2013-10-22 Jan-Benedict Glaw <jbglaw@lug-owl.de>
[official-gcc.git] / gcc / ChangeLog
blobcfb618710e67cf56c6278a8995c0cfe4a148779a
1 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3         * config/tilepro/tilepro.c: Include "tree.h".
5 2013-10-22  Andreas Schwab  <schwab@suse.de>
7         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
8         with PRE_DEC.
10 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
12         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
13         [Ed Smith-Rowland]): New entries.
14         ([Stephen M. Webb]): Update.
16 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
18         * tree-ssa-ter.h: Remove duplicate copy of file contents.
20 2013-10-21  Marek Polacek  <polacek@redhat.com>
22         PR middle-end/58809
23         * fold-const.c (fold_range_test): Return 0 if the type is not
24         an integral type.
26 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
28         * system.h: Move hwint.h include further down.
29         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
30         gcc_checking_asserts.
31         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
33 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
35         Fix volatile issues in optimize_bit_field_compare.
36         * fold-const.c (optimize_bit_field_compare): Bail out if
37         lvolatilep or rvolatilep.
39 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
41         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
42         and get_inner_reference returning different pmode for non-volatile
43         bit-field members dependent on flag_strict_volatile_bitfields.
44         * stor-layout.c (layout_decl): Remove special handling of
45         flag_strict_volatile_bitfields.
46         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
47         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
49 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
51         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
52         calculations.
54 2013-10-21  Jeff Law  <law@redhat.com>
56         * tree-ssa-threadedge.c (thread_through_normal_block): New
57         argument VISITED.  Remove VISISTED as a local variable.  When we
58         have a threadable jump, verify the destination of the jump has not
59         been visised.
60         (thread_across_edge): Allocate VISITED bitmap once at function
61         scope and use it throughout.  Make sure to set appropriate bits in
62         VISITED for E (start of jump thread path).
63         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
64         through a joiner if any edge on the path has a recorded jump thread.
66 2013-10-21  Ian Lance Taylor  <iant@google.com>
68         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
69         don't imply that attributes can solve all problems.
70         (Directory Options): Fix typo.
72 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
74         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
75         extend and extend_add.
77 2013-10-21  Richard Biener  <rguenther@suse.de>
79         PR tree-optimization/58794
80         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
81         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
83 2013-10-21  Richard Biener  <rguenther@suse.de>
85         PR middle-end/58742
86         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
87         to (T) X for sign-changing conversions (or no conversion).
89 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
91         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
93 2013-10-20  Jan Hubicka  <jh@suse.cz>
95         * config/i386/i386-tune.def: Add comment; organize into categories
97 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
99         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
100         argument.  Update function comment.
101         (expand_set_or_movmem_via_rep): New function combining
102         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
103         (expand_movmem_via_rep_mov): Remove.
104         expand_setmem_via_rep_stos): Remove.
105         (expand_movmem_epilogue): Update calls correspondingly.
106         (expand_setmem_epilogue_via_loop): Likewise.
107         (emit_memset): New.
108         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
109         (expand_set_or_movmem_prologue): New function combining
110         expand_movmem_prologue and expand_setmem_prologue.
111         (expand_movmem_prologue): Remove.
112         (expand_setmem_prologue): Remove.
113         (expand_set_or_movmem_constant_prologue): New function combining
114         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
115         (expand_constant_movmem_prologue): Remove.
116         (expand_constant_setmem_prologue): Remove.
117         (promote_duplicated_reg): Allow vector-const0 value.
118         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
119         and ix86_expand_setmem.
120         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
121         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
123 2013-10-21  Diego Novillo  <dnovillo@google.com>
125         * asan.c: Include tree.h
126         * bb-reorder.c: Likewise.
127         * cfgcleanup.c: Likewise.
128         * cfgloopmanip.c: Likewise.
129         * data-streamer-in.c: Likewise.
130         * data-streamer-out.c: Likewise.
131         * data-streamer.c: Likewise.
132         * dwarf2cfi.c: Likewise.
133         * graphite-blocking.c: Likewise.
134         * graphite-clast-to-gimple.c: Likewise.
135         * graphite-dependences.c: Likewise.
136         * graphite-interchange.c: Likewise.
137         * graphite-optimize-isl.c: Likewise.
138         * graphite-poly.c: Likewise.
139         * graphite-scop-detection.c: Likewise.
140         * graphite-sese-to-poly.c: Likewise.
141         * graphite.c: Likewise.
142         * ipa-devirt.c: Likewise.
143         * ipa-profile.c: Likewise.
144         * ipa.c: Likewise.
145         * ira.c: Likewise.
146         * loop-init.c: Likewise.
147         * loop-unroll.c: Likewise.
148         * lower-subreg.c: Likewise.
149         * lto/lto-object.c: Likewise.
150         * recog.c: Likewise.
151         * reginfo.c: Likewise.
152         * tree-loop-distribution.c: Likewise.
153         * tree-parloops.c: Likewise.
154         * tree-ssa-strlen.c: Likewise.
155         * tree-streamer.c: Likewise.
156         * value-prof.c: Likewise.
157         * target-globals.c: Likewise.
158         * expr.h: Include tree-core.h instead of tree.h.
159         * gimple.h: Likewise.
160         * ipa-prop.h: Likewise.
161         * ipa-utils.h: Likewise.
162         * lto-streamer.h: Likewise.
163         * streamer-hooks.h: Likewise.
164         * ipa-reference.h: Include cgraph.h instead of tree.h.
165         * cgraph.h: Include basic-block.h instead of tree.h.
166         * tree-streamer.h: Do not include tree.h.
167         * genattrtab.c (write_header): Generate inclusion of
168         tree.h.
169         * genautomata.c (main): Likewise.
170         * genemit.c: Likewise.
171         * genopinit.c: Likewise.
172         * genoutput.c (output_prologue): Likewise.
173         * genpeep.c: Likewise.
175 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
177         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
178         little endian.
179         (vec_unpacku_hi_v8hi): Likewise.
180         (vec_unpacku_lo_v16qi): Likewise.
181         (vec_unpacku_lo_v8hi): Likewise.
183 2013-10-20  Jan Hubicka  <jh@suse.cz>
185         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
186         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
187         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
189 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
191         * config/mips/mips.h (ISA_HAS_WSBH): Define.
192         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
193         constants.
194         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
196 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
198         PR target/58603
199         * system.h: Undef m_slot.
201 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
203         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
204         all elements for both endian flavors.
206 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
208         PR target/58792
209         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
210         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
211         and SI_REG for 64bit SYSV ABI targets.
213 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
215         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
216         to multi_reg_return.  Clarify that we are skipping USEs of multiple
217         return registers.  Use bool type where appropriate.
219 2013-10-18  Jan Hubicka  <jh@suse.cz>
221         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
222         for cold functions.
223         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
224         (X86_TUNE_PUSH_MEMORY): Likewise.
225         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
226         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
227         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
228         New.
229         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
230         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
231         Remove.
232         (ix86_option_override_internal): Update to use tune features instead
233         of variables.
235 2013-10-18  Cong Hou  <congh@google.com>
237         PR tree-optimization/58508
238         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
239         statement that contains data refs with zero-step.
241 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
243         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
244         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
245         * gimple-low.c (gimple_check_call_arg,
246         gimple_check_call_matching_types): Move to cgraph.c.
247         * gimple-low.h: Remove prototype.
248         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
249         Relocate from gimple-low.c.
250         * cgraph.h: Add prototype.  Don't include basic-block.h.
251         * gimplify.c: Add gimple-low to include list.
252         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
253         * tree-eh.c: Add gimple-low to include list.
254         * tree-nested.c: Likewise.
255         * cfgexpand.c: Add tree-ssa-address.h to include list.
256         * expr.c: Likewise.
257         * gimple-fold.c: Likewise.
258         * gimple-ssa-strength-reduction.c: Likewise.
259         * trans-mem.c: Likewise.
260         * tree-mudflap.c: Likewise.
261         * tree-ssa-loop-ivopts.c: Likewise.
262         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
263         (degenerate_phi_result): Move to tree-phinodes.c.
264         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
265         * tree-ssa-threadedge.c: Likewise.
266         * tree-vrp.c: Likewise.
267         * tree-phinodes.c (degenerate_phi_result): Relocate here.
268         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
269         * tree-phinodes.h (degenerate_phi_result): Add prototype.
270         * tree-ssa-copy.c: Include tree-ssa-dom.h.
271         * tree-ssa-forwprop.c: Likewise.
272         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
273         pass_data_cleanup_cfg_post_optimizing,
274         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
275         * tree-optimize.c: Delete File.
276         * graphite.c: Include tree-cfgcleanup.h.
277         * passes.c: Likewise.
278         * tree-cfg.c: Likewise.
279         * tree-profile.c: Likewise.
280         * tree-ssa-dse.c: Likewise.
281         * tree-ssa-loop-ivcanon.c: Likewise.
282         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
283         * tree-outof-ssa.c: Include sbitmap.h.
284         * tree-ssa-live.c: Likewise.
285         * tree-ssa-propagate.c: Likewise.
286         * tree-ssa-structalias.c: Likewise.
287         * tree-stdarg.c: Likewise.
288         * Makefile.in (OBJS): Delete tree-optimize.o.
289         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
290         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
291         * varasm.c: Include basic-block.h.
292         * cfgloop.h: Include function.h instead of basic-block.h
293         (bb_loop_depth): Move to cfgloop.c.
294         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
296 2013-10-18  Teresa Johnson  <tejohnson@google.com>
298         * predict.c (probably_never_executed): Compare frequency-based
299         count to number of training runs.
300         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
302 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
304         * config/arm/arm.c (cortexa9_extra_costs): New table.
305         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
307 2013-10-18  Jeff Law  <law@redhat.com>
309         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
311         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
312         (register_jump_thread): Add "registered_jump_thread" debug
313         counter support.
314         * dbgcnt.def (registered_jump_thread): New debug counter.
316 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
318         * config/rs6000/rs6000.c: Include cgraph.h.
320 2013-10-18  Teresa Johnson  <tejohnson@google.com>
322         * tree-ssa-tail-merge.c (replace_block_by): Update edge
323         weights during merging.
325 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
327         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
328         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
329         * tree-ssa-operands.h: Remove prototype.
330         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
331         * gimple.c (virtual_operand_p): Relocate from gimple.c.
332         * gimple.h: Add prototype.
333         * gimple-ssa.h: Include tree-ssa-operands.h.
334         * tree-dump.c: Add tree-cfg.h to include list.
335         * tree-ssa-alias.c: Add ipa-reference.h to include list.
336         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
337         * config/i386/i386.c: Don't include tree-flow.h.
338         * config/rs6000/rs6000.c: Likewise.
340 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
342         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
344 2013-10-18  Richard Biener  <rguenther@suse.de>
346         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
347         or TYPE_UNSIGNED of integral types.
348         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
349         NULL_TREE for zero-precision integral types.
351 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
353         * config/aarch64/arm_neon.h
354         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
356 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
358         * opth-gen.awk: Define target_flags_explicit.
360 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
362         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
363         fields to the reg_addr array that describes the valid addressing
364         mode for any register, general purpose registers, floating point
365         registers, and Altivec registers.
366         (FIRST_RELOAD_REG_CLASS): Likewise.
367         (LAST_RELOAD_REG_CLASS): Likewise.
368         (struct reload_reg_map_type): Likewise.
369         (reload_reg_map_type): Likewise.
370         (RELOAD_REG_VALID): Likewise.
371         (RELOAD_REG_MULTIPLE): Likewise.
372         (RELOAD_REG_INDEXED): Likewise.
373         (RELOAD_REG_OFFSET): Likewise.
374         (RELOAD_REG_PRE_INCDEC): Likewise.
375         (RELOAD_REG_PRE_MODIFY): Likewise.
376         (reg_addr): Likewise.
377         (mode_supports_pre_incdec_p): New helper functions to say whether
378         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
379         (mode_supports_pre_modify_p): Likewise.
380         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
381         print the valid address mode bits for each mode.
382         (rs6000_debug_print_mode): Likewise.
383         (rs6000_debug_reg_global): Likewise.
384         (rs6000_setup_reg_addr_masks): New function to set up the address
385         mask bits for each type.
386         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
387         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
388         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
389         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
390         PRE_MODIFY are supported.
391         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
392         registers, instead of {src,dest}_av_p.
393         (rs6000_print_options_internal): Tweak the debug output slightly.
395 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
397         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
398         isa attribute.
400 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
402         * tree-flow.h (struct omp_region): Move to omp-low.c.
403         Remove omp_ prototypes and variables.
404         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
405         (copy_var_decl): Relocate prototype from tree-flow.h.
406         * gimple.c (copy_var_decl): Relocate from omp-low.c.
407         * tree.h: Move prototype to omp-low.h.
408         * omp-low.h: New File.  Relocate prototypes here.
409         * omp-low.c (struct omp_region): Make local here.
410         (root_omp_region): Make static.
411         (copy_var_decl) Move to gimple.c.
412         (new_omp_region): Make static.
413         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
414         * tree-cfg.c: Include omp-low.h.
415         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
416         * gimplify.c: Include omp-low.h.
417         * tree-parloops.c: Likewise.
419 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
421         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
422         operands are in memory, prefer to force non-matched operand 1 to
423         the register.
425 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
427         PR target/58673
428         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
429         restrict TImode addresses to single indirect registers if both
430         -mquad-memory and -mvsx-timode are used.
431         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
432         we should emit load/store quad.  Remove using %y for quad memory
433         addresses.
435         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
436         constraints to allow load/store quad on machines where TImode is
437         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
438         for TImode to load integer constants.
440 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
442         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
444 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
446         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
447         handling of STACK_REG.
449 2013-10-17  Richard Biener  <rguenther@suse.de>
451         PR tree-optimization/58143
452         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
453         New function.
454         (rewrite_to_defined_overflow): Likewise.
455         (move_computations_dom_walker::before_dom): Rewrite stmts
456         with undefined signed overflow that are not always executed
457         into unsigned arithmetic.
459 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
461         PR target/57756
462         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
463         explicit isa flag to be an options variable, instead of using
464         global_options_set.  Remove define from rs6000.h.
465         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
467         * config/rs6000/rs6000.c (rs6000_option_override_internal):
468         Initialize rs6000_isa_flags_explicit.
469         (rs6000_function_specific_save): Add gcc_options* parameter, so
470         that the powerpc builds after the 2013-10-15 changes.
471         (rs6000_function_specific_restore): Likewise.
473 2013-10-16  DJ Delorie  <dj@redhat.com>
475         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
476         op is a REG before checking REGNO.
477         (rl78_alloc_physical_registers): Verify pattern is a SET before
478         checking SET_SRC.
480 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
482         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
483         endianness.
484         (vec_unpacks_lo_v4sf): Likewise.
485         (vec_unpacks_float_hi_v4si): Likewise.
486         (vec_unpacks_float_lo_v4si): Likewise.
487         (vec_unpacku_float_hi_v4si): Likewise.
488         (vec_unpacku_float_lo_v4si): Likewise.
490 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
492         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
493         (vsx_concat_v2sf): Likewise.
495 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
497         * config/aarch64/aarch64.md
498         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
500 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
502         PR tree-optimization/58697
503         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
504         estimated_loop_iterations_int.
505         (max_stmt_executions_int): Call get_max_loop_iterations_int.
506         (get_max_loop_iterations_int): New.  HWINT version of
507         get_max_loop_iterations.
508         * cfgloop.h: Add prototypes.
509         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
510         * loop-unroll.c (decide_peel_once_rolling): Call
511         get_estimated_loop_iterations_int.
512         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
513         * tree-ssa-loop-niter.h: Tweak prototypes.
515 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
517         * gengtype-parse.c (struct_field_seq): Ignore access-control
518         keywords ("public:" etc).
520 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
522         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
523         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
525 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
527         * config/arm/arm.opt (mlra): New option.
528         * config/arm/arm.c (arm_lra_p): New function.
529         (TARGET_LRA_P): Define.
531 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
533         * tree-core.h (tree_code_name): Remove.
534         * tree.h (get_tree_code_name): New prototype.
535         * tree.c (tree_code_name): Make static.
536         (get_tree_code_name): New function.
537         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
538         tree_class_check_failed, tree_range_check_failed,
539         tree_not_class_check_failed, omp_clause_check_failed,
540         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
541         wrapper get_tree_code_name instead of calling tree_code_name directly.
542         * tree-vrp.c (dump_asserts_for): Likewise.
543         * tree-dump.c (dequeue_and_dump): Likewise.
544         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
545         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
546         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
547         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
548         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
549         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
550         dump_gimple_omp_for): Likewise.
551         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
552         * tree-ssa-pre.c (print_pre_expr): Likewise.
553         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
554         * print-tree.c (print_node_brief, print_node): Likewise.
555         * gimple.c (gimple_check_failed): Likewise.
556         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
557         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
558         * config/mep/mep.c (mep_validate_vliw): Likewise.
559         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
560         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
562 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
564         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
565         for AMD bdver3.
567 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
569         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
570         (MULTILIB_MATCHES): Add multilib for -march=v8.
572 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
574         PR target/57756
575         * optc-save-gen.awk: Add extra parameter to the save and restore
576         target calls.
