_loaddi, unaligned_storedi): New patterns.
17161 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17163         * config.gcc: Obsolete *-*-solaris2.9*.
17164         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17166 2013-05-16  Richard Biener  <rguenther@suse.de>
17168         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17169         earlier, after GRAPHITE transforms and IV canonicalization.
17171 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17173         * omp-low.c (extract_omp_for_data): For collapsed loops,
17174         if at least one of the loops is known at compile time to
17175         iterate zero times, set count to 0.
17176         (expand_omp_regimplify_p): New function.
17177         (expand_omp_for_generic): For collapsed loops, if at least
17178         one of the loops isn't known to iterate at least once,
17179         add runtime check with setting count to 0.
17180         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17181         For unsigned types if it isn't known at compile time that
17182         the loop will iterate at least once, add runtime check to bypass
17183         the whole loop if initial condition isn't true.
17185 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17187         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17189 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17191         PR middle-end/57286
17192         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17193         transformations to avoid an infinite loop.
17195 2013-05-16  Marek Polacek  <polacek@redhat.com>
17197         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17199 2013-05-15  Leif Ekblad  <leif@rdos.net>
17201         * config/i386/i386.c (ix86_decompose_address): Use
17202         DEFAULT_TLS_SEG_REG to access TLS segment register.
17203         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17204         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17205         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17207 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17209         PR target/57260
17210         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17211         sibling calls to functions that would normally be lazily bound,
17212         unless $gp is call-clobbered.
17214 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17216         * config/i386/i386.c (ix86_option_override_internal): Update
17217         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17218         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17219         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17220         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17221         of TARGET_3DNOW.
17222         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17224 2013-05-15  Andreas Schwab  <schwab@suse.de>
17226         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17227         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17228         third operand.
17230 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17232         * loop-unroll.c (report_unroll_peel): Check decision before
17233         emitting unroll/peel message.
17235 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17237         * function.h (has_bb_partition): New rtl_data flag.
17238         (bb_reorder_complete): Ditto.
17239         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17240         instead of flag_reorder_blocks_and_partition.
17241         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17242         with some enhancements.
17243         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17244         * bb-reorder.c (connect_traces): Check for has_bb_partition
17245         instead of flag_reorder_blocks_and_partition.
17246         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17247         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17248         verify_hot_cold_block_grouping.
17249         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17251 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17253         PR target/19599
17254         * config/arm/predicates.md (call_insn_operand): New predicate.
17255         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17256         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17257         if insn is not a tail call.
17258         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17259         registers.
17260         * config/arm/arm.h (enum reg_class): New caller save register class.
17261         (REG_CLASS_NAMES): Likewise.
17262         (REG_CLASS_CONTENTS): Likewise.
17263         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17264         without decls.
17266 2013-05-15  Richard Biener  <rguenther@suse.de>
17268         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17269         of MSG_OPTIMIZED_LOCATIONS.
17270         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17271         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17272         message.
17273         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17274         of MSG_OPTIMIZED_LOCATIONS.
17275         (execute_vect_slp): Likewise.
17276         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17277         (vect_create_cond_for_alias_checks): Likewise.
17278         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17279         (vect_recog_widen_mult_pattern): Likewise.
17280         (vect_recog_widen_sum_pattern): Likewise.
17281         (vect_recog_over_widening_pattern): Likewise.
17282         (vect_recog_widen_shift_pattern): Likewise.
17283         (vect_recog_vector_vector_shift_pattern): Likewise.
17284         (vect_recog_divmod_pattern): Likewise.
17285         (vect_recog_mixed_size_cond_pattern): Likewise.
17286         (vect_recog_bool_pattern): Likewise.
17287         (vect_pattern_recog_1): Likewise.
17289 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17291         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17292         non-functions to builtin_unreachable.
17293         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17294         correct when new direct edges were discovered.
17296 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17298         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17299         header, print symbol order instead of node uid, print more information
17300         about indirect edge targets.
17301         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17302         uids.
17303         (ipa_make_edge_direct_to_target): Likewise.
17304         (remove_described_reference): Likewise.
17305         (propagate_controlled_uses): Likewise.
17306         (ipa_print_node_params): Also print symbol order.
17307         (ipcp_transform_function): Print symbol order instead of node uids.
17308         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17309         (cgraph_get_create_real_symbol_node): Likewise.
17310         * ipa-cp.c (print_lattice): Likewise.
17311         (print_all_lattices): Likewise.
17312         (determine_versionability): Likewise.
17313         (initialize_node_lattices): Likewise.
17314         (estimate_local_effects): Likewise.
17315         (update_profiling_info): Likewise.
17316         (create_specialized_node): Likewise.
17317         (perhaps_add_new_callers): Likewise.
17318         (decide_about_value): Likewise.
17319         (decide_whether_version_node): Likewise.
17320         (identify_dead_nodes): Likewise.
17321         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17322         (dump_inline_summary): Likewise.
17323         (estimate_node_size_and_time): Likewise.
17324         (inline_analyze_function): Likewise.
17325         * ipa-inline.c (report_inline_failed_reason): Likewise.
17326         (want_early_inline_function_p): Likewise.
17327         (edge_badness): Likewise.
17328         (update_edge_key): Likewise.
17329         (inline_small_functions): Likewise.  Add dumping of order to two other
17330         dumps.
17331         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17332         instead of node uids.
17333         (propagate_pure_const): Likewise.
17334         (propagate_pure_const): Likewise.
17335         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17336         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17337         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17338         of node uids.
17339         * tree-pretty-print.c (dump_function_header): Likewise.
17340         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17341         Print symbol order instead of node uids.
17343 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17345         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17346         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17348 2013-05-15  Richard Biener  <rguenther@suse.de>
17350         PR tree-optimization/57275
17351         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17352         return value for fail to do runtime alias checks for gather loads.
17354 2013-05-15  Jan Hubicka  <jh@suse.cz>
17356         PR lto/57038
17357         PR lto/47375
17358         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17359         weakrefs are not external.
17360         (lto_symtab_merge_decls): Fix thinko when dealing with
17361         non-lto_symtab decls.
17362         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17363         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17364         * varpool.c (dump_varpool_node): Dump more flags.
17366 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17368         * config/i386/i386.c (processor_alias_table): Add instruction
17369         FSGSBASE for AMD bdver3 architecture.
17371 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17373         * tree.c (warn_deprecated_use): Print file:line using locus color.
17374         * diagnostic.c (diagnostic_report_current_module): Print file:line
17375         and file:line:column using locus color.
17377 2013-05-14  Mike Stump  <mikestump@comcast.net>
17379         * gdbinit.in: Add __null.
17381 2013-05-14  Mike Stump  <mikestump@comcast.net>
17383         * recog.h: Rename struct recog_data to Recog_data.
17384         * recog.c: Likewise.
17385         * reload.c (can_reload_into): Likewise.
17386         * config/picochip/picochip.c: Likewise.
17388 2013-05-14  Mike Stump  <mikestump@comcast.net>
17390         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17392 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17394         * resource.h (struct resources): Remove unch_memory member.
17395         (CLEAR_RESOURCE): Don't clear unch_memory.
17396         * resource.c (mark_referenced_resources): Don't set it.
17397         (mark_set_resources): Likewise.
17398         (mark_target_live_regs): Don't clear it.
17399         (init_resource_info): Likewise.
17400         * reorg.c (resource_conflicts_p): Don't compare it.
17401         (redundant_insn): Don't set it.
17403         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17404         Remove prototypes.
17405         * emit-rtl.c (next_label): Remove unused function.
17406         (skip_consecutive_labels, link_cc0_insns): Move to ...
17407         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17408         only place where these functions are used, and make them static.
17410 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17412         * fold-const.c (fold_negate_expr): Handle vectors.
17413         (fold_truth_not_expr): Make it static.
17414         (fold_invert_truthvalue): New static function.
17415         (invert_truthvalue_loc): Handle vectors. Do not call
17416         fold_truth_not_expr directly.
17417         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17418         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17419         (fold_comparison): Handle vector constants.
17420         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17421         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17422         * tree.h (fold_truth_not_expr): Remove declaration.
17424 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17426         * config/aarch64/aarch64-simd.md
17427         (aarch64_vcond_internal<mode>): Rename to...
17428         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17429         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17430         float modes. Clarify all iterator modes.
17431         (vcond<mode><mode>): Use new name for vcond expanders.
17432         (vcond<v_cmp_result><mode>): Likewise.
17433         (vcondu<mode><mode>: Likewise.
17434         * config/aarch64/iterators.md (VDQF_COND): New.
17436 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17438         PR bootstrap/57266
17439         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17440         variable for the shift amount. Check that we shift by non-negative
17441         amounts.
17443 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17445         PR target/42017
17446         * config/arm/arm.h (EPILOGUE_USES): Only return true
17447         for LR_REGNUM after epilogue_completed.
17449 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17451         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17452         is error_mark_node.
17454 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17456         PR target/57261
17457         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17458         and Solaris 11+/x86 with gld.
17459         * configure: Regenerate.
17461 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17463         * expmed.c (expand_shift_1): Canonicalize rotates by
17464         constant bitsize / 2 to bitsize - 1.
17465         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17466         case ROTATERT>: Likewise.
17468         Revert:
17469         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17471         * config/i386/i386.md (rotateinv): New code attr.
17472         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17473         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17474         roll $31, %eax, etc.
17476 2013-05-14  Richard Biener  <rguenther@suse.de>
17478         PR middle-end/57235
17479         * tree-eh.c (sink_clobbers): Give up for successors with
17480         multiple predecessors and no virtual uses.
17482 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17484         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17485         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17487 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17489         PR middle-end/57251
17490         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17491         the case when both op0 and op1 have VOIDmode.
17493 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17495         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17496         in multiply-accumulate mode.
17498 2013-05-13  Guozhi Wei  <carrot@google.com>
17500         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17502 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17504         PR target/56975
17505         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17506         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17507         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17508         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17509         (ix86_expand_prologue): Likewise.
17510         (ix86_expand_split_stack_prologue): Likewise.
17511         (legitimate_pic_address_disp_p): Likewise.
17512         (legitimize_pic_address): Likewise.
17513         (legitimize_tls_address): Likewise.
17514         (legitimize_pe_coff_symbol): Likewise.
17515         (output_pic_addr_const): Likewise.
17516         (construct_plt_address): Likewise.
17517         (ix86_expand_call): Likewise.
17518         (x86_output_mi_thunk): Likewise.
17519         (x86_function_profiler): Likewise.
17521 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17523         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17524         similar switch cases.
17525         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17526         (aarch64_simd_mov_to_<mode>low): Delete.
17527         (aarch64_simd_mov_to_<mode>high): Delete.
17528         (move_lo_quad_<mode>): Add w<-r alternative.
17529         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17530         (aarch64_simd_mov_from_*): Update type attribute.
17531         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17532         statement.
17534 2013-05-13  Jan Hubicka  <jh@suse.cz>
17536         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17537         * config/i386/i386.c (ix86_compute_frame_layout,
17538         ix86_expand_epilogue, emit_i387_cw_initialization,
17539         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17540         ix86_local_alignment): Fix use of size/speed predicates.
17542 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17544         PR tree-optimization/45216
17545         PR tree-optimization/57157
17546         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17547         the (-Y) & (B - 1) variant if OP is |.
17548         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17549         return shifted.  Use (-op1) & (prec - 1) as other_amount
17550         instead of prec - op1.
17552 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17554         PR middle-end/42371
17555         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17556         (ipa_constant_data): New type.
17557         (ipa_jump_func): Use ipa_constant_data to hold information about
17558         constant jump functions.
17559         (ipa_get_jf_constant): Adjust to jump function type changes.
17560         (ipa_get_jf_constant_rdesc): New function.
17561         (ipa_param_descriptor): New field controlled_uses.
17562         (ipa_get_controlled_uses): New function.
17563         (ipa_set_controlled_uses): Likewise.
17564         * ipa-ref.h (ipa_find_reference): Declare.
17565         * ipa-prop.c (ipa_cst_ref_desc): New type.
17566         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17567         changes.
17568         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17569         New parameter cs.  Adjust all callers.
17570         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17571         (remove_described_reference): New function.
17572         (jfunc_rdesc_usable): Likewise.
17573         (try_make_edge_direct_simple_call): Decrement controlled use count,
17574         attempt to remove reference if it hits zero.
17575         (combine_controlled_uses_counters): New function.
17576         (propagate_controlled_uses): Likewise.
17577         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17578         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17579         (ipa_print_node_params): Print described use counter.
17580         (ipa_write_jump_function): Adjust to jump function type changes.
17581         (ipa_read_jump_function): New parameter CS, pass it to
17582         ipa_set_jf_constant.  Adjust caller.
17583         (ipa_write_node_info): Stream controlled use count
17584         (ipa_read_node_info): Likewise.
17585         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17586         asserting.
17587         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17588         count.  Remove cloning-added reference if it reaches zero.
17589         * ipa-ref.c (ipa_find_reference): New function.
17591 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17593         * config/i386/i386.c (processor_target_table): Modified default
17594         alignment values for AMD BD and BT architectures.
17596 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17598         * tree-vect-generic.c (uniform_vector_p): Move ...
17599         * tree.c (uniform_vector_p): ... here.
17600         * tree.h (uniform_vector_p): Declare it.
17601         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17602         into a scalar.
17604 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17606         PR tree-optimization/57230
17607         * tree-ssa-strlen.c (handle_char_store): Record length for
17608         array store from STRING_CST.
17610         PR tree-optimization/57230
17611         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17612         check.
17614 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17616         * config/epiphany/epiphany.c (epiphany_init): Check size of
17617         NUM_MODES_FOR_MODE_SWITCHING.
17618         (epiphany_expand_prologue):
17619         Remove CONFIG_REGNUM initial value handling code.
17620         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17621         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17622         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17623         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17624         Don't return 1 for FP_MODE_NONE.
17625         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17626         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17627         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17628         * config/epiphany/epiphany.md (save_config): New pattern.
17630 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17632         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17634 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17636         * config/i386/i386.md (memory): Handle sseishft1.
17637         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17638         (*vec_extractv2di_1): Ditto.
17640 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17642         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17643         saved registers.
17645 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17647         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17648         Add mthumb/march=armv7-a multilib.
17649         Add mthumb/march=armv7-r multilib.
17650         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17652 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17654         * config/v850/t-rtems: Add more multilibs.
17656 2013-05-10  Richard Biener  <rguenther@suse.de>
17658         PR tree-optimization/57214
17659         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17660         not propagate from SSA names that occur in abnormal PHI nodes.
17662 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17664         * stor-layout.c (element_precision): New function.
17665         * machmode.h (element_precision): Declare it.
17666         * tree.c (build_minus_one_cst): New function.
17667         (element_precision): Likewise.
17668         * tree.h (build_minus_one_cst): Declare new function.
17669         (element_precision): Likewise.
17670         * fold-const.c (operand_equal_p): Use element_precision.
17671         (fold_binary_loc): Handle vector types.
17672         * convert.c (convert_to_integer): Use element_precision.
17673         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17674         separately.
17676 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17678         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17679         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17680         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17681         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17682         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17683         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17684         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17685         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17686         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17687         (Uuw8): New constraints.
17688         (Usb4): Move into alphabetical order.
17689         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17690         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17691         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17692         previously unnamed patterns.
17693         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17694         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17695         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17696         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17697         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17698         of specific lengths.
17700 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17702         * config/i386/i386.md (rotateinv): New code attr.
17703         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17704         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17705         roll $31, %eax, etc.
17707         PR tree-optimization/45216
17708         PR tree-optimization/57157
17709         * tree-ssa-forwprop.c (simplify_rotate): New function.
17710         (ssa_forward_propagate_and_combine): Call it.
17712 2013-05-10  Richard Biener  <rguenther@suse.de>
17714         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17715         disable peeling when we version for aliasing.
17716         (vector_alignment_reachable_p): Honor explicit user alignment.
17717         (vect_supportable_dr_alignment): Likewise.
17718         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17719         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17720         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17721         then peeling to arrange for the cost-model check to come first.
17723 2013-05-10  Alan Modra  <amodra@gmail.com>
17725         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17726         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17727         * configure: Regenerate.
17729 2013-05-10  Alan Modra  <amodra@gmail.com>
17731         PR target/55033
17732         * varasm.c (default_elf_select_section): Move !DECL_P check..
17733         (get_named_section): ..to here before calling get_section_name.
17734         Adjust assertion.
17735         (default_section_type_flags): Add DECL_P check.
17736         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17737         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17739 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17741         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17742         When using gen_stack_adjust_str with a register offset, add a
17743         REG_FRAME_RELATED_EXPR note.
17745 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17747         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17748         (*vec_extractv4si_zext_mem): Ditto.
17749         (*vec_extractv2di): Add 0->x and x->x alternatives.
17750         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17751         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17753 2013-05-09  Jason Merrill  <jason@redhat.com>
17755         N3639 C++1y VLA support
17756         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17757         DECL_VALUE_EXPR.
17759         * tree.c (build_constructor_va): New.
17760         * tree.h: Declare it.
17762 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17764         PR lto/57084
17765         * gimple-fold.c (canonicalize_constructor_val): Call
17766         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17768 2013-05-09  Jan Hubicka  <jh@suse.cz>
17769             Richard Biener  <rguenther@suse.de>
17771         PR lto/54095
17772         * symtab.c (symtab_make_decl_local): Do not add private names.
17774 2013-05-09  Jan Hubicka  <jh@suse.cz>
17776         PR lto/54095
17777         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17778         (unlink_from_assembler_name_hash): Likewise.
17779         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17780         symtab_unregister_node, symtab_initialize_asm_name_hash,
17781         change_decl_assembler_name): Update.
17783 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17785         * config/aarch64/aarch64.md: New movtf split.
17786         (*movtf_aarch64): Update.
17787         (aarch64_movdi_tilow): Handle TF modes and rename to
17788         aarch64_movdi_<mode>low.
17789         (aarch64_movdi_tihigh): Handle TF modes and rename to
17790         aarch64_movdi_<mode>high
17791         (aarch64_movtihigh_di): Handle TF modes and rename to
17792         aarch64_mov<mode>high_di
17793         (aarch64_movtilow_di): Handle TF modes and rename to
17794         aarch64_mov<mode>low_di
17795         (aarch64_movtilow_tilow): Remove spurious whitespace.
17796         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17797         splits.
17798         (aarch64_print_operand): Update.
17800 2013-05-09  Alan Modra  <amodra@gmail.com>
17802         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17803         powerpc64le.
17804         * configure: Regenerate.
17806 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17808         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17809         splitter preparation statements.
17810         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17811         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17812         adjust_address_nv.
17814 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17816         * gimple-ssa-strength-reduction.c (count_candidates): Change
17817         return value to int.
17818         (analyze_candidates_and_replace): Change type of length to int.
17820 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17822         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17823         (*vec_extract<mode>): Use VI12_128 mode iterator.
17824         (*vec_extract<mode>_mem): Ditto.
17825         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17826         attribute.
17828 2013-05-08  Diego Novillo  <dnovillo@google.com>
17830         PR bootstrap/54659
17832         Revert:
17833         2012-08-17  Diego Novillo  <dnovillo@google.com>
17835         PR bootstrap/54281
17836         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17837         * config.in: Regenerate.
17838         * configure: Regenerate.
17839         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17841 2013-05-08  Jan Hubicka  <jh@suse.cz>
17843         PR lto/54095
17844         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17845         * cgraph.h (symtab_node_base): Add unique_name.
17846         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17847         input_overwrite_node, input_varpool_node): Stream unique_name.
17848         * cgraphclones.c (cgraph_create_virtual_clone,
17849         cgraph_function_versioning): Set unique_name.
17850         * ipa.c (function_and_variable_visibility): Set unique_name.
17852 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17854         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17855         (alloc_cand_and_find_basis): Restrict conditional candidate
17856         processing to CAND_MULTs.
17858 2013-05-08  Jan Hubicka  <jh@suse.cz>
17860         PR lto/54095
17861         lto-symtab.c (lto_symtab_symbol_p): New function.
17862         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17863         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17864         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17865         Skip static symbols.
17867 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17869         PR tree-optimization/57200
17870         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17871         Only call inform if the preceding warning_at returns true.
17873 2013-05-07  Han Shen  <shenhan@google.com>
17875         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17876         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17877         * common.opt (fstack-protector-strong): New option.
17878         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17879         * doc/invoke.texi (Optimization Options): Document
17880         "-fstack-protector-strong".
17881         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17883 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17885         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17887 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17889         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17890         it is not lhs of the insn.
17891         (setup_reg_equiv): Remove insn having equiv memory which it is not
17892         lhs of the insn.
17893         * lra-constraints.c (process_address): Try to improve generation
17894         code for address base + disp.
17895         (lra_constraints): Make correct the code for checking insn setting
17896         up backward equivalence.  Remove insn only if it is in the init
17897         insn list.
17898         * lra-eliminations.c (update_reg_eliminate): Change return value.
17899         (lra_eliminate): Use the result.
17901 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17903         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17904         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17905         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17906         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17907         register target operands.
17908         (*vec_extractv8hi_sse2): New pattern.
17909         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17910         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17911         (*vec_extract<mode>_mem): New insn and split pattern.
17913 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17915         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17916         (TARGET_ASAN_SHADOW_OFFSET): Define.
17917         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17918         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17920 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17922         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17923         (incr_vec_index): Return -1 if increment not found.
17924         (create_add_on_incoming_edge): Assert if increment not found.
17925         (record_increment): Limit number of increments recorded.
17926         (all_phi_incrs_profitable): Return false if an increment not found.
17927         (replace_profitable_candidates): Don't process increments that were
17928         not recorded.
17929         (analyze_candidates_and_replace): Limit size of incr_vec.
17931 2013-05-07  Richard Biener  <rguenther@suse.de>
17933         * calls.c (special_function_p): setjmp-like functions are leaf.
17934         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17935         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17937 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17939         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17940         (aarch64_simd_mov<mode>): New expander.
17941         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17942         (aarch64_simd_mov_to_<mode>high): Likewise.
17943         (aarch64_simd_mov_from_<mode>low): Likewise.
17944         (aarch64_simd_mov_from_<mode>high): Likewise.
17945         (aarch64_dup_lane<mode>): Update.
17946         (aarch64_dup_lanedi): New instruction pattern.
17947         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17948         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17950 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17952         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17953         (replace_mult_candidate): Remove unnecessary argument; remove
17954         unnecessary parameter from call to introduce_cast_before_cand.
17955         (replace_unconditional_candidate): Remove unnecessary parameter
17956         from call to replace_mult_candidate.
17957         (replace_conditional_candidate): Likewise.
17958         (insert_initializers): Use make_temp_ssa_name.
17959         (introduce_cast_before_cand): Remove unnecessary argument; use
17960         make_temp_ssa_name.
17961         (replace_one_candidate): Remove unnecessary argument; remove
17962         unnecessary parameter from calls to introduce_cast_before_cand.
17963         (replace_profitable_candidates): Remove unnecessary parameters
17964         from calls to replace_one_candidate.
17966 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17968         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17969         phi def as possibly hiding a basis for a CAND_ADD whose operands
17970         have been commuted in the analysis.
17971         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17973 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17975         * config/aarch64/aarch64.md
17976         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17977         shift value between 0-4.
17979 2013-05-07  Richard Biener  <rguenther@suse.de>
17981         * double-int.h (rshift): New overload.
17982         * double-int.c (rshift): New function.
17983         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17984         (create_reference_ops_from_ref): Remove.
17985         (vn_reference_insert): Use shared ops for constructing the
17986         reference and copy it.
17988 2013-05-07  Richard Biener  <rguenther@suse.de>
17990         PR middle-end/57190
17991         * tree-eh.c (sink_clobbers): Properly propagate
17992         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17994 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17996         PR tree-optimization/57149
17997         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17998         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17999         collect_phi_def_edges, execute_late_warn_uninitialized): Use
18000         uninit_undefined_value_p instead of ssa_undefined_value_p.
18002         PR debug/57184
18003         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
18004         for modifier == EXPAND_INITIALIZER.
18006 2013-05-07  Anton Blanchard  <anton@samba.org>
18008         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
18009         for powerpc64 little endian.
18010         * configure: Regenerate.
18012 2013-05-06  Graham Stott  <grahams@btinternet.com>
18014         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
18015         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
18016         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
18017         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
18019 2013-05-06  Graham Stott  <grahams@btinternet.com>
18021         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
18022         codes which allow non-lvalues.
18024 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
18026         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
18027         components are all 1s.
18028         (integer_minus_onep): New function.
18029         * tree.h (integer_minus_onep): Declare it.
18030         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
18031         integer_minus_onep instead of integer_all_onesp.
18033 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18035         PR target/52933
18036         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
18037         variations of these patterns.
18039 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18041         * config/i386/i386.md (isa): Add x64_sse4 member.
18042         (enabled): Handle x64_sse4.
18043         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
18044         instruction for 64bit SSE4_1 targets.  Update insn attributes.
18045         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
18046         instruction for SSE4_1 targets.  Update insn attributes.
18047         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18048         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
18049         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
18050         const_1 selector.
18051         (*vec_extractv4si): Rename from *sse4_1_pextrd.
18052         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
18053         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
18055 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18057         PR target/57108
18058         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
18060 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
18062         * final.c (do_assembler_dialects): Don't handle curly braces and
18063         vertical bar escaped by % as dialect delimiters.
18064         (output_asm_insn): Print curly braces and vertical bar if escaped
18065         by % and ASSEMBLER_DIALECT defined.
18066         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
18067         * doc/tm.texi: Regenerated.
18069 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18071         * config/mips/mips.c: Include tree-pass.h.
18072         (mips_reorg): Split in pre- and post-dbr_schedule parts.
18073         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
18074         (pass_mips_machine_reorg2): New machine specific pass.
18075         (insert_pass_mips_machine_reorg2): New pass plugin definition.
18076         (mips_option_override): Register the new pass.
18077         * rtl.h (cleanup_barriers): Remove prototype.
18078         (dbr_schedule): Likewise.
18079         * jump.c (cleanup_barriers): Make static.
18080         * reorg.c (dbr_schedule): Likewise.
18082 2013-05-06  Richard Biener  <rguenther@suse.de>
18084         PR tree-optimization/57185
18085         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18086         SSA names properly.
18088 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18090         PR target/57106
18091         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18092         between "first_arg" and "insn", not "last" and "insn".
18094 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18096         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18097         (find_candidates_in_block): Re-enable slsr_process_phi.
18098         (create_phi_basis): Fix double counting of candidate adjustment.
18100 2013-05-06  Richard Biener  <rguenther@suse.de>
18102         PR middle-end/57147
18103         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18104         the edge is also fallthru, preserve it and just clear the
18105         abnormal flag.
18106         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18107         also complex, preserve that and just clear the fallthru flag.
18108         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18109         update virtual operands.
18111 2013-05-06  Alan Modra  <amodra@gmail.com>
18113         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18114         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18115         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18116         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18117         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18118         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18120 2013-05-06  Alan Modra  <amodra@gmail.com>
18122         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18123         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18124         (DEFAULT_ASM_ENDIAN): Define.
18125         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18126         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18127         Update -K PIC clause from sysv4.h.
18128         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18129         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18131 2013-05-06  Alan Modra  <amodra@gmail.com>
18133         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18134         twice for little-endian.
18135         (ashrdi3_no_power, ashrdi3): Support little-endian.
18137 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18139         PR target/55303
18140         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18141         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18142         related expanders.
18143         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18144         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18145         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18146         New predicates.
18148 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18149             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18151         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18152         * config/pa/pa.opt: Make mbig-switch a no-op.
18153         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18154         (CASE_VECTOR_MODE): Always return SImode.
18155         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18156         for the !TARGET_BIG_SWITCH case.
18157         * config/pa/pa-linux.h: Likewise.
18158         * config/pa/pa-openbsd.h: Likewise.
18159         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18160         * config/pa/pa.md (short_jump): Remove define_insn.
18161         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18162         (casesi0): Remove define_insn.
18163         (type): Remove btable_branch.
18164         (pa_combine_type): Likewise.
18165         (in_nullified_branch_delay): Likewise.
18166         (in_call_delay): Likewise.
18167         (define_delay): Likewise.
18168         (define_insn_reservation "Z3"): Likewise.
18169         (define_insn_reservation "Z4"): Likewise.
18170         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18171         (pa_adjust_insn_length): Remove adjustment for btable branches.
18172         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18173         and mno-big-switch
18175 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18177         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18178         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18179         Add m->r,x alternatives.
18180         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18181         splitters using SWI48x mode iterator.
18182         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18183         TARGET_64BIT.  Add m->x alternative.
18184         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18185         Add o->x alternative.  Enable for TARGET_SSE.
18186         (sse_storeq): Remove expander.
18187         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18188         with memory input operand.
18189         (*vec_extractv2di_1 splitter): New.
18190         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18191         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18193 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18195         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18196         (INT_LOWPART): Delete.
18197         (extract_MB): Adjust.
18198         (extract_ME): Adjust.
18199         (print_operand): Adjust.
18201 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18203         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18204         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18205         (reg_or_logical_cint_operand, easy_fp_constant,
18206         logical_const_operand): Delete "CONST_DOUBLE" case.
18207         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18208         "HOST_BITS_PER_WIDE_INT == 64" test.
18209         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18210         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18211         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18212         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18213         test.
18214         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18215         CONST_DOUBLE DImode/VOIDmode case.
18216         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18217         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18218         CONST_DOUBLE VOIDmode case.
18219         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18220         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18221         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18222         Delete CONST_DOUBLE case.
18223         (splitters for mov FMOVE64 const_double): Delete
18224         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18225         "HOST_BITS_PER_WIDE_INT >= 64" test.
18226         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18227         case.
18228         (mov DI const_double): Delete.
18230 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18232         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18233         on op shows all bits zero in mode of a lowpart subreg, return zero.
18235 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18237         PR target/57150
18238         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18239         to save TFmode registers and DImode to save TImode registers for
18240         caller save operations.
18241         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18242         mark being partially clobbered since they only use the first
18243         double word.
18245         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18246         and TDmode only use the upper 64-bits of each VSX register.
18248 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18250         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18251         (find_candidates_in_block): Disable slsr_process_phi.
18253 2013-05-03  Guozhi Wei  <carrot@google.com>
18255         * coverage.c (coverage_obj_init): Move the construction of gcov
18256         constructor to ...
18257         (build_init_ctor): ... here.
18259 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18261         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18262         (slsr_cand_d): Redefine def_phi.
18263         (stride_status, phi_adjust_status, count_phis_status): New enums.
18264         (find_phi_def): New.
18265         (find_basis_for_base_expr): New.
18266         (find_basis_for_candidate): Handle hidden bases.
18267         (alloc_cand_and_find_basis): Handle phi candidates.
18268         (slsr_process_phi): New.
18269         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18270         (create_mul_imm_cand): Likewise.
18271         (create_add_ssa_cand): Exclude phi base candidates.
18272         (create_add_imm_cand): Likewise.
18273         (slsr_process_cast): Likewise.
18274         (slsr_process_copy): Likewise.
18275         (find_candidates_in_block): Handle phi candidates.
18276         (dump_candidate): Likewise.
18277         (unconditional_cands): Delete.
18278         (unconditional_cands_with_known_stride_p): Delete.
18279         (phi_dependent_cand_p): New.
18280         (cand_increment): Handle phi-dependent candidates.
18281         (replace_dependent): Delete.
18282         (replace_mult_candidate): New.
18283         (replace_unconditional_candidate): New.
18284         (incr_vec_index): Move to avoid forward reference.
18285         (create_add_on_incoming_edge): New.
18286         (create_phi_basis): New.
18287         (replace_dependents): Delete.
18288         (replace_conditional_candidate): New.
18289         (phi_add_costs): New.
18290         (replace_uncond_cands_and_profitable_phis): New.
18291         (record_increment): Handle phi adjustments.
18292         (record_phi_increments): New.
18293         (record_increments): Handle phi adjustments.
18294         (phi_incr_cost): New.
18295         (lowest_cost_path): Handle phis.
18296         (total_savings): Likewise.
18297         (analyze_increments): Likewise.
18298         (ncd_with_phi): New.
18299         (ncd_of_cand_and_phis): New.
18300         (nearest_common_dominator_for_cands): Handle phi increments.
18301         (all_phi_incrs_profitable): New.
18302         (replace_profitable_candidates): Handle phi-dependent candidates.
18303         (analyze_candidates_and_replace): Likewise.
18305 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18307         PR bootstrap/57154
18308         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18309         do not exceed REG_BR_PROB_BASE.
18311 2013-05-03  Jeff Law  <law@redhat.com>
18313         PR tree-optimization/57144
18314         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18315         operand of the condition will bit into the new type when eliminating
18316         a cast feeding a condition.
18318 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18320         PR rtl-optimization/57130
18321         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18322         of COMPARE as in_code to the recursive call if needed.
18324 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18326         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18327         (enabled): Handle new members.
18328         * config/i386/sse.md (*vec_concatv2si): Merge from
18329         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18330         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18332 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18334         PR tree-optimization/57027
18335         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18336         for fnms opportunity, check we got the prerequisite kind
18337         of tree / gimple before using accessor functions.
18339 2013-05-03  Richard Biener  <rguenther@suse.de>
18341         * double-int.h (lshift): New overload without precision
18342         and arith argument.
18343         (operator *=, operator +=, operator -=): Move ...
18344         * double-int.c (operator *=, operator +=, operator -=): ... here
18345         and implement more efficiently.
18346         (mul_double_with_sign): Remove.
18347         (lshift_double): Adjust to take unsinged shift argument, push
18348         dispatching code to callers.
18349         (mul_double_wide_with_sign): Add early out for callers that
18350         are not interested in high parts or overflow.
18351         (lshift): New function.
18352         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18353         dispatch code here.
18354         (lrotate, rrotate): Use logical shifts.
18355         * expr.c (get_inner_reference): Use lshift.
18356         * fixed-value.c (do_fixed_divide): Likewise.
18357         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18358         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18359         (indirect_refs_may_alias_p): Likewise.
18360         (stmt_kills_ref_p_1): Likewise.
18362 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18364         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18366 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18368         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18369         scalar form of FABD instruction.
18371 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18373         * lra-constraints.c (process_alt_operands): Add checking alt
18374         number to choose the best alternative.
18376 2013-05-02  Richard Biener  <rguenther@suse.de>
18378         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18379         bitmap and its handling.
18380         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18382 2013-05-02  Richard Biener  <rguenther@suse.de>
18384         PR middle-end/57140
18385         * tree-inline.c (copy_loops): Properly handle removed loops.
18386         (copy_cfg_body): Mark destination loops for fixup if source
18387         loops needed fixup.
18389 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18391         PR target/56732
18392         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18393         generating simple_return for naked functions.
18395 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18397         PR middle-end/56988
18398         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18399         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18400         flags match.
18401         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18402         ipa_agg_replacement_value structures.
18403         (known_aggs_to_agg_replacement_list): Likewise.
18404         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18405         (read_agg_replacement_chain): Likewise.
18406         (ipcp_transform_function): Also check that by_ref flags match.
18408 2013-05-02  Richard Biener  <rguenther@suse.de>
18410         * graphds.h (struct graph): Add obstack member.
18411         * graphds.c (new_graph): Initialize obstack and allocate
18412         vertices from it.
18413         (add_edge): Allocate edge from the obstack.
18414         (free_graph): Free the obstack instead of all edges and vertices.
18416 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18418         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18419         divides.
18420         * cfg.c (update_bb_profile_for_threading): Ditto.
18421         * tree-inline.c (copy_bb): Ditto.
18422         (copy_edges_for_bb): Ditto.
18423         (initialize_cfun): Ditto.
18424         (copy_cfg_body): Ditto.
18425         (expand_call_inline): Ditto.
18426         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18427         (estimate_node_size_and_time): Ditto.
18428         (inline_merge_summary): Ditto.
18429         * cgraphclones.c (cgraph_clone_edge): Ditto.
18430         (cgraph_clone_node): Ditto.
18431         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18432         (compute_trg_info): Ditto.
18434 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18436         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18437         S reg when fp attribute set.
18438         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18440 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18442         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18443         New pattern.
18444         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18445         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18446         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18448 2013-05-02  Richard Biener  <rguenther@suse.de>
18450         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18451         (struct instantiate_cache_entry): New type.
18452         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18453         (struct instantiate_cache_type): New type.
18454         (set_instantiated_value, get_instantiated_value): Remove.
18455         (get_instantiated_value_entry): New function.
18456         (instantiate_scev_name): Use the new cache and adjust.
18457         (instantiate_scev_poly): Adjust.
18458         (instantiate_scev_binary): Likewise.
18459         (instantiate_array_ref): Likewise.
18460         (instantiate_scev_convert): Likewise.
18461         (instantiate_scev_not): Likewise.
18462         (instantiate_scev_3): Likewise.
18463         (instantiate_scev_2): Likewise.
18464         (instantiate_scev_r): Likewise.
18465         (instantiate_scev): Likewise.
18466         (resolve_mixers): Likewise.
18468 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18470         PR target/57091
18471         * lra-constraints.c (best_small_class_operands_num): Remove.
18472         (process_alt_operands): Remove small_class_operands_num.  Take
18473         small classes operands into losers and only if the operand is not
18474         matched.  Modify debugging output.
18475         (curr_insn_transform): Remove best_small_class_operands_num.
18476         Print insn name.
18478 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18480         * config/aarch64/aarch64-builtins.c
18481         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18482         * config/aarch64/aarch64-simd-builtins.def
18483         (reduc_splus_): Add new modes.
18484         (reduc_uplus_): New.
18485         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18486         (reduc_uplus_v4sf): Likewise.
18487         (reduc_splus_v4sf): Likewise.
18488         (aarch64_addv<mode>): Likewise.
18489         (reduc_uplus_<mode>): Likewise.
18490         (reduc_splus_<mode>): Likewise.
18491         (aarch64_addvv2di): Likewise.
18492         (reduc_uplus_v2di): Likewise.
18493         (reduc_splus_v2di): Likewise.
18494         (aarch64_addvv2si): Likewise.
18495         (reduc_uplus_v2si): Likewise.
18496         (reduc_splus_v2si): Likewise.
18497         (reduc_<sur>plus_<mode>): New.
18498         (reduc_<sur>plus_v2di): Likewise.
18499         (reduc_<sur>plus_v2si): Likewise.
18500         (reduc_<sur>plus_v4sf): Likewise.
18501         (aarch64_addpv4sf): Likewise.
18502         * config/aarch64/arm_neon.h
18503         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18504         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18505         add UNSPEC_SADDV, UNSPEC_UADDV.
18506         (SUADDV): New.
18507         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18509 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18511         * config/aarch64/arm_neon.h
18512         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18514 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18516         * config/aarch64/aarch64-builtins
18517         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18519 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18521         * config/aarch64/aarch64-simd-builtins.def
18522         (reduc_smax_): New.
18523         (reduc_smin_): Likewise.
18524         (reduc_umax_): Likewise.
18525         (reduc_umin_): Likewise.
18526         (reduc_smax_nan_): Likewise.
18527         (reduc_smin_nan_): Likewise.
18528         (fmax): Remove.
18529         (fmin): Likewise.
18530         (smax): Update for V2SF, V4SF and V2DF modes.
18531         (smin): Likewise.
18532         (smax_nan): New.
18533         (smin_nan): Likewise.
18534         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18535         (<su><maxmin><mode>3): ...This, refactor.
18536         (s<maxmin><mode>3): New.
18537         (<maxmin_uns><mode>3): Likewise.
18538         (reduc_<maxmin_uns>_<mode>): Refactor.
18539         (reduc_<maxmin_uns>_v4sf): Likewise.
18540         (reduc_<maxmin_uns>_v2si): Likewise.
18541         (aarch64_<fmaxmin><mode>: Remove.
18542         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18543         new builtin names.
18544         (vmin<q>_f<32,64>): Likewise.
18545         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18546         (FMAXMIN): New.
18547         (su): Add mappings for smax, smin, umax, umin.
18548         (maxmin): New.
18549         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18550         (FMAXMIN): Rename as...
18551         (FMAXMIN_UNS): ...This.
18552         (maxminv): Remove.
18553         (fmaxminv): Likewise.
18554         (fmaxmin): Likewise.
18555         (maxmin_uns): New.
18556         (maxmin_uns_op): Likewise.
18558 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18560         * config/aarch64/arm_neon.h
18561         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18562         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18563         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18565 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18567         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18568         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18570 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18572         * config/aarch64/aarch64-simd.md
18573         (vcond<mode>_internal): Handle special cases for constant masks.
18574         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18575         (vcondu<mode><mode>): Likewise.
18576         (vcond<v_cmp_result><mode>): New.
18578 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18580         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18581         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18582         * config/aarch64/aarch64-simd-builtins.def
18583         (cmeq): Update to BUILTIN_VALLDI.
18584         (cmgt): Likewise.
18585         (cmge): Likewise.
18586         (cmle): Likewise.
18587         (cmlt): Likewise.
18588         * config/aarch64/arm_neon.h
18589         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18590         to builtins or C as appropriate.
18592 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18594         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18595         (cmgeu): ...This.
18596         (cmhi): Rename to...
18597         (cmgtu): ...This.
18598         * config/aarch64/aarch64-simd.md
18599         (simd_mode): Add SF.
18600         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18601         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18602         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18603         (cstore<mode>_neg): ...This.
18604         * config/aarch64/iterators.md
18605         (VALLF): new.
18606         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18607         (COMPARISONS): New.
18608         (UCOMPARISONS): Likewise.
18609         (optab): Add missing comparisons.
18610         (n_optab): New.
18611         (cmp_1): Likewise.
18612         (cmp_2): Likewise.
18613         (CMP): Likewise.
18614         (cmp): Remove.
18615         (VCMP_S): Likewise.
18616         (VCMP_U): Likewise.
18617         (V_cmp_result): Add DF, SF modes.
18618         (v_cmp_result): Likewise.
18619         (v): Likewise.
18620         (vmtype): Likewise.
18621         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18623 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18625         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18626         define_insn to define_insn_and_split.
18627         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18628         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18629         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18630         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18631         (thumb2_negscc): Likewise.
18633 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18635         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18637 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18639         * config/arm/thumb2.md: Remove trailing whitespaces.
18641 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18643         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18644         Use gen_int_mode rather than GEN_INT.
18646 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18648         * value-prof.c (stream_in_histogram_value): Remove the strayed
18649         debug_gimple_stmt.
18651 2013-04-30  Richard Biener  <rguenther@suse.de>
18653         PR middle-end/57122
18654         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18656 2013-04-30  Richard Biener  <rguenther@suse.de>
18658         PR middle-end/57107
18659         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18661 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18663         PR rtl-optimization/56957
18664         PR rtl-optimization/57105
18665         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18666         variable.  Use just INSN_UID for determining whether an insn
18667         should be only disconnected from the insn stream.
18668         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18670 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18672         PR tree-optimization/57104
18673         * tsan.c (instrument_expr): Don't instrument accesses to
18674         DECL_HARD_REGISTER VAR_DECLs.
18676 2013-04-30  Richard Biener  <rguenther@suse.de>
18678         * function.h (loops_for_fn): New inline function.
18679         (set_loops_for_fn): Likewise.
18680         * cfgloop.h (place_new_loop): Add struct function parameter.
18681         (get_loop): Likewise.
18682         (get_loops): Likewise.
18683         (number_of_loops): Likewise.
18684         (fel_next): Adjust.
18685         (fel_init): Likewise.
18686         * cfg.c (get_loop_copy): Adjust.
18687         * cfgloop.c (flow_loops_dump): Likewise.
18688         (record_loop_exits): Likewise.
18689         (verify_loop_structure): Likewise.
18690         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18691         (estimate_reg_pressure_cost): Likewise.
18692         (mark_loop_exit_edges): Likewise.
18693         * cfgloopmanip.c (place_new_loop): Likewise.
18694         (add_loop): Likewise.
18695         (duplicate_loop): Likewise.
18696         * graph.c (draw_cfg_nodes): Likewise.
18697         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18698         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18699         (extract_affine_chrec): Likewise.
18700         (build_scop_iteration_domain): Likewise.
18701         * graphite.c (graphite_initialize): Likewise.
18702         * ira-build.c (create_loop_tree_nodes): Likewise.
18703         (more_one_region_p): Likewise.
18704         (rebuild_regno_allocno_maps): Likewise.
18705         (mark_loops_for_removal): Likewise.
18706         (mark_all_loops_for_removal): Likewise.
18707         (remove_unnecessary_regions): Likewise.
18708         (ira_build): Likewise.
18709         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18710         * loop-init.c (fix_loop_structure): Likewise.
18711         (gate_rtl_move_loop_invariants): Likewise.
18712         (gate_rtl_unswitch): Likewise.
18713         (gate_rtl_unroll_and_peel_loops): Likewise.
18714         (rtl_doloop): Likewise.
18715         * lto-streamer-in.c (input_cfg): Likewise.
18716         * lto-streamer-out.c (output_cfg): Likewise.
18717         * modulo-sched.c (sms_schedule): Likewise.
18718         * predict.c (tree_estimate_probability): Likewise.
18719         (tree_estimate_probability_driver): Likewise.
18720         (estimate_loops): Likewise.
18721         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18722         (move_sese_region_to_fn): Likewise.
18723         (debug_loop_num): Likewise.
18724         * tree-chrec.c (chrec_evaluate): Likewise.
18725         (hide_evolution_in_other_loops_than_loop): Likewise.
18726         (chrec_component_in_loop_num): Likewise.
18727         (reset_evolution_in_loop): Likewise.
18728         (evolution_function_is_invariant_rec_p): Likewise.
18729         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18730         * tree-inline.c (copy_loops): Likewise.
18731         (copy_cfg_body): Likewise.
18732         (tree_function_versioning): Likewise.
18733         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18734         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18735         Likewise.
18736         (add_to_evolution_1): Likewise.
18737         (scev_const_prop): Likewise.
18738         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18739         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18740         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18741         (tree_ssa_lim_initialize): Likewise.
18742         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18743         (verify_loop_closed_ssa): Likewise.
18744         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18745         (tree_ssa_loop_im): Likewise.
18746         (tree_ssa_loop_unswitch): Likewise.
18747         (tree_vectorize): Likewise.
18748         (check_data_deps): Likewise.
18749         (tree_ssa_loop_ivcanon): Likewise.
18750         (tree_ssa_loop_bounds): Likewise.
18751         (tree_complete_unroll): Likewise.
18752         (tree_complete_unroll_inner): Likewise.
18753         (tree_parallelize_loops): Likewise.
18754         (tree_ssa_loop_prefetch): Likewise.
18755         (tree_ssa_loop_ivopts): Likewise.
18756         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18757         * tree-vectorizer.c (vectorize_loops): Likewise.
18759 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18761         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18762         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18763         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18764         with EABI_LINK_SPEC.
18766 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18768         PR target/44578
18769         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18770         alternative.
18772 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18774         PR target/57097
18775         * lra-constraints.c (process_alt_operands): Discourage a bit more
18776         using memory for pseudos.  Print cost dump for alternatives.
18777         Modify cost values for conflicts with early clobbers.
18778         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18780 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18782         PR target/57098
18783         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18785 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18787         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18788         from/to S register.
18789         (movdi_aarch64): Support LDR/STR from/to D register.
18791 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18793         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18794         or higher by default.
18796 2013-04-29  Richard Biener  <rguenther@suse.de>
18798         PR middle-end/57075
18799         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18800         even if not adding abnormal edges for calls that can make
18801         abnormal gotos.
18803 2013-04-29  Richard Biener  <rguenther@suse.de>
18805         PR middle-end/57103
18806         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18807         TREE_BLOCK.
18808         (move_stmt_r): Remove redundant checking.
18810 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18812         PR bootstrap/57077
18813         * basic-block.h (apply_scale): New function.
18814         (apply_probability): Use apply_scale.
18815         * gimple-streamer-in.c (input_bb): Ditto.
18816         * lto-streamer-in.c (input_cfg): Ditto.
18817         * lto-cgraph.c (merge_profile_summaries): Ditto.
18818         * tree-optimize.c (execute_fixup_cfg): Ditto.
18819         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18820         (copy_edges_for_bb): Ditto.
18821         (copy_cfg_body): Ditto.
18823 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18825         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18826         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18827         (tail_merge_optimize): Handle current_loops == NULL.
18829 2013-04-26  Jeff Law  <law@redhat.com>
18831         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18832         (simplify_cond_using_ranges): Generalize code to simplify
18833         COND_EXPRs where one argument is a constant and the other
18834         is an SSA_NAME created by an integral type conversion.
18836 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18838         * config/arm/arm.md (store_minmaxsi): Use only when
18839         optimize_insn_for_size_p.
18841 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18843         PR target/57108
18844         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18846 2013-04-29  Richard Biener  <rguenther@suse.de>
18848         PR middle-end/57089
18849         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18850         loop tree make sure to schedule a fixup for the child as well.
18851         (expand_omp_for_generic): Properly add loops.
18852         (expand_omp_for_static_nochunk): Likewise.
18853         (expand_omp_for_static_chunk): Likewise.
18854         (expand_omp_for): For the degenerate case fixup loops.
18855         (expand_omp_sections): Fix default bb placement in loops.
18856         (expand_omp_atomic_pipeline): Properly add loops.
18858 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18860         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18862 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18864         * tree-ssa-tail-merge.c: Update header comment.
18866 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18868         * config/aarch64/arm_neon.h
18869         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18870         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18871         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18872         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18873         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18874         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18875         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18876         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18878 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18880         * config/aarch64/aarch64-simd.md
18881         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18882         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18883         fix_trunc, fixuns_trunc.
18884         (ftrunc<VDQF:mode>2): New.
18885         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18886         (fix_trunc_optab): New.
18888 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18890         * config/aarch64/aarch64-builtins.c
18891         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18892         iceilf, lround, iroundf.
18894 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18896         PR target/54349
18897         * config/i386/i386.h (enum ix86_tune_indices)
18898         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18899         New, split from X86_TUNE_INTER_UNIT_MOVES.
18900         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18901         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18902         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18903         (TARGET_INTER_UNIT_MOVES): Remove.
18904         * config/i386/i386.c (initial_ix86_tune_features): Update.
18905         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18906         (ix86_expand_convert_uns_didf_sse): Use
18907         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18908         (ix86_expand_vector_init_one_nonzero): Ditto.
18909         (ix86_expand_vector_init_interleave): Ditto.
18910         (inline_secondary_memory_needed): Return true for moves from SSE class
18911         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18912         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18913         * config/i386/constraints.md (Yi, Ym): Depend on
18914         TARGET_INTER_UNIT_MOVES_TO_VEC.
18915         (Yj, Yn): New constraints.
18916         * config/i386/i386.md (*movdi_internal): Change constraints of
18917         operand 1 from Yi to Yj and from Ym to Yn.
18918         (*movsi_internal): Ditto.
18919         (*movdf_internal): Ditto.
18920         (*movsf_internal): Ditto.
18921         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18922         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18923         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18924         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18925         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18926         * config/i386/sse.md (movdi_to_sse): Ditto.
18927         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18928         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18929         TARGET_INTER_UNIT_MOVES.
18930         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18931         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18932         instead of TARGET_INTER_UNIT_MOVES.
18933         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18934         operand 1 from Yi to Yj and from Ym to Yn.
18936 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18938         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18939         (float_truncate_hi_): Likewise.
18940         (float_extend_lo_): Likewise.
18941         (float_truncate_lo_): Likewise.
18942         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18943         (aarch64_float_extend_lo_v2df): Likewise.
18944         (vec_unpacks_hi_v4sf): Likewise.
18945         (aarch64_float_truncate_lo_v2sf): Likewise.
18946         (aarch64_float_truncate_hi_v4sf): Likewise.
18947         (vec_pack_trunc_v2df): Likewise.
18948         (vec_pack_trunc_df): Likewise.
18950 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18952         * config/aarch64/aarch64-builtins.c
18953         (aarch64_fold_builtin): Fold float conversions.
18954         * config/aarch64/aarch64-simd-builtins.def
18955         (floatv2si, floatv4si, floatv2di): New.
18956         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18957         * config/aarch64/aarch64-simd.md
18958         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18959         * config/aarch64/iterators.md (FLOATUORS): New.
18960         (optab): Add float, floatuns.
18961         (su_optab): Likewise.
18963 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18965         * config/aarch64/aarch64-builtins.c
18966         (aarch64_builtin_vectorized_function): Use new names for
18967         fcvt builtins.
18968         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18969         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18970         (fcvtzu): Split as...
18971         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18972         (fcvtas): Split as...
18973         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18974         (fcvtau): Split as...
18975         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18976         (fcvtps): Split as...
18977         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18978         (fcvtpu): Split as...
18979         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18980         (fcvtms): Split as...
18981         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18982         (fcvtmu): Split as...
18983         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18984         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18985         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18986         (lfrintnusf, lfrintnudf): Likewise.
18987         * config/aarch64/aarch64-simd.md
18988         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18989         define_insn.
18990         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18991         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18992         (fcvt_pattern): Likewise.
18994 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18996         * config/aarch64/aarch64-simd.md
18997         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18998         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
19000 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19002         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
19003         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
19004         (vrnd<a,m,n,p>_f32): Implement using builtins.
19005         (vrnd<i,x><q>_f<32, 64>): New.
19007 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19009         * config/aarch64/aarch64-builtins.c
19010         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
19011         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
19012         (frintz): Rename to...
19013         (btrunc): ...this.
19014         (frintp): Rename to...
19015         (ceil): ...this.
19016         (frintm): Rename to...
19017         (floor): ...this.
19018         (frinti): Rename to...
19019         (nearbyint): ...this.
19020         (frintx): Rename to...
19021         (rint): ...this.
19022         (frinta): Rename to...
19023         (round): ...this.
19024         * config/aarch64/aarch64-simd.md
19025         (aarch64_frint<frint_suffix><mode>): Delete.
19026         (<frint_pattern><mode>2): Convert to insn.
19027         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
19028         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
19029         (frint_pattern): Likewise.
19030         (frint_suffix): Likewise.
19032 2013-04-29  Richard Biener  <rguenther@suse.de>
19034         PR tree-optimization/57081
19035         * loop-init.c: Include tree-flow.h.
19036         (loop_optimizer_finalize): Free number of iteration estimates.
19037         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
19039 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
19041         PR tree-optimization/57083
19042         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
19043         non-singleton shift count range, zero extend low_bound for uns case.
19045         * config/i386/predicates.md (general_vector_operand): New predicate.
19046         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
19047         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
19048         if they aren't nonimmediate operands.  If their original values
19049         satisfy const_vector_equal_evenodd_p, don't shift them.
19050         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
19051         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
19052         if not nonimmediate_operand.
19053         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
19054         instead of register_operand.
19055         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
19057 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
19059         * stor-layout.c (finalize_size_functions): Allocate a structure and
19060         reset cfun before dumping the functions.
19062 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
19064         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
19066         PR target/56866
19067         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
19068         use xop_pmacsdqh if uns_p.
19069         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
19070         the immediate rotate count.
19072 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
19074         * rtl.h (struct rtx_def): Add comment for field jump.
19075         (LRA_SUBREG_P): New macro.
19076         * recog.c (register_operand): Check LRA_SUBREG_P.
19077         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19078         stack frame.
19079         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19080         (lra_final_code_change): Skip subreg change for operators.
19081         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19082         if there are no operand changes.
19083         * lra-constraints.c (curr_insn_set): New.
19084         (match_reload): Set LRA_SUBREG_P.
19085         (emit_spill_move): Ditto.
19086         (check_and_process_move): Use curr_insn_set. Process only single
19087         set insns.  Don't initialize sec_mem_p and change_p.
19088         (simplify_operand_subreg): Use LRA_SUBREG_P.
19089         (reg_in_class_p): New function.
19090         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19091         of #ifdef.  Add code to remove cycling.
19092         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19093         non-null disp.  Reload inner instead of disp when base and index
19094         are null.  Try to put lo_sum into register.
19095         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19096         (check_and_process_move): Move code for move cost check to
19097         simple_move_p.  Remove equiv_substitution.
19098         (simple_move_p): New function.
19099         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19100         curr_insn_set.  Call check_and_process_move only for single set
19101         insns.  Use the new function.  Move call of check_and_process_move
19102         after operand equiv substitution and address process.
19104 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19106         PR go/57045
19107         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19108         with nonlocal goto receivers or returns twice calls, ignore
19109         unininitialized values from abnormal edges to nl goto receiver
19110         or returns twice call.
19112 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19114         PR tree-optimization/57051
19115         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19116         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19117         bitsize.
19119 2013-04-26  Richard Biener  <rguenther@suse.de>
19121         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19122         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19123         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19124         (fixup_loop_arrays_after_move): New function.
19125         (move_sese_region_to_fn): Properly outline the loop tree parts
19126         of the SESE region.
19128 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19130         * config/i386/i386.md (type, unit): Fix long lines.
19132 2013-04-26  Richard Biener  <rguenther@suse.de>
19134         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19135         (lto-streamer-out.o): Likewise.
19136         * cfgloop.c (init_loops_structure): Export, add struct function
19137         argument and adjust.
19138         (flow_loops_find): Adjust.
19139         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19140         (init_loops_structure): Declare.
19141         * lto-streamer-in.c: Include cfgloop.h.
19142         (input_cfg): Input the loop tree.
19143         * lto-streamer-out.c: Include cfgloop.h.
19144         (output_cfg): Output the loop tree.
19145         (output_struct_function_base): Do not drop PROP_loops.
19147 2013-03-26  Richard Biener  <rguenther@suse.de>
19149         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19150         (pass_build_cfg): Provide PROP_loops.
19151         (move_sese_region_to_fn): Remove loops that are outlined into fn
19152         for now.
19153         * tree-inline.c: Include cfgloop.h.
19154         (initialize_cfun): Do not drop PROP_loops.
19155         (copy_loops): New function.
19156         (copy_cfg_body): Copy loop structure.
19157         (tree_function_versioning): Initialize destination loop tree.
19158         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19159         (pass_parallelize_loops): Do IL verification.
19160         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19161         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19162         the CFG make sure we fixup loops as well.
19163         * tree-ssa-tail-merge.c: Include cfgloop.h.
19164         (replace_block_by): When merging loop latches mark loops for fixup.
19165         * lto-streamer-out.c (output_struct_function_base): Drop
19166         PROP_loops for now.
19167         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19168         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19169         * ipa-split.c: Include cfgloop.h.
19170         (split_function): Add the new return block to the loop tree root.
19171         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19172         whether we have removed the forwarder block.
19173         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19174         * cfgloop.h (place_new_loop): Declare.
19175         * cfgloopmanip.c (place_new_loop): Export.
19176         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19177         (tree-switch-conversion.o): Likewise.
19178         (tree-complex.o): Likewise.
19179         (tree-inline.o): Likewise.
19180         (tree-ssa-tailmerge.o): Likewise.
19181         (ipa-split.o): Likewise.
19182         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19183         (tree-ssa-copy.o): Likewise.
19184         * tree-switch-conversion.c: Include cfgloop.h
19185         (process_switch): If we emit a bit-test cascade, schedule loops
19186         for fixup.
19187         * tree-complex.c: Include cfgloop.h.
19188         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19189         * asan.c: Include cfgloop.h.
19190         (create_cond_insert_point): Properly add new basic-blocks to
19191         loops, schedule loop fixup.
19192         * cfgloop.c (verify_loop_structure): Check that looks are not
19193         marked for fixup.
19194         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19195         to loops.
19196         (expand_omp_for_generic): Likewise.
19197         (expand_omp_sections): Likewise.
19198         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19199         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19200         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19201         is initialized, not when loops are present.
19202         * tree-parloops.c (parallelize_loops): Remove checking here.
19203         * passes.c (init_optimization_passes): Schedule a copy-propagation
19204         pass before complete unrolling of inner loops.
19206 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19208         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19209         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19210         (colorize_init): Add argument to _WIN32 version.
19211         * toplev.c: Include diagnostic-color.h.
19212         (process_options): Default to -fdiagnostics-color=auto if
19213         GCC_COLORS env var is in the environment.
19214         * common.opt (fdiagnostics-color=): Add Var and Init.
19215         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19216         env var is in the environment, the default is auto rather than never.
19218         * diagnostic.h (file_name_as_prefix): Add context argument.
19219         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19220         the string as locus.
19221         * langhooks.c (lhd_print_error_function): Adjust caller.
19223 2013-04-25  Lawrence Crowl  <crowl@google.com>
19225         * var-tracking.c (shared_hash_def::htab):
19226         Change type to hash_table.  Update dependent calls and types.
19228 2013-04-25  Lawrence Crowl  <crowl@google.com>
19230         * Makefile.in: Update as needed below.
19232         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19233         Move declaration to after the type's method definitons.
19235         * attribs.c (htab_t scoped_attributes::attribute_hash):
19236         Change type to hash_table.  Update dependent calls and types.
19238         * bitmap.c (htab_t bitmap_desc_hash):
19239         Change type to hash_table.  Update dependent calls and types.
19241         * cselib.c (htab_t cselib_hash_table):
19242         Change type to hash_table.  Update dependent calls and types.
19244         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19245         (hash_string_slot_node): Move implementation into lto-streamer.h
19246         struct string_slot_hasher.
19247         (eq_string_slot_node): Likewise.
19249         * data-streamer-out.c: Update output_block::string_hash_table
19250         dependent calls and types.
19252         * dwarf2cfi.c (htab_t trace_index):
19253         Change type to hash_table.  Update dependent calls and types.
19255         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19256         Change type to hash_table.  Update dependent calls and types.
19257         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19258         (htab_t optimize_external_refs::map): Likewise.
19259         (htab_t output_comp_unit::extern_map): Likewise.
19260         (htab_t output_comdat_type_unit::extern_map): Likewise.
19261         (htab_t output_macinfo::macinfo_htab): Likewise.
19262         (htab_t optimize_location_lists::htab): Likewise.
19263         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19265         * except.c (htab_t ehspec_hash_type):
19266         Change type to hash_table.  Update dependent calls and types.
19267         (assign_filter_values::ttypes): Likewise.
19268         (assign_filter_values::ehspec): Likewise.
19269         (sjlj_assign_call_site_values::ar_hash): Likewise.
19270         (convert_to_eh_region_ranges::ar_hash): Likewise.
19272         * gcse.c (htab_t pre_ldst_table):
19273         Change type to hash_table.  Update dependent calls and types.
19275         * ggc-common.c (htab_t saving_htab):
19276         Change type to hash_table.  Update dependent calls and types.
19277         (htab_t loc_hash): Likewise.
19278         (htab_t ptr_hash): Likewise.
19279         (call_count): Rename ggc_call_count.
19280         (call_alloc): Rename ggc_call_alloc.
19281         (loc_descriptor): Rename make_loc_descriptor.
19282         (add_statistics): Rename ggc_add_statistics.
19284         * ggc-common.c (saving_htab):
19285         Change type to hash_table.  Update dependent calls and types.
19287         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19288         (push_gimplify_context): Likewise.
19289         (pop_gimplify_context): Likewise.
19290         (struct gimple_temp_hash_elt): Added.
19291         (struct gimplify_hasher): Likewise.
19292         (struct gimplify_ctx.temp_htab):
19293         Change type to hash_table.  Update dependent calls and types.
19295         * gimple-fold.c: Include gimplify-ctx.h.
19297         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19298         Change type to hash_table.  Update dependent calls and types.
19299         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19300         avoid potential global name collision.
19302         * gimplify.c: Include gimplify-ctx.h.
19303         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19304         (htab_t gimplify_ctx::temp_htab):
19305         Update dependent calls and types for new type hash_table.
19306         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19307         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19309         * gimplify-ctx.h: New.
19310         (struct gimple_temp_hash_elt): Move from gimplify.c.
19311         (class gimplify_hasher): New.
19312         (struct gimplify_ctx): Move from gimple.h.
19313         (htab_t gimplify_ctx::temp_htab):
19314         Change type to hash_table.  Update dependent calls and types.
19316         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19317         (htab_t ivs_params::newivs_index):
19318         Change type to hash_table.  Update dependent calls and types.
19319         (htab_t ivs_params::params_index): Likewise.
19320         (htab_t print_generated_program::params_index): Likewise.
19321         (htab_t gloog::newivs_index): Likewise.
19322         (htab_t gloog::params_index): Likewise.
19324         * graphite.c: Include graphite-htab.h.
19325         4htab_t graphite_transform_loops::bb_pbb_mapping):
19326         Change type to hash_table.  Update dependent calls and types.
19328         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19329         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19330         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19332         * graphite-dependences.c: Include graphite-htab.h.
19333         (loop_is_parallel_p): Change hash table type of parameter.
19335         * graphite-htab.h: New.
19336         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19337         (extern find_pbb_via_hash): Move from graphite-poly.h.
19338         (extern loop_is_parallel_p): Move from graphite-poly.h.
19339         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19341         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19342         (extern loop_is_parallel_p): Move to graphite-htab.h.
19343         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19345         * haifa-sched.c (htab_t delay_htab):
19346         Change type to hash_table.  Update dependent calls and types.
19347         (htab_t delay_htab_i2): Likewise.
19349         * ira-color.c (htab_t allocno_hard_regs_htab):
19350         Change type to hash_table.  Update dependent calls and types.
19352         * ira-costs.c (htab_t cost_classes_htab):
19353         Change type to hash_table.  Update dependent calls and types.
19355         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19356         Change type to hash_table.  Update dependent calls and types.
19358         * loop-iv.c (htab_t bivs):
19359         Change type to hash_table.  Update dependent calls and types.
19361         * loop-unroll.c (htab_t opt_info::insns_to_split):
19362         Change type to hash_table.  Update dependent calls and types.
19363         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19365         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19366         (struct string_slot_hasher): New.
19367         (htab_t output_block::string_hash_table):
19368         Change type to hash_table.  Update dependent calls and types.
19370         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19371         (htab_t file_name_hash_table):
19372         Change type to hash_table.  Update dependent calls and types.
19374         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19375         calls and types.
19377         * lto-streamer.c (htab_t tree_htab):
19378         Change type to hash_table.  Update dependent calls and types.
19380         * omp-low.c: Include gimplify-ctx.h.
19382         * passes.c (htab_t name_to_pass_map):
19383         Change type to hash_table.  Update dependent calls and types.
19384         (pass_traverse): Rename to passes_pass_traverse.
19386         * plugin.c (htab_t event_tab):
19387         Change type to hash_table.  Update dependent calls and types.
19389         * postreload-gcse.c (htab_t expr_table):
19390         Change type to hash_table.  Update dependent calls and types.
19391         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19393         * sese.c (debug_rename_map_1): Make extern.
19394         (htab_t copy_bb_and_scalar_dependences::rename_map):
19395         Change type to hash_table.  Update dependent calls and types.
19397         * sese.h (extern debug_rename_map): Move to .c file.
19399         * store-motion.c (htab_t store_motion_mems_table):
19400         Change type to hash_table.  Update dependent calls and types.
19402         * trans-mem.c (htab_t tm_new_mem_hash):
19403         Change type to hash_table.  Update dependent calls and types.
19405         * tree-browser.c (htab_t TB_up_ht):
19406         Change type to hash_table.  Update dependent calls and types.
19408         * tree-cfg.c (htab_t discriminator_per_locus):
19409         Change type to hash_table.  Update dependent calls and types.
19411         * tree-complex.c: Include tree-hasher.h
19412         (htab_t complex_variable_components):
19413         Change type to hash_table.  Update dependent calls and types.
19415         * tree-eh.c (htab_t finally_tree):
19416         Change type to hash_table.  Update dependent calls and types.
19418         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19419         struct int_tree_hasher.
19420         (extern int_tree_map_eq): Likewise.
19421         (uid_decl_map_hash): Removed.
19422         (extern decl_tree_map_eq): Likewise.
19424         * tree-hasher.h: New.
19425         (struct int_tree_hasher): New.
19426         (typedef int_tree_htab_type): New.
19428         * tree-inline.c: Include gimplify-ctx.h.
19430         * tree-mudflap.c: Include gimplify-ctx.h.
19432         * tree-parloops.c: Include tree-hasher.h.
19433         (htab_t eliminate_local_variables_stmt::decl_address):
19434         Change type to hash_table.  Update dependent calls and types.
19435         (htab_t separate_decls_in_region::decl_copies): Likewise.
19437         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19438         Change type to hash_table.  Update dependent calls and types.
19440         * tree-sra.c (candidates):
19441         Change type to hash_table.  Update dependent calls and types.
19443         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19444         in tree-flow.h.
19445         (int_tree_map_hash): Likewise.
19447         * tree-ssa-dom.c (htab_t avail_exprs):
19448         Change type to hash_table.  Update dependent calls and types.
19450         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19451         Change type to hash_table.  Update dependent calls and types.
19453         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19454         Change type to hash_table.  Update dependent calls and types.
19456         * tree-ssa-phiopt.c (seen_ssa_names):
19457         Change type to hash_table.  Update dependent calls and types.
19459         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19460         Change type to hash_table.  Update dependent calls and types.
19462         * tree-ssa-uncprop.c (equiv):
19463         Change type to hash_table.  Update dependent calls and types.
19465 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19467         PR rtl-optimization/57003
19468         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19469         call note_stores with kill_clobbered_value callback again after
19470         killing regs_invalidated_by_call.
19472 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19474         * config/aarch64/aarch64-simd.md
19475         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19476         (aarch64_simd_bsl<mode>): Likewise.
19477         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19479 2013-04-25  Marek Polacek  <polacek@redhat.com>
19481         PR tree-optimization/57066
19482         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19484 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19486         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19488 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19490         * config/aarch64/aarch64-builtins.c
19491         (aarch64_fold_builtin): New.
19492         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19493         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19494         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19495         * config/aarch64/arm_neon.h
19496         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19498 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19499             Tejas Belagod  <tejas.belagod@arm.com>
19501         * config/aarch64/aarch64-builtins.c
19502         (aarch64_gimple_fold_builtin): New.
19503         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19504         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19505         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19506         (addvv4sf): Update.
19507         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19509 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19511         * config/aarch64/aarch64.md
19512         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19514 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19516         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19517         (*ngcsi_uxtw): New pattern.
19519 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19520             Julian Brown  <julian@codesourcery.com>
19522         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19523         (TB_DREG): Add T_V4HF.
19524         (v4hf_UP): New macro.
19525         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19526         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19527         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19528         built-ins.
19529         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19530         (arm_vector_mode_supported_p): Handle V4HF.
19531         (arm_mangle_map): Handle V4HFmode.
19532         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19533         * config/arm/arm_neon_builtins.def: Add entries for
19534         vcvtv4hfv4sf, vcvtv4sfv4hf.
19535         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19536         (neon_vcvtv4hfv4sf): Likewise.
19537         * config/arm/neon-gen.ml: Handle half-precision floating point
19538         features.
19539         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19540         * config/arm/arm_neon.h: Regenerate.
19541         * config/arm/neon.ml (type elts): Add F16.
19542         (type vectype): Add T_float16x4, T_floatHF.
19543         (type vecmode): Add V4HF.
19544         (type features): Add Requires_FP_bit feature.
19545         (elt_width): Handle F16.
19546         (elt_class): Likewise.
19547         (elt_of_class_width): Likewise.
19548         (mode_of_elt): Refactor.
19549         (type_for_elt): Handle F16, fix error messages.
19550         (vectype_size): Handle T_float16x4.
19551         (vcvt_sh): New function.
19552         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19553         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19554         (string_of_mode): Handle V4HF.
19555         * doc/arm-neon-intrinsics.texi: Regenerate.
19557 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19559         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19560         format specifier in 'X' case.
19562 2013-04-25  Alan Modra  <amodra@gmail.com>
19564         PR target/57052
19565         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19566         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19567         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19568         Repeat for many other rotate/shift and mask patterns using subregs.
19569         Name lshiftrt insns.
19570         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19571         on WORDS_BIG_ENDIAN.
19573 2013-04-25  Alan Modra  <amodra@gmail.com>
19575         * config.gcc: Support little-endian powerpc-linux targets.
19576         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19577         (LINK_OS_LINUX_SPEC): Define.
19578         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19579         Preserve MASK_LITTLE_ENDIAN.
19580         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19581         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19582         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19583         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19584         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19585         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19586         Correct fp word order for little-endian.  Don't shift toc entries
19587         smaller than a word for little-endian.
19588         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19589         (bswapdi2 splits): Correct low-part subreg for little-endian.
19590         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19591         low/high where such is correct only for be.
19592         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19593         little-endian for -mcall-aixdesc.
19595 2013-04-25  Alan Modra  <amodra@gmail.com>
19597         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19598         replace_equiv_address_nv.
19600 2013-04-25  Alan Modra  <amodra@gmail.com>
19602         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19604 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19606         Revert:
19607         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19608         * rtl.h (struct rtx_def): ...
19610 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19612         PR rtl-optimizations/57046
19613         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19614         for multi-reg splits.
19616 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19618         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19620 2013-04-24  Sterling Augustine  <saugustine@google.com>
19622         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19623         (comp_dir_string, debug_str_dwo_section): New.
19624         (DEBUG_STR_DWO_SECTION): Rename to ...
19625         (DEBUG_DWO_STR_SECTION): ... this.
19626         (DEBUG_NORM_STR_SECTION): Delete.
19627         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19628         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19629         (find_AT_string): Move most logic to ...
19630         (find_AT_string_in_table): ... here.  New.
19631         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19632         add_skeleton_AT_string.  Delete logic.
19633         (output_skeleton_debug_sections): Remove call to
19634         add_top_level_skeleton_die_attrs.
19635         (add_comp_dir_attribute): Move logic to comp_dir_string.
19636         (dwarf2out_init): Initialize debug_str_dwo_section.
19637         (output_indirect_string): Call find_string_form.
19638         (output_indirect_strings): Rewrite.
19639         (prune_unused_types): Empty skeleton_debug_str_hash.
19640         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19641         (dwarf2out_finish):  Call output_indirect_strings.
19643 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19645         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19647 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19649         * rtl.h (struct rtx_def): Add comment for field jump.
19650         (LRA_SUBREG_P): New macro.
19651         * recog.c (register_operand): Check LRA_SUBREG_P.
19652         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19653         stack frame.