577         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
578         * tree.c (build_optimization_node): New parameter.  Add extra parameter
579         to call to cl_optimization_save.
580         (build_target_option_node): New parameter. Add extra parameter
581         to call to cl_target_option_save.
582         * tree.h (build_optimization_node): New parameter.
583         (build_target_option_node): New parameter.
584         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
585         build_optimization_node and build_target_option_node.
586         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
587         (handle_pragma_push_options): Ditto.
588         * toplev.c (process_options): Ditto.
589         * opts.c (init_options_struct): Check for opts_set non-null.
590         * target.def (target_option.save): New parameter.
591         (target_option.restore): New parameter.
592         * tm.texi: Generate.
593         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
594         (ix86_pragma_target_parse): Ditto.
595         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
596         parameters.
597         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
598         to  build_optimization_node and build_target_option_node.
599         (rs6000_valid_attribute_p): Ditto.
600         (rs6000_pragma_target_parse): Ditto.
601         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
602         data.
603         * config/i386/i386.h:
604         TARGET_64BIT_P: New Macro
605         TARGET_MMX_P: New Macro.
606         TARGET_3DNOW_P: New Macro.
607         TARGET_3DNOW_A_P: New Macro.
608         TARGET_SSE_P: New Macro.
609         TARGET_SSE2_P: New Macro.
610         TARGET_SSE3_P: New Macro.
611         TARGET_SSSE3_P: New Macro.
612         TARGET_SSE4_1_P: New Macro.
613         TARGET_SSE4_2_P: New Macro.
614         TARGET_AVX_P: New Macro.
615         TARGET_AVX2_P: New Macro.
616         TARGET_AVX512F_P: New Macro.
617         TARGET_AVX512PF_P: New Macro.
618         TARGET_AVX512ER_P: New Macro.
619         TARGET_AVX512CD_P: New Macro.
620         TARGET_FMA_P: New Macro.
621         TARGET_SSE4A_P: New Macro.
622         TARGET_FMA4_P: New Macro.
623         TARGET_XOP_P: New Macro.
624         TARGET_LWP_P: New Macro.
625         TARGET_ABM_P: New Macro.
626         TARGET_BMI_P: New Macro.
627         TARGET_BMI2_P: New Macro.
628         TARGET_LZCNT_P: New Macro.
629         TARGET_TBM_P: New Macro.
630         TARGET_POPCNT_P: New Macro.
631         TARGET_SAHF_P: New Macro.
632         TARGET_MOVBE_P: New Macro.
633         TARGET_CRC32_P: New Macro.
634         TARGET_AES_P: New Macro.
635         TARGET_PCLMUL_P: New Macro.
636         TARGET_CMPXCHG16B_P: New Macro.
637         TARGET_FSGSBASE_P: New Macro.
638         TARGET_RDRND_P: New Macro.
639         TARGET_F16C_P: New Macro.
640         TARGET_RTM_P: New Macro.
641         TARGET_HLE_P: New Macro.
642         TARGET_RDSEED_P: New Macro.
643         TARGET_PRFCHW_P: New Macro.
644         TARGET_ADX_P: New Macro.
645         TARGET_FXSR_P: New Macro.
646         TARGET_XSAVE_P: New Macro.
647         TARGET_XSAVEOPT_P: New Macro.
648         TARGET_LP64_P: New Macro.
649         TARGET_X32_P: New Macro.
650         TARGET_FPMATH_DEFAULT_P: New Macro.
651         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
652         * config/i386/i386.c (ix86_option_override_internal): New parameters.
653         opts and opts_set.
654         Change ix86_tune_string to access opts->x_ix86_tune_string.
655         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
656         Change ix86_arch_string to access opts->x_ix86_arch_string.
657         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
658         Change ix86_pmode to access opts->x_ix86_pmode.
659         Change ix86_abi to access opts->x_ix86_abi.
660         Change ix86_cmodel to access opts->x_ix86_cmodel.
661         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
662         Change ix86_isa_flags_explicit to access
663         opts->x_ix86_isa_flags_explicit.
664         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
665         Change ix86_regparm to access opts->x_ix86_regparm.
666         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
667         Change ix86_preferred_stack_boundary_arg to access
668         opts->x_ix86_preferred_stack_boundary_arg.
669         Change ix86_force_align_arg_pointer to access
670         opts->x_ix86_force_align_arg_pointer.
671         Change ix86_incoming_stack_boundar_arg to access
672         opts->x_ix86_incoming_stack_boundar_arg.
673         Change ix86_fpmath to access opts->x_ix86_fpmath.
674         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
675         Change ix86_recip_name to access opts->x_ix86_recip_name.
676         Change ix86_stack_protector_guard to access
677         opts->x_ix86_stack_protector_guard.
678         Change ix86_tune_memcpy_strategy to access
679         opts->x_ix86_tune_memcpy_strategy.
680         Change ix86_tune_memset_strategy to access
681         opts->x_ix86_tune_memset_strategy.
682         Change global_options to access opts.
683         Change global_options_set to access opts_set.
684         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
685         Change TARGET_MMX to TARGET_MMX_P (opts->...).
686         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
687         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
688         Change TARGET_SSE to TARGET_SSE_P (opts->...).
689         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
690         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
691         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
692         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
693         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
694         Change TARGET_AVX to TARGET_AVX_P (opts->...).
695         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
696         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
697         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
698         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
699         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
700         Change TARGET_FMA to TARGET_FMA_P (opts->...).
701         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
702         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
703         Change TARGET_XOP to TARGET_XOP_P (opts->...).
704         Change TARGET_LWP to TARGET_LWP_P (opts->...).
705         Change TARGET_ABM to TARGET_ABM_P (opts->...).
706         Change TARGET_BMI to TARGET_BMI_P (opts->...).
707         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
708         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
709         Change TARGET_TBM to TARGET_TBM_P (opts->...).
710         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
711         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
712         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
713         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
714         Change TARGET_AES to TARGET_AES_P (opts->...).
715         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
716         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
717         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
718         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
719         Change TARGET_F16C to TARGET_F16C_P (opts->...).
720         Change TARGET_RTM to TARGET_RTM_P (opts->...).
721         Change TARGET_HLE to TARGET_HLE_P (opts->...).
722         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
723         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
724         Change TARGET_ADX to TARGET_ADX_P (opts->...).
725         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
726         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
727         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
728         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
729         Change TARGET_X32 to TARGET_X32_P (opts->...).
730         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
731         Change TARGET_FLOAT_RETURNS_IN_80387 to
732         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
733         (ix86_function_specific_save): New parameter. Use opts-> fields
734         to replace global fields.
735         (ix86_function_specific_restore): Ditto.
736         (ix86_valid_target_attribute_inner_p): New parameters.
737         Fix recursive call.
738         Fix call to ix86_handle_option and set_option.
739         (ix86_valid_target_attribute_tree): New parameters.
740         Change global_options to access opts.
741         Change global_options_set to access opts_set.
742         Fix call to ix86_valid_target_attribute_inner_p.
743         Change ix86_tune_string to access opts->x_ix86_tune_string.
744         Change ix86_arch_string to access opts->x_ix86_arch_string.
745         Change ix86_fpmath to access opts->x_ix86_fpmath
746         Fix call to ix86_option_override_internal.
747         Fix call to ix86_add_new_builtins.
748         Fix calls to build_optimization_node and build_target_option_node.
749         (ix86_valid_target_attribute_p): Remove access to global_options.
750         Use new gcc_options structure func_options.
751         Fix call to ix86_valid_target_attribute_tree.
752         Fix call to  build_optimization_node.
753         (get_builtin_code_for_version): Fix call to
754         ix86_valid_target_attribute_tree.
756 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
758         * Makefile.in (PICFLAG): New.
759         (enable_host_shared): New.
760         (INTERNAL_CFLAGS): Use PICFLAG.
761         (LIBIBERTY): Use pic build of libiberty.a if configured with
762         --enable-host-shared.
763         * configure.ac: Add --enable-host-shared, setting up new
764         PICFLAG variable.
765         * configure: Regenerate.
766         * doc/install.texi (--enable-shared): Add note contrasting it with ...
767         (--enable-host-shared): New option.
769 2013-10-15  Richard Biener  <rguenther@suse.de>
771         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
772         for built-in functions.
774 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
776         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
777         (optimize_range_tests_1): New function,
778         extracted from optimize_range_tests.
779         (optimize_range_tests_xor): Similarly.
780         (optimize_range_tests_diff): New function.
781         (optimize_range_tests): Use optimize_range_tests_1.
783 2013-10-15  Cong Hou  <congh@google.com>
785         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
786         requirement of the reduction pattern so that one operand of the
787         reduction operation can come from outside of the loop.
789 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
791         * config/arm/neon-schedgen.ml: Remove.
792         * config/arm/cortex-a9-neon.md: Remove comment regarding
793         neon-schedgen.ml.
795 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
797         * config/arm/types: Remove old neon types.
799 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
801         * config/arm/cortex-a7.md
802         (cortex_a7_neon_type): New.
803         (cortex_a7_neon_mul): Update for new types.
804         (cortex_a7_neon_mla): Likewise.
805         (cortex_a7_neon): Likewise.
807 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
809         * config/arm/cortex-a15-neon.md
810         (cortex_a15_neon_type): New,
812         (cortex_a15_neon_int_1): Remove.
813         (cortex_a15_neon_int_2): Likewise.
814         (cortex_a15_neon_int_3): Likewise.
815         (cortex_a15_neon_int_4): Likewise.
816         (cortex_a15_neon_int_5): Likewise.
817         (cortex_a15_neon_vqneg_vqabs): Likewise.
818         (cortex_a15_neon_vmov): Likewise.
819         (cortex_a15_neon_vaba): Likewise.
820         (cortex_a15_neon_vaba_qqq): Likewise.
821         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
822         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
823         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
824         Likewise.
825         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
826         (cortex_a15_neon_mla_qqq_8_16): Likewise.
827         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
828         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
829         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
830         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
831         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
832         (cortex_a15_neon_shift_1): Likewise.
833         (cortex_a15_neon_shift_2): Likewise.
834         (cortex_a15_neon_shift_3): Likewise.
835         (cortex_a15_neon_vshl_ddd): Likewise.
836         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
837         (cortex_a15_neon_vsra_vrsra): Likewise.
838         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
839         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
840         (cortex_a15_neon_bp_3cycle): Likewise.
841         (cortex_a15_neon_ldm_2): Likewise.
842         (cortex_a15_neon_stm_2): Likewise.
843         (cortex_a15_neon_mcr): Likewise.
844         (cortex_a15_neon_mrc): Likewise.
845         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
846         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
847         (cortex_a15_neon_fp_vmul_ddd): Likewise.
848         (cortex_a15_neon_fp_vmul_qqd): Likewise.
849         (cortex_a15_neon_fp_vmla_ddd): Likewise.
850         (cortex_a15_neon_fp_vmla_qqq): Likewise.
851         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
852         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
853         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
854         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
855         (cortex_a15_neon_bp_simple): Likewise.
856         (cortex_a15_neon_bp_2cycle): Likewise.
857         (cortex_a15_neon_bp_3cycle): Likewise.
858         (cortex_a15_neon_vld1_1_2_regs): Likewise.
859         (cortex_a15_neon_vld1_3_4_regs): Likewise.
860         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
861         (cortex_a15_neon_vld2_4_regs): Likewise.
862         (cortex_a15_neon_vld3_vld4): Likewise.
863         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
864         (cortex_a15_neon_vst1_3_4_regs): Likewise.
865         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
866         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
867         (cortex_a15_neon_vst3_vst4): Rename to...
868         (cortex_a15_neon_vst4): This, update for new attributes.
869         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
870         (cortex_a15_neon_vld3_vld4_lane): Likewise.
871         (cortex_a15_neon_vst1_vst2_lane): Likewise.
872         (cortex_a15_neon_vst3_vst4_lane): Likewise.
873         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
874         (cortex_a15_neon_ldm_2): Likewise.
875         (cortex_a15_neon_stm_2): Likewise.
876         (cortex_a15_neon_mcr): Likewise.
877         (cortex_a15_neon_mcr_2_mcrr): Likewise.
878         (cortex_a15_neon_mrc): Likewise.
879         (cortex_a15_neon_mrrc): Likewise.
881         (cortex_a15_neon_abd): New.
882         (cortex_a15_neon_abd_q): Likewise.
883         (cortex_a15_neon_aba): Likewise.
884         (cortex_a15_neon_aba_q): Likewise.
885         (cortex_a15_neon_acc): Likewise.
886         (cortex_a15_neon_acc_q): Likewise.
887         (cortex_a15_neon_arith_basic): Likewise.
888         (cortex_a15_neon_arith_complex): Likewise.
889         (cortex_a15_neon_multiply): Likewise.
890         (cortex_a15_neon_multiply_q): Likewise.
891         (cortex_a15_neon_mla): Likewise.
892         (cortex_a15_neon_mla_q): Likewise.
893         (cortex_a15_neon_sat_mla_long): Likewise.
894         (cortex_a15_neon_shift_acc): Likewise.
895         (cortex_a15_neon_shift_imm_basic): Likewise.
896         (cortex_a15_neon_shift_imm_complex): Likewise.
897         (cortex_a15_neon_shift_reg_basic): Likewise.
898         (cortex_a15_neon_shift_reg_basic_q): Likewise.
899         (cortex_a15_neon_shift_reg_complex): Likewise.
900         (cortex_a15_neon_shift_reg_complex_q): Likewise.
901         (cortex_a15_neon_fp_negabs): Likewise
902         (cortex_a15_neon_fp_arith): Likewise
903         (cortex_a15_neon_fp_arith_q): Likewise
904         (cortex_a15_neon_fp_cvt_int): Likewise
905         (cortex_a15_neon_fp_cvt_int_q): Likewise
906         (cortex_a15_neon_fp_cvt_16): Likewise
907         (cortex_a15_neon_fp_mul): Likewise
908         (cortex_a15_neon_fp_mul_q): Likewise
909         (cortex_a15_neon_fp_mla): Likewise
910         (cortex_a15_neon_fp_mla_q): Likewise
911         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
912         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
913         (cortex_a15_neon_bitops): Likewise.
914         (cortex_a15_neon_bitops_q): Likewise.
915         (cortex_a15_neon_from_gp): Likewise.
916         (cortex_a15_neon_from_gp_q): Likewise.
917         (cortex_a15_neon_tbl3_tbl4): Likewise.
918         (cortex_a15_neon_zip_q): Likewise.
919         (cortex_a15_neon_to_gp): Likewise.
920         (cortex_a15_neon_load_a): Likewise.
921         (cortex_a15_neon_load_b): Likewise.
922         (cortex_a15_neon_load_c): Likewise.
923         (cortex_a15_neon_load_d): Likewise.
924         (cortex_a15_neon_load_e): Likewise.
925         (cortex_a15_neon_load_f): Likewise.
926         (cortex_a15_neon_store_a): Likewise.
927         (cortex_a15_neon_store_b): Likewise.
928         (cortex_a15_neon_store_c): Likewise.
929         (cortex_a15_neon_store_d): Likewise.
930         (cortex_a15_neon_store_e): Likewise.
931         (cortex_a15_neon_store_f): Likewise.
932         (cortex_a15_neon_store_g): Likewise.
933         (cortex_a15_neon_store_h): Likewise.
934         (cortex_a15_vfp_to_from_gp): Likewise.
936 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
938         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
940         (cortex_a9_neon_vshl_ddd): Remove.
941         (cortex_a9_neon_vst3_vst4): Likewise.
942         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
944         (cortex_a9_neon_bit_ops_q): New.
946         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
947         (cortex_a9_neon_int_2): Likewise.
948         (cortex_a9_neon_int_3): Likewise.
949         (cortex_a9_neon_int_4): Likewise.
950         (cortex_a9_neon_int_5): Likewise.
951         (cortex_a9_neon_vqneg_vqabs): Likewise.
952         (cortex_a9_neon_vmov): Likewise.
953         (cortex_a9_neon_vaba): Likewise.
954         (cortex_a9_neon_vaba_qqq): Likewise.
955         (cortex_a9_neon_shift_1): Likewise.
956         (cortex_a9_neon_shift_2): Likewise.
957         (cortex_a9_neon_shift_3): Likewise.
958         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
959         (cortex_a9_neon_vsra_vrsra): Likewise.
960         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
961         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
962         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
963         Likewise.
964         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
965         (cortex_a9_neon_mla_qqq_8_16): Likewise.
966         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
967         Likewise.
968         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
969         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
970         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
971         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
972         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
973         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
974         (cortex_a9_neon_fp_vsum): Likewise.
975         (cortex_a9_neon_fp_vmul_ddd): Likewise.
976         (cortex_a9_neon_fp_vmul_qqd): Likewise.
977         (cortex_a9_neon_fp_vmla_ddd): Likewise.
978         (cortex_a9_neon_fp_vmla_qqq): Likewise.
979         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
980         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
981         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
982         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
983         (cortex_a9_neon_bp_simple): Likewise.
984         (cortex_a9_neon_bp_2cycle): Likewise.
985         (cortex_a9_neon_bp_3cycle): Likewise.
986         (cortex_a9_neon_ldr): Likewise.