19654         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19655         (lra_final_code_change): Skip subreg change for operators.
19656         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19657         if there are no operand changes.
19658         * lra-constraints.c (curr_insn_set): New.
19659         (match_reload): Set LRA_SUBREG_P.
19660         (emit_spill_move): Ditto.
19661         (check_and_process_move): Use curr_insn_set. Process only single
19662         set insns.  Don't initialize sec_mem_p and change_p.
19663         (simplify_operand_subreg): Use LRA_SUBREG_P.
19664         (reg_in_class_p): New function.
19665         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19666         of #ifdef.  Add code to remove cycling.
19667         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19668         non-null disp.  Reload inner instead of disp when base and index
19669         are null.  Try to put lo_sum into register.
19670         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19671         (check_and_process_move): Move code for move cost check to
19672         simple_move_p.  Remove equiv_substitution.
19673         (simple_move_p): New function.
19674         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19675         curr_insn_set.  Call check_and_process_move only for single set
19676         insns.  Use the new function.  Move call of check_and_process_move
19677         after operand equiv substitution and address process.
19679 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19681         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19682         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19683         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19685 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19687         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19689 2013-04-24  Marek Polacek  <polacek@redhat.com>
19691         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19692         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19693         (select_loops_exit_conditions): Likewise.
19694         (number_of_iterations_for_all_loops): Likewise.
19695         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19696         (scev_analysis): Likewise.
19698 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19699             Chao-ying Fu  <fu@mips.com>
19701         * config/mips/micromips.md (jraddiusp): New pattern.
19702         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19703         instruction if possible.
19705 2013-04-24  Alan Modra  <amodra@gmail.com>
19707         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19709 2013-04-24  Julian Brown  <julian@codesourcery.com>
19710             Chung-Lin Tang  <cltang@codesourcery.com>
19712         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19713         dependency behavior in enumeration type DIE generation. Add TODO note
19714         to comments about future DW_FORM_sdata/udata re-work of related code.
19716 2013-04-23  Lawrence Crowl  <crowl@google.com>
19718         * Makefile.in: Update as needed below.
19720         * hash-table.h (class hash_table):
19721         Correct many methods with parameter types compare_type to the correct
19722         value_type.  (Correct code was unlikely to notice the change.)
19723         (hash_table::elements_with_deleted) New.
19724         (class hashtable::iterator): New.
19725         (hashtable::begin()): New.
19726         (hashtable::end()): New.
19727         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19729         * statistics.c (statistics_hashes):
19730         Change type to hash_table.  Update dependent calls and types.
19732         * tree-into-ssa.c (var_infos):
19733         Change type to hash_table.  Update dependent calls and types.
19735         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19736         Change type to hash_table.  Update dependent calls and types.
19738         * tree-ssa-loop-im.c (struct mem_ref.refs):
19739         Change type to hash_table.  Update dependent calls and types.
19741         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19742         Change type to hash_table.  Update dependent calls and types.
19744         * tree-ssa-sccvn.c (vn_tables_s::nary):
19745         Change type to hash_table.  Update dependent calls and types.
19746         (vn_tables_s::phis): Likewise.
19747         (vn_tables_s::references): Likewise.
19749         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19750         (vn_reference_eq): Update parameter and return types.
19752         * tree-ssa-structalias.c (pointer_equiv_class_table):
19753         Change type to hash_table.  Update dependent calls and types.
19754         (location_equiv_class_table): Likewise.
19756         * tree-vect-data-refs.c: Consequential changes for making
19757         peeling a hash_table.
19759         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19760         (destroy_loop_vec_info): Dependent hash_table update.
19762         * tree-vectorizer.h (peeling_htab):
19763         Change type to hash_table.  Update dependent calls and types.
19765 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19767         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19768         to check the register content is equal or not.
19769         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19770         to assign register content record.
19771         * lra-eliminations.c (update_reg_eliminate): Use
19772         lra_update_reg_val_offset to update register content offset.
19773         * lra-int.h (struct lra_reg): Add offset member.
19774         (lra_reg_val_equal_p): New static inline function.
19775         (lra_update_reg_val_offset): New static inline function.
19776         (lra_assign_reg_val): New static inline function.
19777         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19778         to assign register content record.
19779         (initialize_lra_reg_info_element): Initial offset to zero.
19781 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19783         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19784         operands.  Record compression.
19786 2013-04-23  Xinliang David Li  <davidxl@google.com>
19788         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19790 2013-04-23  Richard Biener  <rguenther@suse.de>
19792         PR middle-end/57036
19793         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19794         parameter, only add abnormal goto edges from the copied body
19795         if the call could perform abnormal gotos.
19796         (copy_cfg_body): Adjust.
19798 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19800         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19802 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19804         * coretypes.h (gimple_stmt_iterator): Add struct to make
19805         compatible with C.
19807 2013-04-23  Richard Biener  <rguenther@suse.de>
19809         PR tree-optimization/57026
19810         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19811         from SSA names occuring in abnormal PHI nodes.
19813 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19815         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19817 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19819         * lto/lto.c (print_lto_report_1): Declare early.
19820         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19822 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19824         * common.opt (-flto-report-wpa): Add.
19825         * doc/invoke.texi (-flto-report-wpa): Add.
19826         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19827         (lto_main): dito.
19829 2013-04-22  Xinliang David Li  <davidxl@google.com>
19831         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19832         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19833         * Makefile.in: New dependency
19835             David Daney <ddaney.cavm@gmail.com>
19837         * configure.ac (gcc_cv_as_micromips_support):  Use the
19838         --fatal-warnings option.
19839         * configure: Regenerate.
19841 2013-04-22  Marek Polacek  <polacek@redhat.com>
19843         PR sanitizer/56990
19844         * tsan.c (instrument_expr): Don't instrument expression
19845         in case its size is zero.
19847 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19849         PR target/57032
19850         Revert:
19851         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19853         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19855 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19857         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19858         (gimple_stmt_iterator): New typedef.
19859         * gimple.h (gimple_stmt_iterator): Rename to...
19860         (gimple_stmt_iterator_d): ... This.
19861         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19862         trees be valid for GIMPLE and GENERIC.
19863         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19864         * gimple-fold.c (gimple_fold_call): Call target hook
19865         gimple_fold_builtin.
19866         * hooks.c (hook_bool_gsiptr_false): New.
19867         * hooks.h (hook_bool_gsiptr_false): New.
19868         * target.def (fold_stmt): New.
19869         * doc/tm.texi: Regenerate.
19871 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19873         PR target/57018
19874         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19875         a set sp if no stack realignment.
19877 2013-04-22  Nick Clifton  <nickc@redhat.com>
19879         * config.gcc (tilegx-linux): Extend extra_objs rather than
19880         overwriting it.
19881         (tilepro-linux): Likewise.
19883 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19885         * config/aarch64/aarch64-builtins.c
19886         (CF): Remove.
19887         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19888         (VAR<1-12>): Add MAP parameter.
19889         (BUILTIN_*): Likewise.
19890         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19891         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19892         (aarch64_ushl_n<mode>): Likewise.
19893         (aarch64_sshr_n<mode>): Likewise.
19894         (aarch64_ushr_n<mode>): Likewise.
19895         (aarch64_<maxmin><mode>): Likewise.
19896         (aarch64_sqrt<mode>): Likewise.
19897         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19898         (vshr<q>_n_*): Likewise.
19900 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19902         * config/aarch64/aarch64-builtins.c
19903         (aarch64_simd_builtin_type_mode): Handle SF types.
19904         (sf_UP): Define.
19905         (BUILTIN_GPF): Define.
19906         (aarch64_init_simd_builtins): Handle SF types.
19907         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19908         (frecps): Likewise.
19909         (frecpx): Likewise.
19910         * config/aarch64/aarch64-simd.md
19911         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19912         (aarch64_frecpe<mode>): New.
19913         (aarch64_frecps<mode>): Likewise.
19914         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19915         (v8type): Add frecp<esx>.
19916         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19917         (aarch64_frecps<mode>): Likewise.
19918         * config/aarch64/iterators.md (FRECP): New.
19919         (frecp_suffix): Likewise.
19920         * config/aarch64/arm_neon.h
19921         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19923 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19925         PR target/56995
19926         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19927         (REG_CLASS_NAMES): Idem.
19928         (REG_CLASS_CONTENTS): Idem.
19929         (REGCLASS_HAS_FP_REG): Idem.
19930         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19931         (sh_conditional_register_usage): Idem.
19933 2013-04-21  Jeff Law  <law@redhat.com>
19935         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19936         (ssa_forward_propagate_and_combine): Use it.
19938 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19940         * lra.c: Update the flow chart diagram.
19942 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19944         PR rtl-optimization/56847
19945         * lra-constraints.c (process_alt_operands): Discourage alternative
19946         with non-matche doffsettable memory constraint fro memory with
19947         known offset.
19949 2013-04-19  Richard Biener  <rguenther@suse.de>
19951         PR tree-optimization/56982
19952         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19953         function.
19954         * gimplify.c (gimplify_call_expr): Notice special calls.
19955         (gimplify_modify_expr): Likewise.
19956         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19957         abnormal control flow receivers.
19958         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19959         in the same way as cfun->has_nonlocal_labels.
19960         (gimple_purge_dead_abnormal_call_edges): Likewise.
19961         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19962         receivers start a basic-block.
19964 2013-04-19  Richard Biener  <rguenther@suse.de>
19966         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19967         member ...
19968         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19969         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19970         (SLP_TREE_LOAD_PERMUTATION): Add.
19971         (vect_transform_slp_perm_load): Adjust prototype.
19972         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19973         (vect_free_slp_instance): Likewise.
19974         (vect_create_new_slp_node): Likewise.
19975         (vect_supported_slp_permutation_p): Remove.
19976         (vect_slp_rearrange_stmts): Adjust.
19977         (vect_supported_load_permutation_p): Likewise.  Inline
19978         vect_supported_slp_permutation_p here.
19979         (vect_analyze_slp_instance): Compute load permutations per
19980         slp node instead of per instance.
19981         (vect_get_slp_defs): Adjust.
19982         (vect_transform_slp_perm_load): Likewise.
19983         (vect_schedule_slp_instance): Remove redundant code.
19984         (vect_schedule_slp): Remove hack for PR56270, add it ...
19985         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19986         CSE loads for SLP.  Adjust.
19988 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19990         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19991         spelling in two comments.
19993 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19995         PR target/56797
19996         * config/arm/arm.c (load_multiple_sequence): Require SP
19997         as base register for loads if SP is in the register list.
19999 2013-04-19  Martin Jambor  <mjambor@suse.cz>
20001         PR tree-optimization/56718
20002         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
20003         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
20004         and made public.  Adjusted all callers.
20005         (ipa_intraprocedural_devirtualization): New function.
20006         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
20007         (ipa_intraprocedural_devirtualization): Likewise.
20008         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
20010 2013-04-19  Richard Biener  <rguenther@suse.de>
20012         PR tree-optimization/57000
20013         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
20015 2013-04-19  Terry Guo  <terry.guo@arm.com>
20017         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
20018         Replace with ...
20019         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
20020         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
20021         (cortex_m4_fmacs): Use new reservations.
20022         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
20024 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
20026         PR rtl-optimization/56999
20027         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
20028         related code.
20029         (lra_coalesce): Remove split_origin_bitmap and related code.
20030         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
20031         ranges if necessary.
20033 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
20035         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
20036         New array.
20037         (ix86_expand_call): Remove clobbered_registers array and use
20038         x86_64_ms_sysv_extra_clobbered_registers instead.
20039         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
20040         Declare here.
20041         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
20042         predicate.
20043         * config/i386/i386.md (*call_rex64_ms_sysv): Use
20044         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
20045         (*call_value_rex64_ms_sysv): Ditto.
20047 2013-04-18  Cary Coutant  <ccoutant@google.com>
20049         * dwarf2out.c (output_pubnames): Check die_perennial_p of
20050         parent instead of die_mark.
20052 2013-04-18  Diego Novillo  <dnovillo@google.com>
20054         * gimple.c (create_gimple_tmp): New.
20055         (get_expr_type): New.
20056         (build_assign): New.
20057         (build_type_cast): New.
20058         * gimple.h (enum ssa_mode): Define.
20059         (gimple_seq_set_location): New.
20060         * asan.c (build_check_stmt): Change some gimple_build_* calls
20061         to use build_assign and build_type_cast.
20063 2013-04-18  Richard Biener  <rguenther@suse.de>
20065         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
20066         handle negative step.  Remove redundant checks.
20067         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
20068         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
20069         for negative step and grouped loads fail to vectorize.
20071 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20073         * emit-rtl.c (reset_insn_used_flags): New function.
20074         (reset_all_used_flags): Use it.
20075         (verify_insn_sharing): New function.
20076         (verify_rtl_sharing): Fix verification for SEQUENCEs.
20078 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
20080         PR tree-optimization/56984
20081         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20082         and (x >> M) >= N don't register any assertion if N << M is the
20083         minimum value.
20085 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20087         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20088         do not delete_insn insns that have not yet been emitted, only
20089         unlink them with remove_insn.
20090         * df-scan.c (df_insn_delete): Revert r197492.
20092 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20094         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20095         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20097 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20099         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20100         define_insn_and_split.
20101         (and_scc,ior_scc,negscc): Likewise.
20102         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20104 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20106         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20107         can benefit from using a sequence of LDRD instructions in epilogue
20108         instead of a single LDM instruction.
20110 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20112         PR 45688
20113         * doc/extend.texi: Fix typo.
20115 2013-04-17  Richard Biener  <rguenther@suse.de>
20117         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20118         (vect_build_slp_tree): ... here.
20119         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20120         match.  Remove special-casing of mismatched complex loads.
20121         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20122         re-try the match with swapped commutative operands.
20123         (vect_supported_load_permutation_p): Remove special-casing of
20124         mismatched complex loads.
20125         (vect_analyze_slp_instance): Adjust.
20127 2013-04-17  Richard Biener  <rguenther@suse.de>
20129         PR rtl-optimization/56921
20130         * cfgloop.h (struct loop): Add simple_loop_desc member.
20131         (struct niter_desc): Mark with GTY(()).
20132         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20133         * loop-iv.c (get_simple_loop_desc): Likewise.
20134         (free_simple_loop_desc): Likewise.
20136         Revert
20137         2013-04-16  Richard Biener  <rguenther@suse.de>
20139         PR rtl-optimization/56921
20140         * loop-init.c (pass_rtl_move_loop_invariants): Add
20141         TODO_do_not_ggc_collect to todo_flags_finish.
20142         (pass_rtl_unswitch): Same.
20143         (pass_rtl_unroll_and_peel_loops): Same.
20144         (pass_rtl_doloop): Same.
20146 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20148         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20149         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20150         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20151         references.
20152         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20153         * tree-streamer.c (record_common_node): Adjust reference in comment.
20155 2013-04-17  Terry Guo  <terry.guo@arm.com>
20157         * config/arm/cortex-m4.md: Add a new bypass.
20159 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20161         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20162         New pattern.
20163         (*subs_<optab><mode>_multp2): New pattern.
20164         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20165         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20167 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20169         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20170         (*subs_mul_imm_<mode>): New pattern.
20172 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20174         PR target/56948
20175         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20176         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20177         (vsx_movti_32bit): Change j->wa to O->wa.
20179 2013-04-16  Richard Biener  <rguenther@suse.de>
20181         PR rtl-optimization/56921
20182         * loop-init.c (pass_rtl_move_loop_invariants): Add
20183         TODO_do_not_ggc_collect to todo_flags_finish.
20184         (pass_rtl_unswitch): Same.
20185         (pass_rtl_unroll_and_peel_loops): Same.
20186         (pass_rtl_doloop): Same.
20188 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20190         * config/arm/arm.c (emit_multi_reg_push): New declaration
20191         for an existing function.
20192         (arm_emit_strd_push): New function.
20193         (arm_expand_prologue): Used here.
20194         (arm_emit_ldrd_pop): New function.
20195         (arm_expand_epilogue): Used here.
20196         (arm_get_frame_offsets): Update condition.
20197         (arm_emit_multi_reg_pop): Add a special case for load of a single
20198         register with writeback.
20200 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20202         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20203         description.
20205 2013-04-16  Richard Biener  <rguenther@suse.de>
20207         PR tree-optimization/56756
20208         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20209         (first_mem_ref_loc): New.
20210         (execute_sm): Place the load temporarily before a previous
20211         access instead of in the latch edge to ensure its SSA dependencies
20212         are defined at points dominating the load.
20214 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20216         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20217         correct fix by moving header and footer insn to the footer of
20218         the merged basic block.  Clear BB_END of the merged-away block.
20220         PR middle-end/43631
20221         * emit-rtl.c (make_note_raw): New function.
20222         (link_insn_into_chain): New static inline function.
20223         (add_insn): Use it.
20224         (add_insn_before, add_insn_after): Factor insn chain linking code...
20225         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20226         using link_insn_into_chain.
20227         (note_outside_basic_block_p): New helper function for emit_note_after
20228         and emit_note_before.
20229         (emit_note_after): Use nobb variant of add_insn_after if the note
20230         should not be contained in a basic block.
20231         (emit_note_before): Use nobb variant of add_insn_before if the note
20232         should not be contained in a basic block.
20233         (emit_note_copy): Use make_note_raw.
20234         (emit_note): Likewise.
20235         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20236         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20237         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20238         the moved barrier the tail of the basic block it follows.
20239         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20241 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20243         PR tree-optimization/56962
20244         * gimple-ssa-strength-reduction.c (record_increment): Only set
20245         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20246         either rhs1 or rhs2 is equal to c->base_expr.
20248 2013-04-15  Richard Biener  <rguenther@suse.de>
20250         PR tree-optimization/56933
20251         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20252         member.
20253         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20254         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20255         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20256         dependence check ...
20257         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20258         ... here.
20259         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20260         GROUP_READ_WRITE_DEPENDENCE.
20262 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20264         * emit-rtl.c (reset_all_used_flags): New function.
20265         (verify_rtl_sharing): Call reset_all_used_flags before and after
20266         performing the checks.
20268 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20270         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20271         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20272         * config/arm/constraints.md (De): New constraint.
20273         * config/arm/neon.md (anddi3_neon): Delete.
20274         (neon_vand<mode>): Expand to standard anddi3 pattern.
20275         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20276         Move earlier in the file.
20277         (neon_inv_logic_op2): Likewise.
20278         (arm_anddi_operand_neon): New predicate.
20280 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20282         * configure.ac (gcc_cv_ld_as_needed): Set
20283         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20284         Use -z ignore, -z record on *-*-solaris2*.
20285         (HAVE_LD_AS_NEEDED): Update comment.
20286         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20287         * configure: Regenerate.
20288         * config.in: Regenerate.
20289         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20290         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20291         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20292         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20293         equivalents.  Fix markup.
20294         * doc/tm.texi: Regenerate.
20296 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20298         * config/i386/i386.opt: New option mstack-protector-guard=.
20299         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20300         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20301         TARGET_SSP_TLS_GUARD.
20302         * config/i386/i386.c (ix86_option_override_internal): Set
20303         ix86_stack_protector_guard.
20304         * config/i386/i386.md (stack_protect_set): Enable for
20305         TARGET_SSP_TLS_GUARD only.
20306         (stack_protect_set_<mode>): Ditto.
20307         (stack_protect_test): Ditto.
20308         (stack_protect_test_<mode>): Ditto.
20309         * doc/invoke.texi (i386 Option): Document.
20311 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20313         PR target/56890
20314         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20315         (S_MODES): Set H_MODE bit.
20316         (SF_MODES): Set only S_MODE and SF_MODE bits.
20317         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20318         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20319         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20320         <MODE_FLOAT>: Likewise.
20322 2013-04-15  Joey Ye  <joey.ye@arm.com>
20324         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20326 2013-04-15  Joey Ye  <joey.ye@arm.com>
20328         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20329         for real far jump.
20330         (thumb_far_jump_used_p): Count instruction size and set
20331         far_jump_used.
20333 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20335         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20336         * resource.c (mark_target_live_regs): Reformat conditional block.
20338 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20340         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20341         notes, they are emitted only just before final.
20342         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20344 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20346         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20347         * cfgrtl.c (delete_insn): Call it here instead.
20348         * lra-spills.c (lra_final_code_change): Use delete_insn.
20349         * haifa-sched.c (sched_remove_insn): Likewise.
20350         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20351         returning to the nop pool.
20352         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20353         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20355 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20357         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20358         * doc/tm.texi: Regenerated.
20360 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20362         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20363         QImode checks.
20365 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20367         * df-core.c (df_find_def): Compare register numbers.
20368         (df_find_use): Likewise.
20370 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20372         PR target/56903
20373         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20374         lra_in_progress for return.
20376 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20378         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20379         define_insn into define_insn_and_split and emit movsicc patterns.
20381 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20383         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20385 2013-04-12  Richard Biener  <rguenther@suse.de>
20387         * tree-pass.h (TODO_do_not_ggc_collect): New.
20388         * passes.c (execute_one_ipa_transform_pass): Honor
20389         TODO_do_not_ggc_collect.
20390         (execute_one_pass): Likewise.
20392         Revert
20393         2013-04-10  Richard Biener  <rguenther@suse.de>
20395         * passes.c (init_optimization_passes): Remove reload pass.
20396         * ira.c (do_reload): Merge into ...
20397         (ira): ... this.
20398         (rest_of_handle_reload): Remove.
20399         (pass_reload): Likewise.
20400         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20401         of reload for vzeroupper pass placement.
20403 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20405         PR tree-optimization/56918
20406         PR tree-optimization/56920
20407         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20408         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20409         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20410         use wide_mul_with_sign method.
20412 2013-04-12  Richard Biener  <rguenther@suse.de>
20414         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20415         not be considered a gimple constant.
20417 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20419         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20420         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20421         operations.
20423 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20424             Jakub Jelinek  <jakub@redhat.com>
20426         * opts.c: Include diagnostic-color.h.
20427         (common_handle_option): Handle OPT_fdiagnostics_color_.
20428         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20429         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20430         (diagnostic-color.o): New.
20431         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20432         (diagnostic_color_rule): New enum.
20433         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20434         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20435         the location string.
20436         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20437         either NULL, or color kind.
20438         * diagnostic-color.c: New file.
20439         * diagnostic-color.h: New file.
20440         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20441         arguments.
20442         * doc/invoke.texi (-fdiagnostics-color): Document.
20443         * pretty-print.h (pp_show_color): Define.
20444         (struct pretty_print_info): Add show_color field.
20445         * diagnostic.c: Include diagnostic-color.h.
20446         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20447         macros.  Colorize error:, warning: etc. strings and also the location
20448         string.
20449         (diagnostic_show_locus): Colorize the caret line.
20450         * pretty-print.c: Include diagnostic-color.h.
20451         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20452         inside of %< %> quotes or quoted through q format modifier.
20454 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20456         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20458 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20460         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20461         code in CC_NZ mode.
20462         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20463         pattern.
20465 2013-04-11  Marek Polacek  <polacek@redhat.com>
20467         PR tree-optimization/48184
20468         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20470 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20472         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20473         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20474         (skip_simple_arithmetic): Tidy up.
20475         * tree.h (skip_simple_constant_arithmetic): Declare.
20477 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20479         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20481 2013-04-11  Richard Biener  <rguenther@suse.de>
20483         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20484         generate vector constants.
20486 2013-04-11  Richard Biener  <rguenther@suse.de>
20488         PR tree-optimization/56878
20489         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20490         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20491         New function.
20492         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20493         Prefer to align the DR with the most invariant base address.
20495 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20497         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20498         comment.
20500 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20502         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20503         floating-point vector comparisons against 0.
20505 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20507         PR tree-optimization/56899
20508         * fold-const.c (extract_muldiv_1): Apply distributive law
20509         only if TYPE_OVERFLOW_WRAPS (ctype).
20511 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20513         PR target/56124
20514         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20515         loading has side effect.
20517 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20519         * config/sparc/sparc.c: Include tree-pass.h.
20520         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20521         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20522         head of file.  Change return type.  Split off gate function.
20523         (sparc_gate_work_around_errata): New function.
20524         (pass_work_around_errata): New pass definition.
20525         (insert_pass_work_around_errata) New pass insert definition to
20526         insert pass_work_around_errata just after delayed-branch scheduling.
20527         (sparc_option_override): Insert the pass.
20528         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20530 2013-04-10  David S. Miller  <davem@davemloft.net>
20532         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20533         or -mcpu=hypersparc.
20535         * target.def (cstore_mode): New hook.
20536         * target.h: Include insn-codes.h
20537         * targhooks.c: Likewise.
20538         (default_cstore_mode): New function.
20539         * targhooks.h: Declare it.
20540         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20541         * doc/tm.texi: Rebuild.
20542         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20543         target hook, rather than inspecting the insn_data.
20544         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20545         (TARGET_CSTORE_MODE): Redefine.
20546         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20547         result patterns.
20548         * config/sparc/predicates.md (cstore_result_operand): New special
20549         predicate.
20550         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20551         Use it for operand 0.
20552         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20553         (*snesi_special): Likewise.
20554         (*snesi_zero): Likewise.
20555         (*seqsi_zero): Likewise.
20556         (*sltu_insn): Likewise.
20557         (*sgeu_insn): Likewise.
20558         (*seqdi_special): Make operand 0 and comparison operation be of
20559         DImode.
20560         (*snedi_special): Likewise.
20561         (*snedi_special_vis3): Likewise.
20562         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20563         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20564         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20565         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20566         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20567         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20568         (*sltu_extend_sp64): Likewise.
20569         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20570         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20571         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20572         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20573         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20575 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20577         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20578         (aarch64_start_file): Use the new function.
20580 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20581             Jason Merrill  <jason@redhat.com>
20583         * common.opt: Add -gdwarf.
20584         * opts.c (common_handle_option): Handle it.
20585         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20587 2013-04-10  Richard Biener  <rguenther@suse.de>
20589         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20590         (execute_one_ipa_transform_pass): But unconditionally here.
20591         (execute_one_pass): And here.
20592         (init_optimization_passes): Remove reload pass.
20593         * tree-pass.h (TODO_ggc_collect): Remove.
20594         (pass_reload): Likewise.
20595         * ira.c (do_reload): Merge into ...
20596         (ira): ... this.
20597         (rest_of_handle_reload): Remove.
20598         (pass_reload): Likewise.
20599         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20600         of reload for vzeroupper pass placement.
20601         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20602         and todo_flags_finish of all passes.
20604 2013-04-10  Richard Biener  <rguenther@suse.de>
20606         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20607         first_const_oprnd field, rename first_def_type to first_op_type.
20608         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20609         (vect_get_and_check_slp_defs): Always use the type of the
20610         operand.  Allow mixed vect_external_def, vect_constant_def types.
20611         (vect_get_constant_vectors): Handle mixed vect_external_def,
20612         vect_constant_def types.
20614 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20616         PR tree-optimization/55524
20617         * tree-ssa-math-opts.c
20618         (convert_mult_to_fma): Don't use an fms construct
20619         when we don't have an fms operation, but fnma, and it looks
20620         likely that we'll be able to use the latter.
20622 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20624         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20625         function.
20626         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20627         inline fail caused by overwritable functions.
20629 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20631         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20632         unnecessary bits in the constant power of two case.
20634 2013-04-10  Richard Biener  <rguenther@suse.de>
20636         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20637         broken code swapping operands.
20638         (vect_build_slp_tree): Do not compute load permutations here.
20639         (vect_analyze_slp_instance): Compute load permutations here,
20640         after building the SLP tree.
20642 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20644         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20645         of next/prev_real_insn.
20647 2013-04-09  Jan Hubicka  <jh@suse.cz>
20649         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20650         Drop aliased parameter.
20651         (function_and_variable_visibility): Do not handle alias pairs.
20652         * cgraph.c (varpool_externally_visible_p): Update prototype.
20653         * varpool.c (varpool_add_new_variable): Update.
20655 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20657         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20659 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20661         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20663         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20665 2013-04-09  Marek Polacek  <polacek@redhat.com>
20667         PR tree-optimization/48762
20668         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20670 2013-04-09  Richard Biener  <rguenther@suse.de>
20672         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20673         dealing with cost.
20674         (vect_build_slp_tree): Likewise.
20675         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20676         calculating the cost of a SLP instance.
20677         (vect_analyze_slp_instance): Use it from here, after building
20678         the SLP tree.
20680 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20682         PR middle-end/56883
20683         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20684         expand_omp_for_static_chunk): Use simple_p = true in
20685         force_gimple_operand_gsi calls when assigning to addressable decls.
20687 2013-04-09  Jeff Law  <law@redhat.com>
20689         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20690         when the boolean was created by converting a wider object which
20691         had a boolean range.
20693 2013-04-09  Richard Biener  <rguenther@suse.de>
20695         * tree-vectorizer.h (slp_void_p): Remove.
20696         (slp_tree): Typedef before _slp_tree declaration.
20697         (struct _slp_tree): Use a vector of slp_tree as children.
20698         (vect_get_place_in_interleaving_chain): Remove.
20699         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20700         Move ...
20701         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20702         and make static.
20703         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20704         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20705         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20706         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20707         Use slp_node instead of slp_void_p and adjust.
20709 2013-04-09  Richard Biener  <rguenther@suse.de>
20711         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20712         work that is not necessary.
20714 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20716         PR tree-optimization/56854
20717         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20718         forward into clobber stmts if it would change MEM_REF lhs into
20719         non-MEM_REF.
20721 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20723         * tree.c (type_hash_lookup, type_hash_add): Make static.
20724         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20726 2013-04-09  Richard Biener  <rguenther@suse.de>
20728         * tree.h (unsave_expr_now): Remove.
20729         * tree-inline.c (mark_local_for_remap_r): Remove.
20730         (unsave_expr_1): Likewise.
20731         (unsave_r): Likewise.
20732         (unsave_expr_now): Likewise.
20733         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20734         (propagate_tree_value): Likewise.
20736 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20738         * doc/rtl.texi (sequence): Rewrite documentation to match the
20739         current use of SEQUENCE rtl objects.
20740         * rtl.def (SEQUENCE): Likewise.
20742         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20743         Update documentation.
20744         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20745         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20747         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20749 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20751         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20752         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20753         methods.
20754         (estimate_edge_size_and_time): Add comment to suggest using rounding
20755         methods.
20756         (estimate_node_size_and_time): Ditto.
20757         (remap_edge_change_prob): Use helper rounding divide methods.
20758         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20759         (gimple_mod_pow2_value_transform): Ditto.
20760         (gimple_mod_subtract_transform): Ditto.
20761         (gimple_ic_transform): Ditto.
20762         (gimple_stringops_transform): Ditto.
20763         * stmt.c (conditional_probability): Ditto.
20764         (emit_case_dispatch_table): Ditto.
20765         * lto-cgraph.c (merge_profile_summaries): Ditto.
20766         * tree-optimize.c (execute_fixup_cfg): Ditto.
20767         * cfgcleanup.c (try_forward_edges): Ditto.
20768         * cfgloopmanip.c (scale_loop_profile): Ditto.
20769         (loopify): Ditto.
20770         (duplicate_loop_to_header_edge): Ditto.
20771         (lv_adjust_loop_entry_edge): Ditto.
20772         * tree-vect-loop.c (vect_transform_loop): Ditto.
20773         * profile.c (compute_branch_probabilities): Ditto.
20774         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20775         * lto-streamer-in.c (input_cfg): Ditto.
20776         * gimple-streamer-in.c (input_bb): Ditto.
20777         * ipa-cp.c (update_profiling_info): Ditto.
20778         (update_specialized_profile): Ditto.
20779         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20780         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20781         rounding methods.
20782         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20783         (compute_trg_info): Ditto.
20784         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20785         (purge_dead_edges): Ditto.
20786         * loop-unswitch.c (unswitch_loop): Ditto.
20787         * cgraphclones.c (cgraph_clone_edge): Ditto.
20788         (cgraph_clone_node): Ditto.
20789         * tree-inline.c (copy_bb): Ditto.
20790         (copy_edges_for_bb): Ditto.
20791         (initialize_cfun): Ditto.
20792         (copy_cfg_body): Ditto.
20793         (expand_call_inline): Ditto.
20795 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20797         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20798         TARGET_CYGWIN64 by TARGET_64BIT.
20800 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20802         * config/epiphany/epiphany.md (GPR_1): New constant.
20803         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20804         * config/epiphany/epiphany.c (gen_compare_reg):
20805         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20806         is already in place.
20807         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20808         Don't require being called during rtl expansion; If y operlaps r0,
20809         return 0.
20810         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20811         (epiphany_expand_epilogue): Likewise.
20813         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20814         Don't use CC_FPmode for ORDERED / UNORDERED.
20815         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20817         * config/epiphany/constraints.md (CnL): New constraint.
20818         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20819         * config/epiphany/predicates.md (add_operand): Allow 1024.
20821         * config/epiphany/epiphany.md (logical_op): New code iterator.
20822         (op_mnc): New code attribute.
20823         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20824         (mov_f+1, mov_f+2): New peephole2 patterns.
20826         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20827         (cstoresi4): Also allow re-use of zero result when doing a NE
20828         comparison to a non-zero operand.
20829         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20831         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20832         Use gen_addsi3_i / gen_subsi3_i.
20834 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20836         PR c++/34949
20837         PR c++/50243
20838         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20839         contain anything but clobbers, at most one __builtin_stack_restore,
20840         optionally debug stmts and final resx, and if it has at least one
20841         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20842         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20843         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20844         which isn't defaut definition, remove them.
20845         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20846         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20847         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20848         with MEM_REF LHS with SSA_NAME address.
20850 2013-04-08  Jeff Law  <law@redhat.com>
20852         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20854 2013-04-08  Richard Biener  <rguenther@suse.de>
20856         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20857         extra newline.
20858         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20859         determined vector type.
20860         (vect_analyze_data_refs): Likewise.
20861         (vect_get_new_vect_var): Adjust.
20862         (vect_create_destination_var): Preserve SSA name versions.
20863         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20864         not dump anything here.
20866 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20868         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20869         Add member lr_slot_known.
20870         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20871         if necessary.
20872         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20873         Remove code that sets lr_slot_offset according to what a previous
20874         version of epiphany_emit_save_restore used to do.
20875         (epiphany_emit_save_restore): When doing an lr save or restore,
20876         set/verify lr_slot_known and lr_slot_offset.
20878 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20880         PR target/54338
20881         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20882         in ALL_REGS.
20884 2013-04-08  Richard Biener  <rguenther@suse.de>
20886         * alias.c (find_base_term): Fix thinko in previous change.
20888 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20890         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20891         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20892         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20893         if possible to compute val.
20894         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20895         For QImode integers don't require anything about precision.  Use
20896         const_with_all_bytes_same to find out if the constant doesn't have
20897         repeated bytes in it.
20899 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20901         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20902         within mode size.
20904 2013-04-08  Marek Polacek  <polacek@redhat.com>
20906         PR rtl-optimization/48182
20907         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20908         value to 1.
20910 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20912         PR target/55487
20913         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20914         nuses, make sure we have a label.
20916 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20918         PR target/56843
20919         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20920         (rs6000_emit_swdiv_low_precision): Remove.
20921         (rs6000_emit_swdiv): Rewrite to handle between one and four
20922         iterations of Newton-Raphson generally; modify required number of
20923         iterations for some cases.
20924         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20926 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20928         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20929         set-but-unused variable.
20931         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20932         basic blocks of released function bodies garbage-collectable.
20934         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20935         (struct rtl_opt_pass): Add TODO_df_finish.
20937         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20939 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20941         * config/arm/constraints.md (q): New constraint.
20942         * config/arm/ldrdstrd.md: New file.
20943         * config/arm/arm.md (ldrdstrd.md) New include.
20944         (arm_movdi): Use "q" instead of "r" constraint
20945         for double-word memory access.
20946         (movdf_soft_insn): Likewise.
20947         * config/arm/vfp.md (movdi_vfp): Likewise.
20948         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20949         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20950         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20951         (mem_ok_for_ldrd_strd): Likewise.
20952         (output_move_double): Update assertion.
20954 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20956         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20958 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20960         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20961         define_insn_and_split.
20962         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20964 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20966         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20967         define_insn_and_split.
20968         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20969         (shiftsi3_compare): New pattern.
20970         (rrx): New pattern.
20971         * config/arm/unspecs.md (UNSPEC_RRX): New.
20973 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20975         * config/arm/arm.md (negdi_extendsidi): New pattern.
20976         (negdi_zero_extendsidi): Likewise.
20978 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20980         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20981         define_insn_and_split.
20982         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20983         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20985 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20987         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20988         define_insn_and_split.
20989         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20990         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20992 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20994         * config/arm/arm.md (subsi3_carryin): New pattern.
20995         (subsi3_carryin_const): Likewise.
20996         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20997         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20999 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21001         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
21003 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21005         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
21006         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
21008 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21010         * config/arm/arm.c (arm_expand_builtin): Change fcode
21011         type to unsigned int.
21013 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21015         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
21017 2013-04-04  Ian Lance Taylor  <iant@google.com>
21019         * doc/standards.texi (Standards): The Go frontend supports the Go 1
21020         language standard.
21022 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
21024         PR middle-end/56729
21025         * df-scan.c (df_insn_delete): Disable failing assert.
21027 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21029         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
21030         New function prototype.
21031         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
21032         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
21033         (arm_builtin_vectorized_function): New function.
21035 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21037         * config/arm/arm_neon_builtins.def: New file.
21038         * config/arm/arm.c (neon_builtin_data): Move contents to
21039         arm_neon_builtins.def.
21040         (enum arm_builtins): Include neon builtin definitions.
21041         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
21042         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
21044 2013-04-04  Marek Polacek  <polacek@redhat.com>
21046         PR tree-optimization/48186
21047         * predict.c (maybe_hot_frequency_p): Return false if
21048         HOT_BB_FREQUENCY_FRACTION is 0.
21049         (cgraph_maybe_hot_edge_p): Likewise.
21051 2013-04-04  Richard Biener  <rguenther@suse.de>
21053         PR tree-optimization/56826
21054         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
21055         more accurately.
21057 2013-04-04  Richard Biener  <rguenther@suse.de>
21059         PR tree-optimization/56213
21060         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
21061         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
21063 2013-04-04  Richard Biener  <rguenther@suse.de>
21065         PR tree-optimization/56837
21066         * tree-loop-distribution.c (classify_partition): For non-zero
21067         values require that the value has the same precision as its
21068         mode to be useful as memset value.
21070 2013-04-03  Nick Clifton  <nickc@redhat.com>
21072         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
21073         (fmssf4): Use fmsf.s on E3V5 architectures.
21074         (fnmasf4): Use fnmaf.s on E3V5 architectures.
21075         (fnmssf4): Use fnmsf.s on E3V5 architectures.
21077 2013-04-03  Jeff Law  <law@redhat.com>
21079         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
21080         (lra-eliminations.o): Likewise.
21082 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21084         * gcov-io.c (compute_working_sets): Moved most of body of old
21085         compute_working_sets here from profile.c.
21086         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21087         (gcov_working_set_t): Moved typedef here from basic-block.h
21088         (compute_working_set): Declare.
21089         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21090         (get_working_sets): Renamed from compute_working_set,
21091         replace most of body with call to new compute_working_sets.
21092         (get_exec_counts): Replace call to compute_working_sets
21093         to get_working_sets.
21094         * profile.h (get_working_sets): Renamed from compute_working_set.
21095         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21096         to get_working_sets.
21097         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21098         * gcov-dump.c (dump_working_sets): New function.
21100 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21102         * hwint.c (sext_hwi, zext_hwi): New functions.
21103         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21104         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21105         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21106         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21107         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21108         (sext_hwi, zext_hwi): New functions.
21110 2013-04-03  Jeff Law  <law@redhat.com>
21112         PR tree-optimization/56799
21113         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21114         back test for widening conversion erroneously dropped in prior change.
21116 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21118         PR target/56809
21119         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21120         instead of next_real_insn.
21122 2013-04-03  Marek Polacek  <polacek@redhat.com>
21124         PR sanitizer/55702
21125         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21127 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21129         PR target/56809
21130         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21131         next_real_insn.
21132         (thumb1_output_casesi): Likewise.
21133         (thumb2_output_casesi): Likewise.
21135 2013-04-03  Richard Biener  <rguenther@suse.de>
21137         PR tree-optimization/56817
21138         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21139         Split out ...
21140         (tree_unroll_loops_completely_1): ... new function to manually
21141         walk the loop tree, properly defering outer loops of unrolled
21142         loops to later iterations.
21144 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21146         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21147         (vectorizable_load): Likewise.
21148         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21149         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21151 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21153         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21154         BIT_FIELD_REF.
21156 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21158         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21160 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21162         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21164 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21166         PR tree-optimization/56790
21167         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21168         folding.
21170 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21172         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21173         Handle VEC_MERGE.
21174         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21175         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21176         equal arguments.
21178 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21180         PR c/19449
21181         * tree.h (force_folding_builtin_constant_p): New decl.
21182         * builtins.c (force_folding_builtin_constant_p): New variable.
21183         (fold_builtin_constant_p): Fold immediately also if
21184         force_folding_builtin_constant_p.
21186 2013-04-03  Richard Biener  <rguenther@suse.de>
21188         PR tree-optimization/56812
21189         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21190         DRs of the same interleaving chain are independent.
21192 2013-04-02  Jason Merrill  <jason@redhat.com>
21194         * gdbinit.in (pbb): Use debug fn.
21196 2013-04-02  Lawrence Crowl  <crowl@google.com>
21198         * sese.h (struct ivtype_map_elt_s): Remove unused.
21199         (extern debug_ivtype_map): Remove unused.
21200         (extern eq_ivtype_map_elts): Remove unused.
21201         * sese.c (debug_ivtype_map): Removed unused.
21202         (debug_ivtype_map_1): Removed unused.
21203         (debug_ivtype_elt): Remove unused.
21204         (eq_ivtype_map_elts): Remove unused.
21207 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21209         PR target/52790
21210         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21211         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21212         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21213         function.
21214         (legitimize_pe_coff_symbol): Likewise.
21215         (is_imported_p): New helper-function.
21216         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21217         for Windows x64 targets.
21218         (ix86_expand_prologue): Optimize for pe-coff targets.
21219         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21220         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21221         medium/large code-model.
21222         (legitimize_pic_address): Likewise.
21223         (legitimize_tls_address): Likewise.
21224         (ix86_expand_call): Likewise.
21225         (x86_output_mi_thunk): Likewise.
21226         (get_dllimport_decl): Add new beimport argument.
21227         (construct_plt_address): Don't assert for x64 pe-coff targets.
21228         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21229         targets.
21230         (SYMBOL_FLAG_STUBVAR): New macro.
21231         (SYMBOL_REF_STUBVAR_P): Likewise.
21232         * config/i386/winnt.c (stub_list): New structure.
21233         (stub_head): New local variable.
21234         (i386_pe_record_stub): New function.
21235         (i386_pe_file_end): Emit refptr-stubs.
21237 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21239         PR rtl-optimization/56745
21240         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21241         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21243         PR c++/34949
21244         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21245         and both of them are MEM_REFs, just compare first argument for
21246         equality and attempt to deal even with differing offsets.
21248         PR c++/34949
21249         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21250         of gimple_clobber_p to be MEM_REF.
21251         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21252         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21253         after gimplification.
21254         * asan.c (get_mem_ref_of_assignment): Don't instrument
21255         gimple_clobber_p stmts.
21256         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21257         gimple_clobber_p stmt if they have MEM_REF lhs and
21258         are dead because of another gimple_clobber_p stmt.
21259         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21260         gimple_clobber_p stmts like debug stmts.
21261         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21262         that refer to unused VAR_DECLs or uninitialized values.
21263         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21264         gimple_clobber_p stmts if they refer to removed parameters.
21265         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21266         formatting.
21268 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21270         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21271         using SWI48 mode attribute.
21273 2013-04-02  Wei Mi  <wmi@google.com>
21275         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21276         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21277         *<rotate_insn><mode>3_mask in i386.md.
21279 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21281         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21283 2013-04-02  Richard Biener  <rguenther@suse.de>
21285         PR tree-optimization/56778
21286         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21287         Runtime alias tests are not supported for gather loads.
21288         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21289         stmts referenced from SSA operands before updating SSA form.
21291 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21292             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21294         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21295         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21296         * config/arm/cortex-a53.md: New file.
21297         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21298         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21299         * config/arm/arm.c (arm_issue_rate): Likewise.
21300         * config/arm/arm-tune.md: Regenerate
21301         * config/arm/arm-tables.opt: Regenerate.
21302         * config/arm/arm-cores.def: Add cortex-a53.
21304 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21306         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21307         non-static link.
21309 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21311         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21312         scalar load/store operations using B/H registers.
21313         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21315 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21317         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21318         scalar move.
21319         * config/aarch64/aarch64.c
21320         (aarch64_simd_scalar_immediate_valid_for_move): New.
21321         * config/aarch64/aarch64-protos.h
21322         (aarch64_simd_scalar_immediate_valid_for_move): New.
21323         * config/aarch64/constraints.md (Dh, Dq): New.
21324         * config/aarch64/iterators.md (hq): New.
21326 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21328         * reorg.c (get_branch_condition): Deal with conditional returns.
21329         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21331 2013-04-01  Wei Mi  <wmi@google.com>
21333         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21334         Truncate operand 2 using %b asm operand modifier.
21335         (*<shift_insn><mode>3_mask): Ditto.
21336         (*<rotate_insn><mode>3_mask): Ditto.
21338 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21340         PR middle-end/56798
21341         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21343 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21345         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21346         of next_real_insn.
21347         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21349 2013-03-30  Lawrence Crowl  <crowl@google.com>
21351         * dse.c (clear_alias_sets): Remove never set.
21352         (disqualified_clear_alias_sets): Remove never set.
21353         (clear_alias_mode_pool): Remove never set.
21354         (dse_step0): Remove condition that is never true.
21355         (canon_address): Remove condition that is never true.
21356         (dse_step7): Remove condition that is never true.
21357         (rest_of_handle_dse): Remove condition that is never true.
21358         (rest_of_handle_dse::did_global): Remove never read from above.
21359         (dse_step2_spill): Remove never called from above.
21360         (dse_step5_spill): Remove never called from above.
21362 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21364         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21365         JUMP_TABLE_DATA changes.
21366         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21367         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21368         (Insns) <jump_table_data>: New entry.
21369         * doc/tm.texi: Regenerate.
21371         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21373         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21374         for table jump at the end of a basic block using tablejump_p.
21375         * targhooks.c (default_invalid_within_doloop): Likewise.
21376         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21377         target hook implementation that is identical to the default hook.
21378         (rs6000_invalid_within_doloop): Remove.
21380         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21381         unused variable from tablejump_p call.
21383         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21384         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21385         (INSN_DELETED_P): Likewise.
21386         (emit_jump_table_data): New prototype.
21387         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21388         after 4th as unused.
21389         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21390         * sched-vis.c (print_insn): Likewise.
21391         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21392         insn for compatibility with back ends that use next_active_insn to
21393         identify jump table data.
21394         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21395         (remove_insn): Likewise.
21396         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21397         to be emitted.
21398         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21399         (emit_jump_table_data): New function.
21401         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21402         basic block, a JUMP_TABLE_DATA never is.
21403         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21404         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21405         off from code handling real insns.
21406         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21407         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21408         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21409         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21410         is not a NONDEBUG_INSN_P.
21411         * ira-costs.c (scan_one_insn): Likewise.
21412         * jump.c (mark_all_labels): Likewise.
21413         (mark_jump_label_1): Likewise.
21414         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21415         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21416         a basic block.
21417         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21418         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21419         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21420         JUMP_TABLE_DATA_P insns.
21421         (calculate_elim_costs_all_insns): Likewise.
21422         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21423         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21424         (delete_output_reload): Code style fixups.
21425         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21426         insn flags on this non-insn.
21427         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21428         as scheduling barriers, for pre-change compatibility.
21429         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21430         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21432         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21433         redundant JUMP_TABLE_DATA_P test.
21434         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21435         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21436         (frv_for_each_packet): Likewise.
21437         * config/i386/i386.c (min_insn_size): Likewise.
21438         (ix86_avoid_jump_mispredicts): Likewise.
21439         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21440         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21441         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21442         (mips16_insn_length): Robustify.
21443         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21444         (mips16_split_long_branches): Likewise.
21445         * config/pa/pa.c (pa_combine_instructions): Likewise.
21446         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21447         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21448         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21449         as contributing to pool range lengths.
21450         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21451         Remove redundant JUMP_TABLE_DATA_P test.
21452         (sh_loop_align): Likewise.
21453         (split_branches): Likewise.
21454         (sh_insn_length_adjustment): Likewise.
21455         * config/spu/spu.c (get_branch_target): Likewise.
21457 2013-03-29  Jan Hubicka  <jh@suse.cz>
21459         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21460         gcov streaming; stream hot bb threshold to ltrans.
21461         * predict.c (get_hot_bb_threshold): Break out from ....
21462         (maybe_hot_count_p): ... here.
21463         (set_hot_bb_threshold): New function.
21464         * lto-section-in.c (lto_section_name): Add profile.
21465         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21466         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21467         and data-streamer.h
21468         (histogram_entry): New structure.
21469         (histogram, histogram_pool): New global vars.
21470         (histogram_hash): New structure.
21471         (histogram_hash::hash): New method.
21472         (histogram_hash::equal): Likewise.
21473         (account_time_size): New function.
21474         (cmp_counts): New function.
21475         (dump_histogram): New function.
21476         (ipa_profile_generate_summary): New function.
21477         (ipa_profile_write_summary): New function.
21478         (ipa_profile_read_summary): New function.
21479         (ipa_profile): Decide on threshold.
21480         (pass_ipa_profile): Add ipa_profile_write_summary and
21481         ipa_profile_read_summary.
21482         * Makefile.in (ipa.o): Update dependencies.
21483         * lto-streamer.h (LTO_section_ipa_profile): New section.
21485 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21487         * tree.h (VAR_P): New.
21489 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21491         PR lto/56777
21492         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21494 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21496         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21497         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21498         (control_flow_insn_p): Likewise.
21499         * cfgrtl.c (duplicate_insn_chain): Likewise.
21500         * final.c (get_attr_length_1): Likewise.
21501         (shorten_branches): Likewise.
21502         (final_scan_insn): Likewise.
21503         * function.c (instantiate_virtual_regs): Likewise.
21504         * gcse.c (insert_insn_end_basic_block): Likewise.
21505         * ira-costs.c (scan_one_insn): Likewise.
21506         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21507         * lra.c (check_rtl): Likewise.
21508         * reload1.c (elimination_costs_in_insn): Likewise.
21509         * reorg.c (follow_jumps): Likewise.
21511         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21512         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21513         (thumb_far_jump_used_p): Likewise.
21514         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21515         (workaround_speculation): Likewise.
21516         (add_sched_insns_for_speculation): Likewise.
21517         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21518         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21519         (frv_for_each_packet): Likewise.
21520         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21521         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21522         (final_emit_insn_group_barriers): Likewise.
21523         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21524         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21525         (mips16_insn_length): Likewise.
21526         * config/pa/pa.c (pa_reorg): Likewise.
21527         (pa_combine_instructions): Likewise.
21528         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21529         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21530         (sh_reorg): Likewise.
21531         (split_branches): Likewise.
21532         * config/spu/spu.c (get_branch_target): Likewise.
21534         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21535         JUMP_TABLE_DATA_P.
21537 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21539         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21540         Fix declaration name.
21542 2013-03-28  Lawrence Crowl  <crowl@google.com>
21544         * graphds.h (struct graph.indicies): Remove unused.
21545         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21546         (SCOP_ORIGINAL_PDDRS): Remove unused.
21547         * sese.h (extern insert_loop_close_phis): Removed unused.
21548         (extern insert_guard_phis): Removed unused.
21549         (extern ivtype_map_elt_info): Removed unused.
21550         (new_ivtype_map_elt): Removed unused.
21551         * sese.c (ivtype_map_elt_info): Removed unused.
21553 2013-03-28  Lawrence Crowl  <crowl@google.com>
21555         * Makefile.in: Add several missing include dependences.
21556         (DUMPFILE_H): New.
21557         (test-dump.o): New.  This object is not added to any executable,
21558         but is present for ad-hoc testing.
21559         * bitmap.c
21560         (debug (const bitmap_head_def &)): New.
21561         (debug (const bitmap_head_def *)): New.
21562         * bitmap.h
21563         (extern debug (const bitmap_head_def &)): New.
21564         (extern debug (const bitmap_head_def *)): New.
21565         * cfg.c
21566         (debug (edge_def &)): New.
21567         (debug (edge_def *)): New.
21568         * cfghooks.c
21569         (debug (basic_block_def &)): New.
21570         (debug (basic_block_def *)): New.
21571         * dumpfile.h
21572         (dump_node (const_tree, int, FILE *)): Correct source file.
21573         * dwarf2out.c
21574         (debug (die_struct &)): New.
21575         (debug (die_struct *)): New.
21576         * dwarf2out.h
21577         (extern debug (die_struct &)): New.
21578         (extern debug (die_struct *)): New.
21579         * gimple-pretty-print.c
21580         (debug (gimple_statement_d &)): New.
21581         (debug (gimple_statement_d *)): New.
21582         * gimple-pretty-print.h
21583         (extern debug (gimple_statement_d &)): New.
21584         (extern debug (gimple_statement_d *)): New.
21585         * ira-build.c
21586         (debug (ira_allocno_copy &)): New.
21587         (debug (ira_allocno_copy *)): New.
21588         (debug (ira_allocno &)): New.
21589         (debug (ira_allocno *)): New.
21590         * ira-int.h
21591         (extern debug (ira_allocno_copy &)): New.
21592         (extern debug (ira_allocno_copy *)): New.
21593         (extern debug (ira_allocno &)): New.
21594         (extern debug (ira_allocno *)): New.
21595         * ira-lives.c
21596         (debug (live_range &)): New.
21597         (debug (live_range *)): New.
21598         * lra-int.h
21599         (debug (lra_live_range &)): New.
21600         (debug (lra_live_range *)): New.
21601         * lra-lives.c
21602         (debug (lra_live_range &)): New.
21603         (debug (lra_live_range *)): New.
21604         * omega.c
21605         (debug (omega_pb_d &)): New.
21606         (debug (omega_pb_d *)): New.
21607         * omega.h
21608         (extern debug (omega_pb_d &)): New.
21609         (extern debug (omega_pb_d *)): New.
21610         * print-rtl.c
21611         (debug (const rtx_def &)): New.
21612         (debug (const rtx_def *)): New.
21613         * print-tree.c
21614         (debug_tree (tree): Move within file.
21615         (debug_raw (const tree_node &)): New.
21616         (debug_raw (const tree_node *)): New.
21617         (dump_tree_via_hooks (const tree_node *, int)): New.
21618         (debug (const tree_node &)): New.
21619         (debug (const tree_node *)): New.
21620         (debug_verbose (const tree_node &)): New.
21621         (debug_verbose (const tree_node *)): New.
21622         (debug_head (const tree_node &)): New.
21623         (debug_head (const tree_node *)): New.
21624         (debug_body (const tree_node &)): New.
21625         (debug_body (const tree_node *)): New.
21626         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21627         (debug (vec<tree, va_gc> &)): New.
21628         (debug (vec<tree, va_gc> *)): New.
21629         * rtl.h
21630         (extern debug (const rtx_def &)): New.
21631         (extern debug (const rtx_def *)): New.
21632         * sbitmap.c
21633         (debug_raw (simple_bitmap_def &)): New.
21634         (debug_raw (simple_bitmap_def *)): New.
21635         (debug (simple_bitmap_def &)): New.
21636         (debug (simple_bitmap_def *)): New.
21637         * sbitmap.h
21638         (extern debug (simple_bitmap_def &)): New.
21639         (extern debug (simple_bitmap_def *)): New.
21640         (extern debug_raw (simple_bitmap_def &)): New.
21641         (extern debug_raw (simple_bitmap_def *)): New.
21642         * sel-sched-dump.c
21643         (debug (vinsn_def &)): New.
21644         (debug (vinsn_def *)): New.
21645         (debug_verbose (vinsn_def &)): New.
21646         (debug_verbose (vinsn_def *)): New.
21647         (debug (expr_def &)): New.
21648         (debug (expr_def *)): New.
21649         (debug_verbose (expr_def &)): New.
21650         (debug_verbose (expr_def *)): New.
21651         (debug (vec<rtx> &)): New.
21652         (debug (vec<rtx> *)): New.
21653         * sel-sched-dump.h
21654         (extern debug (vinsn_def &)): New.
21655         (extern debug (vinsn_def *)): New.
21656         (extern debug_verbose (vinsn_def &)): New.
21657         (extern debug_verbose (vinsn_def *)): New.
21658         (extern debug (expr_def &)): New.
21659         (extern debug (expr_def *)): New.
21660         (extern debug_verbose (expr_def &)): New.
21661         (extern debug_verbose (expr_def *)): New.
21662         (extern debug (vec<rtx> &)): New.
21663         (extern debug (vec<rtx> *)): New.
21664         * sel-sched-ir.h
21665         (_list_iter_cond_expr): Make inline instead of static.
21666         * sreal.c
21667         (debug (sreal &)): New.
21668         (debug (sreal *)): New.
21669         * sreal.h
21670         (extern debug (sreal &)): New.
21671         (extern debug (sreal *)): New.
21672         * tree.h
21673         (extern debug_raw (const tree_node &)): New.
21674         (extern debug_raw (const tree_node *)): New.
21675         (extern debug (const tree_node &)): New.
21676         (extern debug (const tree_node *)): New.
21677         (extern debug_verbose (const tree_node &)): New.
21678         (extern debug_verbose (const tree_node *)): New.
21679         (extern debug_head (const tree_node &)): New.
21680         (extern debug_head (const tree_node *)): New.
21681         (extern debug_body (const tree_node &)): New.
21682         (extern debug_body (const tree_node *)): New.
21683         (extern debug (vec<tree, va_gc> &)): New.
21684         (extern debug (vec<tree, va_gc> *)): New.
21685         * tree-cfg.c
21686         (debug (struct loop &)): New.
21687         (debug (struct loop *)): New.
21688         (debug_verbose (struct loop &)): New.
21689         (debug_verbose (struct loop *)): New.
21690         * tree-dump.c: Add header dependence.
21691         * tree-flow.h
21692         (extern debug (struct loop &)): New.
21693         (extern debug (struct loop *)): New.
21694         (extern debug_verbose (struct loop &)): New.
21695         (extern debug_verbose (struct loop *)): New.
21696         * tree-data-ref.c
21697         (debug (data_reference &)): New.
21698         (debug (data_reference *)): New.
21699         (debug (vec<data_reference_p> &)): New.
21700         (debug (vec<data_reference_p> *)): New.
21701         (debug (vec<ddr_p> &)): New.
21702         (debug (vec<ddr_p> *)): New.
21703         * tree-data-ref.h
21704         (extern debug (data_reference &)): New.
21705         (extern debug (data_reference *)): New.
21706         (extern debug (vec<data_reference_p> &)): New.
21707         (extern debug (vec<data_reference_p> *)): New.
21708         (extern debug (vec<ddr_p> &)): New.
21709         (extern debug (vec<ddr_p> *)): New.
21710         * tree-ssa-alias.c
21711         (debug (pt_solution &)): New.
21712         (debug (pt_solution *)): New.
21713         * tree-ssa-alias.h
21714         (extern debug (pt_solution &)): New.
21715         (extern debug (pt_solution *)): New.
21716         * tree-ssa-alias.c
21717         (debug (_var_map &)): New.
21718         (debug (_var_map *)): New.
21719         (debug (tree_live_info_d &)): New.
21720         (debug (tree_live_info_d *)): New.
21721         * tree-ssa-alias.h
21722         (extern debug (_var_map &)): New.
21723         (extern debug (_var_map *)): New.
21724         (extern debug (tree_live_info_d &)): New.
21725         (extern debug (tree_live_info_d *)): New.
21727 2013-03-28  Jan Hubicka  <jh@suse.cz>
21729         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21731 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21733         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21734         record only when desired or required.
21736 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21738         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21739         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21741 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21743         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21744         (*andsi3_compare0_uxtw): New pattern.
21745         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21746         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21748 2013-03-28  Jan Hubicka  <jh@suse.cz>
21750         * data-streamer-in.c (streamer_read_gcov_count): New function.
21751         * gimple-streamer-out.c: Include value-prof.h.
21752         (output_gimple_stmt): Output histogram.
21753         (output_bb): Use streamer_write_gcov_count.
21754         * value-prof.c: Include data-streamer.h
21755         (dump_histogram_value): Add HIST_TYPE_MAX.
21756         (stream_out_histogram_value): New function.
21757         (stream_in_histogram_value): New function.
21758         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21759         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21760         * data-streamer-out.c (streamer_write_gcov_count): New function.
21761         (streamer_write_gcov_count_stream): New function.
21762         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21763         (lto_output_node): Likewise.
21764         (input_node, input_edge): Likewise.
21765         * lto-streamer-out.c (output_cfg): Update streaming.
21766         * lto-streamer-in.c (input_cfg): Likewise.
21767         * data-streamer.h (streamer_write_gcov_count,
21768         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21769         * gimple-streamer-in.c: Include value-prof.h
21770         (input_gimple_stmt): Input histograms.
21771         (input_bb): Update profile streaming.
21773 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21775         * genmodes.c (emit_max_int): New function.
21776         (emit_insn_modes_h): Added call to emit_max_function.
21777         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21778         Added doc.
21779         * machmode.def: Fixed comment.
21781 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21783         * combine.c (try_combine): Removed useless assert.
21784         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21786 2013-03-28  Marek Polacek  <polacek@redhat.com>
21787             Richard Biener  <rguenther@suse.de>
21789         PR tree-optimization/56695
21790         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21791         build signed result of a vector comparison.
21792         * tree-cfg.c (verify_gimple_comparison): Check that a result
21793         of a vector comparison has signed type.
21795 2013-03-28  Richard Biener  <rguenther@suse.de>
21797         PR tree-optimization/37021
21798         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21799         do not restrict gaps between groups.
21800         * tree-vect-stmts.c (vectorizable_load): Properly account for
21801         a gap between groups.
21803 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21805         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21806         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21807         is not enabled.
21809 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21811         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21812         * doc/extend.texi (Named Address Spaces): Ditto.
21813         (Variable Attributes): Ditto.
21815 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21817         * config.build: Add support for cygwin x64 target.
21818         * config.gcc: Likewise.
21819         * config.host: Likewise.
21820         * configure.ac: Likewise
21821         * configure: Regenerated.
21823 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21825         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21826         * config/i386/t-cygwin-w64: New file.
21827         * config/i386/cygwin-w64.h: New file.
21828         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21829         and add support for x64-cygwin target.
21830         (CPP_SPEC): Likewise.
21831         (CXX_WRAP_SPEC_LIST): Undefine before define.
21832         (LIBGCJ_SONAME): Use 15 as version.
21834 2013-03-27  Richard Biener  <rguenther@suse.de>
21836         PR tree-optimization/56716
21837         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21838         dumping for ref nodes.
21840 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21842         PR tree-optimization/55334
21843         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21844         restricted pointers to arrays.
21846 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21848         * Makefile.in (.SUFFIXES): Add .cc.
21849         (.c.o): Apply same recipe for implicit rule .cc.o.
21851 2013-03-27  Richard Biener  <rguenther@suse.de>
21853         PR tree-optimization/37021
21854         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21855         REALPART/IMAGPART_EXPRs around the supported refs.
21856         * tree-ssa-structalias.c (find_func_aliases): Assume that
21857         floating-point values are not used to transfer pointers.
21859 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21861         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21862         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21863         * doc/tm.texi: Regenerate.
21864         * targhooks.h (default_has_ifunc_p): New.
21865         * targhooks.c (default_has_ifunc_p): Ditto.
21866         * config/linux-protos.h: New file.
21867         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21868         hook for linux which disables support of indirect functions in android.
21869         * config/linux-android.c: New file.
21870         * config/t-linux-android.c: Ditto.
21871         * config.gcc: Added new object file linux-android.o.
21872         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21873         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21874         * varasm.c (do_assemble_alias): Likewise.
21875         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21876         doesn't support indirect functions.
21877         * configure: Regenerate.
21879 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21881         PR target/56102
21882         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21883         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21884         mult-word mode.
21886 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21888         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21890 2013-03-27  Terry Guo  <terry.guo@arm.com>
21892         * config/arm/arm-cores.def: Added core cortex-r7.
21893         * config/arm/arm-tune.md: Regenerated.
21894         * config/arm/arm-tables.opt: Regenerated.
21895         * doc/invoke.texi: Added entry for core cortex-r7.
21897 2013-03-27  Walter Lee  <walt@tilera.com>
21899         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21900         double-decrement of next_scratch_regno.
21902 2013-03-27  Walter Lee  <walt@tilera.com>
21904         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21905         input operands.
21906         (insn_v1mulus): Ditto.
21907         (insn_v2muls): Ditto.
21909 2013-03-27  Walter Lee  <walt@tilera.com>
21911         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21912         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21914 2013-03-27  Walter Lee  <walt@tilera.com>
21916         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21917         (*sibcall_value): Ditto.
21919 2013-03-27  Walter Lee  <walt@tilera.com>
21921         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21922         (insn_mnz_v8qi): ... this ...
21923         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21924         vector equivalent.
21925         (insn_v<n>mnz): Replaced by ...
21926         (insn_v1mnz): ... this ...
21927         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21928         equivalent.
21929         (insn_mz_<mode>): Replaced by ...
21930         (insn_mz_v8qi): ... this ...
21931         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21932         vector equivalent.
21933         (insn_v<n>mz): Replaced by ...
21934         (insn_v1mz): ... this ...
21935         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21936         equivalent.
21938 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21940         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21942 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21944         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21945         than fprintf with a non-constant, non-format string.
21947 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21949         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21950         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21951         operand 0 predicate.
21952         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21953         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21954         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21955         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21956         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21957         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21958         (mov<mode>_insv_1): Remove expander.  Merge insn with
21959         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21960         Use general_x64nomem_operand as operand 1 predicate.
21961         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21962         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21963         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21964         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21965         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21966         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21967         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21968         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21969         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21970         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21971         (general_x64nomem_operand): Ditto.
21973 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21975         * config/rtems.opt: Add -pthread option.
21977 2013-03-26  Richard Biener  <rguenther@suse.de>
21979         * alias.c (find_base_term): Avoid redundant and not used recursion.
21980         (base_alias_check): Get the initial base term from the caller.
21981         (true_dependence_1): Compute and pass base terms to base_alias_check.
21982         (write_dependence_p): Likewise.
21983         (may_alias_p): Likewise.
21985 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21987         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21988         PC-relative load in SI modes and above only.
21990 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21992         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21993         * config/arm/iwmmxt.md (WCGR0): Update.
21994         (WCGR1, WCGR2, WCGR3): Likewise.
21996 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21998         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21999         Use x64 and nox64 isa attributes.
22001 2013-03-26  Richard Biener  <rguenther@suse.de>
22003         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
22004         alignment computations and rely on get_object_alignment_1
22005         for the !TYPE_P case.
22006         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
22008 2013-03-26  Walter Lee  <walt@tilera.com>
22010         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
22011         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
22013 2013-03-25  Jeff Law  <law@redhat.com>
22015         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
22016         check for INTEGRAL_TYPE_P that was missing due to checking in
22017         wrong version of prior patch.
22019 2013-03-25  Walter Lee  <walt@tilera.com>
22021         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
22022         TILEGX_INSN_SHUFFLEBYTES1.
22023         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
22024         shufflebytes1.
22025         (tilegx_builtins): Ditto.
22026         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
22028 2013-03-25  Walter Lee  <walt@tilera.com>
22030         * config/tilegx/tilegx.md (floatsisf2): New pattern.
22031         (floatunssisf2): New pattern.
22032         (floatsidf2): New pattern.
22033         (floatunssidf2): New pattern.
22035 2013-03-25  Walter Lee  <walt@tilera.com>
22037         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
22038         tests for constraint J, K, N, P.
22040 2013-03-25  Walter Lee  <walt@tilera.com>
22042         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
22043         Use indirect/pcrel encoding.
22044         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
22045         Ditto.
22047 2013-03-25  Steve Ellcey  <sellcey@mips.com>
22049         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
22050         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
22051         * config/mips/mips.c (mips_option_override): Set IMADD default.
22052         * config/mips/mips.h (PTF_AVOID_IMADD): New.
22053         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
22054         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
22055         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
22056         * doc/invoke.texi (-mimadd/-mno-imadd): New.
22058 2013-03-25  Jeff Law  <law@redhat.com>
22060         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
22061         slightly to avoid creating and folding useless trees.  Simplify
22062         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
22064 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
22066         * config/i386/i386.md (*zero_extendsidi2): Merge with
22067         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
22068         * config/i386/predicates.md (x86_64_zext_operand): Rename from
22069         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
22070         targets.  Clarify comment.
22072 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22074         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
22075         pass-through jump functions differently.
22076         (ipa_read_jump_function): Likewise.  Also use setter functions to set
22077         up jump functions.
22079 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22081         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22082         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22083         process it.
22084         (ipa_get_indirect_edge_target): New function.
22085         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22086         ipa_get_indirect_edge_target.  Update all callers.
22087         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22088         ipa_get_indirect_edge_target_1 instead of calling
22089         ipa_get_indirect_edge_target.
22090         (create_specialized_node): Pass aggvlas to
22091         ipcp_discover_new_direct_edges.
22093 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22095         * config/arm/arm.md (f_sels, f_seld): New types.
22096         (*cmov<mode>): New pattern.
22097         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22098         predicate.
22100 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22102         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22103         POSIX-printf for mingw-hosted  builds.
22105 2013-03-25  Richard Biener  <rguenther@suse.de>
22107         PR middle-end/56694
22108         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22109         must-not-throw stmt location.
22111 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22113         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22114         Emit load-acquire versions when acq is true.
22115         (arm_emit_store_exclusive): Add rel parameter.
22116         Emit store-release versions when rel is true.
22117         (arm_split_compare_and_swap): Use acquire-release instructions
22118         instead.
22119         of barriers when appropriate.
22120         (arm_split_atomic_op): Likewise.
22121         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22122         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22123         (VUNSPEC_SLX): Likewise.
22124         (VUNSPEC_LDA): Likewise.
22125         (VUNSPEC_STL): Likewise.
22126         * config/arm/sync.md (atomic_load<mode>): New pattern.
22127         (atomic_store<mode>): Likewise.
22128         (arm_load_acquire_exclusive<mode>): Likewise.
22129         (arm_load_acquire_exclusivesi): Likewise.
22130         (arm_load_acquire_exclusivedi): Likewise.
22131         (arm_store_release_exclusive<mode>): Likewise.
22133 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22135         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22136         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22137         * config/mip/predicates.md (lwsp_swsp_operand,
22138         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22139         sb16_operand, db4_operand, db7_operand, ib3_operand,
22140         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22141         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22142         andi16_operand): New predicates.
22143         * config/mips/mips.md (compression): New attribute.
22144         (enabled): New attribute.
22145         (length): Consider compression in computing length.
22146         (shift_compression): New code attribute.
22147         (*add<mode>3): New operands. Record compression.
22148         (sub<mode>3): Likewise.
22149         (one_cmpl<mode>2): Likewise.
22150         (*and<mode>3): Likewise.
22151         (*ior<mode>3): Likewise.
22152         (unnamed pattern for xor): Likewise.
22153         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22154         (*<optab><mode>3): Likewise.
22155         (*mov<mode>_internal: Likewise.
22156         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22157         (mips_unsigned_immediate_p): New.
22158         (umips_lwsp_swsp_address_p): New.
22159         (m16_based_address_p): New.
22160         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22161         (mips_unsigned_immediate_p): New prototype.
22162         (lwsp_swsp_address_p): New prototype.
22163         (m16_based_address_p): New prototype.
22164         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22165         (mips_signed_immediate_p): New function.
22166         (m16_based_address_p): New function.
22167         (lwsp_swsp_address_p): New function.
22168         (mips_print_operand_punctuation): Recognize short delay slot insns
22169         for microMIPS.add<mode>3"
22171 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22173         PR target/56720
22174         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22175         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22177 2013-03-25  Richard Biener  <rguenther@suse.de>
22179         PR tree-optimization/56689
22180         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22181         any edge.
22183 2013-03-25  Richard Biener  <rguenther@suse.de>
22185         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22186         of bitmap.
22187         (memory_references): Likewise.
22188         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22189         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22190         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22191         (gather_mem_refs_in_loops): Fold into ...
22192         (analyze_memory_references): ... this.  Move initialization
22193         to tree_ssa_lim_initialize.
22194         (fill_always_executed_in): Rename to ...
22195         (fill_always_executed_in_1): ... this.
22196         (fill_always_executed_in): Move contains_call computation to
22197         this new function from ...
22198         (tree_ssa_lim_initialize): ... here.
22199         (tree_ssa_lim): Call fill_always_executed_in.
22201 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22203         * postreload.c (reload_combine): Fix code detecting returns.
22205 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22207         * function.c (emit_use_return_register_into_block): On cc0 targets,
22208         do not emit the sequence between cc0 setter and user.
22210 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22212         * config/i386/predicates.md (local_symbolic_operand): Interpret
22213         dll-imported symbols as none-local.
22215 2013-03-25  Richard Biener  <rguenther@suse.de>
22217         * tree-ssa-loop-im.c (struct depend): Remove.
22218         (struct lim_aux_data): Make depends a vec of gimples.
22219         (free_lim_aux_data): Adjust.
22220         (add_dependency): Likewise.
22221         (set_level): Likewise.
22223 2013-03-25  Richard Biener  <rguenther@suse.de>
22225         PR middle-end/56434
22226         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22227         the pointer returned by calls with ECF_MALLOC set.
22229 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22231         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22233 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22235         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22236         using MMXMODE mode iterator.
22237         (*move<mode>_internal): Merge with *movv2sf_internal and
22238         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22240 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22242         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22243         (record_last_mem_set_info): Likewise.
22245         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22246         of XNEWVEC followed by memset.