987         (cortex_a9_neon_str): Likewise.
988         (cortex_a9_neon_vld1_1_2_regs): Likewise.
989         (cortex_a9_neon_vld1_3_4_regs): Likewise.
990         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
991         (cortex_a9_neon_vld2_4_regs): Likewise.
992         (cortex_a9_neon_vld3_vld4): Likewise.
993         (cortex_a9_neon_vld1_vld2_lane): Likewise.
994         (cortex_a9_neon_vld3_vld4_lane): Likewise.
995         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
996         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
997         (cortex_a9_neon_vst1_3_4_regs): Likewise.
998         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
999         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1000         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1001         (cortex_a9_neon_mcr): Likewise.
1002         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1003         (cortex_a9_neon_mrc): Likewise.
1004         (cortex_a9_neon_mrrc): Likewise.
1006 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1008         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
1010         (cortex_a8_neon_vshl_ddd): Remove.
1011         (cortex_a8_neon_vst3_vst4): Likewise.
1012         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1014         (cortex_a8_neon_bit_ops_q): New.
1016         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
1017         (cortex_a8_neon_int_2): Likewise..
1018         (cortex_a8_neon_int_3): Likewise.
1019         (cortex_a8_neon_int_5): Likewise.
1020         (cortex_a8_neon_vqneg_vqabs): Likewise.
1021         (cortex_a8_neon_int_4): Likewise.
1022         (cortex_a8_neon_vaba): Likewise.
1023         (cortex_a8_neon_vaba_qqq): Likewise.
1024         (cortex_a8_neon_shift_1): Likewise.
1025         (cortex_a8_neon_shift_2): Likewise.
1026         (cortex_a8_neon_shift_3): Likewise.
1027         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1028         (cortex_a8_neon_vsra_vrsra): Likewise.
1029         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1030         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1031         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1032         Likewise.
1033         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1034         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1035         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1036         Likewise.
1037         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1038         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1039         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1040         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1041         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1042         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1043         (cortex_a8_neon_fp_vsum): Likewise.
1044         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1045         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1046         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1047         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1048         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1049         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1050         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1051         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1052         (cortex_a8_neon_bp_simple): Likewise.
1053         (cortex_a8_neon_bp_2cycle): Likewise.
1054         (cortex_a8_neon_bp_3cycle): Likewise.
1055         (cortex_a8_neon_ldr): Likewise.
1056         (cortex_a8_neon_str): Likewise.
1057         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1058         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1059         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1060         (cortex_a8_neon_vld2_4_regs): Likewise.
1061         (cortex_a8_neon_vld3_vld4): Likewise.
1062         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1063         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1064         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1065         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1066         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1067         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1068         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1069         (cortex_a8_neon_mcr): Likewise.
1070         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1071         (cortex_a8_neon_mrc): Likewise.
1072         (cortex_a8_neon_mrrc): Likewise.
1074 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1076         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1077         (fp): New.
1078         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1079         (aarch64_simd_dup<mode>): Add "type" attribute.
1080         (aarch64_dup_lane<mode>): Likewise.
1081         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1082         (*aarch64_simd_mov<mode>): Likewise.
1083         (aarch64_simd_mov_from_<mode>low): Likewise.
1084         (aarch64_simd_mov_from_<mode>high): Likewise.
1085         (orn<mode>3): Likewise.
1086         (bic<mode>3): Likewise.
1087         (add<mode>3): Likewise.
1088         (sub<mode>3): Likewise.
1089         (mul<mode>3): Likewise.
1090         (*aarch64_mul3_elt<mode>): Likewise.
1091         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1092         (*aarch64_mul3_elt_to_128df): Likewise.
1093         (*aarch64_mul3_elt_to_64v2df): Likewise.
1094         (neg<mode>2): Likewise.
1095         (abs<mode>2): Likewise.
1096         (abd<mode>_3): Likewise.
1097         (aba<mode>_3): Likewise.
1098         (fabd<mode>_3): Likewise.
1099         (*fabd_scalar<mode>3): Likewise.
1100         (and<mode>3): Likewise.
1101         (ior<mode>3): Likewise.
1102         (xor<mode>3): Likewise.
1103         (one_cmpl<mode>2): Likewise.
1104         (aarch64_simd_vec_set<mode>): Likewise.
1105         (aarch64_simd_lshr<mode>): Likewise.
1106         (aarch64_simd_ashr<mode>): Likewise.
1107         (aarch64_simd_imm_shl<mode>): Likewise.
1108         (aarch64_simd_reg_sshl<mode): Likewise.
1109         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1110         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1111         (aarch64_simd_vec_setv2di): Likewise.
1112         (aarch64_simd_vec_set<mode>): Likewise.
1113         (aarch64_mla<mode>): Likewise.
1114         (*aarch64_mla_elt<mode>): Likewise.
1115         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1116         (aarch64_mls<mode>): Likewise.
1117         (*aarch64_mls_elt<mode>): Likewise.
1118         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1119         (<su><maxmin><mode>3): Likewise.
1120         (move_lo_quad_<mode>): Likewise.
1121         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1122         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1123         (vec_pack_trunc_<mode>): Likewise.
1124         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1125         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1126         (*aarch64_<su>mlal_lo<mode>): Likewise.
1127         (*aarch64_<su>mlal_hi<mode>): Likewise.
1128         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1129         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1130         (*aarch64_<su>mlal<mode>): Likewise.
1131         (*aarch64_<su>mlsl<mode>): Likewise.
1132         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1133         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1134         (add<mode>3): Likewise.
1135         (sub<mode>3): Likewise.
1136         (mul<mode>3): Likewise.
1137         (div<mode>3): Likewise.
1138         (neg<mode>2): Likewise.
1139         (abs<mode>2): Likewise.
1140         (fma<mode>4): Likewise.
1141         (*aarch64_fma4_elt<mode>): Likewise.
1142         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1143         (*aarch64_fma4_elt_to_128df): Likewise.
1144         (*aarch64_fma4_elt_to_64v2df): Likewise.
1145         (fnma<mode>4): Likewise.
1146         (*aarch64_fnma4_elt<mode>): Likewise.
1147         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1148         (*aarch64_fnma4_elt_to_128df): Likewise.
1149         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1150         (<frint_pattern><mode>2): Likewise.
1151         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1152         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1153         (vec_unpacks_lo_v4sf): Likewise.
1154         (aarch64_float_extend_lo_v2df): Likewise.
1155         (vec_unpacks_hi_v4sf): Likewise.
1156         (aarch64_float_truncate_lo_v2sf): Likewise.
1157         (aarch64_float_truncate_hi_v4sf): Likewise.
1158         (aarch64_vmls<mode>): Likewise.
1159         (<su><maxmin><mode>3): Likewise.
1160         (<maxmin_uns><mode>3): Likewise.
1161         (reduc_<sur>plus_<mode>): Likewise.
1162         (reduc_<sur>plus_v2di): Likewise.
1163         (reduc_<sur>plus_v2si): Likewise.
1164         (reduc_<sur>plus_<mode>): Likewise.
1165         (aarch64_addpv4sf): Likewise.
1166         (clz<mode>2): Likewise.
1167         (reduc_<maxmin_uns>_<mode>): Likewise.
1168         (reduc_<maxmin_uns>_v2di): Likewise.
1169         (reduc_<maxmin_uns>_v2si): Likewise.
1170         (reduc_<maxmin_uns>_<mode>): Likewise.
1171         (reduc_<maxmin_uns>_v4sf): Likewise.
1172         (aarch64_simd_bsl<mode>_internal): Likewise.
1173         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1174         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1175         (aarch64_get_lane<mode>): Likewise.
1176         (*aarch64_combinez<mode>): Likewise.
1177         (aarch64_combine<mode>): Likewise.
1178         (aarch64_simd_combine<mode>): Likewise.
1179         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1180         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1181         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1182         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1183         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1184         (aarch64_<sur>h<addsub><mode>): Likewise.
1185         (aarch64_<sur><addsub>hn<mode>): Likewise.
1186         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1187         (aarch64_pmul<mode>): Likewise.
1188         (aarch64_<su_optab><optab><mode>): Likewise.
1189         (aarch64_<sur>qadd<mode>): Likewise.
1190         (aarch64_sqmovun<mode>): Likewise.
1191         (aarch64_<sur>qmovn<mode>): Likewise.
1192         (aarch64_s<optab><mode>): Likewise.
1193         (aarch64_sq<r>dmulh<mode>): Likewise.
1194         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1195         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1196         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1197         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1198         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1199         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1200         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1201         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1202         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1203         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1204         (aarch64_sqdmull<mode>): Likewise.
1205         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1206         (aarch64_sqdmull_n<mode>): Likewise.
1207         (aarch64_sqdmull2<mode>_internal): Likewise.
1208         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1209         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1210         (aarch64_<sur>shl<mode>): Likewise.
1211         (aarch64_<sur>q<r>shl<mode>
1212         (aarch64_<sur>shll_n<mode>): Likewise.
1213         (aarch64_<sur>shll2_n<mode>): Likewise.
1214         (aarch64_<sur>shr_n<mode>): Likewise.
1215         (aarch64_<sur>sra_n<mode>): Likewise.
1216         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1217         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1218         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1219         (aarch64_cm<optab><mode>): Likewise.
1220         (aarch64_cm<optab>di): Likewise.
1221         (aarch64_cm<optab><mode>): Likewise.
1222         (aarch64_cm<optab>di): Likewise.
1223         (aarch64_cmtst<mode>): Likewise.
1224         (aarch64_cmtstdi): Likewise.
1225         (aarch64_cm<optab><mode>): Likewise.
1226         (*aarch64_fac<optab><mode>): Likewise.
1227         (aarch64_addp<mode>): Likewise.
1228         (aarch64_addpdi): Likewise.
1229         (sqrt<mode>2): Likewise.
1230         (vec_load_lanesoi<mode>): Likewise.
1231         (vec_store_lanesoi<mode>): Likewise.
1232         (vec_load_lanesci<mode>): Likewise.
1233         (vec_store_lanesci<mode>): Likewise.
1234         (vec_load_lanesxi<mode>): Likewise.
1235         (vec_store_lanesxi<mode>): Likewise.
1236         (*aarch64_mov<mode>): Likewise.
1237         (aarch64_ld2<mode>_dreg): Likewise.
1238         (aarch64_ld2<mode>_dreg): Likewise.
1239         (aarch64_ld3<mode>_dreg): Likewise.
1240         (aarch64_ld3<mode>_dreg): Likewise.
1241         (aarch64_ld4<mode>_dreg): Likewise.
1242         (aarch64_ld4<mode>_dreg): Likewise.
1243         (aarch64_tbl1<mode>): Likewise.
1244         (aarch64_tbl2v16qi): Likewise.
1245         (aarch64_combinev16qi): Likewise.
1246         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1247         (aarch64_st2<mode>_dreg): Likewise.
1248         (aarch64_st2<mode>_dreg): Likewise.
1249         (aarch64_st3<mode>_dreg): Likewise.
1250         (aarch64_st3<mode>_dreg): Likewise.
1251         (aarch64_st4<mode>_dreg): Likewise.
1252         (aarch64_st4<mode>_dreg): Likewise.
1253         (*aarch64_simd_ld1r<mode>): Likewise.
1254         (aarch64_frecpe<mode>): Likewise.
1255         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1256         (aarch64_frecps<mode>): Likewise.
1258 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1260         * config/arm/iterators.md (V_elem_ch): New.
1261         (q): Likewise.
1262         (VQH_type): Likewise.
1263         * config/arm/arm.md (is_neon_type): New.
1264         (conds): Use is_neon_type.
1265         (anddi3_insn): Update type attribute.
1266         (xordi3_insn): Likewise.
1267         (one_cmpldi2): Likewise.
1268         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1269         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1270         (*movmisalign<mode>_neon_store): Likewise.
1271         (*movmisalign<mode>_neon_load): Likewise.
1272         (vec_set<mode>_internal): Likewise.
1273         (vec_set<mode>_internal): Likewise.
1274         (vec_setv2di_internal): Likewise.
1275         (vec_extract<mode>): Likewise.
1276         (vec_extract<mode>): Likewise.
1277         (vec_extractv2di): Likewise.
1278         (*add<mode>3_neon): Likewise.
1279         (adddi3_neon): Likewise.
1280         (*sub<mode>3_neon): Likewise.
1281         (subdi3_neon): Likewise.
1282         (fma<VCVTF:mode>4): Likewise.
1283         (fma<VCVTF:mode>4_intrinsic): Likewise.
1284         (*fmsub<VCVTF:mode>4): Likewise.
1285         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1286         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1287         (ior<mode>3): Likewise.
1288         (and<mode>3): Likewise.
1289         (orn<mode>3_neon): Likewise.
1290         (orndi3_neon): Likewise.
1291         (bic<mode>3_neon): Likewise.
1292         (bicdi3_neon): Likewise.
1293         (xor<mode>3): Likewise.
1294         (one_cmpl<mode>2): Likewise.
1295         (abs<mode>2): Likewise.
1296         (neg<mode>2): Likewise.
1297         (negdi2_neon): Likewise.
1298         (*umin<mode>3_neon): Likewise.
1299         (*umax<mode>3_neon): Likewise.
1300         (*smin<mode>3_neon): Likewise.
1301         (*smax<mode>3_neon): Likewise.
1302         (vashl<mode>3): Likewise.
1303         (vashr<mode>3_imm): Likewise.
1304         (vlshr<mode>3_imm): Likewise.
1305         (ashl<mode>3_signed): Likewise.
1306         (ashl<mode>3_unsigned): Likewise.
1307         (neon_load_count): Likewise.
1308         (ashldi3_neon_noclobber): Likewise.
1309         (ashldi3_neon): Likewise.
1310         (signed_shift_di3_neon): Likewise.
1311         (unsigned_shift_di3_neon): Likewise.
1312         (ashrdi3_neon_imm_noclobber): Likewise.
1313         (lshrdi3_neon_imm_noclobber): Likewise.
1314         (<shift>di3_neon): Likewise.
1315         (widen_ssum<mode>3): Likewise.
1316         (widen_usum<mode>3): Likewise.
1317         (quad_halves_<code>v4si): Likewise.
1318         (quad_halves_<code>v4sf): Likewise.
1319         (quad_halves_<code>v8hi): Likewise.
1320         (quad_halves_<code>v16qi): Likewise.
1321         (reduc_splus_v2di): Likewise.
1322         (neon_vpadd_internal<mode>): Likewise.
1323         (neon_vpsmin<mode>): Likewise.
1324         (neon_vpsmax<mode>): Likewise.
1325         (neon_vpumin<mode>): Likewise.
1326         (neon_vpumax<mode>): Likewise.
1327         (*ss_add<mode>_neon): Likewise.
1328         (*us_add<mode>_neon): Likewise.
1329         (*ss_sub<mode>_neon): Likewise.
1330         (*us_sub<mode>_neon): Likewise.
1331         (neon_vadd<mode>_unspec): Likewise.
1332         (neon_vaddl<mode>): Likewise.
1333         (neon_vaddw<mode>): Likewise.
1334         (neon_vhadd<mode>): Likewise.
1335         (neon_vqadd<mode>): Likewise.
1336         (neon_vaddhn<mode>): Likewise.
1337         (neon_vmul<mode>): Likewise.
1338         (neon_vfms<VCVTF:mode>): Likewise.
1339         (neon_vmlal<mode>): Likewise.
1340         (neon_vmls<mode>): Likewise.
1341         (neon_vmlsl<mode>): Likewise.
1342         (neon_vqdmulh<mode>): Likewise.
1343         (neon_vqdmlal<mode>): Likewise.
1344         (neon_vqdmlsl<mode>): Likewise.
1345         (neon_vmull<mode>): Likewise.
1346         (neon_vqdmull<mode>): Likewise.
1347         (neon_vsub<mode>_unspec): Likewise.
1348         (neon_vsubl<mode>): Likewise.
1349         (neon_vsubw<mode>): Likewise.
1350         (neon_vqsub<mode>): Likewise.
1351         (neon_vhsub<mode>): Likewise.
1352         (neon_vsubhn<mode>): Likewise.
1353         (neon_vceq<mode>): Likewise.
1354         (neon_vcge<mode>): Likewise.
1355         (neon_vcgeu<mode>): Likewise.
1356         (neon_vcgt<mode>): Likewise.
1357         (neon_vcgtu<mode>): Likewise.
1358         (neon_vcle<mode>): Likewise.
1359         (neon_vclt<mode>): Likewise.
1360         (neon_vcage<mode>): Likewise.
1361         (neon_vcagt<mode>): Likewise.
1362         (neon_vtst<mode>): Likewise.
1363         (neon_vabd<mode>): Likewise.
1364         (neon_vabdl<mode>): Likewise.
1365         (neon_vaba<mode>): Likewise.
1366         (neon_vabal<mode>): Likewise.
1367         (neon_vmax<mode>): Likewise.
1368         (neon_vmin<mode>): Likewise.
1369         (neon_vpaddl<mode>): Likewise.
1370         (neon_vpadal<mode>): Likewise.
1371         (neon_vpmax<mode>): Likewise.
1372         (neon_vpmin<mode>): Likewise.
1373         (neon_vrecps<mode>): Likewise.