22247         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22249 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22251         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22252         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22253         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22254         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22255         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22256         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22257         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22258         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22259         BARRIER_P instead of GET_CODE.
22261 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22263         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22264         inaccuracy in the probing code.
22266         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22267         (ctrapdi4): Likewise.
22269 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22271         * calls.c (expand_call): Add missing guard to code handling return
22272         of non-BLKmode structures in MSB.
22273         * function.c (expand_function_end): Likewise.
22275 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22277         * combine.c (try_combine): Adjust comment.  Do not add the set of
22278         insn #0 if the destination indirectly is set or dies in insn #2.
22279         Tidy up code to distribute a new note.
22281 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22283         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22284         also for alternatives 16 and 17.
22286 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22288         * config/i386/sse.md (*mov<mode>_internal): Merge with
22289         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22290         Emit insn template depending on type attribute.  Use
22291         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22292         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22293         attribute calculation.  Remove unit attribute calculation.
22294         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22295         Set prefix_data16 attribute for DImode ssemov types.
22296         Use Ym instead of y for SSE-MMX conversion alternatives.
22297         Reorder operand constraints.
22299 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22301         * df.h (df_insn_delete): Adjust prototype.
22302         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22303         and let it decide whether mark the basic block dirty.
22304         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22305         * df-scan.c (df_insn_info_delete): New helper function, split
22306         off from df_insn_delete.
22307         (df_scan_free_bb_info): Use it.
22308         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22309         Likewise.
22310         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22311         that the insn is actually an insn and it has a non-NULL basic block.
22312         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22314 2013-03-22  Richard Biener  <rguenther@suse.de>
22316         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22317         dep_ref members.
22318         (mem_ref_alloc): Do not allocate them.
22319         (refs_independent_p): Do not query or maintain a cache.
22321 2013-03-22  Richard Biener  <rguenther@suse.de>
22323         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22324         (gather_mem_refs_in_loops): Do not compute it.
22325         (analyze_memory_references): Do not allocate it.
22326         (tree_ssa_lim_finalize): Do not free it.
22327         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22329 2013-03-22  Richard Biener  <rguenther@suse.de>
22331         * is-a.h (as_a): Use gcc_checking_assert.
22333 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22335         * config/aarch64/aarch64.c (aarch64_print_operand): New
22336         format specifier for printing a constant in hex.
22337         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22338         format specifier for printing second operand.
22340 2013-03-22  Richard Biener  <rguenther@suse.de>
22342         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22343         bitmaps.
22344         (gather_mem_refs_in_loops): Perform store accumulation here.
22345         (create_vop_ref_mapping_loop): Remove.
22346         (create_vop_ref_mapping): Likewise.
22347         (analyze_memory_references): Initialize refs_stored_in_loop.
22348         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22349         (record_indep_loop): Remove.
22350         (record_dep_loop): New function.
22351         (ref_indep_loop_p_1): Adjust to only walk over references
22352         in the loop, not its subloops.
22353         (ref_indep_loop_p): Rename to ...
22354         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22355         maintaining a more fine-grained cache.
22356         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22357         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22359 2013-03-22  Richard Biener  <rguenther@suse.de>
22361         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22362         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22363         aggregate mem_ref_loc.
22364         (free_mem_ref_locs): Inline into ...
22365         (memref_free): ... this and adjust.
22366         (mem_ref_alloc): Adjust.
22367         (mem_ref_locs_alloc): Remove.
22368         (record_mem_ref_loc): Adjust.
22369         (get_all_locs_in_loop): Rewrite into ...
22370         (for_all_locs_in_loop): ... this iterator.
22371         (rewrite_mem_ref_loc): New functor.
22372         (rewrite_mem_refs): Use for_all_locs_in_loop.
22373         (sm_set_flag_if_changed): New functor.
22374         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22375         (ref_always_accessed): New functor.
22376         (ref_always_accessed_p): Use for_all_locs_in_loop.
22378 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22380         * tree-pass.h (PROP_gimple_lvec): New.
22381         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22382         (init_optimization_passes): Move pass_lower_vector.
22383         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22384         PROP_gimple_lvec.
22385         (pass_lower_vector): Provide PROP_gimple_lvec.
22386         (pass_lower_vector_ssa): Likewise.
22387         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22389 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22391         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22393 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22395         * config/i386/i386.md (*movdi_internal): Disparage slightly
22396         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22397         conversion alternatives.
22399 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22401         PR middle-end/48087
22402         * diagnostic.def (DK_WERROR): New kind.
22403         * diagnostic.h (werrorcount): Define.
22404         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22405         promoted to DK_ERROR, increment DK_WERROR counter instead of
22406         DK_ERROR counter.
22407         * toplev.c (toplev_main): Call print_ignored_options even if
22408         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22409         even if just werrorcount is non-zero.
22411         PR debug/55608
22412         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22413         on failure.
22414         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22415         (string_cst_pool_decl): New function.
22416         (optimize_one_addr_into_implicit_ptr): New function.
22417         (resolve_addr_in_expr): Optimize DWARF location expression
22418         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22419         which doesn't live in memory, but has DW_AT_location or
22420         DW_AT_const_value, or refers to a string literal, into
22421         DW_OP_GNU_implicit_pointer.
22422         (optimize_location_into_implicit_ptr): New function.
22423         (resolve_addr): If removing DW_AT_location of a variable because
22424         it was DW_OP_addr of address of the variable, but the variable doesn't
22425         live in memory, try to emit const value attribute for the initializer.
22427 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22429         * tree.h (VECTOR_TYPE_P): New macro.
22430         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22431         TYPE_MODE): Use it.
22432         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22433         VEC_COND_EXPR cannot be lvalues.
22434         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22436 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22438         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22439         Restrict the transformation to equal modes.
22441 2013-03-21  Richard Biener  <rguenther@suse.de>
22443         PR tree-optimization/39326
22444         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22445         (MEM_ANALYZABLE): Adjust.
22446         (record_mem_ref_loc): Move bitmap ops ...
22447         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22448         unanalyzable refs, do not record locations for it.
22449         (analyze_memory_references): Allocate ref zero as shared
22450         unanalyzable ref.
22451         (refs_independent_p): Do not test for unanalyzed mems here.
22452         (ref_indep_loop_p_1): Special-case disambiguation against
22453         the unanalyzed ref.
22454         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22456 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22458         * config/arm/arm-protos.h (tune_params): Add
22459         prefer_neon_for_64bits field.
22460         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22461         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22462         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22463         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22464         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22465         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22466         (arm_option_override): Handle -mneon-for-64bits new option.
22467         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22468         (prefer_neon_for_64bits): Declare new variable.
22469         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22470         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22471         (arch_enabled): Handle new arch types. Remove support for onlya8
22472         and nota8.
22473         (one_cmpldi2): Use new arch names.
22474         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22475         * config/arm/arm.opt (mneon-for-64bits): Add option.
22476         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22477         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22478         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22479         of onlya8.
22480         * doc/invoke.texi (-mneon-for-64bits): Document.
22482 2013-03-21  Richard Biener  <rguenther@suse.de>
22484         PR tree-optimization/39326
22485         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22486         (sort_bbs_in_loop_postorder_cmp): New function.
22487         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22489 2013-03-21  Richard Biener  <rguenther@suse.de>
22491         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22492         (vect_insert_into_interleaving_chain): Likewise.
22493         (vect_drs_dependent_in_basic_block): Inline ...
22494         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22495         split out from ...
22496         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22497         (vect_check_interleaving): Simplify.
22498         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22499         (vect_slp_analyze_data_ref_dependences): ... this new function.
22500         (dr_group_sort_cmp): New function.
22501         (vect_analyze_data_ref_accesses): Compute data-reference groups
22502         here instead of in vect_analyze_data_ref_dependence.  Use
22503         a more efficient algorithm.
22504         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22505         vect_slp_analyze_data_ref_dependences.  Call
22506         vect_analyze_data_ref_accesses earlier.
22507         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22508         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22509         (vect_slp_analyze_data_ref_dependences): New prototype.
22511 2013-03-21  Richard Biener  <rguenther@suse.de>
22513         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22514         ref is stored in the loop.
22515         (find_refs_for_sm): Walk only over all stores.
22516         (store_motion_loop): Allocate from lim_bitmap_obstack.
22517         (store_motion): Likewise.
22519 2013-03-21  Richard Biener  <rguenther@suse.de>
22521         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22522         Update virtual SSA form.
22524 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22526         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22527         * configure: Regenerate.
22528         * config.in: Regenerate.
22529         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22530         if !HAVE_LD_EH_FRAME_CIEV3.
22532 2013-03-21  Richard Biener  <rguenther@suse.de>
22534         * tree-cfg.c (verify_expr_no_block): New function.
22535         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22536         nor DECL_VALUE_EXPR have locations with associated blocks.
22537         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22538         (clear_unused_block_pointer): Remove code dealing with
22539         blocks in DECL_DEBUG_EXPR locations.
22541 2013-03-21  Richard Biener  <rguenther@suse.de>
22543         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22544         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22545         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22546         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22547         instead of DECL_DEBUG_EXPR_IS_FROM.
22548         * gimplify.c (gimplify_modify_expr): Likewise.
22549         * tree-cfg.c (verify_expr_location_1): Likewise.
22550         * tree-complex.c (create_one_component_var): Likewise.
22551         * tree-sra.c (create_access_replacement): Likewise.
22552         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22553         (clear_unused_block_pointer): Likewise.
22554         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22555         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22556         * var-tracking.c (var_debug_decl): Likewise.
22557         (track_expr_p): Likewise.
22558         * tree-inline.c (add_local_variables): Likewise.  Set
22559         DECL_HAS_DEBUG_EXPR_P after copying it.
22560         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22561         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22563 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22565         PR bootstrap/56656
22566         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22567         * configure: Regenerate.
22568         * config.in: Regenerate.
22569         * config/i386/i386.md (*movdf_internal): Use
22570         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22571         movd instead of movq mnemonic for interunit moves.
22572         (*movdi_internal): Ditto.
22574 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22576         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22577         (abd<mode>_3): New pattern.
22578         (aba<mode>_3): New pattern.
22579         (fabd<mode>_3): New pattern.
22581 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22583         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22584         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22585         occurrence of REGISTER_PREFIX as its empty string.
22587 2013-03-20  Jeff Law  <law@redhat.com>
22589         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22590         addititional equivalences for equality comparisons between an SSA_NAME
22591         and a constant where the SSA_NAME was set from a widening conversion.
22593 2013-03-20  Walter Lee  <walt@tilera.com>
22595         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22597 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22599         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22600         depending on type attribute.
22601         (*movti_internal): Ditto.
22602         (*movtf_internal): Ditto.
22603         (*movxf_internal): Ditto.
22604         (*movdf_internal): Ditto.
22605         (*movsf_internal): Ditto.
22607 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22609         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22610         maybe_vex for sselog1 and ssemov types.
22611         (*movdi_internal): Reorder operand constraints.
22612         (*movsi_internal): Ditto.  Set prefix attribute to
22613         maybe_vex for sselog1 and ssemov types.
22614         (*movtf_internal): Set prefix attribute to maybe_vex
22615         for sselog1 and ssemov types.
22616         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22617         DImode ssemov types.  Reorder operand constraints.
22618         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22619         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22620         attribute for SImode ssemov types.  Reorder operand constraints.
22622 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22624         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22625         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22627 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22629         * config/rs6000/predicates.md (indexed_address, update_address_mem
22630         update_indexed_address_mem): New predicates.
22631         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22632         attribute for load/store instructions.
22633         * config/rs6000/dfp.md (movsd_store): Likewise.
22634         (movsd_load): Likewise.
22635         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22636         (unnamed HI->DI extend define_insn): Likewise.
22637         (unnamed SI->DI extend define_insn): Likewise.
22638         (unnamed QI->SI extend define_insn): Likewise.
22639         (unnamed QI->HI extend define_insn): Likewise.
22640         (unnamed HI->SI extend define_insn): Likewise.
22641         (unnamed HI->SI extend define_insn): Likewise.
22642         (extendsfdf2_fpr): Likewise.
22643         (movsi_internal1): Likewise.
22644         (movsi_internal1_single): Likewise.
22645         (movhi_internal): Likewise.
22646         (movqi_internal): Likewise.
22647         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22648         attribute for load/store instructions.
22649         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22650         instructions.
22651         (mov<mode>_softfloat): Likewise.
22652         (mov<mode>_hardfloat32): Likewise.
22653         (mov<mode>_hardfloat64): Likewise.
22654         (mov<mode>_softfloat64): Likewise.
22655         (movdi_internal32): Likewise.
22656         (movdi_internal64): Likewise.
22657         (probe_stack_<mode>): Likewise.
22659 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22661         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22662         floating point, and decimal floating point to reload iterator.
22664         * config/rs6000/constraints.md (wl constraint): New constraints to
22665         return FLOAT_REGS if certain options are used to reduce the number
22666         of separate patterns that exist in the file.
22667         (wx constraint): Likewise.
22668         (wz constraint): Likewise.
22670         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22671         -mdebug=reg, print wg, wl, wx, and wz constraints.
22672         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22673         Initialize the reload functions for 64-bit binary/decimal floating
22674         point types.
22675         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22676         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22677         create the buffer on the stack to overcome not having a 32-bit
22678         load and store.
22679         (rs6000_emit_move): Likewise.
22680         (rs6000_secondary_memory_needed_rtx): Likewise.
22681         (rs6000_alloc_sdmode_stack_slot): Likewise.
22682         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22683         via xxlxor, just like DFmode 0.0.
22685         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22686         define as 1 if we are running on a power7 or newer.
22687         (enum r6000_reg_class_enum): Add new constraints.
22689         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22690         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22691         with other moves by using conditional constraits (wg).  Use LFIWZX
22692         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22693         (movsd splitter): Likewise.
22694         (movsd_hardfloat): Likewise.
22695         (movsd_softfloat): Likewise.
22697         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22698         binary and decimal floating point moves.
22699         (fmove_ok): New attributes to combine binary and decimal floating
22700         point moves, and to combine power6x (mfpgpr) moves along normal
22701         floating moves.
22702         (real_value_to_target): Likewise.
22703         (f32_lr): Likewise.
22704         (f32_lm): Likewise.
22705         (f32_li): Likewise.
22706         (f32_sr): Likewise.
22707         (f32_sm): Likewise.
22708         (f32_si): Likewise.
22709         (movsf): Combine binary and decimal floating point moves.  Combine
22710         power6x (mfpgpr) moves with other moves by using conditional
22711         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22712         (mov<mode> for SFmode/SDmode); Likewise.
22713         (SFmode/SDmode splitters): Likewise.
22714         (movsf_hardfloat): Likewise.
22715         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22716         (movsf_softfloat): Likewise.
22717         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22719         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22720         wx and wz constraints.
22722         * config/rs6000/constraints.md (wg constraint): New constraint to
22723         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22725         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22726         constraint.
22728         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22729         -mdebug=reg, print wg, wl, wx, and wz constraints.
22730         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22731         Initialize the reload functions for 64-bit binary/decimal floating
22732         point types.
22733         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22734         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22735         create the buffer on the stack to overcome not having a 32-bit
22736         load and store.
22737         (rs6000_emit_move): Likewise.
22738         (rs6000_secondary_memory_needed_rtx): Likewise.
22739         (rs6000_alloc_sdmode_stack_slot): Likewise.
22740         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22741         via xxlxor, just like DFmode 0.0.
22743         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22744         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22745         with other moves by using conditional constraits (wg).  Use LFIWZX
22746         and STFIWX for loading SDmode on power7.
22747         (movdd splitters): Likewise.
22748         (movdd_hardfloat32): Likewise.
22749         (movdd_softfloat32): Likewise.
22750         (movdd_hardfloat64_mfpgpr): Likewise.
22751         (movdd_hardfloat64): Likewise.
22752         (movdd_softfloat64): Likewise.
22754         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22755         64-bit binary and decimal floating point moves.
22756         (FMOVE64X): Likewise.
22757         (movdf): Combine 64-bit binary and decimal floating point moves.
22758         Combine power6x (mfpgpr) moves with other moves by using
22759         conditional constraits (wg).
22760         (mov<mode> for DFmode/DDmode): Likewise.
22761         (DFmode/DDmode splitters): Likewise.
22762         (movdf_hardfloat32): Likewise.
22763         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22764         (movdf_softfloat32): Likewise.
22765         (movdf_hardfloat64_mfpgpr): Likewise.
22766         (movdf_hardfloat64): Likewise.
22767         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22768         (movdf_softfloat64): Likewise.
22769         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22770         (reload_<mode>_load): Move to later in the file so they aren't in
22771         the middle of the floating point move insns.
22772         (reload_<mode>_store): Likewise.
22774         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22775         constraint.
22777         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22778         constraint if -mdebug=reg.
22779         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22780         Enable using dd reload support if needed.
22782         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22783         binary and decimal floating point moves in rs6000.md.
22784         (movtd_internal): Likewise.
22786         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22787         decimal floating point moves.
22788         (movtf): Likewise.
22789         (movtf_internal): Likewise.
22790         (mov<mode>_internal, TDmode/TFmode): Likewise.
22791         (movtf_softfloat): Likewise.
22792         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22794         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22795         movdi_internal64, using wg constraint for move direct operations.
22796         (movdi_internal64): Likewise.
22798         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22799         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22800         the various virtual registers. Use GPR/FPR first/last values,
22801         instead of hard coding the register numbers.  Print which modes
22802         have reload functions registered.
22803         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22804         settings before/after setting cpu, target and subtarget settings.
22805         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22806         and for secondary reload failures in rs6000_secondary_reload_inner.
22807         (rs6000_secondary_reload_fail): Likewise.
22808         (rs6000_secondary_reload_inner): Likewise.
22810         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22811         macros for first/last GPR and FPR registers.
22812         (LAST_GPR_REGNO): Likewise.
22813         (FIRST_FPR_REGNO): Likewise.
22814         (LAST_FPR_REGNO): Likewise.
22816         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22817         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22818         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22819         (vcond<mode><mode>): Likewise.
22820         (vcondu<mode><mode>): Likewise.
22821         (vector_gtu<mode>): Likewise.
22822         (vector_gte<mode>): Likewise.
22823         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22824         to prevent the compiler from converting DImode operations to TImode.
22825         (ior<mode>3): Likewise.
22826         (and<mode>3): Likewise.
22827         (one_cmpl<mode>2): Likewise.
22828         (nor<mode>3): Likewise.
22829         (andc<mode>3): Likewise.
22831         * config/rs6000/constraints.md (wt constraint): New constraint
22832         that returns VSX_REGS if TImode is allowed in VSX registers.
22834         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22835         constant under VSX.
22837         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22838         similar to TImode, but it is restricted to being in the GPRs.
22840         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22841         TImode to occupy a single VSX register.
22843         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22844         -mvsx-timode for power7/power8.
22845         (power7 cpu): Likewise.
22846         (power8 cpu): Likewise.
22848         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22849         sure that TFmode/TDmode take up two registers if they are ever
22850         allowed in the upper VSX registers.
22851         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22852         registers.
22853         (rs6000_init_hard_regno_mode_ok): Likewise.
22854         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22855         constraint.  Print if LRA is turned on.
22856         (rs6000_option_override_internal): Give an error if -mvsx-timode
22857         and VSX is not enabled.
22858         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22859         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22860         to reg+offset addressing.  Use PTImode when checking offset
22861         addresses for validity.
22862         (reg_offset_addressing_ok_p): Likewise.
22863         (rs6000_legitimate_offset_address_p): Likewise.
22864         (rs6000_legitimize_address): Likewise.
22865         (rs6000_legitimize_reload_address): Likewise.
22866         (rs6000_legitimate_address_p): Likewise.
22867         (rs6000_eliminate_indexed_memrefs): Likewise.
22868         (rs6000_emit_move): Likewise.
22869         (rs6000_secondary_reload): Likewise.
22870         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22871         reloads to fpr registers to continue to use reg+offset addressing,
22872         but 64-bit reloads to altivec registers need reg+reg addressing.
22873         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22874         it.  Treat LO_SUM like a PLUS operation.
22875         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22876         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22877         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22878         registers to share a register with a smaller sized type, since VSX
22879         puts scalars in the upper 64-bits.
22880         (print_operand): Add support for PTImode.
22881         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22882         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22883         registers, but don't have arithmetic support.
22884         (rs6000_memory_move_cost): Add test for VSX.
22885         (rs6000_opt_masks): Add -mvsx-timode.
22887         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22888         for TImode.
22889         (VSs): Likewise.
22890         (VSr): Use wt constraint for TImode.
22891         (VSv): Drop TImode support.
22892         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22893         (vsx_movti_64bit): Likewise.
22894         (vsx_movti_32bit): Likewise.
22895         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22896         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22897         one '?' on the appropriate output constraint.  Do not allow TImode
22898         logical operations on 32-bit systems.
22899         (vsx_ior<mode>3): Likewise.
22900         (vsx_xor<mode>3): Likewise.
22901         (vsx_one_cmpl<mode>2): Likewise.
22902         (vsx_nor<mode>3): Likewise.
22903         (vsx_andc<mode>3): Likewise.
22904         (vsx_concat_<mode>): Likewise.
22905         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22907         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22908         OPTION_MASK_VSX_TIMODE.
22909         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22910         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22912         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22913         (TI2 iterator): New iterator for TImode, PTImode.
22914         (wd mode attribute): Add values for vector types.
22915         (movti_string): Replace TI move operations with operations for TImode
22916         and PTImode.  Add support for TImode being allowed in VSX registers.
22917         (mov<mode>_string, TImode/PTImode): Likewise.
22918         (movti_ppc64): Likewise.
22919         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22920         (TI mode splitters): Likewise.
22922         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22923         constraint.
22925 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22927         PR tree-optimization/56355
22928         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22929         Also handle integers with undefined overflow.
22931 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22932             Maciej W. Rozycki  <macro@codesourcery.com>
22933             Tom de Vries  <tom@codesourcery.com>
22934             Nathan Sidwell  <nathan@codesourcery.com>
22935             Iain Sandoe  <iain@codesourcery.com>
22936             Nathan Froyd  <froydnj@codesourcery.com>
22937             Chao-ying Fu  <fu@mips.com>
22939         * doc/extend.texi: (micromips, nomicromips, nocompression):
22940         Document new function attributes.
22941         * doc/invoke.texi (minterlink-compressed, mmicromips,
22942         m14k, m14ke, m14kec): Document new options.
22943         (minterlink-mips16): Update documentation.
22944         * doc/md.texi (ZC, ZD): Document new constraints.
22945         * configure.ac (gcc_cv_as_micromips): Check if linker
22946         supports the .set micromips directive.
22947         * configure: Regenerate.
22948         * config.in: Regenerate.
22949         * config/mips/mips-tables.opt: Regenerate.
22950         * config/mips/micromips.md: New file.
22951         * constraints.md (ZC, ZD): New constraints.
22952         * config/mips/predicates.md (movep_src_register): New predicate.
22953         (movep_src_operand): New predicate.
22954         (non_volatile_mem_operand): New predicate.
22955         * config/mips/mips.md (multimem): New type.
22956         (length): Differentiate between 17-bit and 18-bit branch offsets.
22957         (MOVEP1, MOVEP2): New mode iterator.
22958         (mov_<load>l): Use ZC constraint.
22959         (mov_<load>r): Likewise.
22960         (mov_<store>l): Likewise.
22961         (mov_<store>r): Likewise.
22962         (*branch_equality<mode>_inverted): Add microMIPS support.
22963         (*branch_equality<mode>): Likewise.
22964         (*jump_absolute): Likewise.
22965         (indirect_jump_<mode>): Likewise.
22966         (tablejump_<mode>): Likewise.
22967         (<optab>_internal): Likewise.
22968         (sibcall_internal): Likewise.
22969         (sibcall_value_internal): Likewise.
22970         (prefetch): Use constraint ZD.
22971         * config/mips/mips.opt (minterlink-compressed): New option.
22972         (minterlink-mips16): Now an alias for minterlink-compressed.
22973         (mmicromips): New option.
22974         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22975         (compare_and_swap_12): Likewise.
22976         (sync_add<mode>): Likewise.
22977         (sync_<optab>_12): Likewise.
22978         (sync_old_<optab>_12): Likewise.
22979         (sync_new_<optab>_12): Likewise.
22980         (sync_nand_12): Likewise.
22981         (sync_old_nand_12): Likewise.
22982         (sync_new_nand_12): Likewise.
22983         (sync_sub<mode>): Likewise.
22984         (sync_old_add<mode>): Likewise.
22985         (sync_old_sub<mode>): Likewise.
22986         (sync_new_add<mode>): Likewise.
22987         (sync_new_sub<mode>): Likewise.
22988         (sync_<optab><mode>): Likewise.
22989         (sync_old_<optab><mode>): Likewise.
22990         (sync_new_<optab><mode>): Likewise.
22991         (sync_nand<mode>): Likewise.
22992         (sync_old_nand<mode>): Likewise.
22993         (sync_new_nand<mode>): Likewise.
22994         (sync_lock_test_and_set<mode>): Likewise.
22995         (test_and_set_12): Likewise.
22996         (atomic_compare_and_swap<mode>): Likewise.
22997         (atomic_exchange<mode>_llsc): Likewise.
22998         (atomic_fetch_add<mode>_llsc): Likewise.
22999         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23000         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23001         (umips_save_restore_pattern_p): Likewise.
23002         (umips_load_store_pair_p): Likewise.
23003         (umips_output_load_store_pair): Likewise.
23004         (umips_movep_target_p): Likewise.
23005         (umips_12bit_offset_address_p): Likewise.
23006         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23007         (mips_base_mips16): Rename this...
23008         (mips_base_compression_flags): ...to this. Update all uses.
23009         (mips_attribute_table): Add micromips, nomicromips and nocompression.
23010         (mips_mips16_decl_p): Delete.
23011         (mips_nomips16_decl_p): Delete.
23012         (mips_get_compress_on_flags): New function.
23013         (mips_get_compress_off_flags): New function.
23014         (mips_get_compress_mode): New function.
23015         (mips_get_compress_on_name): New function.
23016         (mips_get_compress_off_name): New function.
23017         (mips_insert_attributes): Support multiple compression types.
23018         (mips_merge_decl_attributes): Likewise.
23019         (umips_12bit_offset_address_p): New function.
23020         (mips_start_function_definition): Emit .set micromips directive.
23021         (mips_call_may_need_jalx_p): New function.
23022         (mips_function_ok_for_sibcall): Add microMIPS support.
23023         (mips_print_operand_punctuation): Support short delay slots and
23024         compact jumps.
23025         (umips_swm_mask, umips_swm_encoding): New.
23026         (umips_build_save_restore): New function.
23027         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23028         (was_mips16_p): Remove.
23029         (old_compression_mode): New.
23030         (mips_set_compression_mode): New function.
23031         (mips_set_current_function): Add microMIPS support.
23032         (mips_option_override): Likewise.
23033         (umips_save_restore_pattern_p): New function.
23034         (umips_output_save_restore): New function.
23035         (umips_load_store_pair_p_1): New function.
23036         (umips_load_store_pair_p): New function.
23037         (umips_output_load_store_pair_1): New function.
23038         (umips_output_load_store_pair): New function.
23039         (umips_movep_target_p) New function.
23040         (mips_prepare_pch_save): Add microMIPS support.
23041         * config/mips/mips.h (TARGET_COMPRESSION): New.
23042         (TARGET_CPU_CPP_BUILTINS): Update macro
23043         to use new compression flags and to support microMIPS.
23044         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23045         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23046         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23047         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23048         (ASM_SPEC): Support mmicromips and mno-micromips.
23049         (M16STORE_REG_P): New macro.
23050         (MIPS_CALL): Support TARGET_MICROMIPS.
23051         (MICROMIPS_J): New macro.
23052         (mips_base_mips16): Rename this...
23053         (mips_base_compression_flags): ...to this.
23054         (UMIPS_12BIT_OFFSET_P): New macro.
23055         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23056         (MULTILIB_DIRNAMES): Likewise.
23057 2013-03-20  Richard Biener  <rguenther@suse.de>
23059         PR tree-optimization/56661
23060         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
23061         the result does not have to be distinct.
23063 2013-03-20  Richard Biener  <rguenther@suse.de>
23065         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
23066         remap_gimple_op_r.
23068 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23069             Steven Bosscher  <steven@gcc.gnu.org>
23071         PR rtl-optimization/56605
23072         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
23074 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23076         PR bootstrap/56656
23077         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
23078         that require movd instead of movq.
23080 2013-03-20  Richard Biener  <rguenther@suse.de>
23082         * tree-ssa-structalias.c (struct variable_info): Add pointer
23083         to the first field of an aggregate with sub-vars.  Make
23084         this and the pointer to the next subfield its ID.
23085         (vi_next): New function.
23086         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23087         storedanything_id, integer_id): Increment by one.
23088         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23089         get_call_clobber_vi): Adjust.
23090         (solution_set_expand): Simplify and speedup.
23091         (solution_set_add): Inline into ...
23092         (set_union_with_increment): ... this.  Adjust accordingly.
23093         (do_sd_constraint): Likewise.
23094         (do_ds_constraint): Likewise.
23095         (do_complex_constraint): Simplify.
23096         (build_pred_graph): Adjust.
23097         (solve_graph): Likewise.  Simplify and speedup.
23098         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23099         get_constraint_for_component_ref, get_constraint_for_1,
23100         first_vi_for_offset, first_or_preceding_vi_for_offset,
23101         create_function_info_for, create_variable_info_for_1,
23102         create_variable_info_for, intra_create_variable_infos): Adjust.
23103         (init_base_vars): Push NULL for ID zero.
23104         (compute_points_to_sets): Adjust.
23106 2013-03-20  Richard Biener  <rguenther@suse.de>
23108         * cfgloop.c (verify_loop_structure): Streamline and avoid
23109         ICEing on corrupt loop tree.
23110         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23111         loop tree.
23113 2013-03-20  Richard Biener  <rguenther@suse.de>
23115         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23116         check whether an SSA update is needed.
23118 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23120         * config/mips/constraints.md (T): Rename to...
23121         (Yf): ...this.
23122         (U): Rename to...
23123         (Yd): ...this.
23124         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23125         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23127 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23129         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23130         (*subsi3_carryin_uxtw): Likewise.
23132 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23134         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23135         (*rorsi3_insn_uxtw): Likewise.
23137 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23139         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23140         (*extrsi5_insn_uxtw): Likewise.
23142 2013-03-19  Richard Biener  <rguenther@suse.de>
23144         PR tree-optimization/56273
23145         * passes.c (init_optimization_passes): Move second VRP after DOM.
23147 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23149         * config/i386/i386.md (*movti_internal): Merge from
23150         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23151         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23152         nox64 isa attributes.
23154 2013-03-18  Richard Biener  <rguenther@suse.de>
23156         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23157         (unite): Likewise.
23158         (merge_node_constraints): Likewise.
23159         (build_succ_graph): Likewise.
23160         (valid_graph_edge): Inline into single caller.
23161         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23162         and cache varinfo.
23163         (scc_visit): Fix formatting and variable use.
23164         (do_sd_constraint): Use gcc_checking_assert.
23165         (do_ds_constraint): Likewise.
23166         (do_complex_constraint): Likewise.
23167         (condense_visit): Likewise.  Cleanup.
23168         (dump_pred_graph): New function.
23169         (perform_var_substitution): Dump the pred-graph before
23170         variable substitution.
23171         (find_equivalent_node): Use gcc_checking_assert.
23172         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23174 2013-03-18  Richard Biener  <rguenther@suse.de>
23176         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23177         Remove cond_expr_stmt_list argument and do not gimplify the
23178         built expression.
23179         (vect_loop_versioning): Adjust.
23180         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23181         Cleanup to use less temporaries.
23182         (vect_create_data_ref_ptr): Cleanup.
23184 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23186         PR tree-optimization/56635
23187         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23188         require types_compatible_p types.
23190 2013-03-18  Nick Clifton  <nickc@redhat.com>
23192         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23193         spurious backslash.
23195         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23196         Add missing line to comment describing function.
23198 2013-03-18  Richard Biener  <rguenther@suse.de>
23200         PR tree-optimization/56210
23201         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23202         Handle string / character search functions.
23203         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23205 2013-03-18  Richard Biener  <rguenther@suse.de>
23207         PR middle-end/56483
23208         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23209         and implement properly.
23210         * gimple.h (gimple_cond_single_var_p): Remove.
23212 2013-03-18  Richard Biener  <rguenther@suse.de>
23214         * tree-data-ref.h (find_data_references_in_loop): Declare.
23215         * tree-data-ref.c (get_references_in_stmt): Use a stack
23216         vector pre-allocated in the callers.
23217         (find_data_references_in_stmt): Adjust.
23218         (graphite_find_data_references_in_stmt): Likewise.
23219         (create_rdg_vertices): Likewise.
23220         (find_data_references_in_loop): Export.
23221         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23222         Compute dependences here...
23223         (vect_analyze_data_refs): ...not here.  When we encounter
23224         a non-vectorizable data reference in basic-block vectorization
23225         truncate the data reference vector.  Do not bother to
23226         fixup data-dependence information for gather loads.
23227         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23228         of data references, as reported.
23230 2013-03-18  Richard Biener  <rguenther@suse.de>
23232         PR tree-optimization/3713
23233         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23234         has_constants and expr.
23235         (stmt_has_constants): Properly valueize SSA names when deciding
23236         whether the stmt has constants.
23238 2013-03-18  Richard Biener  <rguenther@suse.de>
23240         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23241         whole function when there is nothing to do.
23242         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23243         * tree-vectorizer.c (vectorize_loops): Update virtual and
23244         loop-closed SSA once.
23245         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23247 2013-03-18  Richard Biener  <rguenther@suse.de>
23249         PR middle-end/56113
23250         * domwalk.c (bb_postorder): New global static.
23251         (cmp_bb_postorder): New function.
23252         (walk_dominator_tree): Replace scheme imposing an order for
23253         visiting dominator sons by one sorting them at the time they
23254         are pushed on the stack.
23256 2013-03-18  Richard Biener  <rguenther@suse.de>
23258         PR tree-optimization/39326
23259         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23260         (struct mem_ref): Replace mem member with ao_ref typed member.
23261         (MEM_ANALYZABLE): Adjust.
23262         (memref_eq): Likewise.
23263         (mem_ref_alloc): Likewise.
23264         (gather_mem_refs_stmt): Likewise.
23265         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23266         (execute_sm_if_changed_flag_set): Adjust.
23267         (execute_sm): Likewise.
23268         (ref_always_accessed_p): Likewise.
23269         (refs_independent_p): Likewise.
23270         (can_sm_ref_p): Likewise.
23272 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23274         PR c/56566
23275         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23276         return 1 even for !unsignedp.
23278 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23280         * config/i386/i386.md (isa): Add x64 and nox64.
23281         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23282         (*pushtf): Enable *roF alternative for x64 isa only.
23283         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23284         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23285         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23286         (*movtf_internal): Merge from *movtf_internal_rex64 and
23287         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23288         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23289         nox64 isa attributes.
23290         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23291         nox64 isa attributes.
23292         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23294 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23296         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23298 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23300         PR target/56640
23301         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23302         class names.  Remove trailing comma after "ALL_REGS".
23304 2013-03-16  Jan Hubicka  <jh@suse.cz>
23306         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23307         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23308         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23309         of cgraph_get_create_node.
23310         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23312 2013-03-16  Jason Merrill  <jason@redhat.com>
23314         PR debug/49090
23315         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23316         with DW_AT_default_value.
23318 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23320         * BASE-VER: Set to 4.9.0.
23322 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23324         PR target/56619
23325         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23326         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23327         Document _x* TSX intrinsics.
23329 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23330             David Holsgrove  <david.holsgrove@xilinx.com>
23332         * configure.ac: Add MicroBlaze TLS support detection.
23333         * configure: Regenerate.
23334         * config/microblaze/microblaze-protos.h
23335         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23336         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23337         * config/microblaze/microblaze.c (microblaze_address_type): Add
23338         ADDRESS_TLS and tls_reloc address types.
23339         (microblaze_address_info): Add tls_reloc.
23340         (TARGET_HAVE_TLS): Define.
23341         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23342          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23343          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23344         load_tls_operand,  microblaze_call_tls_get_addr,
23345         microblaze_legitimize_tls_address): New functions.
23346         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23347         (get_base_reg): Use microblaze_tls_symbol_p.
23348         (microblaze_classify_address): Handle TLS.
23349         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23350         label_mentioned_p and microblaze_tls_referenced_p.
23351         (microblaze_legitimize_address): Handle TLS.
23352         (microblaze_address_insns): Handle ADDRESS_TLS.
23353         (pic_address_needs_scratch): Handle TLS.
23354         (print_operand_address): Handle TLS.
23355         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23356         (microblaze_expand_move): Handle TLS.
23357         (microblaze_legitimate_constant_p): Check
23358         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23359         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23360         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23361         (PIC_OFFSET_TABLE_REGNUM): Set.
23362         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23363         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23364         (addsi3, movsi_internal2, movdf_internal): Update constraints
23365         * config/microblaze/predicates.md (arith_plus_operand): Define
23366         (move_operand): Redefine as move_src_operand,
23367         check microblaze_tls_referenced_p.
23369 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23371         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23372         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23374 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23376         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23377         CC mode for AND.
23379 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23381         PR tree-optimization/53265
23382         * common.opt (Waggressive-loop-optimizations): New option.
23383         * tree-ssa-loop-niter.c: Include tree-pass.h.
23384         (do_warn_aggressive_loop_optimizations): New function.
23385         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23386         if number_of_latch_executions returned constant.
23387         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23388         early.  If number_of_latch_executions returned constant, set
23389         nb_iterations_upper_bound back to it.
23390         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23391         field.
23392         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23393         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23395         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23396         (MULTILIB_OSDIRNAMES): Set.
23397         * genmultilib: If defaultosdirname doesn't start with :: , set
23398         defaultosdirname2 instead, clear it and emit two . multilib_raw
23399         entries instead of just one.
23401 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23403         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23404         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23405         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23406         (SUBTARGET_OVERRIDE_OPTIONS): New.
23408 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23410         PR target/49880
23411         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23412         (musermode): Convert to Var(TARGET_USERMODE).
23413         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23414         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23415         * config/sh/sh.c (sh_option_override): Use
23416         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23417         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23418         condition.
23419         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23420         TARGET_SH4.
23421         (udivsi3_i4_single, divsi3_i4_single): Use
23422         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23424 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23426         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23427         default setting.
23429 2013-03-13  Richard Biener  <rguenther@suse.de>
23431         PR tree-optimization/56608
23432         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23433         calls when vectorizing basic-blocks.
23435 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23437         PR plugins/45078
23438         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23439         tm_file.
23441 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23443         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23445 2013-03-11  Jan Hubicka  <jh@suse.cz>
23447         PR lto/56557
23448         * lto-streamer-out.c (output_symbol_p): Skip references from
23449         constructors of external variables.
23451 2013-03-11  Jan Hubicka  <jh@suse.cz>
23453         PR middle-end/56571
23454         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23455         from pseudos.
23456         * emit-rtl.c (verify_rtx_sharing): Likewise.
23457         (copy_insn_1): Likewise.
23458         * rtl.c (copy_rtx): Likewise.
23460 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23462         PR target/56591
23463         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23464         output_operand_lossage message.
23466 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23468         PR target/56470
23469         * arm.c (shift_op): Validate RTL pattern on the fly.
23470         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23471         the RTL.
23473 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23475         PR target/56347
23476         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23477         new call patterns that clobber %fr12.
23478         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23479         split and postreload patterns.
23480         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23481         registers %fr12 and %fr12R as call used.
23483 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23485         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23486         (canon_address, record_store, replace_read, check_mem_read_rtx,
23487         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23488         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23489         rest_of_handle_dse): Likewise.
23491 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23493         PR middle-end/56524
23494         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23495         Add base_optabs.
23496         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23497         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23498         (save_optabs_if_changed): Replace with...
23499         (init_tree_optimization_optabs): ...this.
23500         * optabs.c (save_optabs_if_changed): Rename to...
23501         (init_tree_optimization_optabs): ...this.  Take the optimization node
23502         as argument.  Do nothing if the base optabs are already correct.
23503         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23504         to recompute optabs.
23505         * function.h (function): Remove optabs field.
23506         * function.c (invoke_set_current_function_hook): Call
23507         init_tree_optimization_optabs.  Use the result to initialize
23508         this_fn_optabs.
23510 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23512         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23513         if GTMA_HAS_NO_INSTRUMENTATION.
23514         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23515         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23516         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23517         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23518         GTMA_HAS_NO_INSTRUMENTATION.
23520 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23522         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23523         libasan_preinit.o.
23525 2013-03-08  Marek Polacek  <polacek@redhat.com>
23526             Jakub Jelinek  <jakub@redhat.com>
23528         PR tree-optimization/56478
23529         * predict.c (is_comparison_with_loop_invariant_p): Change the
23530         type of loop_step to tree.
23531         (predict_loops): Adjust.
23532         (predict_iv_comparison): Perform the computations on double_ints.
23534 2013-03-08  Richard Biener  <rguenther@suse.de>
23536         PR tree-optimization/56570
23537         * tree-cfg.c (verify_expr_location_1): Verify locations for
23538         DECL_DEBUG_EXPR.
23539         * tree-sra.c (create_access_replacement): Strip locations
23540         from DECL_DEBUG_EXPRs.
23542 2013-03-08  Richard Biener  <rguenther@suse.de>
23544         * tree-inline.c (expand_call_inline): Do not associate
23545         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23546         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23548 2013-03-08  Richard Biener  <rguenther@suse.de>
23550         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23551         or block changes with -Og.  Fix for location / block encoding
23552         changes and PHI arguments with locations.
23554 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23556         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23557         for all counters.
23558         (struct output_info): Likewise.
23559         (register_overhead): Remove bad gcc_assert.
23560         (bitmap_find_bit): If there is only a single bitmap element, do not
23561         count a miss as a search.
23562         (print_statistics): Update for counter type changes.
23563         (dump_bitmap_statistics): Likewise.  Print headers such that they
23564         are properly lined up with the printed counters.
23566 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23568         PR tree-optimization/56559
23569         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23570         check that it has only a single use.
23572 2013-03-07  Richard Biener  <rguenther@suse.de>
23574         * doc/invoke.texi (fwhole-program): Discourage use in combination
23575         with -flto.
23577 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23579         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23581         PR tree-optimization/56539
23582         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23583         instead of GSI_CONTINUE_LINKING as last argument to
23584         force_gimple_operand_gsi.  Adjust function comment.
23586         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23587         aarch64-cores.def.
23589         PR middle-end/56548
23590         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23591         promoted mode, convert the result back to the original mode.
23593 2013-03-06  Richard Biener  <rguenther@suse.de>
23595         PR middle-end/56294
23596         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23597         (insert_updated_phi_nodes_compare_uids): New function.
23598         (update_ssa): Sort symbols_to_rename after UID before
23599         traversing it to insert PHI nodes.
23601 2013-03-06  Richard Biener  <rguenther@suse.de>
23603         PR middle-end/50494
23604         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23605         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23607         Revert
23608         2013-02-13  Richard Biener  <rguenther@suse.de>
23610         PR lto/50494
23611         * varasm.c (output_constant_def_1): Get the decl representing
23612         the constant as argument.
23613         (output_constant_def): Wrap output_constant_def_1.
23614         (make_decl_rtl): Use output_constant_def_1 with the decl
23615         representing the constant.
23616         (build_constant_desc): Optionally re-use a decl already
23617         representing the constant.
23618         (tree_output_constant_def): Adjust.
23620 2013-03-06  Joey Ye  <joey.ye@arm.com>
23622         PR lto/50293
23623         * gcc.c (convert_white_space): New function.
23624         (main): Handles white space in function name.
23626 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23628         PR target/56529
23629         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23630         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23631         to SH_DIV_CALL_TABLE for TARGET_SH2.
23632         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23633         list.
23634         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23635         call-table options.
23637 2013-03-05  Sterling Augustine  <saugustine@google.com>
23638             Cary Coutant  <ccoutant@google.com>
23640         PR debug/55364
23641         * dwarf2out.c (resolve_addr): Don't call
23642         remove_loc_list_addr_table_entries a second time for the same
23643         expression.
23645 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23647         PR debug/56510
23648         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23649         (avoid_complex_debug_insns): New function.
23650         (expand_debug_locations): Call it.
23652         PR rtl-optimization/56484
23653         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23654         lifetimes of hard registers on small register class machines.
23656 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23658         * config/microblaze/microblaze-protos.h: Rename
23659         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23660         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23661         fast_interrupt.
23662         (microblaze_fast_interrupt_function_p): New function.
23663         (microblaze_is_interrupt_handler): Rename to
23664         microblaze_is_interrupt_variant and add fast_interrupt check.
23665         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23666         (save_restore_insns): Likewise.
23667         (compute_frame_size): Likewise.
23668         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23669         (microblaze_globalize_label): Likewise.
23670         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23671         * config/microblaze/microblaze.md: Use wrapper
23672         microblaze_is_interrupt_variant.
23674 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23676         * sdbout.c (sdbout_one_type): Switch to current function's section
23677         supporting cold/hot.
23679 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23681         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23682         -mxl-reorder.
23684 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23686         PR middle-end/56461
23687         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23688         if VALGRIND_GET_VBITS is defined, temporarily make object
23689         memory all defined, and restore previous valgrind addressability
23690         and definability afterwards.  Free this_object at the end.
23692         PR middle-end/56461
23693         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23694         right before calling lra_create_live_ranges, also call it
23695         when clearing live_p.  Only call lra_clear_live_ranges
23696         at the end if live_p.
23698         PR middle-end/56461
23699         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23701 2013-03-05  Richard Biener  <rguenther@suse.de>
23703         PR tree-optimization/56521
23704         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23705         value-id.
23707 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23709         PR c++/55135
23710         * except.h (remove_unreachable_eh_regions): New prototype.
23711         * except.c (remove_eh_handler_splicer): New function, split out
23712         of remove_eh_handler.
23713         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23714         warning about running it on many EH regions one at a time.
23715         (remove_unreachable_eh_regions_worker): New function, walk the
23716         EH tree in depth-first order and remove non-marked regions.
23717         (remove_unreachable_eh_regions): New function.
23718         * tree-eh.c (mark_reachable_handlers): New function, split out
23719         from remove_unreachable_handlers.
23720         (remove_unreachable_handlers): Use mark_reachable_handlers and
23721         remove_unreachable_eh_regions.
23722         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23723         and remove_unreachable_eh_regions.
23725 2013-03-05  Richard Biener  <rguenther@suse.de>
23727         PR middle-end/56525
23728         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23729         not freeing them until the end.
23731 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23733         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23735 2013-03-05  Richard Biener  <rguenther@suse.de>
23737         PR tree-optimization/56270
23738         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23739         of loads after scheduling an SLP instance.
23741 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23743         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23744         tic6x.exp.
23745         (check_gcc_parallelize): Run guality.exp as a separate job from
23746         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23747         struct-layout-1.exp with stackalign.exp.
23749         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23751         PR middle-end/56461
23752         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23753         load_index sbitmap even if some bit in it isn't set.
23755         PR middle-end/56461
23756         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23757         (discover_iteration_bound_by_body_walk): Change queues to
23758         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23759         spelling in comment.  Call safe_push on queues[bound_index] directly.
23760         Release queues[queue_index] in every iteration unconditionally.
23761         Release bounds vector.
23763         PR middle-end/56461
23764         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23765         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23766         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23767         inner_phis vector.
23769 2013-03-05  Richard Biener  <rguenther@suse.de>
23771         PR lto/56515
23772         * tree-inline.c (remap_blocks_to_null): New function.
23773         (expand_call_inline): When expanding a call stmt without
23774         an associated block inline remap all callee blocks to NULL.
23776 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23778         PR rtl-optimization/56494
23779         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23780         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23781         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23783         PR middle-end/56461
23784         * sel-sched-ir.c (free_sched_pools): Release
23785         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23786         if succs_info_pool.max_top isn't -1.
23788         PR bootstrap/56509
23789         * opts.c (opts_obstack, opts_concat): Moved to...
23790         * opts-common.c (opts_obstack, opts_concat): ... here.
23792 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23794         PR middle-end/56461
23795         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23797 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23799         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23800         all appropriate places.
23802 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23804         PR tree-optimization/56424
23805         * ipa-split.c (split_function): Do not set the RSO flag if result is
23806         not by reference and its type is a register type.
23808 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23810         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23811         (microblaze_legitimate_pic_operand): Likewise
23812         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23813         new function microblaze_legitimate_pic_operand
23814         * config/microblaze/microblaze-protos.h
23815         (microblaze_legitimate_pic_operand): Declare.
23817 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23819         * config/microblaze/predicates.md (call_insn_simple_operand):
23820         New predicate for supported rtx code types.
23821         * config/microblaze/microblaze.md (call_internal1): Use
23822         call_insn_simple_operand predicate.
23824 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23826         PR middle-end/56461
23827         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23828         partitions.ordered_remove.
23830         PR middle-end/56461
23831         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23832         vec_oprnds0.create (1) for modifier == NONE.
23834         PR middle-end/56461
23835         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23836         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23837         vec_oprnds1 right before pushing anything to it for
23838         scalar_shift_arg.
23840         PR middle-end/56461
23841         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23842         set nbbs to 0 instead of having separate code path.
23843         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23844         instead of false as last argument if returning NULL.
23846 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23848         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23849         the attribute is now called "target" instead of "option".
23850         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23851         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23852         attribute/pragma name for TARGET_OPTION_VALID_P and
23853         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23854         * doc/tm.texi: Regenerated.
23856 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23858         * config/microblaze/microblaze.c:
23859         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23860         * config/microblaze/microblaze.h: Add -mxl-reorder to
23861         DRIVER_SELF_SPECS.
23862         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23863         instructions emitted if TARGET_REORDER.
23864         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23865         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23866         separately.
23868 2013-03-01  Xinliang David Li  <davidxl@google.com>
23870         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23871         walk length.
23873 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23875         PR middle-end/56461
23876         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23877         vector even when returning true.  Fix up function comment formatting.
23879         PR middle-end/56461
23880         * ira-build.c (ira_loop_nodes_count): New variable.
23881         (create_loop_tree_nodes): Initialize it.
23882         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23884         PR middle-end/56461
23885         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23886         method on dr_chain and result_chain.
23887         * tree-vect-stmts.c (vectorizable_store): Only call
23888         result_chain.create if j == 0.
23890         PR middle-end/56461
23891         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23892         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23893         before overwriting it.
23895 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23897         * doc/extended.texi (C Extensions): Change order in @menu
23898         to match @node.
23899         (Other MIPS Built-in Functions): Move last MIPS entry before
23900         "picoChip Built-in Functions".
23901         (SH Built-in Functions): Move after RX Built-in Functions.
23902         * doc/gcc.texi (Introduction): Change order in @menu
23903         to match @node.
23904         * doc/md.texi (Constraints): Ditto.
23905         * gty.texi (Type Information): Ditto.
23906         (User-provided marking routines for template types): Make
23907         subsection.
23908         * doc/invoke.texi (AArch64 Options): Move before
23909         "Adapteva Epiphany Options".
23911 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23912             Jakub Jelinek  <jakub@redhat.com>
23914         PR sanitizer/56454
23915         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23916         no_address_safety_analysis attribute.
23917         * doc/extend.texi (no_address_safety_attribute): Rename to
23918         no_sanitize_address attribute, mention no_address_safety_analysis
23919         attribute as deprecated alias.
23921 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23923         PR middle-end/56461
23924         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23925         type to vec<vec<tree> > *.
23926         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23927         to be vec<tree> instead of vec<tree> *, set vec_defs
23928         to vNULL and call vec_defs.create (number_of_vects), adjust other
23929         uses of vec_defs.
23930         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23931         vectorizable_condition): Adjust vect_get_slp_defs callers.
23933 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23935         * config/aarch64/aarch64.c
23936         (aarch64_float_const_representable): Remove unused variable.
23938 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23940         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23942 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23944         * config/aarch64/aarch64-builtins.c
23945         (aarch64_init_simd_builtins): Make static.
23947 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23949         * config/aarch64/aarch64.c
23950         (aarch64_simd_make_constant): Make static.
23952 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23954         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23955         with no initialization to the RHS of debug statements.
23957 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23959         PR tree-optimization/56294
23960         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23961         Adjust dumping.
23962         (get_access_replacement): Do not call create_access_replacement.
23963         Assert a replacement exists.
23964         (get_repl_default_def_ssa_name): Create the replacement declaration
23965         itself.
23967 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23969         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23970         final_end_function.
23972 2013-02-28  Marek Polacek  <polacek@redhat.com>
23974         PR rtl-optimization/56466
23975         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23976         if we're changing a loop.
23977         (peel_loops_completely): Likewise.
23979 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23981         PR c++/55813
23982         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23984 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23986         PR target/56445
23987         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23988         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23989         INTX_FTYPE_FX, FX_FTYPE_INTX.
23990         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23992 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23994         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23995         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23996         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23997         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23998         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23999         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
24000         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
24001         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
24002         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
24003         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
24004         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
24005         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
24006         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
24007         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
24008         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
24009         (avrxmega6): Increase max flash segments from 5 to 6.
24010         * config/avr/t-multilib: Regenerate.
24011         * config/avr/avr-tables.opt: Regenerate.
24012         * doc/avr-mmcu.texi: Regenerate.
24014 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24016         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
24017         (avr_device_to_arch): Rename to avr_device_to_ld.
24018         (avr_device_to_as): New prototype.
24019         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
24020         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
24021         * config/avr/driver-avr.c (avr_device_to_as): New.
24022         (avr_device_to_arch): Rename to avr_device_to_ld.
24024 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24026         PR middle-end/56461
24027         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
24028         method on dr_chain and result_chain.
24030         PR middle-end/56461
24031         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
24032         pointer_set_destroy on not_executed_last_iteration.
24034         PR middle-end/56461
24035         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
24037         PR middle-end/56461
24038         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
24039         FOR_EACH_DEFINED_FUNCTION when freeing state.
24041         PR middle-end/56461
24042         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
24043         pool_free.
24044         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
24045         overwriting it.
24047         PR middle-end/56461
24048         * ipa-cp.c (decide_whether_version_node): Call vec_free on
24049         known_aggs[i].items and release known_aggs vector.
24051         PR middle-end/56461
24052         * ipa-reference.c (propagate): Free node_info even for alias nodes.
24054 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24056         * config/microblaze/microblaze.c (microblaze_emit_compare):
24057         Use xor for EQ/NE comparisions.
24058         * config/microblaze/microblaze.md (cstoresf4): Add constraints
24059         (cbranchsf4): Adjust operator to comparison_operator.
24061 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24063         PR middle-end/56461
24064         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
24065         vector.
24066         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
24067         vec_safe_push, always update *slot.
24068         (redirect_edge_var_map_clear): Use vec_free.
24069         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
24070         (free_var_map_entry): Use vec_free.
24071         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
24072         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
24074 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
24076         PR middle-end/45472
24077         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
24078         when the may_trap_p bit of the exprs being merged differs.
24079         Reorder tests for speculativeness in the logical and operator.
24081 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24083         * incpath.c (add_standard_paths): Use reconcat instead of concat
24084         where appropriate and avoid leaking memory.
24086         * opts.h: Include obstack.h.
24087         (opts_concat): New prototype.
24088         (opts_obstack): New declaration.
24089         * opts.c (opts_concat): New function.
24090         (opts_obstack): New variable.
24091         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24092         (finish_options): Use opts_concat instead of concat
24093         and XOBNEWVEC instead of XNEWVEC.
24094         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24095         generate_option): Likewise.
24096         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24097         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24099         PR target/56455
24100         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24101         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24103 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24105         PR middle-end/56461
24106         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24108 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24110         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24111         (arm_block_move_unaligned_straight): Likewise.
24112         (arm_adjust_block_mem): Likewise.
24114 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24116         PR target/48901
24117         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24118         temp, cond and label.
24119         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24121         PR target/52500
24122         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24123         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24125         PR target/52501
24126         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24127         of prologue/epilogue functions.
24129         PR target/52550
24130         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24131         Remove unused variable cfa_offset.
24132         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24134         PR target/54639
24135         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24136         type promotion to unsigned.
24138         PR target/54640
24139         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24140         for HOST_WIDE_INT of 32 bit / same size as int.
24141         (arm_block_move_unaligned_straight): Likewise.
24142         (arm_adjust_block_mem): Likewise.
24144         PR target/54662
24145         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24146         ALL_CFLAGS.
24148 2013-02-26  Marek Polacek  <polacek@redhat.com>
24150         PR tree-optimization/56426
24151         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24153 2013-02-26  Richard Biener  <rguenther@suse.de>
24155         PR target/56444
24156         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24157         unused variable loops.
24159 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24161         PR tree-optimization/56448
24162         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24163         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24164         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24165         later operands of the references, or even first operand for
24166         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24168         PR tree-optimization/56443
24169         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24170         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24171         to type_for_mode langhook.
24173 2013-02-25  Matt Turner  <mattst88@gmail.com>
24175         * doc/invoke.texi: Document r4700.
24177 2013-02-25  Richard Biener  <rguenther@suse.de>
24179         PR tree-optimization/56175
24180         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24181         split out from ...
24182         (simplify_bitwise_binary): ... here.  Also guard the conversion
24183         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24185 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24187         Revert:
24188         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24189             Maciej W. Rozycki  <macro@codesourcery.com>
24190             Tom de Vries  <tom@codesourcery.com>
24191             Nathan Sidwell  <nathan@codesourcery.com>
24192             Iain Sandoe  <iain@codesourcery.com>
24193             Nathan Froyd  <froydnj@codesourcery.com>
24194             Chao-ying Fu  <fu@mips.com>
24196         * doc/extend.texi: (micromips, nomicromips, nocompression):
24197         Document new function attributes.
24198         * doc/invoke.texi (minterlink-compressed, mmicromips,
24199         m14k, m14ke, m14kec): Document new options.
24200         (minterlink-mips16): Update documentation.
24201         * doc/md.texi (ZC, ZD): Document new constraints.
24202         * configure.ac (gcc_cv_as_micromips): Check if linker
24203         supports the .set micromips directive.
24204         * configure: Regenerate.
24205         * config.in: Regenerate.
24206         * config/mips/mips-tables.opt: Regenerate.
24207         * config/mips/micromips.md: New file.
24208         * constraints.md (ZC, AD): New constraints.
24209         * config/mips/predicates.md (movep_src_register): New predicate.
24210         (movep_src_operand): New predicate.
24211         (non_volatile_mem_operand): New predicate.
24212         * config/mips/mips.md (multimem): New type.
24213         (length): Differentiate between 17-bit and 18-bit branch offsets.
24214         (MOVEP1, MOVEP2): New mode iterator.
24215         (mov_<load>l): Use ZC constraint.
24216         (mov_<load>r): Likewise.
24217         (mov_<store>l): Likewise.
24218         (mov_<store>r): Likewise.
24219         (*branch_equality<mode>_inverted): Add microMIPS support.
24220         (*branch_equality<mode>): Likewise.
24221         (*jump_absolute): Likewise.
24222         (indirect_jump_<mode>): Likewise.
24223         (tablejump_<mode>): Likewise.
24224         (<optab>_internal): Likewise.
24225         (sibcall_internal): Likewise.
24226         (sibcall_value_internal): Likewise.
24227         (prefetch): Use constraint ZD.
24228         * config/mips/mips.opt (minterlink-compressed): New option.
24229         (minterlink-mips16): Now an alias for minterlink-compressed.
24230         (mmicromips): New option.
24231         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24232         (compare_and_swap_12): Likewise.
24233         (sync_add<mode>): Likewise.
24234         (sync_<optab>_12): Likewise.
24235         (sync_old_<optab>_12): Likewise.
24236         (sync_new_<optab>_12): Likewise.
24237         (sync_nand_12): Likewise.
24238         (sync_old_nand_12): Likewise.
24239         (sync_new_nand_12): Likewise.
24240         (sync_sub<mode>): Likewise.
24241         (sync_old_add<mode>): Likewise.
24242         (sync_old_sub<mode>): Likewise.
24243         (sync_new_add<mode>): Likewise.
24244         (sync_new_sub<mode>): Likewise.
24245         (sync_<optab><mode>): Likewise.
24246         (sync_old_<optab><mode>): Likewise.
24247         (sync_new_<optab><mode>): Likewise.
24248         (sync_nand<mode>): Likewise.
24249         (sync_old_nand<mode>): Likewise.
24250         (sync_new_nand<mode>): Likewise.
24251         (sync_lock_test_and_set<mode>): Likewise.
24252         (test_and_set_12): Likewise.
24253         (atomic_compare_and_swap<mode>): Likewise.
24254         (atomic_exchange<mode>_llsc): Likewise.
24255         (atomic_fetch_add<mode>_llsc): Likewise.
24256         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24257         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24258         (umips_save_restore_pattern_p): Likewise.
24259         (umips_load_store_pair_p): Likewise.
24260         (umips_output_load_store_pair): Likewise.
24261         (umips_movep_target_p): Likewise.
24262         (umips_12bit_offset_address_p): Likewise.
24263         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24264         (mips_base_mips16): Rename this...
24265         (mips_base_compression_flags): ...to this. Update all uses.
24266         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24267         (mips_mips16_decl_p): Delete.
24268         (mips_nomips16_decl_p): Delete.
24269         (mips_get_compress_on_flags): New function.
24270         (mips_get_compress_off_flags): New function.
24271         (mips_get_compress_mode): New function.
24272         (mips_get_compress_on_name): New function.
24273         (mips_get_compress_off_name): New function.
24274         (mips_insert_attributes): Support multiple compression types.
24275         (mips_merge_decl_attributes): Likewise.
24276         (umips_12bit_offset_address_p): New function.
24277         (mips_start_function_definition): Emit .set micromips directive.
24278         (mips_call_may_need_jalx_p): New function.
24279         (mips_function_ok_for_sibcall): Add microMIPS support.
24280         (mips_print_operand_punctuation): Support short delay slots and
24281         compact jumps.
24282         (umips_swm_mask, umips_swm_encoding): New.
24283         (umips_build_save_restore): New function.
24284         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24285         (was_mips16_p): Remove.
24286         (old_compression_mode): New.
24287         (mips_set_compression_mode): New function.
24288         (mips_set_current_function): Add microMIPS support.
24289         (mips_option_override): Likewise.
24290         (umips_save_restore_pattern_p): New function.
24291         (umips_output_save_restore): New function.
24292         (umips_load_store_pair_p_1): New function.
24293         (umips_load_store_pair_p): New function.
24294         (umips_output_load_store_pair_1): New function.
24295         (umips_output_load_store_pair): New function.
24296         (umips_movep_target_p) New function.
24297         (mips_prepare_pch_save): Add microMIPS support.
24298         * config/mips/mips.h (TARGET_COMPRESSION): New.
24299         (TARGET_CPU_CPP_BUILTINS): Update macro
24300         to use new compression flags and to support microMIPS.
24301         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24302         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24303         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24304         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24305         (ASM_SPEC): Support mmicromips and mno-micromips.
24306         (M16STORE_REG_P): New macro.
24307         (MIPS_CALL): Support TARGET_MICROMIPS.
24308         (MICROMIPS_J): New macro.
24309         (mips_base_mips16): Rename this...
24310         (mips_base_compression_flags): ...to this.
24311         (UMIPS_12BIT_OFFSET_P): New macro.
24312         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24313         (MULTILIB_DIRNAMES): Likewise.
24315 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24317         PR rtl-optimization/56131
24318         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24319         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24320         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24322 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24324         * doc/invoke.texi (-fsanitize=): Move from optimization
24325         to debugging options.
24327 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24329         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24331 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24332             Alexander Monakov  <amonakov@ispras.ru>
24334         PR middle-end/56077
24335         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24336         flush pending lists also on non-jumps.  Adjust comment.
24338 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24339             Maciej W. Rozycki  <macro@codesourcery.com>
24340             Tom de Vries  <tom@codesourcery.com>
24341             Nathan Sidwell  <nathan@codesourcery.com>
24342             Iain Sandoe  <iain@codesourcery.com>
24343             Nathan Froyd  <froydnj@codesourcery.com>
24344             Chao-ying Fu  <fu@mips.com>
24346         * doc/extend.texi: (micromips, nomicromips, nocompression):
24347         Document new function attributes.
24348         * doc/invoke.texi (minterlink-compressed, mmicromips,
24349         m14k, m14ke, m14kec): Document new options.
24350         (minterlink-mips16): Update documentation.
24351         * doc/md.texi (ZC, ZD): Document new constraints.
24352         * configure.ac (gcc_cv_as_micromips): Check if linker
24353         supports the .set micromips directive.
24354         * configure: Regenerate.
24355         * config.in: Regenerate.
24356         * config/mips/mips-tables.opt: Regenerate.
24357         * config/mips/micromips.md: New file.
24358         * constraints.md (ZC, AD): New constraints.
24359         * config/mips/predicates.md (movep_src_register): New predicate.
24360         (movep_src_operand): New predicate.
24361         (non_volatile_mem_operand): New predicate.
24362         * config/mips/mips.md (multimem): New type.
24363         (length): Differentiate between 17-bit and 18-bit branch offsets.
24364         (MOVEP1, MOVEP2): New mode iterator.
24365         (mov_<load>l): Use ZC constraint.
24366         (mov_<load>r): Likewise.
24367         (mov_<store>l): Likewise.
24368         (mov_<store>r): Likewise.
24369         (*branch_equality<mode>_inverted): Add microMIPS support.
24370         (*branch_equality<mode>): Likewise.
24371         (*jump_absolute): Likewise.
24372         (indirect_jump_<mode>): Likewise.
24373         (tablejump_<mode>): Likewise.
24374         (<optab>_internal): Likewise.
24375         (sibcall_internal): Likewise.
24376         (sibcall_value_internal): Likewise.
24377         (prefetch): Use constraint ZD.
24378         * config/mips/mips.opt (minterlink-compressed): New option.
24379         (minterlink-mips16): Now an alias for minterlink-compressed.
24380         (mmicromips): New option.
24381         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24382         (compare_and_swap_12): Likewise.
24383         (sync_add<mode>): Likewise.
24384         (sync_<optab>_12): Likewise.
24385         (sync_old_<optab>_12): Likewise.
24386         (sync_new_<optab>_12): Likewise.
24387         (sync_nand_12): Likewise.
24388         (sync_old_nand_12): Likewise.
24389         (sync_new_nand_12): Likewise.
24390         (sync_sub<mode>): Likewise.
24391         (sync_old_add<mode>): Likewise.
24392         (sync_old_sub<mode>): Likewise.
24393         (sync_new_add<mode>): Likewise.
24394         (sync_new_sub<mode>): Likewise.
24395         (sync_<optab><mode>): Likewise.
24396         (sync_old_<optab><mode>): Likewise.
24397         (sync_new_<optab><mode>): Likewise.
24398         (sync_nand<mode>): Likewise.
24399         (sync_old_nand<mode>): Likewise.
24400         (sync_new_nand<mode>): Likewise.
24401         (sync_lock_test_and_set<mode>): Likewise.
24402         (test_and_set_12): Likewise.
24403         (atomic_compare_and_swap<mode>): Likewise.
24404         (atomic_exchange<mode>_llsc): Likewise.
24405         (atomic_fetch_add<mode>_llsc): Likewise.
24406         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24407         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24408         (umips_save_restore_pattern_p): Likewise.
24409         (umips_load_store_pair_p): Likewise.
24410         (umips_output_load_store_pair): Likewise.
24411         (umips_movep_target_p): Likewise.
24412         (umips_12bit_offset_address_p): Likewise.
24413         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24414         (mips_base_mips16): Rename this...
24415         (mips_base_compression_flags): ...to this. Update all uses.
24416         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24417         (mips_mips16_decl_p): Delete.
24418         (mips_nomips16_decl_p): Delete.
24419         (mips_get_compress_on_flags): New function.
24420         (mips_get_compress_off_flags): New function.
24421         (mips_get_compress_mode): New function.
24422         (mips_get_compress_on_name): New function.
24423         (mips_get_compress_off_name): New function.
24424         (mips_insert_attributes): Support multiple compression types.
24425         (mips_merge_decl_attributes): Likewise.
24426         (umips_12bit_offset_address_p): New function.
24427         (mips_start_function_definition): Emit .set micromips directive.
24428         (mips_call_may_need_jalx_p): New function.
24429         (mips_function_ok_for_sibcall): Add microMIPS support.
24430         (mips_print_operand_punctuation): Support short delay slots and
24431         compact jumps.
24432         (umips_swm_mask, umips_swm_encoding): New.
24433         (umips_build_save_restore): New function.
24434         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24435         (was_mips16_p): Remove.
24436         (old_compression_mode): New.
24437         (mips_set_compression_mode): New function.
24438         (mips_set_current_function): Add microMIPS support.
24439         (mips_option_override): Likewise.
24440         (umips_save_restore_pattern_p): New function.
24441         (umips_output_save_restore): New function.
24442         (umips_load_store_pair_p_1): New function.
24443         (umips_load_store_pair_p): New function.
24444         (umips_output_load_store_pair_1): New function.
24445         (umips_output_load_store_pair): New function.
24446         (umips_movep_target_p) New function.
24447         (mips_prepare_pch_save): Add microMIPS support.
24448         * config/mips/mips.h (TARGET_COMPRESSION): New.
24449         (TARGET_CPU_CPP_BUILTINS): Update macro
24450         to use new compression flags and to support microMIPS.
24451         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24452         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24453         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24454         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24455         (ASM_SPEC): Support mmicromips and mno-micromips.
24456         (M16STORE_REG_P): New macro.
24457         (MIPS_CALL): Support TARGET_MICROMIPS.
24458         (MICROMIPS_J): New macro.
24459         (mips_base_mips16): Rename this...
24460         (mips_base_compression_flags): ...to this.
24461         (UMIPS_12BIT_OFFSET_P): New macro.
24462         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24463         (MULTILIB_DIRNAMES): Likewise.
24465 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24467         PR target/52555
24468         * target-globals.c (save_target_globals): For init_reg_sets and
24469         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24471 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24473         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24474         * config/aarch64/t-aarch64
24475         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24477 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24479         PR inline-asm/56148
24480         * lra-constraints.c (process_alt_operands): Reload operand
24481         conflicting with earlier clobber only if no more other conflicting
24482         operands.
24484 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24486         PR sanitizer/56393
24487         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24488         if not linking a shared library.
24490 2013-02-22  Seth LaForge  <sethml@google.com>
24492         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24494 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24496         * config/arm/arm.md (split for extendsidi): Update condition.
24497         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24498         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24499         (qhs_zextenddi_cstr): Likewise.
24501 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24503         PR middle-end/56420
24504         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24505         avoid signed wrapping.
24506         (expand_mult): Handle properly multiplication by
24507         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24508         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24509         in the compiler if coeff is HOST_WIDE_INT_MIN.
24510         (expand_divmod): Don't make ext_op1 static, change it's type to
24511         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24513         PR rtl-optimization/50339
24514         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24515         field.
24516         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24517         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24518         into splitting_ashiftrt field.
24519         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24520         ASHIFTRT.
24521         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24522         choices.
24524 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24526         PR middle-end/56108
24527         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24528         are sure to go irrevocable.
24530 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24532         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24533         scalars are valid operands.
24535 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24537         PR tree-optimization/56310
24538         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24539         only matching indices and non-negative final offsets.
24540         (intersect_aggregates_with_edge): Pass src_idx to
24541         agg_replacements_to_vector.  Pass src_idx insstead of index to
24542         intersect_with_agg_replacements.
24544 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24546         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24547         instead of hard-wired defaults.
24549 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24551         * doc/invoke.texi (MIPS Options): Update documentation of the
24552         floating-point multiply-accumulate instruction restrictions.
24554 2013-02-21  Kostya Serebryany  <kcc@google.com>
24556         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24557         asan_shadow_offset on x86_64 linux.
24559 2013-02-21  Richard Biener  <rguenther@suse.de>
24561         PR tree-optimization/56415
24562         Revert
24563         2013-02-11  Richard Biener  <rguenther@suse.de>
24565         PR tree-optimization/56273
24566         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24567         first VRP run.
24569 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24571         PR bootstrap/56258
24572         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24573         instead of @itemx.
24575         PR inline-asm/56405
24576         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24577         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24579 2013-02-20  Jan Hubicka  <jh@suse.cz>
24581         PR tree-optimization/56265
24582         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24583         when target is referenced for first time.
24585 2013-02-20  Richard Biener  <rguenther@suse.de>
24587         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24588         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24589         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24590         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24591         not return anything.
24592         (rename_ssa_copies): Do not remove unused locals.
24593         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24594         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24595         * passes.c (execute_function_todo): Do not schedule unused locals
24596         removal if cleanup_tree_cfg did something.
24597         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24598         about the number of removed locals.
24600 2013-02-20  Richard Biener  <rguenther@suse.de>
24602         PR tree-optimization/56398
24603         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24605 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24607         PR tree-optimization/55334
24608         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24609         restricted pointers to arrays.
24611 2013-02-20  Richard Biener  <rguenther@suse.de>
24612             Jakub Jelinek  <jakub@redhat.com>
24614         PR tree-optimization/56396
24615         * tree-ssa-ccp.c (n_const_val): New static variable.
24616         (get_value): Return NULL for SSA names we don't have a lattice
24617         entry for.
24618         (ccp_initialize): Initialize n_const_val.
24619         * tree-ssa-copy.c (n_copy_of): New static variable.
24620         (init_copy_prop): Initialize n_copy_of.
24621         (get_value): Return NULL_TREE for SSA names we don't have a
24622         lattice entry for.
24624 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24626         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24628 2013-02-20  Richard Biener  <rguenther@suse.de>
24630         * genpreds.c (write_lookup_constraint): Do not compare first
24631         letter of the constraint again.
24633 2013-02-20  Richard Biener  <rguenther@suse.de>
24635         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24636         and ceil_log2.
24637         (get_use_iv_cost): Terminate hashtable walk when coming across
24638         an empty entry.
24640 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24642         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24643         reassociation for avx2 targets.