1374         (neon_vrsqrts<mode>): Likewise.
1375         (neon_vqabs<mode>): Likewise.
1376         (neon_vqneg<mode>): Likewise.
1377         (neon_vcls<mode>): Likewise.
1378         (clz<mode>2): Likewise.
1379         (popcount<mode>2): Likewise.
1380         (neon_vrecpe<mode>): Likewise.
1381         (neon_vrsqrte<mode>): Likewise.
1382         (neon_vget_lane<mode>_sext_internal): Likewise.
1383         (neon_vget_lane<mode>_zext_internal): Likewise.
1384         (neon_vdup_n<mode>): Likewise.
1385         (neon_vdup_n<mode>): Likewise.
1386         (neon_vdup_nv2di): Likewise.
1387         (neon_vdup_lane<mode>_interal): Likewise.
1388         (*neon_vswp<mode>): Likewise.
1389         (neon_vcombine<mode>): Likewise.
1390         (float<mode><V_cvtto>2): Likewise.
1391         (floatuns<mode><V_cvtto>2): Likewise.
1392         (fix_trunc<mode><V_cvtto>2): Likewise.
1393         (fixuns_trunc<mode><V_cvtto>2
1394         (neon_vcvt<mode>): Likewise.
1395         (neon_vcvt<mode>): Likewise.
1396         (neon_vcvtv4sfv4hf): Likewise.
1397         (neon_vcvtv4hfv4sf): Likewise.
1398         (neon_vcvt_n<mode>): Likewise.
1399         (neon_vcvt_n<mode>): Likewise.
1400         (neon_vmovn<mode>): Likewise.
1401         (neon_vqmovn<mode>): Likewise.
1402         (neon_vqmovun<mode>): Likewise.
1403         (neon_vmovl<mode>): Likewise.
1404         (neon_vmul_lane<mode>): Likewise.
1405         (neon_vmul_lane<mode>): Likewise.
1406         (neon_vmull_lane<mode>): Likewise.
1407         (neon_vqdmull_lane<mode>): Likewise.
1408         (neon_vqdmulh_lane<mode>): Likewise.
1409         (neon_vqdmulh_lane<mode>): Likewise.
1410         (neon_vmla_lane<mode>): Likewise.
1411         (neon_vmla_lane<mode>): Likewise.
1412         (neon_vmlal_lane<mode>): Likewise.
1413         (neon_vqdmlal_lane<mode>): Likewise.
1414         (neon_vmls_lane<mode>): Likewise.
1415         (neon_vmls_lane<mode>): Likewise.
1416         (neon_vmlsl_lane<mode>): Likewise.
1417         (neon_vqdmlsl_lane<mode>): Likewise.
1418         (neon_vext<mode>): Likewise.
1419         (neon_vrev64<mode>): Likewise.
1420         (neon_vrev32<mode>): Likewise.
1421         (neon_vrev16<mode>): Likewise.
1422         (neon_vbsl<mode>_internal): Likewise.
1423         (neon_vshl<mode>): Likewise.
1424         (neon_vqshl<mode>): Likewise.
1425         (neon_vshr_n<mode>): Likewise.
1426         (neon_vshrn_n<mode>): Likewise.
1427         (neon_vqshrn_n<mode>): Likewise.
1428         (neon_vqshrun_n<mode>): Likewise.
1429         (neon_vshl_n<mode>): Likewise.
1430         (neon_vqshl_n<mode>): Likewise.
1431         (neon_vqshlu_n<mode>): Likewise.
1432         (neon_vshll_n<mode>): Likewise.
1433         (neon_vsra_n<mode>): Likewise.
1434         (neon_vsri_n<mode>): Likewise.
1435         (neon_vsli_n<mode>): Likewise.
1436         (neon_vtbl1v8qi): Likewise.
1437         (neon_vtbl2v8qi): Likewise.
1438         (neon_vtbl3v8qi): Likewise.
1439         (neon_vtbl4v8qi): Likewise.
1440         (neon_vtbl1v16qi): Likewise.
1441         (neon_vtbl2v16qi): Likewise.
1442         (neon_vcombinev16qi): Likewise.
1443         (neon_vtbx1v8qi): Likewise.
1444         (neon_vtbx2v8qi): Likewise.
1445         (neon_vtbx3v8qi): Likewise.
1446         (neon_vtbx4v8qi): Likewise.
1447         (*neon_vtrn<mode>_insn): Likewise.
1448         (*neon_vzip<mode>_insn): Likewise.
1449         (*neon_vuzp<mode>_insn): Likewise.
1450         (neon_vld1<mode>): Likewise.
1451         (neon_vld1_lane<mode>): Likewise.
1452         (neon_vld1_lane<mode>): Likewise.
1453         (neon_vld1_dup<mode>): Likewise.
1454         (neon_vld1_dup<mode>): Likewise.
1455         (neon_vld1_dupv2di): Likewise.
1456         (neon_vst1<mode>): Likewise.
1457         (neon_vst1_lane<mode>): Likewise.
1458         (neon_vst1_lane<mode>): Likewise.
1459         (neon_vld2<mode>): Likewise.
1460         (neon_vld2<mode>): Likewise.
1461         (neon_vld2_lane<mode>): Likewise.
1462         (neon_vld2_lane<mode>): Likewise.
1463         (neon_vld2_dup<mode>): Likewise.
1464         (neon_vst2<mode>): Likewise.
1465         (neon_vst2<mode>): Likewise.
1466         (neon_vst2_lane<mode>): Likewise.
1467         (neon_vst2_lane<mode>): Likewise.
1468         (neon_vld3<mode>): Likewise.
1469         (neon_vld3qa<mode>): Likewise.
1470         (neon_vld3qb<mode>): Likewise.
1471         (neon_vld3_lane<mode>): Likewise.
1472         (neon_vld3_lane<mode>): Likewise.
1473         (neon_vld3_dup<mode>): Likewise.
1474         (neon_vst3<mode>): Likewise.
1475         (neon_vst3qa<mode>): Likewise.
1476         (neon_vst3qb<mode>): Likewise.
1477         (neon_vst3_lane<mode>): Likewise.
1478         (neon_vst3_lane<mode>): Likewise.
1479         (neon_vld4<mode>): Likewise.
1480         (neon_vld4qa<mode>): Likewise.
1481         (neon_vld4qb<mode>): Likewise.
1482         (neon_vld4_lane<mode>): Likewise.
1483         (neon_vld4_lane<mode>): Likewise.
1484         (neon_vld4_dup<mode>): Likewise.
1485         (neon_vst4<mode>): Likewise.
1486         (neon_vst4qa<mode>): Likewise.
1487         (neon_vst4qb<mode>): Likewise.
1488         (neon_vst4_lane<mode>): Likewise.
1489         (neon_vst4_lane<mode>): Likewise.
1490         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1491         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1492         (neon_vec_<US>mult_lo_<mode>): Likewise.
1493         (neon_vec_<US>mult_hi_<mode>): Likewise.
1494         (neon_vec_<US>shiftl_<mode>): Likewise.
1495         (neon_unpack<US>_<mode>): Likewise.
1496         (neon_vec_<US>mult_<mode>): Likewise.
1497         (vec_pack_trunc_<mode>): Likewise.
1498         (neon_vec_pack_trunc_<mode>): Likewise.
1499         (neon_vabd<mode>_2): Likewise.
1500         (neon_vabd<mode>_3): Likewise.
1502 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1504         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1505         (load_pair): Update type attribute.
1506         (store_pair): Update type attribute.
1507         * config/aarch64/iterators.md (q): New.
1509 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1511         * config/arm/types.md: Add new types for Neon insns.
1513 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1514             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1515             Sergey Lega  <sergey.s.lega@intel.com>
1516             Anna Tikhonova  <anna.tikhonova@intel.com>
1517             Ilya Tocar  <ilya.tocar@intel.com>
1518             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1519             Ilya Verbin  <ilya.verbin@intel.com>
1520             Kirill Yukhin  <kirill.yukhin@intel.com>
1521             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1523         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1524         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1525         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1526         (rcp14<mode>): New.
1527         (srcp14<mode>): Ditto.
1528         (rsqrt14<mode>): Ditto.
1529         (rsqrt14<mode>): Ditto.
1530         (avx512f_vmscalef<mode>): Ditto.
1531         (avx512f_scalef<mode>): Ditto.
1532         (avx512f_getexp<mode>): Ditto.
1533         (avx512f_sgetexp<mode>): Ditto.
1534         (avx512f_fixupimm<mode>): Ditto.
1535         (avx512f_sfixupimm<mode>): Ditto.
1536         (avx512f_rndscale<mode>): Ditto.
1537         (*avx512er_exp2<mode>): Ditto.
1538         (*avx512er_rcp28<mode>): Ditto.
1539         (avx512er_rsqrt28<mode>): Ditto.
1540         (avx512f_getmant<mode>): Ditto.
1541         (avx512f_getmant<mode>): Ditto.
1542         (avx512f_rndscale<mode>): Fix formatting.
1544 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1546         * ipa-utils.h (ipa_edge_within_scc): Declare.
1547         * ipa-cp.c (edge_within_scc): Moved...
1548         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1550 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1551             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1552             Sergey Lega  <sergey.s.lega@intel.com>
1553             Anna Tikhonova  <anna.tikhonova@intel.com>
1554             Ilya Tocar  <ilya.tocar@intel.com>
1555             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1556             Ilya Verbin  <ilya.verbin@intel.com>
1557             Kirill Yukhin  <kirill.yukhin@intel.com>
1558             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1560         * config/i386/predicates.md (const_8_to_15_operand): New.
1561         (const_16_to_31_operand): Ditto.
1562         * config/i386/sse.md (V8FI): New.
1563         (V16FI): Ditto.
1564         (reduc_splus_v8df): Ditto.
1565         (reduc_splus_v16sf): Ditto.
1566         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1567         (vec_extract_hi_<mode>): Ditto.
1568         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1569         (vec_set_lo_<mode>): Ditto.
1570         (vec_set_hi_<mode>): Ditto.
1571         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1572         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1573         (avx512f_pshufd_1): Ditto.
1574         (avx512f_broadcast<mode>): Ditto.
1575         (avx512f_broadcast<mode>): Ditto.
1576         (define_split): Split vec_extract_lo into move.
1577         (ssequartermode): Ditto.
1578         (ssedoublemode): Extened with wider modes.
1579         (vec_extract_lo_<mode>): Ditto.
1581 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1582             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1583             Sergey Lega  <sergey.s.lega@intel.com>
1584             Anna Tikhonova  <anna.tikhonova@intel.com>
1585             Ilya Tocar  <ilya.tocar@intel.com>
1586             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1587             Ilya Verbin  <ilya.verbin@intel.com>
1588             Kirill Yukhin  <kirill.yukhin@intel.com>
1589             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1591         * config/i386/predicates.md (register_or_constm1_operand): New.
1592         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1593         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1594         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1595         UNSPEC_SCATTER_PREFETCH
1596         (VI48_512): New.
1597         (avx512f_ucmp<mode>3): Ditto.
1598         (avx512f_vternlog<mode>): Ditto.
1599         (avx512f_align<mode>): Ditto.
1600         (<shift_insn><mode>3): Ditto.
1601         (avx512f_<rotate>v<mode>): Ditto.
1602         (avx512f_<rotate><mode>): Ditto.
1603         (avx512f_eq<mode>3): Ditto.
1604         (avx512f_eq<mode>3_1): Ditto.
1605         (avx512f_gt<mode>3): Ditto.
1606         (avx512f_testm<mode>3): Ditto.
1607         (avx512f_testnm<mode>3): Ditto.
1608         (avx512pf_gatherpf<mode>): Ditto.
1609         (*avx512pf_gatherpf<mode>_mask): Ditto.
1610         (*avx512pf_gatherpf<mode>): Ditto.
1611         (avx512pf_scatterpf<mode>): Ditto.
1612         (*avx512pf_scatterpf<mode>_mask): Ditto.
1613         (*avx512pf_scatterpf<mode>): Ditto.
1614         (avx512f_vec_dup_gpr<mode>): Ditto.
1615         (clz<mode>2): Ditto.
1616         (conflict<mode>): Ditto.
1617         (REDUC_SMINMAX_MODE): Extened with wider modes.
1618         (reduc_<code>_<mode>): Ditto.
1619         (vlshr<mode>3): Ditto.
1620         (vashl<mode>3): Ditto.
1622 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1623             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1624             Sergey Lega  <sergey.s.lega@intel.com>
1625             Anna Tikhonova  <anna.tikhonova@intel.com>
1626             Ilya Tocar  <ilya.tocar@intel.com>
1627             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1628             Ilya Verbin  <ilya.verbin@intel.com>
1629             Kirill Yukhin  <kirill.yukhin@intel.com>
1630             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1632         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1633         UNSPEC_SCATTER.
1634         (VI48F_512): New.
1635         (avx512fmaskmode): Ditto.
1636         (bcstscalarsuff): Ditto.
1637         (avx512f_blendm<mode>): Ditto.
1638         (cmp_imm_predicate): Ditto.
1639         (avx512f_cmp<mode>3): Ditto.
1640         (avx512f_vec_dup<mode>): Ditto.
1641         (avx512f_vec_dup_mem<mode>): Ditto.
1642         (avx512f_vpermi2var<mode>3): Ditto.
1643         (avx512f_vpermt2var<mode>3): Ditto.
1644         (vec_init<mode>): Ditto.
1645         (avx512f_gathersi<mode>): Ditto.
1646         (*avx512f_gathersi<mode>): Ditto.
1647         (*avx512f_gathersi<mode>_2): Ditto.
1648         (avx512f_gatherdi<mode>): Ditto.
1649         (*avx512f_gatherdi<mode>): Ditto.
1650         (*avx512f_gatherdi<mode>_2): Ditto.
1651         (avx512f_scattersi<mode>): Ditto.
1652         (*avx512f_scattersi<mode>): Ditto.
1653         (avx512f_scatterdi<mode>): Ditto.
1654         (*avx512f_scatterdi<mode>): Ditto.
1655         (sseintprefix): Extened with wider modes.
1656         (VEC_GATHER_IDXSI): Ditto.
1657         (VEC_GATHER_IDXDI): Ditto.
1658         (VEC_GATHER_SRCDI): Ditto.
1660 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1661             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1663         * config/arm/t-aprofile: New file.
1664         * config.gcc: Handle --with-multilib-list option.
1666 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1668         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1669         the input for r2.
1671 2013-10-15  Richard Biener  <rguenther@suse.de>
1673         * tree-loop-distribution.c (build_empty_rdg): Inline into
1674         single user.
1675         (rdg_flag_vertex): Inline into single user.
1676         (rdg_flag_vertex_and_dependent): Likewise.
1677         (build_rdg_partition_for_vertex): Remove processed bitmap.
1678         (rdg_build_partitions): Simplify.
1680 2013-10-15  Richard Biener  <rguenther@suse.de>
1682         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1683         Restructure forwarding through conversions and copies to
1684         avoid performing copy-propagation the wrong way.  Adjust
1685         recursion invocations.
1686         (forward_propagate_addr_expr): Add argument stating if we
1687         are recursing from a single-use.
1688         (ssa_forward_propagate_and_combine): Adjust.
1690 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1692         * dumpfile.h (gcc::dump_manager): New class, to hold state
1693         relating to dumpfile management.
1694         (get_dump_file_name): Remove in favor of method of dump_manager.
1695         (dump_initialized_p): Likewise.
1696         (dump_start): Likewise.
1697         (dump_finish): Likewise.
1698         (dump_switch_p): Likewise.
1699         (dump_register): Likewise.
1700         (get_dump_file_info): Likewise.
1701         * context.c (gcc::context::context): Construct the dump_manager
1702         instance.
1703         * context.h (gcc::context::get_dumps): New.
1704         (gcc::context::m_dumps): New.
1705         * coverage.c (coverage_init): Port to dump_manager API.
1706         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1707         (extra_dump_files_in_use): Likewise.
1708         (extra_dump_files_alloced): Likewise.
1709         (gcc::dump_manager::dump_manager): New.
1710         (dump_register): Convert to...
1711         (gcc::dump_manager::dump_register): ...method, replacing
1712         function-static next_dump with m_next_dump field.
1713         (get_dump_file_info): Convert to...
1714         (gcc::dump_manager::get_dump_file_info): ...method.
1715         (get_dump_file_name): Convert to...
1716         (gcc::dump_manager::get_dump_file_name): ...method.
1717         (dump_start): Convert to...
1718         (gcc::dump_manager::dump_start): ...method.
1719         (dump_finish): Convert to...
1720         (gcc::dump_manager::dump_finish): ...method.
1721         (dump_begin): Replace body with...
1722         (gcc::dump_manager::dump_begin): ...new method.
1723         (dump_phase_enabled_p): Convert to...
1724         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1725         (dump_phase_enabled_p): Convert to...
1726         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1727         (dump_initialized_p):  Convert to...
1728         (gcc::dump_manager::dump_initialized_p): ...method.
1729         (dump_flag_name): Replace body with...
1730         (gcc::dump_manager::dump_flag_name): ...new method.
1731         (dump_enable_all): Convert to...
1732         (gcc::dump_manager::dump_enable_all): ...new method.