24645 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24647         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24648         Add version check for v8.10.a to enable microblaze_has_clz
24649         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24650         version and TARGET_PATTERN_COMPARE check
24651         * config/microblaze/microblaze.md: New clzsi2 instruction
24653 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24655         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24656         function before branching.
24658 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24660         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24661         DUMP_INSN_RTX_UID.
24662         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24664 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24666         PR middle-end/55889
24667         * sel-sched.c: Include ira.h.
24668         (implicit_clobber_conflict_p): New function.
24669         (moveup_expr): Use it.
24670         * Makefile.in (sel-sched.o): Depend on ira.h.
24672 2013-02-19  Richard Biener  <rguenther@suse.de>
24674         PR tree-optimization/56384
24675         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24676         (vn_hash_type): Split out from ...
24677         (vn_hash_constant_with_type): ... here.
24678         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24679         (vn_phi_eq): Compare types from vn_phi_s structure.
24680         (vn_phi_lookup): Populate vn_phi_s type.
24681         (vn_phi_insert): Likewise.
24683 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24685         PR tree-optimization/56350
24686         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24687         if haven't found reduction or nested cycle operand, rather than
24688         asserting we must find it.
24690         PR tree-optimization/56381
24691         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24692         to fold_build3.
24694 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24695             Jakub Jelinek  <jakub@redhat.com>
24697         PR target/52555
24698         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24699         (swap_optab_enable): Same.
24700         (init_all_optabs): Use argument instead of global.
24701         * tree.h (struct tree_optimization_option): New field target_optabs.
24702         * expr.h (init_all_optabs): Add argument to prototype.
24703         (TREE_OPTIMIZATION_OPTABS): New.
24704         (save_optabs_if_changed): Protoize.
24705         * optabs.h: Declare this_fn_optabs.
24706         * optabs.c (save_optabs_if_changed): New.
24707         Declare this_fn_optabs.
24708         (init_optabs): Add argument to init_all_optabs() call.
24709         * function.c (invoke_set_current_function_hook): Handle per
24710         function optabs.
24711         * function.h (struct function): New field optabs.
24712         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24713         optimization_current_node has changed.
24714         * target-globals.h (save_target_globals_default_opts): Protoize.
24715         * target-globals.c (save_target_globals_default_opts): New.
24717 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24719         PR target/56347
24720         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24721         registers %fr12 and %fr12R as call used.
24723         PR target/56214
24724         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24725         and HImode, require all displacements to be an integer multiple of
24726         their mode size.
24727         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24728         only allow QImode and HImode when reload is in progress and strict is
24729         true.  Likewise for symbolic addresses.  Use base14_operand to check
24730         displacements in REG+BASE addresses.
24732 2013-02-18  Richard Biener  <rguenther@suse.de>
24734         PR tree-optimization/56366
24735         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24736         handle sign-conversion of outer-loop initial induction value.
24738 2013-02-18  Richard Biener  <rguenther@suse.de>
24740         PR middle-end/56349
24741         * cfghooks.c (merge_blocks): If we merge a latch into another
24742         block adjust references to it.
24743         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24744         (verify_loop_structure): Verify that a recorded latch is in fact
24745         a latch.
24747 2013-02-18  Richard Biener  <rguenther@suse.de>
24749         PR tree-optimization/56321
24750         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24751         order SSA name release and virtual operand unlinking.
24753 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24755         * config/microblaze/microblaze.md (save_stack_block): Define.
24756         (restore_stack_block): Likewise.
24758 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24760         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24761         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24762         * config/microblaze/microblaze.c (microblaze_option_override):
24763         Bail out early for PIC modes when target does not support PIC.
24765 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24767         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24768         Replace with a microblaze version.
24769         (microblaze_trampoline_init): Adapt for microblaze.
24770         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24771         microblaze.
24773 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24774             Dodji Seketeli  <dodji@redhat.com>
24776         PR asan/56330
24777         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24778         (instrument_mem_region_access): Do not forget to always put
24779         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24780         0) statement, even for cases where either 'base' or 'base + len'
24781         are not instrumented -- because they have been previously
24782         instrumented.  Simplify the logic by putting all the statements
24783         instrument 'base + len' inside a sequence, and then insert that
24784         sequence right before the current insertion point.  Then, to
24785         instrument 'base + len', just get an iterator on that statement.
24786         And do not forget to update the pointer to iterator the function
24787         received as argument.
24789 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24791         PR rtl-optimization/56348
24792         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24794 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24796         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24797         (clean_graph_dump_file): Pass base to start_graph_dump.
24799 2013-02-14  Richard Henderson  <rth@redhat.com>
24801         PR target/55941
24802         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24804 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24806         * collect2-aix.h: Define F_LOADONLY.
24808 2013-02-14  Richard Biener  <rguenther@suse.de>
24810         PR lto/50494
24811         * varasm.c (output_constant_def_1): Get the decl representing
24812         the constant as argument.
24813         (output_constant_def): Wrap output_constant_def_1.
24814         (make_decl_rtl): Use output_constant_def_1 with the decl
24815         representing the constant.
24816         (build_constant_desc): Optionally re-use a decl already
24817         representing the constant.
24818         (tree_output_constant_def): Adjust.
24820 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24822         Fix an asan crash
24823         * asan.c (instrument_builtin_call):  Really put the length of the
24824         second source argument into src1_len.
24826 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24828         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24829         argument.  If it is false, don't create edge from then_bb to
24830         fallthru_bb.
24831         (insert_if_then_before_iter): Pass true to it.
24832         (build_check_stmt): Pass false to it.
24833         (transform_statements): Flush hash table only on extended basic
24834         block boundaries, rather than at the beginning of every bb.
24835         Don't flush hash table on nonfreeing_call_p calls.
24836         * tree-flow.h (nonfreeing_call_p): New prototype.
24837         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24839 2013-02-13  David S. Miller  <davem@davemloft.net>
24841         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24843 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24845         PR target/56184
24846         * ira.c (max_regno_before_ira): Move from ...
24847         (ira): ... here.
24848         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24849         vec_safe_length.
24851 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24853         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24855 2013-02-13  Richard Biener  <rguenther@suse.de>
24857         PR lto/56295
24858         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24859         globals in MEM_REFs.
24861 2013-02-13  Richard Biener  <rguenther@suse.de>
24863         * loop-init.c (loop_optimizer_init): Clear loop state when
24864         re-initializing preserved loops.
24865         * loop-unswitch.c (unswitch_single_loop): Return whether
24866         we unswitched the loop.  Do not verify loop state here.
24867         (unswitch_loops): When we unswitched a loop discover new loops.
24869 2013-02-13  Kostya Serebryany  <kcc@google.com>
24871         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24872         on x86_64 linux.
24873         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24875 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24877         Avoid instrumenting duplicated memory access in the same basic block
24878         * Makefile.in (asan.o): Add new dependency on hash-table.h
24879         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24880         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24881         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24882         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24883         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24884         (get_mem_ref_of_assignment): New functions.
24885         (get_mem_refs_of_builtin_call): Extract from
24886         instrument_builtin_call and tweak a little bit to make it fit with
24887         the new signature.
24888         (instrument_builtin_call): Use the new
24889         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24890         of is_gimple_builtin_call.
24891         (instrument_derefs, instrument_mem_region_access): Insert the
24892         instrumented memory reference into the hash table.
24893         (maybe_instrument_assignment): Renamed instrument_assignment into
24894         this, and change it to advance the iterator when instrumentation
24895         actually happened and return true in that case.  This makes it
24896         homogeneous with maybe_instrument_assignment, and thus give a
24897         chance to callers to be more 'regular'.
24898         (transform_statements): Clear the memory reference hash table
24899         whenever we enter a new BB, when we cross a function call, or when
24900         we are done transforming statements.  Use
24901         maybe_instrument_assignment instead of instrumentation.  No more
24902         need to special case maybe_instrument_assignment and advance the
24903         iterator after calling it; it's now handled just like
24904         maybe_instrument_call.  Update comment.
24906 2013-02-13  Richard Biener  <rguenther@suse.de>
24908         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24909         Fix loop discovery code.
24911 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24913         PR inline-asm/56148
24914         * lra-constraints.c (process_alt_operands): Match early clobber
24915         operand with itself.  Check conflicts with earlyclobber only if
24916         the operand is not reloaded.  Prefer to reload conflicting operand
24917         if earlyclobber and matching operands are the same.
24919 2013-02-12  Richard Biener  <rguenther@suse.de>
24921         PR lto/56297
24922         * lto-streamer-out.c (write_symbol): Do not output symbols
24923         for hard register variables.
24925 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24927         PR target/54222
24928         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24929         (umulsidi3_insn, mulsidi3_insn): New insns.
24931 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24933         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24934         (struct tune_params): Add vec_costs field.
24935         * config/arm/arm.c (arm_builtin_vectorization_cost)
24936         (arm_add_stmt_cost): New functions.
24937         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24938         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24939         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24940         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24941         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24942         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24943         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24945 2013-02-12  Richard Biener  <rguenther@suse.de>
24947         PR lto/56295
24948         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24949         decls again if possible.
24951 2013-02-12  Richard Biener  <rguenther@suse.de>
24953         PR middle-end/56288
24954         * tree-ssa.c (verify_ssa_name): Fix check, move
24955         SSA_NAME_IN_FREE_LIST check up.
24957 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24958             Steven Bosscher   <steven@gcc.gnu.org>
24960         PR rtl-optimization/56151
24961         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24962         equal to op0 or op1, and last_insn pattern is CODE operation
24963         with MEM dest and one of the operands matches that MEM.
24965 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24967         * doc/extend.texi: Document Function Multiversioning and "default"
24968         parameter string to target attribute.
24969         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24970         target attribute parameter is "default".
24971         (ix86_compare_version_priority): Remove checks for target attribute.
24972         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24973         Remove check for target attribute equal to NULL. Add assert.
24974         (ix86_generate_version_dispatcher_body): Change error to sorry.
24976 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24977             Jack Howarth  <howarth@bromo.med.uc.edu>
24978             Patrick Marlier  <patrick.marlier@gmail.com>
24980         PR libitm/55693
24981         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24982         define ENDFILE_SPEC as TM_DESTRUCTOR.
24983         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24985 2013-02-11  Alexander Potapenko  <glider@google.com>
24986             Jack Howarth  <howarth@bromo.med.uc.edu>
24987             Jakub Jelinek  <jakub@redhat.com>
24989         PR sanitizer/55617
24990         * config/darwin.c (cdtor_record): Rename ctor_record.
24991         (sort_cdtor_records): Rename sort_ctor_records.
24992         (finalize_dtors): New routine to sort destructors by
24993         priority before use in assemble_integer.
24994         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24996 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24998         PR rtl-optimization/56275
24999         * simplify-rtx.c (avoid_constant_pool_reference): Check that
25000         offset is non-negative and less than cmode size before
25001         calling simplify_subreg.
25003 2013-02-11  Richard Biener  <rguenther@suse.de>
25005         PR tree-optimization/56264
25006         * cfgloop.h (fix_loop_structure): Adjust prototype.
25007         * loop-init.c (fix_loop_structure): Return the number of
25008         newly discovered loops.
25009         * tree-cfgcleanup.c (repair_loop_structures): When new loops
25010         are discovered, do a full loop-closed SSA rewrite.
25012 2013-02-11  Richard Biener  <rguenther@suse.de>
25014         PR tree-optimization/56273
25015         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
25016         first VRP run.
25017         (check_array_ref): Fix missing newline in dumps.
25018         (search_for_addr_array): Likewise.
25020 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
25022         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
25024 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
25026         PR target/56256
25027         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
25029 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
25031         PR rtl-optimization/56246
25032         * lra-constraints.c (simplify_operand_subreg): Try to reuse
25033         reload pseudo.
25034         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
25035         constraints are satisfied.
25037 2013-02-08  Jeff Law  <law@redhat.com>
25039         PR debug/53948
25040         * emit-rtl.c (reg_is_parm_p): New function.
25041         * regs.h (reg_is_parm_p): New prototype.
25042         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
25043         callee-clobbered registers.
25045 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
25047         PR target/56043
25048         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
25049         If there is no implicit builtin declaration, just return NULL.
25051 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25053         * config/i386/sse.md (FMAMODEM): New mode iterator.
25054         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
25055         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
25057 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25059         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
25060         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
25061         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
25063 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25065         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
25066         (microblaze*-*-elf): Likewise.
25067         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
25068         LINK_SPEC.
25069         * config/microblaze/microblaze-c.c: Add builtin defines for
25070         _LITTLE_ENDIAN and _BIG_ENDIAN.
25071         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
25072         add to TARGET_DEFAULT flags.
25073         Expand ASM_SPEC and LINK_SPEC.
25074         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
25075         * config/microblaze/microblaze.md: Update extendsidi2 and
25076         movdi_internal instructions to use low-order / high-order reg
25077         print_operands.
25078         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
25079         options and inversemask / mask of LITTLE_ENDIAN.
25080         * config/microblaze/t-microblaze: Expand multilib options to
25081         include mlittle-endian (le) and update exceptions patterns.
25083 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25085         PR rtl-optimization/56195
25086         * lra-constraints.c (get_reload_reg): Don't reuse regs
25087         if they have smaller mode than requested, if they have
25088         wider mode than requested, try to return a SUBREG.
25090         PR tree-optimization/56250
25091         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25092         if type is unsigned and code isn't MULT_EXPR.
25094 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25096         PR tree-optimization/56064
25097         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25098         bits according to mode.
25099         * fixed-value.h (fixed_from_double_int)
25100         (const_fixed_from_double_int): Adjust comments.
25102 2013-02-08  Richard Biener  <rguenther@suse.de>
25104         PR lto/56231
25105         * lto-streamer.h (struct data_in): Remove current_file, current_line
25106         and current_col members.
25107         * lto-streamer-out.c (lto_output_location): Stream changed bits
25108         en-block for efficiency.
25109         * lto-streamer-in.c (clear_line_info): Remove.
25110         (lto_input_location): Cache current file, line and column
25111         globally via local statics.  Read changed bits en-block.
25112         (input_function): Do not call clear_line_info.
25113         (lto_read_body): Likewise.
25114         (lto_input_toplevel_asms): Likewise.
25116 2013-02-08  Michael Matz  <matz@suse.de>
25118         PR tree-optimization/52448
25119         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25120         (nt_call_phase): New static.
25121         (add_or_mark_expr): Only mark accesses with newer phase than any
25122         call seen.
25123         (nonfreeing_call_p): New.
25124         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25125         (nt_fini_block): Keep blocks marked as visited.
25126         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25128 2013-02-08  Richard Biener  <rguenther@suse.de>
25130         * ira.c (ira): Free broken dominator information.
25132 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25134         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25136 2013-02-08  Marek Polacek  <polacek@redhat.com>
25138         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25140 2013-02-08  Richard Biener  <rguenther@suse.de>
25142         PR middle-end/56181
25143         * cfgloop.h (flow_loops_find): Adjust.
25144         (bb_loop_header_p): Declare.
25145         * cfgloop.c (bb_loop_header_p): New function split out from ...
25146         (flow_loops_find): ... here.  Adjust function signature,
25147         support incremental loop structure update.
25148         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25149         * cfgloopmanip.c (fix_loop_structure): Move ...
25150         * loop-init.c (fix_loop_structure): ... here.
25151         (apply_loop_flags): Split out from ...
25152         (loop_optimizer_init): ... here.
25153         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25154         in incremental mode, only remove dead loops here.
25156 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25158         PR target/54222
25159         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25160         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25161         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25162         (*round<mode>3.libgcc): New insns for fixed-modes.
25163         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25164         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25165         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25166         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25167         implementations.  Define to __builtin_avr_absFX,
25168         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25169         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25170         __builtin_avr_countlsFX, respectively.
25171         * config/avr/avr-c.c (target.h): Include it.
25172         (enum avr_builtin_id): New enum.
25173         (avr_resolve_overloaded_builtin): New static function.
25174         (avr_register_target_pragmas): Use it to set
25175         targetm.resolve_overloaded_builtin.
25176         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25177         tree nodes used by DEF_BUILTIN.
25178         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25179         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25180         <AVR_BUILTIN_xxBITS>: Same.
25182 2013-02-08  Richard Biener  <rguenther@suse.de>
25184         * cfgloop.c (verify_loop_structure): Properly handle
25185         a loop exiting to another loop header.
25186         * ira-int.h (ira_loops): Remove.
25187         * ira.c (ira_loops): Remove.
25188         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25189         (do_reload): Use loop_optimizer_finalize.
25190         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25191         number_of_loops to access the loop tree.
25192         (more_one_region_p): Likewise.
25193         (finish_loop_tree_nodes): Likewise.
25194         (rebuild_regno_allocno_maps): Likewise.
25195         (mark_loops_for_removal): Likewise.
25196         (mark_all_loops_for_removal): Likewise.
25197         (remove_unnecessary_regions): Likewise.
25198         (ira_build): Likewise.
25199         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25201 2013-02-08  Richard Biener  <rguenther@suse.de>
25203         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25204         * ipa-pure-const.c (analyze_function): Avoid calling
25205         mark_irreducible_loops twice.
25206         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25208 2013-02-07  David S. Miller  <davem@davemloft.net>
25210         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25211         on 'reg'.
25212         * var-tracking.c (vt_add_function_parameter): Test the presence of
25213         HAVE_window_save properly and do not remap argument registers when
25214         we have a leaf function.
25216 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25218         PR bootstrap/56227
25219         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25220         instead of "ll".
25221         * config/i386/i386.c (ix86_print_operand): Ditto.
25223 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25225         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25227 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25229         PR rtl-optimization/56225
25230         * lra-constraints.c (process_alt_operands): Check that reload hard
25231         reg can hold value for strict_low_part.
25233 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25235         PR debug/56154
25236         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25237         dwarf2out_end_function.
25238         (in_first_function_p, maybe_at_text_label_p,
25239         first_loclabel_num_not_at_text_label): New variables.
25240         (dwarf2out_var_location): In the first function find out
25241         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25242         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25243         functions.
25245 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25247         PR rtl-optimization/56178
25248         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25249         SUBREG of a register.  Tidy up related block of code.
25250         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25251         note if the source is a register or a SUBREG of a register.
25253 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25255         PR target/56228
25256         * config/rs6000/rs6000.md (ptrm): New mode attr.
25257         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25258         call_value_indirect_aix<pttrsize>,
25259         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25260         m in constraints.
25262 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25264         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25265         if -bnortl. Convert to strcmp and strncmp.
25267 2013-02-07  Alan Modra  <amodra@gmail.com>
25269         PR target/54009
25270         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25271         addresses won't wrap when offsetting.
25272         (rs6000_secondary_reload): Provide secondary reloads needed for
25273         wrapping LO_SUM addresses.
25275 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25277         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25278         MACH, just __MACH__.
25280 2013-02-06  Richard Biener  <rguenther@suse.de>
25282         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25283         instead of calling fix_loop_structure.
25285 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25287         PR middle-end/56217
25288         * omp-low.c (use_pointer_for_field): Return false if
25289         lower_send_shared_vars doesn't generate any copy-out code.
25291 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25293         PR rtl-optimization/56131
25294         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25295         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25296         of the label is NULL.  Add comment.
25298 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25300         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25302         PR sanitizer/55374
25303         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25304         (STATIC_LIBTSAN_LIBS): Likewise.
25305         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25306         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25307         is defined, don't add anything else beyond that.
25308         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25309         (LINK_COMMAND_SPEC): Use them.
25311         PR tree-optimization/56205
25312         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25313         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25314         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25316 2013-02-05  Richard Biener  <rguenther@suse.de>
25318         PR tree-optimization/53342
25319         PR tree-optimization/53185
25320         * tree-vectorizer.h (vect_check_strided_load): Remove.
25321         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25322         not disallow peeling for vectorized strided loads.
25323         (vect_check_strided_load): Make static and simplify.
25324         (vect_analyze_data_refs): Adjust.
25325         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25326         correctly when vectorizing strided loads.
25328 2013-02-05  Richard Biener  <rguenther@suse.de>
25330         * doc/install.texi: Refer to ISL, not PPL.
25332 2013-02-05  Jan Hubicka  <jh@suse.cz>
25334         PR tree-optimization/55789
25335         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25337 2013-02-05  Jan Hubicka  <jh@suse.cz>
25339         PR tree-optimization/55789
25340         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25341         the dead call anyway.
25343 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25345         PR sanitizer/55374
25346         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25348 2013-02-04  Alexander Potapenko  <glider@google.com>
25349             Jack Howarth  <howarth@bromo.med.uc.edu>
25350             Jakub Jelinek  <jakub@redhat.com>
25352         PR sanitizer/55617
25353         * config/darwin.c (sort_ctor_records): Stabilized qsort
25354         on constructor priority by using original position.
25355         (finalize_ctors): New routine to sort constructors by
25356         priority before use in assemble_integer.
25357         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25359 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25361         PR libstdc++/54314
25362         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25363         about visibility on artificial decls.
25364         * config/sol2.c (solaris_assemble_visibility): Likewise.
25366 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25368         PR target/56186
25369         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25370         argument and improve checking of return-argument types for 16-byte
25371         modes.
25372         (ix86_function_value_1): Add additional valtype argument on call
25373         of function_value_64.
25374         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25375         handling infunction_value_64 function.
25377 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25379         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25381 2013-02-04  Richard Biener  <rguenther@suse.de>
25383         PR tree-optimization/56188
25384         * tree-ssa-structalias.c (label_visit): Consider case with
25385         initially non-empty points-to set.
25386         (perform_var_substitution): Dump node mapping and clean up.
25388 2013-02-04  Richard Guenther  <rguenther@suse.de>
25390         PR lto/56168
25391         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25392         node prevail as last resort.
25393         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25394         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25396 2013-02-04  Richard Biener  <rguenther@suse.de>
25398         PR tree-optimization/56113
25399         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25400         Merge into ...
25401         (equiv_class_lookup_or_add): ... this.
25402         (label_visit): Adjust and fix error in previous patch.
25403         (perform_var_substitution): Adjust.
25405 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25407         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25408         * config/sh/sh4-300.md: Likewise.
25409         * config/sh/sh4a.md: Likewise.
25410         * config/sh/constraints.md: Likewise.
25411         * config/sh/sh.md: Likewise.
25412         * config/sh/netbsd-elf.h: Likewise.
25413         * config/sh/predicates.md: Likewise.
25414         * config/sh/sh-protos.h: Likewise.
25415         * config/sh/ushmedia.h: Likewise.
25416         * config/sh/linux.h: Likewise.
25417         * config/sh/sh.c: Likewise.
25418         * config/sh/superh.h: Likewise.
25419         * config/sh/elf.h: Likewise.
25420         * config/sh/sh4.md: Likewise.
25421         * config/sh/sh.h: Likewise.
25423 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25425         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25426         constraint to match_test floating_point_store_memory_operand().
25427         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25428         (base14_operand): New.
25429         (floating_point_store_memory_operand): New.
25430         (integer_store_memory_operand): Revise to use base14_operand and
25431         reg_plus_base_memory_operand.
25432         (move_dest_operand): Allow symbolic_memory_operands.
25433         (symbolic_memory_operand): Check for LO_SOM.
25434         (symbolic_operand): Change default case to break.
25435         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25436         CONST_DOUBLE values to be reloaded by putting them into memory when
25437         the destination is a floating point register.
25438         (movdf): Remove code to handle CONST_DOUBLE.
25439         (movsf): Likewise.
25440         (reload_indf_r1): New.
25441         (reload_insf_r1): New.
25442         Consistently use "Q" and "T" constraints with integer and floating
25443         point move instructions, respectively.
25444         (movdi): Remove FAIL.
25445         Change predicate for source operand unamed DImode move from
25446         general_operand to move_src_operand.
25447         (umulsidi3): Change predicate for destination operand to
25448         register_operand.
25449         Likewise for similar unamed patterns.
25450         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25451         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25452         (hppa_legitimize_address): Simplify mask calculation.
25453         (pa_emit_move_sequence): Revised handling of secondary reloads from
25454         REG+D addresses for floating point loads and stores.  Directly handle
25455         loading CONST0_RTX (mode) to a floating point register.
25456         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25457         to floating point registers.  Don't restrict secondary reloads to
25458         floating point registers to integer modes.  Revise some comments and
25459         cleanup some code.
25460         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25461         (pa_legitimate_address_p): New.
25462         (pa_legitimize_reload_address): New.
25463         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25464         (STRICT_REG_OK_FOR_BASE_P): New.
25465         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25466         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25468 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25469             Andrew Dixie  <andrewd@gentrack.com>
25471         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25472         flag set.
25474 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25476         * expmed.c (extract_bit_field_1): Pass the full width of the
25477         structure to get_best_reg_extraction_insn.
25479 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25481         PR target/54601
25482         * configure.ac (use_cxa_atexit): Add AIX.
25483         * configure: Regenerate.
25485         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25487 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25489         PR debug/54793
25490         * final.c (need_profile_function): New variable.
25491         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25492         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25493         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25494         notes, targetm.asm_out.function_prologue doesn't emit anything,
25495         HAVE_prologue and profiler should be emitted before prologue,
25496         set need_profile_function instead of emitting it.
25497         (final_scan_insn): If need_profile_function, emit
25498         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25499         NOTE_INSN_FUNCTION_BEG note.
25501 2013-02-01  Richard Henderson  <rth@redhat.com>
25503         * config/rs6000/rs6000.md (smulditi3): New.
25504         (umulditi3): New.
25506         * config/alpha/alpha.md (umulditi3): New.
25508 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25510         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25511         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25513 2013-02-01  Richard Biener  <rguenther@suse.de>
25515         PR tree-optimization/56113
25516         * tree-ssa-structalias.c (label_visit): Reduce work for
25517         single-predecessor nodes.
25519 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25521         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25522         range isn't testing for zero.
25524 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25526         PR middle-end/56113
25527         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25529 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25530             Nick Clifton  <nickc@redhat.com>
25532         * config/v850/constraints.md (Q): Define as a memory constraint.
25533         * config/v850/predicates.md (label_ref_operand): New predicate.
25534         (e3v5_shift_operand): New predicate.
25535         (ior_operator): New predicate.
25536         * config/v850/t-v850: Add e3v5 multilib.
25537         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25538         (v850_gen_movdi): Prototype.
25539         * config/v850/v850.c: Add support for e3v5 architecture.
25540         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25541         TARGET_V850E_UP.
25542         (construct_save_jarl): Add e3v5 long JARL support.
25543         (v850_adjust_insn_length): New function.  Adjust length of call
25544         insns when using e3v5 instructions.
25545         (v850_gen_movdi): New function: Generate instructions to move a
25546         DImode value.
25547         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25548         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25549         (TARGET_USE_FPU): Enable for e3v5.
25550         (CONST_OK_FOR_W): New macro.
25551         (ADJUST_INSN_LENGTH): Define.
25552         * config/v850/v850.md (UNSPEC_LOOP): Define.
25553         (attr cpu): Add v850e3v5.
25554         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25555         (movdi): New pattern.
25556         (movdi_internal): New pattern.
25557         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25558         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25559         (cstoresf4): Likewise.
25560         (cstoredf4): Likewise.
25561         (insv): New pattern.
25562         (rotlso3_a): New pattern.
25563         (rotlsi3_b): New pattern
25564         (rotlsi3_v850e3v5): New pattern.
25565         (doloop_begin): New pattern.
25566         (fix_loop_counter): New pattern.
25567         (doloop_end): New pattern.
25568         (branch_normal): Add e3v5 long branch support.
25569         (branch_invert): Likewise.
25570         (branch_z_normal): Likewise.
25571         (branch_z_invert): Likewise.
25572         (branch_nz_normal): Likewise.
25573         (branch_nz_invert): Likewise.
25574         (call_internal_short): Add e3v5 register-indirect JARL support.
25575         (call_internal_long): Likewise.
25576         (call_value_internal_short): Likewise.
25577         (call_value_internal_long): Likewise.
25578         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25579         (mloop): New option.
25580         * config.gcc: Add support for configuring v840e3v5 target.
25581         * doc/invoke.texi: Document new v850 specific command line options.
25583 2013-01-31  Paul Koning  <ni1d@arrl.net>
25585         PR debug/55059
25586         PR debug/54508
25587         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25588         children if parent is a class.
25589         (prune_unused_types_prune): Don't add DW_AT_declaration.
25591 2013-01-31  Richard Biener  <rguenther@suse.de>
25593         PR tree-optimization/56157
25594         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25595         match up operand with SLP child.
25597 2013-01-31  Jason Merrill  <jason@redhat.com>
25599         PR debug/54410
25600         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25601         parameters the first time.
25602         (gen_scheduled_generic_parms_dies): Check completeness here.
25604 2013-01-31  Richard Biener  <rguenther@suse.de>
25606         PR middle-end/53073
25607         * common.opt (faggressive-loop-optimizations): New flag,
25608         enabled by default.
25609         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25610         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25611         infer_loop_bounds_from_undefined by it.
25613 2013-01-31  Richard Biener  <rguenther@suse.de>
25615         PR tree-optimization/56150
25616         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25617         visit virtual operands.
25618         (find_uses_to_rename_bb): Likewise.
25620 2013-01-31  Richard Biener  <rguenther@suse.de>
25622         PR tree-optimization/56150
25623         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25624         mixed store non-store stmts.
25626 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25628         PR sanitizer/55374
25629         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25630         LIBASAN_EARLY_SPEC is defined.
25631         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25632         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25633         before %o.
25634         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25636         PR c++/55742
25637         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25638         invalid args instead of ICEing on it.
25639         (ix86_valid_target_attribute_tree): Return error_mark_node if
25640         ix86_valid_target_attribute_inner_p failed.
25641         (ix86_valid_target_attribute_p): Return false only if
25642         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25643         target("default") attribute.
25644         (sorted_attr_string): Change argument from const char * to tree,
25645         merge in all target attribute arguments rather than just one.
25646         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25647         instead of free.  Avoid using strcat.
25648         (ix86_mangle_function_version_assembler_name): Mangle
25649         target("default") as if no target attribute is present.  Adjust
25650         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25651         instead of xmalloc and XDELETEVEC instead of free.
25652         (ix86_function_versions): Don't return true if one of the decls
25653         doesn't have target attribute.  If they don't and one of the decls
25654         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25655         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25656         (ix86_supports_function_versions): Remove.
25657         (make_name): Fix up formatting.
25658         (make_dispatcher_decl): Remove resolver_name and its initialization.
25659         Avoid leaking memory.
25660         (is_function_default_version): Return true if there is
25661         target("default") attribute rather than no target attribute at all.
25662         (make_resolver_func): Avoid leaking memory.
25663         (ix86_generate_version_dispatcher_body): Likewise.
25664         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25665         * target.def (supports_function_versions): Remove.
25666         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25667         * doc/tm.texi: Regenerated.
25669 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25671         PR rtl-optimization/56144
25672         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25673         for values with side effects.
25675 2013-01-30  Richard Biener  <rguenther@suse.de>
25677         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25678         (sparseset_pop): Likewise.
25679         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25680         to be able to use quick_push in the worker loop.
25682 2013-01-30  Marek Polacek  <polacek@redhat.com>
25684         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25686 2013-01-30  Richard Biener  <rguenther@suse.de>
25688         PR lto/56147
25689         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25691 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25693         PR tree-optimization/56064
25694         * fixed-value.c (fixed_from_double_int): New function.
25695         * fixed-value.h (fixed_from_double_int): New prototype.
25696         (const_fixed_from_double_int): New static inline function.
25697         * fold-const.c (native_interpret_fixed): New static function.
25698         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25699         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25700         (native_encode_fixed): New static function.
25701         (native_encode_expr) <FIXED_CST>: Use it.
25702         (native_interpret_int): Move double_int worker code to...
25703         * double-int.c (double_int::from_buffer): ...this new static method.
25704         * double-int.h (double_int::from_buffer): Prototype it.
25706 2013-01-30  Richard Biener  <rguenther@suse.de>
25708         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25709         New pointer-map and obstack.
25710         (init_alias_vars): Allocate pointer-map and obstack.
25711         (delete_points_to_sets): Free them.
25712         (find_what_var_points_to): Cache result.
25713         (find_what_p_points_to): Adjust for changed interface of
25714         find_what_var_points_to.
25715         (compute_points_to_sets): Likewise.
25716         (ipa_pta_execute): Likewise.
25718 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25720         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25721         * configure: Regenerate.
25722         * config.in: Regenerate.
25723         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25724         #nobits/#progbits if supported.
25726 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25728         PR target/56121
25729         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25730         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25731         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25733 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25735         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25736         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25738 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25740         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25741         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25743 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25745         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25746         declaration.
25747         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25748         * config/arm/cortex-a7.md: New bypasses using
25749         arm_mac_accumulator_is_result.
25751 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25753         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25754         (cortex_a7_neon_mla): Likewise.
25755         (cortex_a7_fpfmad): New reservation.
25756         (cortex_a7_fpmacs): Use ffmas and update required units.
25757         (cortex_a7_fpmuld): Update required units and latency.
25758         (cortex_a7_fpmacd): Likewise.
25759         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25760         (cortex_a7_neon). Likewise.
25761         (bypass) Update participating units.
25763 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25765         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25766         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25767         from fmac to ffma.
25768         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25769         (vfp_fmul): Use ffmad.
25770         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25771         (cortex_r4_fmacd): Use ffmad.
25772         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25773         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25774         (cortex_a9_fmacd): Use ffmad.
25775         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25776         (cortex_a8_vfp_macd): Use ffmad.
25777         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25778         (cortex_a5_fpmacd): Use ffmad.
25779         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25780         (cortex_a15_vfp_macd): Use ffmad.
25781         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25783 2013-01-29  Jason Merrill  <jason@redhat.com>
25785         PR libstdc++/54314
25786         * varasm.c (default_assemble_visibility): Don't warn about
25787         visibility on artificial decls.
25789 2013-01-29  Richard Biener  <rguenther@suse.de>
25791         PR tree-optimization/56113
25792         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25793         the bitmap leader.
25794         (label_visit): Free duplicate bitmaps and record the leader instead.
25795         (perform_var_substitution): Adjust.
25797 2013-01-29  Richard Biener  <rguenther@suse.de>
25799         PR tree-optimization/55270
25800         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25801         the CFG, schedule loops for fixup.
25803 2013-01-29  Nick Clifton  <nickc@redhat.com>
25805         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25806         SP_REG.
25808 2013-01-28  Leif Ekblad  <leif@rdos.net>
25810         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25811         * config/i386/i386.h (TARGET_RDOS): New macro.
25812         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25813         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25814         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25815         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25816         DEFAULT_LARGE_SECTION_THRESHOLD.
25817         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25818         * config/i386/rdos.h: New file.
25819         * config/i386/rdos64.h: New file.
25821 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25823         PR other/54814
25824         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25825         TEST_HARD_REG_BIT.
25827 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25829         PR rtl-optimization/56117
25830         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25831         call cselib_lookup_from_insn on the MEM before calling
25832         add_insn_mem_dependence.
25834 2013-01-28  Richard Biener  <rguenther@suse.de>
25836         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25837         to a stmt that didn't have one.
25838         (copy_phis_for_bb): Likewise for PHI arguments.
25839         (copy_debug_stmt): Likewise for debug stmts.
25841 2013-01-28  Richard Biener  <rguenther@suse.de>
25843         PR tree-optimization/56034
25844         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25845         (partition_builtin_p): Adjust.
25846         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25847         it is the last partition.
25848         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25849         up the vertex for the definition.
25850         (classify_partition): Classify whether a partition is a
25851         PKIND_REDUCTION, thus has uses outside of the loop.
25852         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25853         Merge all PKIND_REDUCTION partitions into the last partition.
25854         (tree_loop_distribution): Seed partitions from reductions as well.
25856 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25858         PR tree-optimization/56125
25859         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25860         pow(x,c) into sqrt(x) * powi(x, n/2) or
25861         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25862         optimizing for size.
25863         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25864         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25865         integer.
25867         PR tree-optimization/56094
25868         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25869         to UNKNOWN_LOCATION while gimplifying expr.
25871 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25873         PR target/56114
25874         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25875         operand 0 in movabs insn template for -masm=intel asm alternative.
25876         (*movabs<mode>_2): Ditto for operand 1.
25878 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25880         PR target/54663
25881         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25882         of microblaze-c.o
25884 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25886         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25887         tm_file.
25889 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25891         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25892         Undef to avoid warning.
25894 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25896         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25897         * configure: Regenerate.
25899 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25901         PR tree-optimization/56098
25902         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25903         for stmts with volatile ops.
25904         (cond_store_replacement): Don't optimize if assign has volatile ops.
25905         (cond_if_else_store_replacement_1): Don't optimize if either
25906         then_assign or else_assign have volatile ops.
25907         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25908         volatile ops.
25910 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25912         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25914 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25916         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25917         missing ':' in asm example.
25919 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25921         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25922         entries into lane and laneq entries.
25923         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25924         Remove AdvSIMD scalar modes.
25925         (aarch64_sq<r>dmulh_laneq<mode>): New.
25926         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25927         modes.
25928         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25929         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25930         * config/aarch64/iterators.md (VCOND): New.
25931         (VCONQ): New.
25933 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25935         PR target/54222
25936         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25937         Add NULL LIBNAME argument to existing definitions.
25938         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25939         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25940         * config/avr/avr.c (DEF_BUILTIN): Same.
25941         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25942         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25943         implementation is available (DECL_ASSEMBLER_NAME is set).
25944         (avr_fold_absfx): New static function.
25945         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25946         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25947         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25948         AVR_BUILTIN_ABSLLK.
25949         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25950         (abshk, absk, abslk, absllk): Provide as static inline functions.
25952 2013-01-25  Marek Polacek  <polacek@redhat.com>
25954         PR tree-optimization/56035
25955         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25957 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25959         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25960         (*movtf_internal_rex64): Add (!o,C) alternative
25961         (*movxf_internal_rex64): Ditto.
25962         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25964 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25966         * doc/invoke.texi: fix typo.
25967         * doc/objc.texi: fix typo.
25969 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25971         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25972         for the first two alternatives.
25974 2013-01-24  Diego Novillo  <dnovillo@google.com>
25976         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25977         (ggc-zone.o): Remove.
25978         * configure.ac: Remove option --with-gc.
25979         * configure: Re-generate.
25980         * doc/install.texi: Remove documentation for --with-gc.
25981         * gengtype.c (write_enum_defn): Remove.  Update all users.
25982         (write_Types_process_field): Remove generation of gt_e_* argument.
25983         (output_type_enum): Remove.  Update all users.
25984         (write_enum_defn): Remove.  Update all users.
25985         (enum alloc_zone): Remove.  Update all users.
25986         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25987         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25988         Update all callers.
25989         (struct ptr_data): Remove field TYPE.  Update all users.
25990         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25991         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25992         Update all users.
25993         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25994         (struct alloc_zone): Remove.
25995         (ggc_internal_alloc_zone_stat): Remove.
25996         (ggc_internal_cleared_alloc_zone_stat): Remove.
25997         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25998         (ggc_pch_count_object): Remove last argument.  Update all users.
25999         (ggc_pch_alloc_object): Remove last argument.  Update all users.
26000         (struct alloc_zone): Remove.
26001         * ggc-zone.c: Remove.
26002         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
26003         (struct alloc_zone): Remove.
26004         (ggc_alloc_typed_stat): Remove.
26005         (ggc_alloc_typed): Remove.
26006         (ggc_splay_alloc): Remove first argument.
26007         (rtl_zone): Remove.  Update all users.
26008         (tree_zone): Remove.  Update all users.
26009         (tree_id_zone): Remove.  Update all users.
26010         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
26011         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
26012         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
26013         * tree-ssanames.c: Remove references to zone allocator in comments.
26015 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
26017         * config/avr/avr.c (avr_out_fract): Make register numbers that
26018         might be outside of source operand signed.
26020 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
26022         * config/i386/constraints.md (Yf): New constraint.
26023         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
26024         of f constraint to conditionaly disable x87 register preferences.
26025         (*movdf_internal): Ditto.
26026         (*movsf_internal): Ditto.
26028 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
26030         PR inline-asm/55934
26031         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
26032         that have operands with impossible constraints.
26033         Add a FIXME for a speed-up opportunity.
26034         * lra-constraints.c (process_alt_operands): Verify that a class
26035         selected from constraints on asms is valid for the operand mode.
26036         (curr_insn_transform): Remove incorrect comment.
26038 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
26040         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
26041         TOC operand is a valid symbol ref in the constant pool.
26043 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26045         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
26047 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
26049         PR target/54222
26050         * config/avr/stdfix.h: New file.
26051         * t-avr (stdfix-gcc.h): New rule to build it.
26052         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
26054 2013-01-23  Kostya Serebryany  <kcc@google.com>
26056         * config/darwin.h: remove dependency on
26057         CoreFoundation (asan on Mac OS).
26059 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
26061         PR target/49069
26062         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
26063         instead of cmpdi_operand for first comparison operand.
26064         Don't assert that comparison operands aren't both constants.
26066 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26068         * doc/install.texi (Downloading the Source): Update references to
26069         downloading separate components.
26071 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26073         * doc/extend.texi (__int128): Improve grammar.
26075 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
26077         PR target/56028
26078         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
26079         alternative to (o,r).
26080         (*movdi_internal_rex64): Remove (!o,n) alternative.
26081         (DImode immediate->memory splitter): Remove.
26082         (DImode immediate->memory peephole2): Remove.
26083         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26084         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26085         alternative to (!o,*r).
26086         (*movtf_internal_sse): New pattern.
26087         (*movxf_internal_rex64): New pattern.
26088         (*movxf_internal): Disable for TARGET_64BIT.
26089         (*movdf_internal_rex64): Remove (!o,F) alternative.
26091 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26093         PR middle-end/56074
26094         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26095         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26096         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26097         stmt locations where LOCATION_LOCUS of the stmt location is
26098         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26100         PR target/55686
26101         * config/i386/i386.md (UNSPEC_STOS): New.
26102         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26103         *strsetqi_1): Add UNSPEC_STOS.
26105 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26107         PR c++/56067
26108         * doc/invoke.texi: Remove left over -Wsynth example.
26110 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26112         PR tree-optimization/56051
26113         * fold-const.c (fold_binary_loc): Don't fold
26114         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26115         a narrowing conversion, or widening conversion from signed
26116         to unsigned.
26118 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26120         PR rtl-optimization/56023
26121         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26122         dependent on debug instruction.
26124 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26126         PR middle-end/56022
26127         * function.c (allocate_struct_function): Call
26128         invoke_set_current_function_hook earlier.
26130 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26132         * reload1.c (init_reload): Only initialize reload_obstack
26133         during the first call.
26135 2013-01-21  Marek Polacek  <polacek@redhat.com>
26137         * cfgloop.c (verify_loop_structure): Fix up grammar.
26139 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26141         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26142         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26144 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26146         PR target/56058
26147         * config/arm/marvell-pj4.md: Update copyright year.
26148         Fix up use of alu to alu_reg and simple_alu_imm.
26150 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26152         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26154 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26156         PR target/55433
26157         * lra-constraints.c (curr_insn_transform): Don't reuse original
26158         insn for secondary memory move when memory mode should be different.
26160 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26162         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26163         atomic_storedi_1): New patterns.
26165 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26167         btver2 pipeline descriptions.
26168         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26169         descriptions.
26170         * config/i386/i386.md (btver2_decode): New type attributes.
26171         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26172         type attributes.
26173         * config/i386/btver2.md: New file describing btver2 pipelines.
26175 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26177         PR tree-optimization/52631
26178         * tree-ssa-sccvn (visit_use): Before looking up the original
26179         statement, try looking up the simplified expression.
26181 2013-01-19  Anthony Green  <green@moxielogic.com>
26183         * config/moxie/moxie.c (moxie_expand_prologue): Set
26184         current_function_static_stack_size.
26186 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26188         PR tree-optimization/56029
26189         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26190         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26192 2013-01-18  Sharad Singhai  <singhai@google.com>
26194         PR tree-optimization/55995
26195         * dumpfile.c (dump_loc): Print location only if available.
26196         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26198 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26200         PR target/55433
26201         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26202         secondary memory move.
26203         (inherit_reload_reg): Use rclass instead of cl for
26204         check_secondary_memory_needed_p.
26206 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26208         PR middle-end/56015
26209         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26210         the case where writing real complex part of target modifies op1.
26212 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26214         * config/aarch64/aarch64-simd.md
26215         (aarch64_vcond_internal<mode>): Handle unordered cases.
26216         * config/aarch64/iterators.md (v_cmp_result): New.
26218 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26219             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26221         * config/arm/marvell-pj4.md: New file.
26222         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26223         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26224         (generic_vfp): Likewise.
26225         * config/arm/arm-cores.def: Add marvell-pj4.
26226         * config/arm/arm-tune.md: Regenerate.
26227         * config/arm/arm-tables.opt: Regenerate.
26228         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26229         * doc/invoke.texi: Document marvell-pj4.
26231 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26233         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26235 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26237         PR debug/54114
26238         PR debug/54402
26239         PR debug/49888
26240         * var-tracking.c (negative_power_of_two_p): New.
26241         (global_get_addr_cache, local_get_addr_cache): New.
26242         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26243         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26244         heading comment.
26245         (vt_stack_offset_p): Remove.
26246         (vt_canon_true_dep): Always canonicalize loc's address.
26247         (clobber_overlapping_mems): Make sure we have a MEM.
26248         (local_get_addr_clear_given_value): New.
26249         (val_reset): Clear local cached entries.
26250         (compute_bb_dataflow): Create and release the local cache.
26251         Disable duplicate MEMs clobbering.
26252         (emit_notes_in_bb): Clobber MEMs likewise.
26253         (vt_emit_notes): Create and release the local cache.
26254         (vt_initialize, vt_finalize): Create and release the global
26255         cache, respectively.
26256         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26258 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26260         PR libmudflap/53359
26261         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26262         not found in the symtab.
26264 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26266         PR debug/56006
26267         PR rtl-optimization/55547
26268         PR rtl-optimization/53827
26269         PR debug/53671
26270         PR debug/49888
26271         * alias.c (offset_overlap_p): New, factored out of...
26272         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26273         the conservative special case for symbolic constants.  Don't
26274         adjust zero sizes on alignment.
26276 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26278         PR rtl-optimization/52573
26279         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26280         REG_UNUSED for the same register.
26282 2013-01-17  Richard Biener  <rguenther@suse.de>
26283             Marek Polacek  <polacek@redhat.com>
26285         PR rtl-optimization/55833
26286         * loop-unswitch.c (unswitch_loops): Move loop verification...
26287         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26288         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26289         Set it to true when we're removing a loop from hierarchy tree in
26290         an irreducible region.
26291         (fix_bb_placements): Adjust caller.
26292         (fix_loop_placements): Likewise.
26294 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26296         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26297         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26298         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26299         Remove ID.  Adjust comments.
26300         * config/avr/avr-c.c (avr_builtin_name): Remove.
26301         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26302         * config/avr/avr.c (avr_tolower): New static function.
26303         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26304         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26305         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26306         default expansion.
26308 2013-01-17  Jan Hubicka  <jh@suse.cz>
26310         PR tree-optimization/55273
26311         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26313 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26315         PR target/55981
26316         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26317         store through atomic_store<mode>_1.
26318         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26320 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26322         PR tree-optimizations/55264
26323         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26324         for virtual methods.
26325         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26326         virtual methods before inlining is over.
26327         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26328         virtual functions.
26329         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26330         non-virtual.
26332 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26334         PR rtl-optimization/56005
26335         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26336         pending reads for prefetch.
26338 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26340         * config/aarch64/aarch64.md
26341         (*cstoresi_neg_uxtw): New pattern.
26342         (*cmovsi_insn_uxtw): New pattern.
26343         (*<optab>si3_uxtw): New pattern.
26344         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26345         (*<optab>si3_insn_uxtw): New pattern.
26346         (*bswapsi2_uxtw): New pattern.
26348 2013-01-16  Richard Biener  <rguenther@suse.de>
26350         * tree-inline.c (tree_function_versioning): Remove set but
26351         never used variable.
26353 2013-01-16  Richard Biener  <rguenther@suse.de>
26355         PR tree-optimization/55964
26356         * tree-flow.h (rename_variables_in_loop): Remove.
26357         (rename_variables_in_bb): Likewise.
26358         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26359         (copy_loop_before): Adjust and delete update-ssa status.
26360         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26361         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26362         (rename_variables_in_loop): Remove.
26363         (slpeel_update_phis_for_duplicate_loop): Likewise.
26364         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26365         use available cfg machinery instead of duplicating it.
26366         Update PHI nodes and perform poor-mans SSA update here.
26367         (slpeel_tree_peel_loop_to_edge): Adjust.
26369 2013-01-16  Richard Biener  <rguenther@suse.de>
26371         PR tree-optimization/54767
26372         PR tree-optimization/53465
26373         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26374         (vrp_visit_phi_node): For PHI arguments coming via backedges
26375         drop all symbolical range information.
26376         (execute_vrp): Compute backedges.
26378 2013-01-16  Richard Biener  <rguenther@suse.de>
26380         * doc/install.texi: Update CLooG and ISL requirements to
26381         0.18.0 and 0.11.1.
26383 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26385         PR target/55301
26386         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26387         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26388         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26390 2013-01-16  DJ Delorie  <dj@redhat.com>
26392         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26393         (UNSPECV_SP_SWITCH_E): New.
26394         (sp_switch_1): Change to an unspec.
26395         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26396         replace $r15.
26398 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26400         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26401         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26402         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26403         (expand_mem_thread_fence): Ditto.
26404         (expand_mem_signal_fence): Ditto.
26405         (expand_atomic_load): Ditto.
26406         (expand_atomic_store): Ditto.
26408 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26410         PR rtl-optimization/55547
26411         PR rtl-optimization/53827
26412         PR debug/53671
26413         PR debug/49888
26414         * alias.c (memrefs_conflict_p): Set sizes to negative after
26415         AND adjustments.
26417 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26419         PR target/55940
26420         * function.c (thread_prologue_and_epilogue_insns): Always
26421         add crtl->drap_reg to set_up_by_prologue.set, even if
26422         stack_realign_drap is false.
26424 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26426         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26427         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26428         *call): Fix indention.
26430 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26432         PR target/55876
26433         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26434         Update comment.
26436 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26438         PR rtl-optimization/55153
26439         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26441 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26443         PR tree-optimization/55920
26444         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26445         accesses as grp_to_be_debug_replaced.
26447 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26449         PR tree-optimization/55920
26450         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26451         there is non-useless type conversion needed from debug rhs to lhs,
26452         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26454 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26455             Mikael Pettersson  <mikpe@it.uu.se>
26457         PR target/43961
26458         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26459         Thumb.
26460         (ASM_OUTPUT_CASE_LABEL): Remove.
26461         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26462         * final.c (shorten_branches): Update alignment of labels before
26463         jump tables if CASE_VECTOR_SHORTEN_MODE.
26465 2013-01-15  Richard Biener  <rguenther@suse.de>
26467         PR bootstrap/55961
26468         * system.h: Do not include gmp.h for building host tools.
26470 2013-01-15  Richard Biener  <rguenther@suse.de>
26472         PR middle-end/55882
26473         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26474         account for bitpos when computing alignment.
26476 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26478         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26479         (ix86_target_macros_internal): Likewise.
26481         * config/i386/i386.c (m_CORE2I7): Removed.
26482         (m_CORE_HASWELL): New macro.
26483         (m_CORE_ALL): Likewise.
26484         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26485         (initial_ix86_arch_features): Likewise.
26486         (processor_target_table): Initializations for Core avx2.
26487         (cpu_names): New names "core-avx2".
26488         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26489         PROCESSOR_CORE_HASWELL.
26490         (ix86_issue_rate): New case.
26491         (ia32_multipass_dfa_lookahead): Likewise.
26492         (ix86_sched_init_global): Likewise.
26494         * config/i386/i386.h (TARGET_HASWELL): New macro.
26495         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26496         (processor_type): New PROCESSOR_HASWELL.
26498 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26500         PR tree-optimization/55955
26501         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26502         *SHIFT_EXPR and *ROTATE_EXPR codes.
26504         PR tree-optimization/48766
26505         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26506         -ftrapv disable -fwrapv.
26508 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26510         PR target/55974
26511         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26512         etc. to 1 and not to __flash.
26513         Use LL suffix for __INT24_MAX__ with -mint8.
26514         Use ULL suffix for __UINT24_MAX__ with -mint8.
26516 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26518         * config/avr/avr-arch.h
26519         (struct base_arch_s): Use typedef avr_arch_t instead.
26520         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26521         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26522         * config/avr/avr.c: Same.
26523         * config/avr/avr-devices.c: Same.
26524         * config/avr/driver-avr.c: Same.
26525         * config/avr/gen-avr-mmcu-texi.c: Same.
26526         * config/avr/avr-mcus.def: Adjust comment.
26528 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26530         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26531         * config/aarch64/iterators.md (VALLDI): New.
26533 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26534             Andi Kleen  <ak@linux.intel.com>
26536         PR target/55948
26537         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26538         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26539         memmodel flag.
26541 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26543         * config/avr/avr-stdint.h: Remove trailing blanks.
26544         * config/avr/avr-log.h: Same.
26545         * config/avr/avr-arch.h: Same.
26546         * config/avr/avr-devices.c: Same.
26547         * config/avr/avr-dimode.md: Same.
26548         * config/avr/predicates.md: Same.
26549         * config/avr/avr-c.c: Same.  And fix typo.
26551         * config/avr/avr-protos.h: Same.  And:
26552         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26553         (init_cumulative_args): Rename to avr_init_cumulative_args.
26554         (expand_prologue): Rename to avr_expand_prologue.
26555         (expand_epilogue): Rename to avr_expand_epilogue.
26556         (adjust_insn_length): Rename to avr_adjust_insn_length.
26557         (notice_update_cc): Rename to avr_notice_update_cc.
26558         (final_prescan_insn): Rename to avr_final_prescan_insn.
26559         * config/avr/avr.c: Same.
26560         * config/avr/avr.h: Same.
26561         * config/avr/avr.md: Remove trailing blanks.
26562         (prologue): Use avr_expand_prologue.
26563         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26565 2013-01-14  Richard Biener  <rguenther@suse.de>
26567         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26568         verify_location, collect_subblocks): New functions.
26569         (verify_gimple_in_cfg): Verify that locations only reference
26570         BLOCKs in the functions BLOCK tree.
26572 2013-01-14  Richard Biener  <rguenther@suse.de>
26574         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26575         PHI argument.
26576         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26577         unshare reference.
26578         (insert_out_of_ssa_copy_on_edge): Likewise.
26579         (rewrite_close_phi_out_of_ssa): Likewise.
26580         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26581         debug expressions.
26582         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26583         propagated constants.
26584         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26585         can not be shared.
26587 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26589         * config/avr/avr-modes.def: Add GPL copyright notice.
26591 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26593         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26594         MEMMODEL_MASK to determine memory model.
26595         (atomic_store<mode>): Ditto from operands[2].
26596         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26598 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26600         PR fortran/55935
26601         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26602         (fold_gimple_assign): Don't call unshare_expr here.
26603         (fold_ctor_reference): Call unshare_expr.
26605 2013-01-13  Terry Guo  <terry.guo@arm.com>
26607         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26608         * doc/fragments.texi: Document MULTILIB_REUSE.
26609         * gcc.c (multilib_reuse): New internal spec.
26610         (set_multilib_dir): Also search multilib from multilib_reuse.
26611         * genmultilib (tmpmultilib3): Refactor code.
26612         (tmpmultilib4): Ditto.
26613         (multilib_reuse): New multilib argument.
26615 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26617         * Makefile.in: Update copyright.
26619 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26621         PR middle-end/55890
26622         * calls.c (expand_call): Check if arg_nr is valid.
26624 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26626         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26627         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26628         documentation.  Add missing '__' in front of
26629         __builtin_ia32_packssdw256.
26631 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26633         PR target/55719
26634         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26635         NO_REGS for larl operands.
26636         (s390_reload_larl_operand): Use s390_load_address instead of
26637         emit_move_insn.
26639 2013-01-11  Richard Biener  <rguenther@suse.de>
26641         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26642         (verify_node_sharing): ... here.
26643         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26645 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26647         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26648         Substitute TREECHECKING.
26649         * configure: Regenerate.
26650         * Makefile.in (TREECHECKING): New.
26652 2013-01-11  Richard Guenther  <rguenther@suse.de>
26654         PR tree-optimization/44061
26655         * tree-vrp.c (extract_range_basic): Compute zero as
26656         value-range for __builtin_constant_p of function parameters.
26658 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26660         Update copyright years.
26662 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26664         PR rtl-optimization/55672
26665         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26666         const to be eliminable.
26668 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26670         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26671         * configure: Regenerate.
26673 2013-01-10  Richard Biener  <rguenther@suse.de>
26675         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26677 2013-01-10  Richard Biener  <rguenther@suse.de>
26679         PR bootstrap/55792
26680         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26681         locations for virtual PHI arguments.
26682         (rewrite_update_phi_arguments): Likewise.
26684 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26686         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26687         on to assembler.
26689 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26691         PR tree-optimization/55921
26692         * tree-complex.c (expand_complex_asm): New function.
26693         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26695 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26697         PR target/55718
26698         * config/s390/s390.c (s390_symref_operand_p)
26699         (s390_loadrelative_operand_p): Merge the two functions.
26700         (s390_check_qrst_address, print_operand_address): Add parameters
26701         to s390_loadrelative_operand_p invokation.
26702         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26703         (s390_reload_larl_operand, s390_secondary_reload): Use
26704         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26705         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26707 2013-01-09  Mike Stump  <mikestump@comcast.net>
26709         * dse.c (record_store): Remove unnecessary assert.
26711 2013-01-09  Jan Hubicka  <jh@suse.cz>
26713         PR tree-optimization/55569
26714         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26715         * cfgloop.h (scale_loop_profile): Likewise.
26717 2013-01-09  Jan Hubicka  <jh@suse.cz>
26719         PR lto/45375
26720         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26721         functions.
26722         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26724 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26726         PR middle-end/55114
26727         * expr.h (maybe_emit_group_store): Declare.
26728         * expr.c (maybe_emit_group_store): New function.
26729         * builtins.c (expand_builtin_int_roundingfn): Call it.
26730         (expand_builtin_int_roundingfn_2): Likewise.
26732 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26734         PR rtl-optimization/55829
26735         * lra-constraints.c (match_reload): Add code for absent output.
26736         (curr_insn_transform): Add code for reloads of matched inputs
26737         without output.
26739 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26741         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26742         attribute of movddup insn to DF.
26743         (*vec_interleave_lowv2df): Ditto.
26744         (vec_dupv2df): Ditto.
26746 2013-01-09  Jan Hubicka  <jh@suse.cz>
26748         PR tree-optimiation/55875
26749         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26750         EVERY_ITERATION parameter.
26751         (number_of_iterations_exit): Check if exit is executed every iteration.
26752         (idx_infer_loop_bounds): Similarly here.
26753         (n_of_executions_at_most): Simplify
26754         to only test for cases where statement is dominated by the
26755         particular bound; handle correctly the "postdominance" test.
26756         (scev_probably_wraps_p): Use max loop iterations info
26757         as a global bound first.
26759 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26760             Nick Clifton  <nickc@redhat.com>
26762         * config/v850/v850.md (cbranchsf4): New pattern.
26763         (cstoresf4): New pattern.
26764         (cbranchdf4): New pattern.
26765         (cstoredf4): New pattern.
26766         (movsicc): Disallow floating point comparisons.
26767         (cmpsf_le_insn): Fix order of operators.
26768         (cmpsf_lt_insn): Likewise.
26769         (cmpsf_eq_insn): Likewise.
26770         (cmpdf_le_insn): Likewise.
26771         (cmpdf_lt_insn): Likewise.
26772         (cmpdf_eq_insn): Likewise.
26773         (cmpsf_ge_insn): Use LE comparison.
26774         (cmpdf_ge_insn): Likewise.
26775         (cmpsf_gt_insn): Use LT comparison.
26776         (cmpdf_gt_insn): Likewise.
26777         (cmpsf_ne_insn): Delete pattern.
26778         (cmpdf_ne_insn): Delete pattern.
26779         * config/v850/v850.c (v850_gen_float_compare): Use
26780         gen_cmpdf_eq_insn for NE comparison.
26781         (v850_float_z_comparison_operator)
26782         (v850_float_nz_comparison_operator): Move from here ...
26783         * config/v850/predicates.md: ... to here.  Move GT and GE
26784         comparisons into v850_float_z_comparison_operator.
26785         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26786         Delete prototype.
26787         (v850_float_nz_comparison_operator): Likewise.
26789 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26791         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26792         with calls to gen_insvsi/gen_insvdi.
26794 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26796         * config/i386/i386.c (initial_ix86_tune_features): Set up
26797         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26799 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26800             Jakub Jelinek  <jakub@redhat.com>
26802         PR tree-optimization/48189
26803         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26804         If nitercst is 0, don't predict the exit edge.
26806 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26808         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26809         in asm_fprintf with reg_names.
26810         (aarch64_print_operand_address): Likewise.
26811         (aarch64_return_addr): Likewise.
26812         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26814 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26816         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26817         (INT_U6_BITS): Likewise.
26818         * config/pa/predicates.md (uint6_operand): New predicate.
26819         (shift5_operand, shift6_operand): Likewise.
26820         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26821         arith32_operand.
26822         (lshrdi3): Use shift6_operand.
26823         (shrpsi4, shrpdi4): New insn patterns.
26824         (extzv): Delete expander.
26825         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26826         predicates in unamed zero extract patterns.  Tighten common constraint.
26827         (extv): Delete expander.
26828         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26829         predicates in unamed sign extract patterns.  Tighten common constraint.
26830         (insv): Delete expander.
26831         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26832         predicates in unamed insert patterns.  Tighten common constraint.
26833         Change uint32_operand predicate to uint6_operand predicate in unamed
26834         DImode pattern to insert constant values of type 1...1xxxx.
26836 2013-01-04  Jan Hubicka  <jh@suse.cz>
26838         PR tree-optimization/55823
26839         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26840         issue.
26842 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26843             Uros Bizjak  <ubizjak@gmail.com>
26845         PR rtl-optimization/55845
26846         * df-problems.c (can_move_insns_across): Stop scanning at
26847         volatile_insn_p source instruction or give up if
26848         across_from .. across_to range contains any volatile_insn_p
26849         instructions.
26851 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26853         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26854         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26855         Declare.
26856         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26857         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26859 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26861         PR fortran/55341
26862         * asan.c (asan_clear_shadow): New function.
26863         (asan_emit_stack_protection): Use it.
26865 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26867         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26868         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26869         with tab instead of space.
26871 2013-01-08  Nick Clifton  <nickc@redhat.com>
26873         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26874         register bank 0 at the start of an interrupt handler.
26875         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26876         MDBH registers.
26878 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26880         * config/aarch64/aarch64-simd.md
26881         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26882         (aarch64_simd_bsl): Likewise.
26883         (aarch64_vcond_internal<mode>): Likewise.
26884         (vcond<mode><mode>): Likewise.
26885         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26886         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26888 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26890         * config/aarch64/aarch64-builtins.c
26891         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26893 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26895         PR debug/55579
26896         * tree-sra.c (analyze_access_subtree): Return true also after
26897         potentially creating a debug-only replacement.
26899 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26901         PR middle-end/55890
26902         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26904         PR tree-optimization/54120
26905         * tree-vrp.c (range_fits_type_p): Don't allow
26906         src_precision < precision from signed vr to unsigned_p
26907         if vr->min or vr->max is negative.
26908         (simplify_float_conversion_using_ranges): Test can_float_p
26909         against CODE_FOR_nothing.
26911 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26912             Richard Biener  <rguenther@suse.de>
26914         PR middle-end/55851
26915         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26916         types instead of just INTEGER_TYPE types.
26918 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26920         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26921         TF_SIZE): Define.
26923 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26925         PR target/42661
26926         * config/mips/mips.opt: Change mad to mmad to match documentation.
26928 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26930         PR target/55897
26931         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26932         .progmemx.data now.
26934 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26936         PR target/55897
26937         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26938         (avr_addrspace_t): Add .section_name field.
26939         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26940         array size.
26941         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26942         NULL entry.  Put __memx into .progmemx.data.
26943         (progmem_section_prefix): Remove.
26944         (avr_asm_init_sections): No need to initialize progmem_section.
26945         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26946         section name prefix.
26947         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26948         retrieve the progmem section.
26949         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26950         boundary to run over avr_addrspace[].
26951         (avr_register_target_pragmas): Ditto.
26953 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26955         * varasm.c (output_constant_def_contents): For asan_protect_global
26956         protected strings, adjust DECL_ALIGN if needed, before testing for
26957         anchored symbols.
26958         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26959         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26960         normal decls.
26961         (output_object_block): For asan protected decls, emit asan padding
26962         after their contents.
26963         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26964         (asan_finish_file): Test it here instead.
26966 2013-01-07  Nick Clifton  <nickc@redhat.com>
26967             Matthias Klose  <doko@debian.org>
26968             Doug Kwan  <dougkwan@google.com>
26969             H.J. Lu  <hongjiu.lu@intel.com>
26971         PR driver/55470
26972         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26974         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26976         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26978         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26980         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26982 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26984         PR target/54461
26985         * doc/install.texi (Cross-Compiler-Specific Options): Document
26986         --with-avrlibc.
26988 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26990         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26991         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26992         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26993         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26994         vqmovun_high_s64): Fix source operand number and update copyright.
26996 2013-01-07  Richard Biener  <rguenther@suse.de>
26998         PR middle-end/55890
26999         * gimple.h (gimple_call_builtin_p): New overload.
27000         * gimple.c (validate_call): New function.
27001         (gimple_call_builtin_p): Likewise.
27002         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27003         Use gimple_call_builtin_p.
27004         (find_func_clobbers): Likewise.
27005         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
27006         (strlen_optimize_stmt): Likewise.
27008 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27010         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
27011         (vld1q_dup_*): Likewise.
27012         (vld1_*): Likewise.
27013         (vld1q_*): Likewise.
27014         (vld1_lane_*): Likewise.
27015         (vld1q_lane_*): Likewise.
27017 2013-01-07  Richard Biener  <rguenther@suse.de>
27019         * lto-streamer.h (LTO_minor_version): Bump to 2.
27021 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27023         * config/aarch64/aarch64-protos.h
27024         (aarch64_const_double_zero_rtx_p): Rename to...
27025         (aarch64_float_const_zero_rtx_p): ...this.
27026         (aarch64_float_const_representable_p): New.
27027         (aarch64_output_simd_mov_immediate): Likewise.
27028         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
27029         move immediate case.
27030         * config/aarch64/aarch64.c
27031         (aarch64_const_double_zero_rtx_p): Rename to...
27032         (aarch64_float_const_zero_rtx_p): ...this.
27033         (aarch64_print_operand): Allow printing of new constants.
27034         (aarch64_valid_floating_const): New.
27035         (aarch64_legitimate_constant_p): Check for valid floating-point
27036         constants.
27037         (aarch64_simd_valid_immediate): Likewise.
27038         (aarch64_vect_float_const_representable_p): New.
27039         (aarch64_float_const_representable_p): Likewise.
27040         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
27041         (aarch64_output_simd_mov_immediate): New.
27042         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
27043         (*movdf_aarch64): Likewise.
27044         * config/aarch64/constraints.md (Ufc): New.
27045         (Y): call aarch64_float_const_zero_rtx.
27046         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
27048 2013-01-07  Richard Biener  <rguenther@suse.de>
27050         PR tree-optimization/55888
27051         PR tree-optimization/55862
27052         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
27053         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
27054         not if it is contained therein.
27056 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27058         * config/avr/t-avr: Typo.
27060 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27062         PR55243
27063         * config/avr/t-avr: Don't automatically rebuild
27064         $(srcdir)/config/avr/t-multilib
27065         $(srcdir)/config/avr/avr-tables.opt
27066         $(srcdir)/doc/avr-mmcu.texi
27067         (avr-mcus): New phony target to build them on request.
27068         (s-avr-mlib, s-avr-mmcu-texi): Remove.
27069         * avr/avr-mcus.def: Adjust comments.
27071 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
27073         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
27075 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27077         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
27079 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27081         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27083 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27085         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27086         to generate profiling.
27087         * config/rs6000/aix64.h (LIB_SPEC): Same.
27089 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27091         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27092         New function.
27093         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27095 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27097         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27098         unconditionally.
27099         (ix86_expand_move): Ditto.
27100         (ix86_zero_extend_to_Pmode): Ditto.
27101         (ix86_expand_call): Ditto.
27102         (ix86_expand_special_args_builtin): Ditto.
27103         (ix86_expand_builtin): Ditto.
27105 2013-01-04  Richard Biener  <rguenther@suse.de>
27107         PR tree-optimization/55862
27108         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27109         translating them through PHI nodes.
27111 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27113         PR tree-optimization/55755
27114         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27115         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27117 2013-01-04  Marek Polacek  <polacek@redhat.com>
27119         PR middle-end/55859
27120         * opts.c (default_options_optimization): Clarify error message.
27122 2013-01-04  Richard Biener  <rguenther@suse.de>
27124         PR middle-end/55863
27125         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27126         reassociation.
27128 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27130         PR target/53789
27131         * config/pa/pa.md (movsi): Revert previous change.
27132         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27133         references.
27135 2013-01-03  Richard Henderson  <rth@redhat.com>
27137         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27138         after eliminating TLS symbols.
27140 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27142         PR bootstrap/50167
27143         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27144         * graphite-poly.c (debug_gmp_value): Likewise.
27146 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27148         PR target/55712
27149         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27150         selected code model, define __code_mode_small__, __code_model_medium__,
27151         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27152         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27153         xchg temporary register with %k.  Declare temporary register as
27154         early clobbered.
27155         [__x86_64__]: For medium and large code models, preserve %rbx register.
27157 2013-01-03  Richard Biener  <rguenther@suse.de>
27159         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27160         (dump_subscript): Adjust.
27161         (finalize_ddr_dependent): Do not dump redundant info.
27162         (analyze_siv_subscript): Adjust.
27163         (subscript_dependence_tester): Likewise.
27164         (compute_affine_dependence): Likewise.
27166 2013-01-03  Richard Biener  <rguenther@suse.de>
27168         Revert
27169         2013-01-03  Richard Biener  <rguenther@suse.de>
27171         PR tree-optimization/55857
27172         * tree-vect-stmts.c (vectorizable_load): Do not setup
27173         re-alignment for invariant loads.
27175         2013-01-02  Richard Biener  <rguenther@suse.de>
27177         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27178         invariant load do not generate a vector load from the scalar location.
27180 2013-01-03  Richard Biener  <rguenther@suse.de>
27182         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27183         for not vectorizing.
27184         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27185         not build INDIRECT_REFs, call get_name once only.
27186         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27187         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27189 2013-01-03  Richard Biener  <rguenther@suse.de>
27191         PR tree-optimization/55857
27192         * tree-vect-stmts.c (vectorizable_load): Do not setup
27193         re-alignment for invariant loads.
27195 2013-01-03  Richard Biener  <rguenther@suse.de>
27197         PR lto/55848
27198         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27199         prefer a built-in decl.
27201 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27203         * gcc.c (process_command): Update copyright notice dates.
27204         * gcov.c (print_version): Likewise.
27205         * gcov-dump.c (print_version): Likewise.
27207         PR rtl-optimization/55838
27208         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27209         iv0.step, iv1.step and step.
27211 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27212             Marc Glisse  <marc.glisse@inria.fr>
27214         PR tree-optimization/55832
27215         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27216         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27217         integer_{one,zero}_node.
27219 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27221         PR debug/54402
27222         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27223         * var-tracking.c (reverse_op): Don't add reverse ops to
27224         VALUEs that have already
27225         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27227 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27229         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27231 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27233         * dumpfile.c (dump_loc): Print filename with location.
27234         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27235         new location_t parameter to emit complete unroll message with
27236         new dump framework.
27237         (canonicalize_loop_induction_variables): Compute loops location
27238         and pass to try_unroll_loop_completely.
27239         * loop-unroll.c (report_unroll_peel): New function.
27240         (peel_loops_completely): Use new dump format with location
27241         for main dumpfile message, and invoke report_unroll_peel on success.
27242         (decide_unrolling_and_peeling): Ditto.
27243         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27244         by report_unroll_peel.
27245         (decide_peel_completely): Ditto.
27246         (decide_unroll_constant_iterations): Ditto.
27247         (decide_unroll_runtime_iterations): Ditto.
27248         (decide_peel_simple): Ditto.
27249         (decide_unroll_stupid): Ditto.
27250         * cfgloop.c (get_loop_location): New function.
27251         * cfgloop.h (get_loop_location): Declare.
27253 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27255         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27256         NULL.
27258 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27260         PR middle-end/55198
27261         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27262         BLKmode objects when EXPAND_MEMORY is specified.
27264 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27266         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27267         in loop predicate.
27268         (fold_builtin_cpu): Do not share cpu model decls across statements.
27270 2013-01-02  Jason Merrill  <jason@redhat.com>
27272         PR c++/55804
27273         * tree.c (build_array_type_1): Revert earlier change.
27275 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27277         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27278         "cortex-a57".
27279         * config/aarch64/aarch64-tune.md: Re-generate.
27281 2013-01-02  Richard Biener  <rguenther@suse.de>
27283         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27284         invariant load do not generate a vector load from the scalar location.
27286 2013-01-02  Richard Biener  <rguenther@suse.de>
27288         PR bootstrap/55784
27289         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27290         * configure: Regenerate.
27292 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27294         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27295         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27296         (expand_builtin_int_roundingfn_2): Keep the original target around
27297         for the fallback case.
27299 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27301         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27302         to be clear for sign changes.
27304 2013-01-01  Jan Hubicka  <jh@suse.cz>
27306         * ipa-inline-analysis.c: Fix formatting.
27308 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27310         PR tree-optimization/55831
27311         * tree-vect-loop.c (get_initial_def_for_induction): Use
27312         gsi_after_labels instead of gsi_start_bb.
27314 Copyright (C) 2013 Free Software Foundation, Inc.
27316 Copying and distribution of this file, with or without modification,
27317 are permitted in any medium without royalty provided the copyright
27318 notice and this notice are preserved.