1733         (opt_info_enable_passes): Convert to...
1734         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1735         (dump_switch_p_1): Convert to...
1736         (gcc::dump_manager::dump_switch_p_1): ...new method.
1737         (dump_switch_p):  Convert to...
1738         (gcc::dump_manager::dump_switch_p): ...new method.
1739         (opt_info_switch_p): Port to dump_manager API.
1740         (enable_rtl_dump_file): Likewise.
1741         * opts-global.c (handle_common_deferred_options): Port to new
1742         dump_manager API.
1743         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1744         (pass_manager::register_one_dump_file): Likewise.
1745         (pass_manager::register_pass): Likewise.
1746         (pass_init_dump_file): Likewise.
1747         (pass_fini_dump_file): Likewise.
1748         * statistics.c (statistics_early_init): Likewise.
1750 2013-10-14  Richard Biener  <rguenther@suse.de>
1752         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1753         iterative_hash_canonical_type, gimple_canonical_type_hash,
1754         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1755         gimple_register_canonical_type, print_gimple_types_stats,
1756         free_gimple_type_tables): Move to lto/lto.c
1757         (gt-gimple.h): Do not include.
1758         * gimple.h (gimple_register_canonical_type,
1759         print_gimple_types_stats, free_gimple_type_tables): Remove.
1760         * Makefile.in (GTFILES): Remove gimple.c.
1762 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1764         PR target/58716
1765         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1766         scanning for msp430x targets.
1768 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1770         PR bootstrap/58509
1771         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1772         (registers_ok_for_ldd_peep): Move around.
1773         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1774         * config/sparc/sparc.md (widening peepholes): Use it.
1776 2013-10-14  Richard Biener  <rguenther@suse.de>
1778         PR middle-end/58712
1779         PR middle-end/55358
1780         * gimple.c (iterative_hash_canonical_type): Make sure to
1781         record the hash into the correct hashtable slot.
1783 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1785         PR rtl-optimization/58662
1786         * combine.c (try_combine): Take into account death nodes on I2 when
1787         splitting a PARALLEL of two independent SETs.  Fix dump message.
1789 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1791         PR target/51244
1792         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1793         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1794         * config/sh/t-sh (sh_treg_combine.o): New entry.
1795         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1796         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1797         (register_sh_passes): New function.  Register sh_treg_combine pass.
1798         (sh_option_override): Invoke it.
1799         (sh_canonicalize_comparison): Handle op0_preserve_value.
1800         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1801         opportunities.  Canonicalize branch condition.
1802         (nott): Allow only if pseudos can be created for non-SH2A.
1804 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1806         PR target/58690
1807         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1808         (ix86_expand_movmem): Replace copy_addr_to_reg with
1809         ix86_copy_addr_to_reg.
1810         (ix86_expand_setmem): Likewise.
1812 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1814         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1815         provided by ix86_fp_compare_mode instead of CCFPUmode.
1817 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1819         * config/aarch64/arm_neon.h
1820         (vtbx<1,3>_<psu>8): Fix register constriants.
1822 2013-10-11  Jeff Law  <law@redhat.com>
1824         PR tree-optimization/58640
1825         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1826         threading paths that cross over two loop entry points.
1828 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1830         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1831         handle vector float as well.
1832         (*vsx_le_perm_load_v4si): Likewise.
1833         (*vsx_le_perm_store_v2di): Likewise.
1834         (*vsx_le_perm_store_v4si): Likewise.
1836 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1838         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1839         directly to circumvent subtract from splat{31} workaround.
1840         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1841         prototype.
1842         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1843         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1844         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1845         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1846         separate big and little endian logic.
1847         (*altivec_vperm_<mode>_internal): New define_insn.
1848         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1849         separate big and little endian logic.
1850         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1851         (vec_permv16qi): Add little endian logic.
1853 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1855         * doc/extend.texi (returns_nonnull): Remove arguments.
1857 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1858             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1859             Sergey Lega  <sergey.s.lega@intel.com>
1860             Anna Tikhonova  <anna.tikhonova@intel.com>
1861             Ilya Tocar  <ilya.tocar@intel.com>
1862             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1863             Ilya Verbin  <ilya.verbin@intel.com>
1864             Kirill Yukhin  <kirill.yukhin@intel.com>
1865             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1867         * config/i386/sse.md (VI48F_256_512): New.
1868         (avx2_permvar<mode>): Change to ...
1869         (<avx2_avx512f>_permvar<mode>): This.
1871 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1872             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1873             Sergey Lega  <sergey.s.lega@intel.com>
1874             Anna Tikhonova  <anna.tikhonova@intel.com>
1875             Ilya Tocar  <ilya.tocar@intel.com>
1876             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1877             Ilya Verbin  <ilya.verbin@intel.com>
1878             Kirill Yukhin  <kirill.yukhin@intel.com>
1879             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1881         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1882         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1883         * config/i386/sse.md (VI4_AVX): New.
1884         (sf2simodelower): Ditto.
1885         (sse2_cvtps2dq): Change to ...
1886         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1888 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1889             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1890             Sergey Lega  <sergey.s.lega@intel.com>
1891             Anna Tikhonova  <anna.tikhonova@intel.com>
1892             Ilya Tocar  <ilya.tocar@intel.com>
1893             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1894             Ilya Verbin  <ilya.verbin@intel.com>
1895             Kirill Yukhin  <kirill.yukhin@intel.com>
1896             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1898         * config/i386/sse.md (V_512): New.
1899         (VI_512): Ditto.
1900         (vcond<V_512:mode><VF_512:mode>): Ditto.
1901         (vcond<V_512:mode><VI_512:mode>): Ditto.
1902         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1904 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1905             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1906             Sergey Lega  <sergey.s.lega@intel.com>
1907             Anna Tikhonova  <anna.tikhonova@intel.com>
1908             Ilya Tocar  <ilya.tocar@intel.com>
1909             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1910             Ilya Verbin  <ilya.verbin@intel.com>
1911             Kirill Yukhin  <kirill.yukhin@intel.com>
1912             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1914         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1915         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1916         (FMAMODE): Ditto.
1917         (fma<mode>4): Removed condition.
1918         (fms<mode>4): Ditto.
1919         (fnma<mode>4): Ditto.
1920         (fnms<mode>4): Ditto.
1921         (fma4i_fmadd_<mode>): Ditto.
1922         (*fma_fmadd_<mode>): Ditto.
1923         (*fma_fmsub_<mode>): Ditto.
1924         (*fma_fnmadd_<mode>): Ditto.
1925         (*fma_fnmsub_<mode>): Ditto.
1926         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1927         (*fma_fmaddsub_<mode>): Ditto.
1928         (*fma_fmsubadd_<mode>): Ditto.
1929         (*fmai_fmadd_<mode>): Ditto.
1930         (*fmai_fmsub_<mode>): Ditto.
1931         (*fmai_fnmadd_<mode>): Ditto.
1932         (*fmai_fnmsub_<mode>): Ditto.
1934 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1935             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1936             Sergey Lega  <sergey.s.lega@intel.com>
1937             Anna Tikhonova  <anna.tikhonova@intel.com>
1938             Ilya Tocar  <ilya.tocar@intel.com>
1939             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1940             Ilya Verbin  <ilya.verbin@intel.com>
1941             Kirill Yukhin  <kirill.yukhin@intel.com>
1942             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1944         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1945         (VI124_256): Changed to ...
1946         (VI124_256_48_512): This.
1947         (ssepackmode): Extended with wider modes.
1948         (<code><mode>3): Changed iterator.
1949         (*avx2_<code><mode>3): Ditto.
1950         (vec_pack_trunc_<mode>): Ditto.
1952 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1953             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1954             Sergey Lega  <sergey.s.lega@intel.com>
1955             Anna Tikhonova  <anna.tikhonova@intel.com>
1956             Ilya Tocar  <ilya.tocar@intel.com>
1957             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1958             Ilya Verbin  <ilya.verbin@intel.com>
1959             Kirill Yukhin  <kirill.yukhin@intel.com>
1960             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1962         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1963         (VI8F_256_512): Ditto.
1964         (abs<mode>2): Changed iterator.
1965         (avx2_perm<mode>): Changed to ...
1966         (<avx2_avx512f>_perm<mode>): This.
1967         (avx2_perm<mode>_1): Changed to ...
1968         (<avx2_avx512f>_perm<mode>_1): This.
1970 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1971             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1972             Sergey Lega  <sergey.s.lega@intel.com>
1973             Anna Tikhonova  <anna.tikhonova@intel.com>
1974             Ilya Tocar  <ilya.tocar@intel.com>
1975             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1976             Ilya Verbin  <ilya.verbin@intel.com>
1977             Kirill Yukhin  <kirill.yukhin@intel.com>
1978             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1980         * config/i386/sse.md (VI48_AVX512F): New.
1981         (VI48_AVX2): Changed to ...
1982         (VI48_AVX2_48_AVX512F): This.
1983         (avx2_ashrv<mode>): Changed to ...
1984         (<avx2_avx512f>_ashrv<mode>): This.
1985         (avx2_<shift_insn>v<mode>): Changed to ...
1986         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1988 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1989             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1990             Sergey Lega  <sergey.s.lega@intel.com>
1991             Anna Tikhonova  <anna.tikhonova@intel.com>
1992             Ilya Tocar  <ilya.tocar@intel.com>
1993             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1994             Ilya Verbin  <ilya.verbin@intel.com>
1995             Kirill Yukhin  <kirill.yukhin@intel.com>
1996             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1998         * config/i386/sse.md (VI4_AVX512F): New.
1999         (VI8_AVX2_AVX512F): Ditto.
2000         (mul<mode>3): Extended with wider modes.
2001         (*<sse4_1_avx2>_mul<mode>3): Ditto.
2002         (mul<mode>3): Ditto.
2003         (vec_widen_<s>mult_odd_<mode>): Ditto.
2005 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2006             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2007             Sergey Lega  <sergey.s.lega@intel.com>
2008             Anna Tikhonova  <anna.tikhonova@intel.com>
2009             Ilya Tocar  <ilya.tocar@intel.com>
2010             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2011             Ilya Verbin  <ilya.verbin@intel.com>
2012             Kirill Yukhin  <kirill.yukhin@intel.com>
2013             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2015         * config/i386/sse.md (VI2_AVX512F): New.
2016         (VI124_AVX512F): Ditto.
2017         (sseunpackmode): Extended with wider modes.
2018         (sseunpackfltmode): Ditto.
2019         (vec_unpacks_float_hi_<mode>): Ditto.
2020         (vec_unpacks_float_lo_<mode>): Ditto.
2021         (vec_unpacku_float_hi_<mode>): Ditto.
2022         (vec_unpacku_float_lo_<mode>): Ditto.
2023         (vec_unpacks_lo_<mode>): Ditto.
2024         (vec_unpacks_hi_<mode>): Ditto.
2025         (vec_unpacku_lo_<mode>): Ditto.
2026         (vec_unpacku_hi_<mode>): Ditto.
2028 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2029             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2030             Sergey Lega  <sergey.s.lega@intel.com>
2031             Anna Tikhonova  <anna.tikhonova@intel.com>
2032             Ilya Tocar  <ilya.tocar@intel.com>
2033             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2034             Ilya Verbin  <ilya.verbin@intel.com>
2035             Kirill Yukhin  <kirill.yukhin@intel.com>
2036             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2038         * config/i386/i386.md (multdiv): New.
2039         (multdiv_mnemonic): Ditto.
2040         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2041         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2042         (<sse>_vmdiv<mode>3): Removed.
2044 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2045             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2046             Sergey Lega  <sergey.s.lega@intel.com>
2047             Anna Tikhonova  <anna.tikhonova@intel.com>
2048             Ilya Tocar  <ilya.tocar@intel.com>
2049             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2050             Ilya Verbin  <ilya.verbin@intel.com>
2051             Kirill Yukhin  <kirill.yukhin@intel.com>
2052             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2054         * config/i386/sse.md (V): Extended with wider modes.
2055         (VF2): Ditto.
2056         (ssehalfvecmode): Ditto.
2057         (i128): Ditto.
2058         (ssepackfltmode): Ditto.
2059         (avx_vec_concat<mode>): Ditto.
2060         (V_256_512): New iterator.
2061         (VF2_512_256): Ditto.
2062         (si2dfmode): New attribute.
2063         (si2dfmodelower): Ditto.
2064         (sf2dfmode): Ditto.
2065         (concat_tg_mode): Ditto.
2066         (floatv4siv4df2): Changed to ...
2067         (float<si2dfmodelower><mode>2): This.
2068         (avx_cvtps2pd256): Changed to ...
2069         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2070         (vec_pack_trunc_v4df): Changed to ...
2071         (vec_pack_trunc_<mode>): This.
2072         (avx_vpermil<mode>): Changed to ...
2073         (<sse2_avx_avx512f>_vpermil<mode>): This.
2074         (<fixsuffix>fix_truncv8dfv8si2): New.
2075         (vec_pack_sfix_trunc_v8df): Ditto.
2076         (avx512f_rndscale<mode>): Ditto.
2077         (avx512f_roundpd512): Ditto.
2078         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2080 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2081             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2082             Sergey Lega  <sergey.s.lega@intel.com>
2083             Anna Tikhonova  <anna.tikhonova@intel.com>
2084             Ilya Tocar  <ilya.tocar@intel.com>
2085             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2086             Ilya Verbin  <ilya.verbin@intel.com>
2087             Kirill Yukhin  <kirill.yukhin@intel.com>
2088             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2090         * config/i386/i386.md (any_fix): New iterator.
2091         (fixsuffix): New attribute.
2092         * config/i386/sse.md (VF1): Extened with wider modes.
2093         (VI): Ditto.
2094         (VI_AVX2): Ditto.
2095         (VI8): Ditto.
2096         (sseintvecmodelower): Ditto.
2097         (ssescalarmode): Ditto.
2098         (ssescalarnum): Ditto.
2099         (VF1_128_256): New.
2100         (ssexmmmode): Ditto.
2101         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2102         (<sse>_rcp<mode>2): Change iterator.
2103         (rsqrt<mode>2): Ditto.
2104         (<sse>_rsqrt<mode>2): Ditto.
2105         (avx2_vec_dup<mode>): Ditto.
2106         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2107         (round<mode>2_sfix): Ditto.
2108         (avx2_pbroadcast<mode>): Ditto.
2109         (*andnot<mode>3): Handle XI mode.
2110         (*<code><mode>3): Ditto.
2111         (AVXTOSSEMODE): Removed.
2112         (avx_vpermil<mode>): Changed to ...
2113         (<sse2_avx_avx512f>_vpermil<mode>): This.
2115 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2116             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2117             Sergey Lega  <sergey.s.lega@intel.com>
2118             Anna Tikhonova  <anna.tikhonova@intel.com>
2119             Ilya Tocar  <ilya.tocar@intel.com>
2120             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2121             Ilya Verbin  <ilya.verbin@intel.com>
2122             Kirill Yukhin  <kirill.yukhin@intel.com>
2123             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2125         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2126         (<sse>_comi): Ditto.
2127         (<sse>_ucomi): Ditto.
2128         (sse_cvtss2siq_2): Ditto.
2129         (sse2_cvtsd2si): Ditto.
2130         (sse2_cvtsd2siq): Ditto.
2131         (sse2_cvttsd2si): Ditto.
2132         (sse2_cvttsd2siq): Ditto.
2133         (<shift_insn><mode>3): Ditto.
2134         (sse2_cvtsi2sdq): Update constraint and prefix.
2135         (sse_cvtsi2ss): Update prefix.
2136         (sse_cvtsi2ssq): Ditto.
2138 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2140         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2141         ignore internal calls.
2143 2013-10-11  Richard Biener  <rguenther@suse.de>
2145         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2146         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2147         TREE_OVERFLOW is set.
2149 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2151         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2153         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2154         * gimple.h: Likewise.
2156         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2158         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2159         macros.
2161         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2163         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2164         * configure: Regenerate.
2166 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2168         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2169         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2170         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2171         similarly pp_right_paren.
2172         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2173         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2174         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2175         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2176         clauses.
2177         (omp_declare_simd_clauses_equal,
2178         omp_remove_redundant_declare_simd_attrs): New functions.
2179         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2180         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2181         * tree.h (OMP_LOOP_CHECK): Define.
2182         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2183         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2184         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2185         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2186         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2187         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2188         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2189         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2190         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2191         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2192         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2193         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2194         (omp_remove_redundant_declare_simd_attrs): New prototype.
2195         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2196         GIMPLE_OMP_TEAMS): New codes.
2197         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2198         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2199         fields.
2200         (target_nesting_level): New variable.
2201         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2202         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2203         for collapse > 1 static schedule unless ordered.
2204         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2205         (determine_parallel_type): Adjust get_ws_args_for caller.
2206         (install_var_field): Handle mask & 4 for double indirection.
2207         (scan_sharing_clauses): Ignore shared clause on teams construct.
2208         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2209         (create_omp_child_function): If inside target or declare target
2210         constructs, set "omp declare target" attribute on the child function.
2211         (find_combined_for): New function.
2212         (scan_omp_parallel): Handle combined loops.
2213         (scan_omp_target, scan_omp_teams): New functions.
2214         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2215         restrictions and set ctx->cancellable for cancellable constructs.
2216         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2217         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2218         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2219         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2220         (omp_clause_aligned_alignment): New function.
2221         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2222         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2223         on teams constructs.  Handle user defined reductions and new
2224         OpenMP 4.0 clauses.
2225         (lower_reduction_clauses): Don't set placeholder to address of ref
2226         if it has already the right type.
2227         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2228         (expand_parallel_call): Use the new non-_start suffixed builtins,
2229         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2230         and GOMP_parallel_end after the call.
2231         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2232         (expand_omp_for_init_counts): Handle combined loops.
2233         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2234         loops.
2235         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2236         end of cancellable loops.
2237         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2238         Likewise.  Handle collapse > 1 loops.
2239         (expand_omp_simd): Handle combined loops.
2240         (expand_omp_for): Add inner_stmt argument, adjust callers of
2241         expand_omp_for* functions, use expand_omp_for_static*chunk even
2242         for collapse > 1 unless ordered.
2243         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2244         of cancellable sections.
2245         (expand_omp_single): Remove need_barrier variable, just rely on
2246         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2247         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2248         (expand_omp_atomic_load, expand_omp_atomic_store,
2249         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2250         (expand_omp_target): New function.
2251         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2252         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2253         (build_omp_regions_1): Immediately close region for
2254         GF_OMP_TARGET_KIND_UPDATE.
2255         (maybe_add_implicit_barrier_cancel): New function.
2256         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2257         cancellation.
2258         (lower_omp_single): Likewise.  Add clobber after the barrier.
2259         (lower_omp_taskgroup): New function.
2260         (lower_omp_for): Handle combined loops.  Adjust
2261         lower_rec_input_clauses caller.  Handle cancellation.
2262         (lower_depend_clauses): New function.
2263         (lower_omp_taskreg): Lower depend clauses.  Adjust
2264         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2265         cancellation.
2266         (lower_omp_target, lower_omp_teams): New functions.
2267         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2268         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2269         and GOMP_cancellation_point calls.
2270         (lower_omp): Fold stmts inside of target region.
2271         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2272         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2273         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2274         (BT_FN_VOID_OMPFN_PTR_UINT,
2275         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2276         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2277         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2278         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2279         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2280         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2281         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2282         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2283         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2284         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2285         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2286         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2287         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2288         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2289         BUILT_IN_GOMP_PARALLEL_END.
2290         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2291         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2292         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2293         GF_OMP_FOR_KIND_DISTRIBUTE.
2294         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2295         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2296         (dump_gimple_omp_return): Print lhs if it has any.
2297         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2298         gimple_omp_atomic_seq_cst_p.
2299         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2300         and GIMPLE_OMP_TEAMS.
2301         * langhooks.c (lhd_omp_mappable_type): New function.
2302         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2303         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2304         hook.
2305         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2306         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2307         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2308         (struct gimplify_omp_ctx): Add combined_loop field.
2309         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2310         on stmts inside of target region.
2311         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2312         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2313         ORT_TARGET and ORT_TARGET_DATA.
2314         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2315         Handle GOVD_MAP.
2316         (omp_notice_threadprivate_variable): Complain about threadprivate
2317         variables in target region.
2318         (omp_notice_variable): Complain about vars with non-mappable type
2319         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2320         (omp_check_private): Ignore ORT_TARGET* regions.
2321         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2322         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2323         (find_combined_omp_for): New function.
2324         (gimplify_omp_for): Handle gimplification of combined loops.
2325         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2326         OMP_TEAMS.
2327         (gimplify_omp_target_update): New function.
2328         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2329         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2330         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2331         (gimplify_body): If fndecl has "omp declare target" attribute, add
2332         implicit ORT_TARGET context around it.
2333         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2334         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2335         * tree-nested.c (convert_nonlocal_reference_stmt,
2336         convert_local_reference_stmt, convert_gimple_call): Handle
2337         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2338         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2339         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2340         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2341         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2342         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2343         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2344         BUILT_IN_GOMP_LOOP_END_CANCEL,
2345         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2346         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2347         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2348         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2349         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2350         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2351         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2352         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2353         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2354         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2355         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2356         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2357         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2358         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2359         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2360         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2361         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2362         gimple_build_omp_teams): New functions.
2363         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2364         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2365         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2366         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2367         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2368         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2369         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2370         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2371         GF_OMP_ATOMIC_SEQ_CST): New.
2372         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2373         gimple_build_omp_teams): New prototypes.
2374         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2375         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2376         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2377         GIMPLE_OMP_SINGLE as end of range.
2378         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2379         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2380         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2381         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2382         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2383         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2384         gimple_omp_target_kind, gimple_omp_target_set_kind,
2385         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2386         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2387         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2388         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2389         gimple_omp_teams_set_clauses): New inlines.
2390         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2391         and GIMPLE_OMP_TASKGROUP.
2392         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2393         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2394         enum omp_clause_proc_bind_kind): New.
2395         (union omp_clause_subcode): Add depend_kind, map_kind and
2396         proc_bind_kind fields.
2397         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2398         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2399         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2400         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2401         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2403 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2405         * predict.c (tree_estimate_probability): Add new parameter
2406         for estimate_bb_frequencies.
2407         (estimate_bb_frequencies): Add new parameter to force estimation.
2408         (rebuild_frequencies): When max frequency in function is small,
2409         recompute counts from frequencies.
2410         * predict.h (estimate_bb_frequencies): New parameter.
2412 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2414         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2415         optimizations are disabled.
2417 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2419         * coverage.c (coverage_finish): Fix leak of da_file_name.
2421 2013-10-10  Jan Hubicka  <jh@suse.cz>
2423         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2424         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2425         Bobcat and generic.
2427 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2429         PR middle-end/58670
2430         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2431         if any labels are in FALLTHRU_BB, use a special label emitted
2432         immediately after the asm goto insn rather than label_rtx
2433         of the LABEL_DECL.
2434         (expand_asm_stmt): Adjust caller.
2435         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2436         edge if the last insn in predecessor is a jump with single successor,
2437         but it isn't simplejump_p.
2439 2013-10-10  Richard Biener  <rguenther@suse.de>
2441         PR tree-optimization/58656
2442         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2444 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2446         * gimplify.c: Include expr.h and tm_p.h for targets with special
2447         va-arg padding requirements.
2449 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2451         * tree-flow.h: Move some prototypes to gimple.h.
2452         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2453         * gimple.h: Relocate some prototypes from tree-flow.h
2454         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2455         Move to gimplify.c.
2456         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2457         (build_va_arg_indirect_ref): Relocate and make static.
2458         (std_gimplify_va_arg_expr): Relocate here.
2459         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2460         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2462 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2464         * doc/md.texi: Document the mnemonic attribute.
2466 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2468         PR target/57377
2469         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2470         (set_attr_alternative x ...) when searching for user defined
2471         mnemonic attribute.
2473 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2475         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2477 2013-10-09  Easwaran Raman <eraman@google.com>
2479         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2480         * cfgexpand.c (defer_stack_allocation): ...use here
2481         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2483 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2485         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2486         (operand_equal_for_value_replacement): New function, extracted from
2487         value_replacement and enhanced to catch more cases.
2488         (value_replacement): Use operand_equal_for_value_replacement.
2490 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2492         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2493         get_max_loop_iterations.
2495 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2497         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2498         Place comment above function.
2500 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2502         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2503         are not related to tree-cfg.c.
2504         * tree-ssa-address.h: New file.  Relocate prototypes.
2505         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2506         (addr_for_mem_ref): New.  Combine call to get_address_description and
2507         return addr_for_mem_ref.
2508         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2509         * tree-ssa-live.h: Adjust prototypes.
2510         * passes.c: Include tree-ssa-live.h.
2511         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2512         * graphite.c (graphite_transform_loops): Make static.
2513         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2514         make_pass_graphite, pass_data_graphite_transforms,
2515         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2516         * ipa-pure-const.c (warn_function_noreturn): Make static.
2517         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2518         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2519         Relocate from tree-cfg.c
2520         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2521         static.
2522         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2523         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2524         Move to ipa-pure-const.c.
2525         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2526         Relocate from tree-optimize.c.
2527         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2528         make_pass_fixup_cfg): Move to tree-cfg.c.
2529         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2530         Relocate some prototypes.
2531         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2532         * tree-dump.c (dump_function_to_file): Remove prototype.
2533         Add tree-flow.h to the include file.
2534         * tree-dump.h: Remove prototype.
2535         * tree-parloops.h: New File.  Add prototypes.
2536         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2537         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2538         from tree-ssa-loop.c.
2539         * tree-predcom.c (run_tree_predictive_commoning,
2540         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2541         Relocate here from tree-ssa-loop.c.
2542         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2543         ssa_name_values.release ().
2544         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2545         (ssa_name_values): Relocate from tree-flow.h.
2546         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2547         * tree-ssa-loop.c (run_tree_predictive_commoning,
2548         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2549         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2550         make_pass_graphite, pass_data_graphite_transforms,
2551         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2552         tree_parallelize_loops, pass_data_parallelize_loops,
2553         make_pass_parallelize_loops): Move to other files.
2554         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2555         moved here.
2556         * tree.h: Remove prototypes from tree-address.c.
2558 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2560         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2561         (struct int_tree_map): Move to tree-hasher.h
2562         (SCALE, LABEL, PERCENT): Move to gimple.h
2563         * tree-flow-inline.h: Delete.  Move functions to other files.
2564         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2565         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2566         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2567         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2568         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2569         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2570         * tree-hasher.h: Don't include tree-flow.h.
2571         (struct int_tree_map): Relocate from tree-flow.h.
2572         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2573         tree-flow-inline.h and make static.
2574         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2575         tree-flow-inline.h.
2576         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2577         tree-flow-inline.h and make static.
2578         * tree.h (is_global_var, may_be_aliased): Relocate from
2579         tree-flow-inline.h.
2580         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2581         * value-prof.c: No longer include tree-flow-inline.h.
2582         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2584 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2586         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2587         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2588         (enum move_pos): Move to tree-ssa-loop-im.h
2589         * cfgloop.h: Move some prototypes.
2590         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2591         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2592         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2593         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2594         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2595         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2596         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2597         (enum move_pos): Relocate here.
2598         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2599         tree-ssa-loop.c.
2600         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2601         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2602         make_pass_lim): Relocate here from tree-ssa-loop.c.
2603         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2604         tree-ssa-loop.c.
2605         (loop_edge_to_cancel, unloop_loops): Make static.
2606         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2607         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2608         (tree_complete_unroll, gate_tree_complete_unroll,
2609         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2610         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2611         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2612         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2613         (stmt_invariant_in_loop_p): Remove unused function.
2614         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2615         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2616         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2617         (gcov_type_to_double_int): Move to cfgloop.h.
2618         (double_int_cmp, bound_index,
2619         estimate_numbers_of_iterations_loop): Make static.
2620         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2621         (max_loop_iterations): Factor out get_max_loop_iterations.
2622         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2623         cfgloop.c.
2624         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2625         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2626         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2627         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2628         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2629         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2630         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2631         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2632         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2633         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2634         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2635         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2636         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2637         pass_data_complete_unroll, make_pass_complete_unroll,
2638         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2639         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2640         tree-ssa-loop-ivcanon.c.
2641         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2642         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2643         tree-ssa-loop-prefetch.c.
2644         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2645         tree-ssa-loop-im.c.
2646         (get_lsm_tmp_name): Relocate and add suffix parameter.
2647         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2648         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2649         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2650         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2651         (get_estimated_loop_iterations): Factor out accessor from
2652         estimated_loop_iterations in tree-ssa-loop-niter.c.
2653         (get_max_loop_iterations): Factor out accessor from
2654         _max_loop_iterations in tree-ssa-niter.c.
2655         * loop-unroll.c (decide_unroll_constant_iterations,
2656         decide_unroll_runtime_iterations, decide_peel_simple,
2657         decide_unroll_stupid): Use new get_* accessors.
2659 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2661         PR tree-optimization/20318
2662         * doc/extend.texi (returns_nonnull): New function attribute.
2663         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2664         attribute.
2665         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2666         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2668 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2670         PR middle-end/58570
2671         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2672         false if both components are bitfields.
2674 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2676         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2677         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2678         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2679         * config/aarch64/aarch64.h
2680         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2681         * config/aarch64/aarch64-simd-builtins.def
2682         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2684 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2686         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2688 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2690         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2692 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2694         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2695         (vneg_f64): New intrinsic.
2696         (vneg_s8): Asm replaced with C.
2697         (vneg_s16): Likewise.
2698         (vneg_s32): Likewise.
2699         (vneg_s64): New intrinsic.
2700         (vnegq_f32): Asm replaced with C.
2701         (vnegq_f64): Likewise.
2702         (vnegq_s8): Likewise.
2703         (vnegq_s16): Likewise.
2704         (vnegq_s32): Likewise.
2705         (vnegq_s64): Likewise.
2707 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2709         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2711 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2713         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2714         packed stack special handling.
2715         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2716         back to fixed stack slots for FPRs saved due to stdarg.
2718 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2720         * config/s390/s390.c (s390_frame_info): Restructure function.
2722 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2724         * config/s390/s390.c (struct s390_frame_layout): New field
2725         gpr_save_slots.
2726         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2727         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2728         regs_ever_clobbered to char*.
2729         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2730         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2731         (s390_register_info): Enable FPR save slots.  Move/Copy some
2732         functionality into ...
2733         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2734         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2735         function.
2736         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2737         not imply a stack frame.
2738         (s390_init_frame_layout): Remove variable clobbered_regs.
2739         (s390_update_register_info): Remove function.
2740         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2741         cannot be used for register renaming.
2742         (s390_hard_regno_scratch_ok): New function.
2743         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2744         (s390_initial_elimination_offset): Change offset calculation of
2745         the return address pointer.
2746         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2747         (restore_gprs): Set frame related flag.
2748         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2749         (s390_emit_prologue): Call s390_register_info instead of
2750         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2751         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2752         (s390_optimize_prologue): Call s390_optimize_register_info.
2753         Try to remove also FPR slot save/restore INSNs.  Remove frame
2754         related flags from restore INSNs.
2756 2013-10-08  DJ Delorie  <dj@redhat.com>
2758         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2759         (movhi): Likewise.
2761         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2762         avoid conflict with the MI use of %c.
2763         * config/rl78/rl78-real.md: change %c to %C throughout.
2764         * config/rl78/rl78-virt.md: Likewise.
2766 2013-10-08  Jan Hubicka  <jh@suse.cz>
2768         * config/i386/i386.c (ix86_option_override_internal): Switch
2769         to SSE math for -ffast-math when target ISA supports SSE2.
2771 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2773         * tree-flow.h: Remove some prototypes.
2774         * tree.h: Remove some protypes, add a couple.
2775         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2776         using_eh_for_cleanups_p): Add interface routines for front ends.
2777         * tree-eh.h: New file.  Add protoptyes.
2778         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2779         (add_stmt_to_eh_lp_fn): Make static.
2780         (lower_try_finally): Use new using_eh_for_cleanups_p.
2781         * emit-rtl.c: Include tree-eh.h.
2782         * gimple.h: Include tree-eh.h.
2784 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2786         PR tree-optimization/58480
2787         * tree-vrp.c (infer_nonnull_range): New function.
2788         (infer_value_range): Call infer_nonnull_range.
2790 2013-10-08  Dehao Chen  <dehao@google.com>
2792         PR tree-optimization/58619
2793         * tree-inline.c (copy_phis_for_bb): Combine location data
2794         only if non-null.
2796 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2798         PR target/58423
2799         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2800         RTX_FRAME_RELATED_P on INSN.
2802 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2804         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2805         (altivec_expand_vec_perm_const): Call it.
2807 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2809         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2810         sequences for LE VSX loads and stores at expand time.
2811         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2812         prototype.
2813         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2814         (rs6000_gen_le_vsx_permute): New.
2815         (rs6000_gen_le_vsx_load): New.
2816         (rs6000_gen_le_vsx_store): New.
2817         (rs6000_gen_le_vsx_move): New.
2818         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2819         (*vsx_le_perm_load_v4si): New.
2820         (*vsx_le_perm_load_v8hi): New.
2821         (*vsx_le_perm_load_v16qi): New.
2822         (*vsx_le_perm_store_v2di): New.
2823         (*vsx_le_perm_store_v4si): New.
2824         (*vsx_le_perm_store_v8hi): New.
2825         (*vsx_le_perm_store_v16qi): New.
2826         (*vsx_xxpermdi2_le_<mode>): New.
2827         (*vsx_xxpermdi4_le_<mode>): New.
2828         (*vsx_xxpermdi8_le_V8HI): New.
2829         (*vsx_xxpermdi16_le_V16QI): New.
2830         (*vsx_lxvd2x2_le_<mode>): New.
2831         (*vsx_lxvd2x4_le_<mode>): New.
2832         (*vsx_lxvd2x8_le_V8HI): New.
2833         (*vsx_lxvd2x16_le_V16QI): New.
2834         (*vsx_stxvd2x2_le_<mode>): New.
2835         (*vsx_stxvd2x4_le_<mode>): New.
2836         (*vsx_stxvd2x8_le_V8HI): New.
2837         (*vsx_stxvd2x16_le_V16QI): New.
2839 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2841         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2843 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2845         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2846         loop to work also for 31bit ABI.
2847         Save the stack pointer for frame_size > 0.
2849 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2851         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2852         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2853         constraint letters from expanders.
2854         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2855         retry count to general_operand.
2856         ("tabort"): Give operand 0 a mode.
2857         ("tabort_1"): Add mode and constraint letter for operand 0.
2858         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2860 2013-10-04  Jeff Law  <law@redhat.com>
2862         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2864         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2865         out of ...
2866         (thread_across_edge): Here.  Call it.
2868 2013-10-04  Cary Coutant  <ccoutant@google.com>
2870         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2871         discarding a location list expression (or a piece of one).
2873 2013-10-03  Jan Hubicka  <jh@suse.cz>
2875         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2876         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2877         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2879 2013-10-03  Jan Hubicka  <jh@suse.cz>
2881         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2882         accumulate-outgoing-args when producing unwind info.
2884 2013-10-03  Wei Mi  <wmi@google.com>
2886         * lra-constraints.c (insert_move_for_subreg): New function
2887         extracted from simplify_operand_subreg.
2888         (simplify_operand_subreg): Add reload for paradoxical subreg.
2890 2013-10-03  Rong Xu  <xur@google.com>
2892         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2893         the candidate of builtin_expect such that we should fix the
2894         size/time estimation.
2895         (estimate_function_body_sizes): Do the acutally size/time fix-up
2896         for builtin_expect.
2898 2013-10-03  Rong Xu  <xur@google.com>
2900         * predict.c (tree_predict_by_opcode): Get the probability
2901         for builtin_expect from param builtin_expect_probability.
2902         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2903         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2904         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2906 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2908         PR c++/19476
2909         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2910         * calls.c (alloca_call_p): Use get_callee_fndecl.
2911         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2912         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2913         Likewise.
2914         (vrp_visit_stmt): Remove duplicated code.
2916 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2918         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2919         ceildf2, btruncdf2, instead of vsx_* name.
2921         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2922         iterators to only do V2DF and V4SF here.  Move the DF code to
2923         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2924         just 'v' since only vector operations are handled with these insns
2925         after moving the DF support to rs6000.md.
2926         (vsx_sub<mode>3): Likewise.
2927         (vsx_mul<mode>3): Likewise.
2928         (vsx_div<mode>3): Likewise.
2929         (vsx_fre<mode>2): Likewise.
2930         (vsx_neg<mode>2): Likewise.
2931         (vsx_abs<mode>2): Likewise.
2932         (vsx_nabs<mode>2): Likewise.
2933         (vsx_smax<mode>3): Likewise.
2934         (vsx_smin<mode>3): Likewise.
2935         (vsx_sqrt<mode>2): Likewise.
2936         (vsx_rsqrte<mode>2): Likewise.
2937         (vsx_fms<mode>4): Likewise.
2938         (vsx_nfma<mode>4): Likewise.
2939         (vsx_copysign<mode>3): Likewise.
2940         (vsx_btrunc<mode>2): Likewise.
2941         (vsx_floor<mode>2): Likewise.
2942         (vsx_ceil<mode>2): Likewise.
2943         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2944         (vsx_sminsf3): Likewise.
2945         (vsx_fmadf4): Likewise.
2946         (vsx_fmsdf4): Likewise.
2947         (vsx_nfmadf4): Likewise.
2948         (vsx_nfmsdf4): Likewise.
2949         (vsx_cmpdf_internal1): Likewise.
2951         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2952         simpler to select whether a target has SPE or traditional floating
2953         point support in iterators.
2954         (TARGET_DF_SPE): Likewise.
2955         (TARGET_SF_FPR): Likewise.
2956         (TARGET_DF_FPR): Likewise.
2957         (TARGET_SF_INSN): Macros to say whether floating point support
2958         exists for a given operation for expanders.
2959         (TARGET_DF_INSN): Likewise.
2961         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2962         combining of SF/DF mode operations, using both traditional and VSX
2963         registers.
2964         (Fvsx): Likewise.
2965         (Ff): Likewise.
2966         (Fv): Likewise.
2967         (Fs): Likewise.
2968         (Ffre): Likewise.
2969         (FFRE): Likewise.
2970         (abs<mode>2): Combine SF/DF modes using traditional floating point
2971         instructions.  Add support for using the upper DF registers with
2972         VSX support, and SF registers with power8-vector support.  Update
2973         expanders for operations supported by both the SPE and traditional
2974         floating point units.
2975         (abs<mode>2_fpr): Likewise.
2976         (nabs<mode>2): Likewise.
2977         (nabs<mode>2_fpr): Likewise.
2978         (neg<mode>2): Likewise.
2979         (neg<mode>2_fpr): Likewise.
2980         (add<mode>3): Likewise.
2981         (add<mode>3_fpr): Likewise.
2982         (sub<mode>3): Likewise.
2983         (sub<mode>3_fpr): Likewise.
2984         (mul<mode>3): Likewise.
2985         (mul<mode>3_fpr): Likewise.
2986         (div<mode>3): Likewise.
2987         (div<mode>3_fpr): Likewise.
2988         (sqrt<mode>3): Likewise.
2989         (sqrt<mode>3_fpr): Likewise.
2990         (fre<Fs>): Likewise.
2991         (rsqrt<mode>2): Likewise.
2992         (cmp<mode>_fpr): Likewise.
2993         (smax<mode>3): Likewise.
2994         (smin<mode>3): Likewise.
2995         (smax<mode>3_vsx): Likewise.
2996         (smin<mode>3_vsx): Likewise.
2997         (negsf2): Delete SF operations that are merged with DF.
2998         (abssf2): Likewise.
2999         (addsf3): Likewise.
3000         (subsf3): Likewise.
3001         (mulsf3): Likewise.
3002         (divsf3): Likewise.
3003         (fres): Likewise.
3004         (fmasf4_fpr): Likewise.
3005         (fmssf4_fpr): Likewise.
3006         (nfmasf4_fpr): Likewise.
3007         (nfmssf4_fpr): Likewise.
3008         (sqrtsf2): Likewise.
3009         (rsqrtsf_internal1): Likewise.
3010         (smaxsf3): Likewise.
3011         (sminsf3): Likewise.
3012         (cmpsf_internal1): Likewise.
3013         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
3014         (negdf2): Delete DF operations that are merged with SF.
3015         (absdf2): Likewise.
3016         (nabsdf2): Likewise.
3017         (adddf3): Likewise.
3018         (subdf3): Likewise.
3019         (muldf3): Likewise.
3020         (divdf3): Likewise.
3021         (fred): Likewise.
3022         (rsqrtdf_internal1): Likewise.
3023         (fmadf4_fpr): Likewise.
3024         (fmsdf4_fpr): Likewise.
3025         (nfmadf4_fpr): Likewise.
3026         (nfmsdf4_fpr): Likewise.
3027         (sqrtdf2): Likewise.
3028         (smaxdf3): Likewise.
3029         (smindf3): Likewise.
3030         (cmpdf_internal1): Likewise.
3031         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3032         (btrunc<mode>2): Delete separate expander, and combine with the
3033         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3034         (btrunc<mode>2_fpr): Likewise.
3035         (ceil<mode>2): Likewise.
3036         (ceil<mode>2_fpr): Likewise.
3037         (floor<mode>2): Likewise.
3038         (floor<mode>2_fpr): Likewise.
3039         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3040         Add support for using the upper registers with VSX and
3041         power8-vector.  Move insns to be closer to the define_expands. On
3042         VSX systems, prefer the traditional form of FMA over the VSX
3043         version, since the traditional form allows the target not to
3044         overlap with the inputs.
3045         (fms<mode>4_fpr): Likewise.
3046         (nfma<mode>4_fpr): Likewise.
3047         (nfms<mode>4_fpr): Likewise.
3049 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3050             Richard Earnshaw  <richard.earnshaw@arm.com>
3052         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3053         (struct mult_cost_table): Likewise.
3054         (struct mem_cost_table): Likewise.
3055         (struct fp_cost_table): Likewise.
3056         (struct vector_cost_table): Likewise.
3057         (cpu_cost_table): Likewise.
3058         * config/arm/arm.opt (mold-rts-costs): New option.
3059         (mnew-generic-costs): Likewise.
3060         * config/arm/arm.c (generic_extra_costs): New table.
3061         (cortexa15_extra_costs): Likewise.
3062         (arm_slowmul_tune): Use NULL as new costs.
3063         (arm_fastmul_tune): Likewise.
3064         (arm_strongarm_tune): Likewise.
3065         (arm_xscale_tune): Likewise.
3066         (arm_9e_tune): Likewise.
3067         (arm_v6t2_tune): Likewise.
3068         (arm_cortex_a5_tune): Likewise.
3069         (arm_cortex_a9_tune): Likewise.
3070         (arm_v6m_tune): Likewise.
3071         (arm_fa726te_tune): Likewise.
3072         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3073         (arm_cortex_tune): Use generict_extra_costs.
3074         (shifter_op_p): New function.
3075         (arm_unspec_cost): Likewise.
3076         (LIBCALL_COST): Define.
3077         (arm_new_rtx_costs): New function.
3078         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3079         table is available. Use old costs otherwise unless mnew-generic-costs
3080         is specified.
3081         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3082         (cpu_cost_table): Declare.
3084 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3086         PR target/58460
3087         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3088         (*subs_mul_imm_<mode>)
3089         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3090         (*sub_<shift>_<mode>)
3091         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3092         Remove k constraint.
3094 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3096         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3097         code.
3098         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3100 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3102         * predict.c (probably_never_executed): New function.
3103         (probably_never_executed_bb_p): Invoke probably_never_executed.
3104         (probably_never_executed_edge_p): Ditto.
3105         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3106         Treat profile insanities conservatively.
3108 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3110         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3112 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3114         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3115         use it.  Use smaller increase for scratch.  Don't increase reject
3116         for early clobber scratch.
3117         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3118         setting eliminated regs except setting fp from hfp.
3119         (lra_eliminate): Check lra_insn_recog_data on NULL.
3121 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3123         PR target/58587
3124         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3125         setting -mvsx-timode by default until the underlying problem is fixed.
3126         (RS6000_CPU, power7 defaults): Likewise.
3128 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3130         * config/x-linux (host-linux.o): Remove header dependencies.
3131         Use $(COMPILE) and $(POSTCOMPILE).
3132         * config/t-linux-android (linux-android.o): Ditto.
3134 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3136         * Makefile.in (expmed.o-warn): Remove.
3138 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3140         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3141         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3143 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3145         * dojump.c (do_jump_1): Divide probability between
3146         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3148 2013-10-02  Tom Tromey  <tromey@redhat.com>
3150         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3152 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3154         * tree-flow.h: Remove some prototypes.
3155         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3156         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3157         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3158         mark_virtual_phi_result_for_renaming): Relocate here.
3159         * tree-into-ssa.h: Add prototypes.
3160         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3161         single_pred_before_succ_order.
3162         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3163         (nonfreeing_call_p) Move to gimple.c.
3164         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3165         tree-ssa-phiopt.c.
3166         * basic-block.h (single_pred_before_succ_order): Add prototype.
3167         * gimple.c (nonfreeing_call_p): Relocate here.
3168         * gimple.h: Add prototype.
3169         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3170         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3171         * tree-ssa.h: Include tree-ssa-dom.h.
3173 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3175         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3176         Use $(COMPILE) and $(POSTCOMPILE).
3178         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3180 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3182         * tree-flow.h: Remove some prototypes.
3183         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3184         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3185         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3186         propagate_tree_value*): Move from here to...
3187         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3188         propagate_tree_value*): Relocate here.
3189         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3190         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3191         * gimple-fold.c: Remove gimple-fold.h from include list.
3192         * tree-vrp.c: Remove gimple-fold.h from include list.
3193         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3194         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3195         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3196         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3197         * sese.c: Add tree-ssa-propagate.h to include list.
3199 2013-10-02  Richard Biener  <rguenther@suse.de>
3201         * tree-loop-distribution.c: Include tree-vectorizer.h for
3202         find_loop_location.
3203         (enum partition_kind): Remove PKIND_REDUCTION.
3204         (struct partition_s): Remove has_writes member, add reduction_p member.
3205         (partition_alloc): Adjust.
3206         (partition_builtin_p): Likewise.
3207         (partition_has_writes): Remove.
3208         (partition_reduction_p): New function.
3209         (partition_merge_into): Likewise.
3210         (generate_code_for_partition): Commonize builtin partition
3211         handling tail.
3212         (rdg_cannot_recompute_vertex_p): Remove.
3213         (already_processed_vertex_p): Likewise.
3214         (rdg_flag_vertex): Do not set has_writes.
3215         (classify_partition): Adjust.
3216         (rdg_build_partitions): Do not set has_writes, treat all
3217         partitions as useful.
3218         (distribute_loop): Record number of library calls generated.  Adjust.
3219         (tree_loop_distribution): Report number of loops and library
3220         calls generated as opt-info.
3222 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3224         * tree-flow.h: Include new .h files.  Move prototypes.
3225         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3226         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3227         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3228         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3229         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3230         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3231         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3232         * tree.h (get_ref_base_and_extent): Move prototype out.
3233         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3234         tree-dfa.h.
3235         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3236         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3237         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3238         * tree-scalar-evolution.c: Include tree.h.
3239         * sese.c: Include tree.h.
3240         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3241         * dwarf2out.c: Include tree-dfa.h.
3242         * tree-chrec.c: Include tree.h.
3243         * tree-data-ref.c: Include tree.h.
3245 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3247         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3248         Fix whitespace.
3250 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3252         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3253         Use $(COMPILE) and $(POSTCOMPILE).
3254         (sol2-cxx.o): Likewise.
3255         (sol2-stubs.o): Likewise.
3256         (sol2.o): Likewise.
3257         * config/x-solaris (host-solaris.o): Likewise.
3259         * config/sparc/t-sparc (sparc.o): Remove.
3260         (sparc-c.o): Remove header dependencies.
3261         Use $(COMPILE) and $(POSTCOMPILE).
3262         * config/sparc/x-sparc: Likewise.
3264 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3266         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3267         * config/arc/arc700.md: Likewise.
3268         * config/arc/arc-modes.def: Likewise.
3269         * config/arc/arc-simd.h: Likewise.
3270         * config/arc/t-arc-uClibc: Likewise.
3271         * config/arc/t-arc-newlib: Likewise.
3273 2013-10-02  Renlin Li  <renlin.li@arm.com>
3275         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3276         plus_constant.
3277         (aarch64_expand_epilogue): Likewise.
3279 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3280             Yufeng Zhang  <yufeng.zhang@arm.com>
3282         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3283         declaration.
3284         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3285         'base_in' represent a conversion and legal_cast_p_1 holds; set
3286         'base_in' with the returned value from get_unwidened.
3288 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3290         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3291         plus_constant is not used.
3293 2013-10-01  Wei Mi  <wmi@google.com>
3295         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3296         * config/i386/i386.md: Add define_peephole2 to
3297         break partial reg stall for cvtss2sd/cvtsd2ss.
3299 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3301         * config/arc/arc.c (pass_arc_ifcvt::clone):
3302         Update for ctxt_ -> m_ctxt change.
3304 2013-10-01  Jeff Law  <law@redhat.com>
3306         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3307         outgoing_edge and intermediate_edge fields.  Instead store the path.
3308         (redirection_data::hash): Hash on the last edge's destination index.
3309         (redirection_data::equal): Check the entire thread path.
3310         (lookup_redirectio_data): Corresponding changes.
3311         (create_edge_and_update_destination_phis): Likewise.
3312         (thread_single_edge): Likewise.
3314 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3315             Diego Novillo <dnovillo@google.com>
3317         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3318         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3319         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3320         versions of these patterns.
3322         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3323         (__builtin_arc_aligned): Likewise.
3325         * config/arc/arc.md: Expand adc_0 comment stating the intended
3326         purpose and why it isn't ready.
3327         Replace commented out call_value_via_label_mixed with a
3328         plain comment about bl_s.
3330         * config/arc/arc.c (stdio.h): Don't include directly.
3331         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3332         Remove if (1) condition.
3333         (arc_encode_section_info): Fix comment.
3335 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3337         * config/arc/arc.c (arc_conditional_register_usage):
3338         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3339         Also set reg_alloc_order for DMA config regs.
3341 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3342             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3344         * doc/install.texi (--with-cpu): Mention ARC.
3345         (arc-*-elf32): New paragraph.
3346         (arc-linux-uclibc): Likewise.
3347         * doc/md.texi (Machine Constraints): Add ARC part.
3348         * doc/invoke.texi: (menu): Add ARC Options.
3349         (Machine Dependent Options) <ARC Options>: Add synopsis.
3350         (node ARC Options): Add.
3351         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3352         (ARC Built-in Functions): New section defining
3353         generic ARC built-in functions.
3354         (ARC SIMD Built-in Functions): New section defining SIMD specific
3355         built-in functions.
3356         (Declaring Attributes of Functions): Extended
3357         description of short_call and long_call attributes for ARC and
3358         added index entries.
3360 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3361             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3362             Joern Rennecke  <joern.rennecke@embecosm.com>
3363             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3364             Brendan Kehoe  <brendan@zen.org>
3365             Michael Eager  <eager@eagercon.com>
3366             Simon Cook  <simon.cook@embecosm.com>
3367             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3369         * config/arc, common/config/arc: New directories.
3371 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3372             Brendan Kehoe  <brendan@zen.org>
3373             Simon Cook  <simon.cook@embecosm.com>
3375         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3377 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3379         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3380         * tree-ssa-coalesce.h: New. Move prototype to here.
3381         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3382         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3383         (gimple_can_coalesce_p): Move to...
3384         * gimple.c (gimple_can_coalesce_p): Here.
3386 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3388         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3389         (dump_decl_set): Move to gimple.c.
3390         * gimple.h: Don't include tree-ssa-operands.h.
3391         (dump_decl_set): Add prototype.
3392         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3393         Move to gimple-ssa.h.
3394         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3395         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3396         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3397         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3398         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3399         than PHI_ARG_DEF.
3400         (dump_decl_set): Relocate here.
3401         * gimple-ssa.h: New file.
3402         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3403         Relocate from gimple.h.
3404         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3405         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3406         swap_tree_operands and remove non-ssa path.
3407         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3408         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3409         swap_ssa_operands.
3410         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3411         vect_is_simple_reduction_1): Use swap_ssa_operands.
3412         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3413         (enum need_phi_state): Move to tree-into-ssa.c.
3414         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3415         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3416         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3417         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3418         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3419         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3420         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3421         num_imm_uses): Move to ssa-iterators.h.
3422         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3423         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3424         tree-phinodes.h.
3425         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3426         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3427         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3428         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3429         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3430         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3431         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3432         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3433         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3434         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3435         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3436         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3437         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3438         (set_phi_nodes): Move to tree-phinodes.h.
3439         * tree-ssa-operands.h (enum ssa_op_iter_type,
3440         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3441         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3442         (dump_decl_set): Remove prototype.
3443         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3444         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3445         (set_phi_nodes): Relocate from tree-flow-inline.h.
3446         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3447         tree-flow-inline.h
3448         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3449         include list.  Temporarily add gimple.h to include list.
3450         * ssa-iterators.h: New file.
3451         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3452         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3453         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3454         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3455         Relocate from tree-ssa-operands.h.
3456         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3457         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3458         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3459         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3460         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3461         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3462         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3463         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3464         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3465         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3466         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3467         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3468         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3469         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3470         Relocate from tree-flow-inline.h.
3471         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3473 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3475         * aarch64-simd.md
3476         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3477         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3478         Insert '\t' to output template.
3479         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3480         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3481         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3482         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3484 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3486         * doc/install.texi (Host/target specific installation notes for GCC):
3487         Put @anchor before @heading.
3488         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3489         Use @email for email addresses.
3491 2013-10-01  Jeff Law  <law@redhat.com>
3493         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3494         a vec.  Only delete the path if we create one without successfully
3495         registering a jump thread.
3496         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3497         as a pointer.
3498         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3499         (paths): New vector of jump threading paths.
3500         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3501         (THREAD_PATH): New accessor macro for the entire thread path.
3502         (lookup_redirection_data): Get intermediate and final outgoing edge
3503         from the thread path.
3504         (create_edge_and_update_destination_phis): Copy the threading path.
3505         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3506         jump threading path.
3507         (ssa_redirect_edges): Get edges and block types from the jump threading
3508         path.  Free the path vector.
3509         (thread_block): Get edges from the jump threading path.  Look at the
3510         entire path to see if we thread to a loop exit.  If we cancel a jump
3511         thread request, then free the path vector.
3512         (thread_single_edge): Get edges and block types from the jump threading
3513         path.  Free the path vector.
3514         (thread_through_loop_header): Get edges and block types from the jump
3515         threading path.  Free the path vector.
3516         (mark_threaded_blocks): Iterate over the vector of paths and store
3517         the path on the appropriate edge.  Get edges and block types from the
3518         jump threading path.
3519         (mark_threaded_blocks): Get edges and block types from the jump
3520         threading path.  Free the path vector.
3521         (thread_through_all_blocks): Use the vector of paths rather than
3522         a vector of 3-edge sets.
3523         (register_jump_thread): Accept pointer to a path vector rather
3524         than the path vector itself.  Store the path vector for later use.
3525         Simplify.
3527 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3528             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3530         PR target/58574
3531         * config/s390/s390.c (s390_split_branches): Modify check for table
3532         jump insns.
3533         (s390_chunkify_start): Rearrange table jump insn check in order to
3534         deal with compare and branch insns correctly.
3536 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3538         PR target/58578
3539         Revert
3540         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3541         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3542         define_insn_and_split.
3543         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3544         (shiftsi3_compare): New pattern.
3545         (rrx): New pattern.
3546         * config/arm/unspecs.md (UNSPEC_RRX): New.
3548 2013-10-01  Alan Modra  <amodra@gmail.com>
3550         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3551         casing inout operands.
3553 2013-10-01  Richard Biener  <rguenther@suse.de>
3555         PR tree-optimization/58553
3556         * tree-loop-distribution.c (struct partition_s): Add niter member.
3557         (classify_partition): Populate niter member for the partition
3558         and properly identify whether the relevant store happens before
3559         or after the loop exit.
3560         (generate_memset_builtin): Use niter member from the partition.
3561         (generate_memcpy_builtin): Likewise.
3563 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3565         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3566         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3568 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3570         * basic-block.h (edge_list): Prefix member names with "m_".
3571         * context.h (context): Likewise.
3572         * domwalk.h (dom_walker): Likewise.
3573         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3574         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3575         * hash-table.h (hash_table): Likewise.
3576         * machmode.h (bit_field_mode_iterator): Likewise.
3577         * pass_manager.h (pass_list): Likewise.
3578         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3579         * tree-pass.h (pass_data): Likewise.
3580         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3581         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3582         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3583         * asan.c (pass_data_asan): Update accordingly.
3584         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3585         (control_dependences::control_dependences): Likewise.
3586         (control_dependences::~control_dependences): Likewise.
3587         (control_dependences::~control_dependences): Likewise.
3588         (control_dependences::get_edges_dependent_on): Likewise.
3589         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3590         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3591         * context.c (gcc::context::context): Likewise.
3592         * cprop.c (pass_rtl_cprop::clone): Likewise.
3593         * domwalk.c (dom_walker::walk): Likewise.
3594         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3595         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3596         * mode-switching.c (pass_mode_switching::clone): Likewise.
3597         * passes.c (opt_pass::opt_pass): Likewise.
3598         (pass_manager::pass_manager): Likewise.
3599         * predict.c (pass_strip_predict_hints::clone): Likewise.
3600         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3601         (pass_split_all_insns::clone): Likewise.
3602         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3603         Likewise.
3604         (bit_field_mode_iterator::next_mode): Likewise.
3605         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3606         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3607         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3608         * tree-complex.c (pass_lower_complex::clone): Likewise.
3609         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3610         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3611         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3612         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3613         (pass_fold_builtins::clone): Likewise.
3614         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3615         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3616         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3617         (pass_cd_dce::clone): Likewise.
3618         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3619         (pass_phi_only_cprop::clone): Likewise.
3620         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3621         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3622         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3623         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3624         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3625         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3626         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3627         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3628         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3629         * tree-vrp.c (pass_vrp::clone): Likewise.
3630         * tsan.c (pass_tsan::clone): Likewise.
3632 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3634         PR middle-end/58564
3635         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3636         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3638         PR middle-end/58564
3639         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3640         optimization, punt if sign_bit_p looked through any zero extension.
3642 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3644         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3645         Update redirected out edge count in joiner case.
3646         (ssa_redirect_edges): Common the joiner and non-joiner cases
3647         so that joiner case gets profile updates.
3649 2013-09-30  Richard Biener  <rguenther@suse.de>
3651         PR tree-optimization/58554
3652         * tree-loop-distribution.c (classify_partition): Require
3653         unconditionally executed stores for memcpy and memset recognition.
3654         (tree_loop_distribution): Calculate dominance info.
3656 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3658         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3659         (NO_PROFILE_COUNTERS): Likewise.
3660         (PROFILE_HOOK): Likewise.
3661         (FUNCTION_PROFILER): Likewise.
3662         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3664 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3666         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3667         calls and defines in TARGET_MACHO conditional.
3668         (load_macho_picbase_di): Likewise.
3669         (reload_macho_picbase): Likewise.
3670         (reload_macho_picbase_si): Likewise.
3671         (reload_macho_picbase_di): Likewise.
3672         (nonlocal_goto_receiver): Likewise.
3674 2013-09-30  Nick Clifton  <nickc@redhat.com>
3676         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3677         that use the MSP430X ISA.
3678         (msp430_option_override): Scan -mmcu command line option for any
3679         MCU name that supports the MSP430X ISA.
3680         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3681         -mmcu options which enable the MSP430X ISA.
3683 2013-09-30  Richard Biener  <rguenther@suse.de>
3685         PR middle-end/58532
3686         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3687         before looking for setjmp-like calls.
3689 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3691         PR target/10901
3692         * config/darwin-protos.h (machopic_get_function_picbase): New.
3693         * config/darwin.c (machopic_get_function_picbase): New.
3694         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3695         label for a new func.  (load_macho_picbase_di): Likewise.
3696         (reload_macho_picbase): New expand.
3697         (reload_macho_picbase_si): New insn.
3698         (reload_macho_picbase_di): New insn.
3699         (nonlocal_goto_receiver): New define and split.
3700         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3701         (unspecv enum): Add UNSPECV_NLGR.
3703 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3705         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3706         that altivec registers are correctly sized on Darwin.
3708 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3710         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3711         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3712         * config/x-darwin (host-darwin.o): Likewise.
3713         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3714         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3715         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3717 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3719         * doc/invoke.texi: Fix usage of @tie{} command.
3721 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3723         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3725 2013-09-28  Jan Hubicka  <jh@suse.cz>
3727         * config/i386/x86-tune.def: Add documentation for each of the options;
3728         add whitespace.
3730 2013-09-28  Jan Hubicka  <jh@suse.cz>
3732         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3733         generic.
3734         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3735         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3736         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3737         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3738         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3740 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3742         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3743         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3744         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3745         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3746         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3747         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3748         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3749         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3750         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3751         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3752         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3753         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3754         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3755         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3756         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3757         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3758         gimple.h, godump.c, graphite-clast-to-gimple.c,
3759         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3760         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3761         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3762         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3763         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3764         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3765         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3766         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3767         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3768         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3769         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3770         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3771         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3772         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3773         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3774         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3775         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3776         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3777         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3778         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3779         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3780         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3781         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3782         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3783         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3784         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3785         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3786         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3787         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3788         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3789         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3790         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3791         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3792         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3793         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3794         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3795         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3796         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3797         whitespace before "(".
3799 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3801         * expr.h (extract_bit_field): Remove packedp parameter.
3802         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3803         from forward declaration.
3804         (store_split_bit_field): Remove packedp arg from calls to
3805         extract_fixed_bit_field.
3806         (extract_bit_field_1): Remove packedp parameter and packedp
3807         argument from recursive calls and calls to extract_fixed_bit_field.
3808         (extract_bit_field): Remove packedp parameter and corresponding
3809         arg to extract_bit_field_1.
3810         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3811         to issue warnings.
3812         (extract_split_bit_field): Remove packedp arg from call to
3813         extract_fixed_bit_field.
3814         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3815         (copy_blkmode_from_reg): Likewise.
3816         (copy_blkmode_to_reg): Likewise.
3817         (read_complex_part): Likewise.
3818         (store_field): Likewise.
3819         (expand_expr_real_1): Likewise.
3820         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3821         to extract_bit_field.
3822         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3823         call to extract_bit_field.
3824         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3825         call to extract_bit_field.
3826         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3827         and special packedp behavior from -fstrict-volatile-bitfields
3828         documentation.
3830 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3832         * lra-eliminations.c (init_elim_table): Guard value_p.
3834 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3836         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3837         DFmode, DImode, and SFmode in the upper VSX registers based on the
3838         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3839         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3840         of the VSX register handling.
3842         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3843         (f32_sv): Likewise.
3844         (zero_extendsidi2_lfiwzx): Add support for loading into the
3845         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3846         only do VSX memory options on Altivec registers.
3847         (extendsidi2_lfiwax): Likewise.
3848         (extendsfdf2_fpr): Likewise.
3849         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3850         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3851         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3852         (movdi_internal64): Likewise.
3854 2013-09-27  Xinliang David Li  <davidxl@google.com>
3856         * opts.c (finish_options): Adjust parameters
3857         according to vect cost model.
3858         (common_handle_option): Set dynamic vect cost
3859         model for FDO.
3860         targhooks.c (default_add_stmt_cost): Compute stmt cost
3861         unconditionally.
3862         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3863         Use helper function.
3864         * tree-vectorizer.h (unlimited_cost_model): New function.
3865         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3866         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3867         function.
3868         (vect_enhance_data_refs_alignment): Ditto.
3869         * flag-types.h: New enum.
3870         * common/config/i386/i386-common.c (ix86_option_init_struct):
3871         No need to initialize vect_cost_model flag.
3872         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3873         unconditionally.
3875 2013-09-27  Diego Novillo  <dnovillo@google.com>
3877         * gimple.h (enum ssa_mode): Remove.
3879 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3881         * cfgloop.h (number_of_loops): Fix typo in check for null.
3883 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3885         PR middle-end/58551
3886         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3887         are children of outermost saved_cfun's loop, and set it up to
3888         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3889         if loop != loop0 and SESE region contains bbs that belong to loop0.
3891 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3893         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3894         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3895         (set_address_segment, set_address_base, set_address_index)
3896         (set_address_disp): Accept the argument unconditionally.
3897         (baseness): Remove must_be_base_p and must_be_index_p checks.
3898         (decompose_normal_address): Classify as much as possible in the
3899         main loop.
3901 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3903         * cse.c (count_reg_usage): Handle INT_LIST.
3904         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3905         * reginfo.c (reg_scan_mark_refs): Likewise.
3906         * reload1.c (eliminate_regs_1): Likewise.
3908 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3910         PR middle-end/58547
3911         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3912         signed.
3914 2013-09-27  Richard Biener  <rguenther@suse.de>
3916         PR tree-optimization/58459
3917         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3918         restriction not propagating into loops.
3920 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3922         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3923         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3924         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3926 2013-09-26  Richard Biener  <rguenther@suse.de>
3928         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3930 2013-09-26  Richard Biener  <rguenther@suse.de>
3932         * alias.h (component_uses_parent_alias_set): Rename to ...
3933         (component_uses_parent_alias_set_from): ... this.
3934         * alias.c (component_uses_parent_alias_set): Rename to ...
3935         (component_uses_parent_alias_set_from): ... this and return
3936         the desired parent.
3937         (reference_alias_ptr_type_1): Use the result from
3938         component_uses_parent_alias_set_from instead of stripping
3939         components one at a time.
3940         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3942 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3944         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3945         Move prototypes to...
3946         * tree-ssa-ter.h: New File.  Move prototypes here.
3947         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3948         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3949         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3950         from is_replaceable_p.
3951         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3952         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3953         (process_replaceable): Use ter_is_replaceable_p.
3954         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3955         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3956         newly refactored ssa_is_replaceable_p.
3957         * cfgexpand.c: Include tree-outof-ssa.h.
3958         * ssaexpand.h: Delete.
3960 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3962         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3963         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3964         tree-ssa.c
3965         (create_gimple_tmp): Delete.
3966         (get_expr_type, build_assign, build_type_cast): Move to...
3967         * gimple-builder.c: New File.
3968         (get_expr_type): Relocate from gimple.c.
3969         (build_assign, build_type_cast): Change to only create ssanames.
3970         * gimple.h: Move prototypes to...
3971         * gimple-builder.h: New File. Here.
3972         * tree-ssa.h: And here.
3973         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3974         count_uses_and_derefs): Relocate from gimple.c.
3975         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3976         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3977         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3978         gimple_replace_ssa_lhs.
3979         * asan.c: Include gimple-builder.h.
3980         * Makefile.in: Add gimple-builder.o.
3982 2013-09-26  Richard Biener  <rguenther@suse.de>
3984         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3985         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3986         (loe_visit_block): Use gcc_checking_assert.
3987         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3988         gimple_assign_ssa_name_copy_p.
3989         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3990         change.
3992 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3994         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3995         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3997 2013-09-26  Richard Biener  <rguenther@suse.de>
3999         PR tree-optimization/58539
4000         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
4001         the fact that debug statements are not taking part in loop-closed
4002         SSA construction.
4004 2013-09-26  Nick Clifton  <nickc@redhat.com>
4006         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4007         time warning message.
4008         (msp430_print_operand_raw): Delete unused letter parameter.
4009         (TARGET_PRINT_OPERAND_ADDRESS): Define.
4010         (msp430_print_operand_address): New function.
4011         (msp430_print_operand): Move address printing code from here to
4012         new function.
4013         * config/msp430/msp430.md (movsipsi2): Add comment in generated
4014         assembler.
4015         (zero_extendpsisi2): Likewise.
4016         (extendpsisi2): New pattern.
4017         (andneghi3): New pattern.
4019 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4021         * config/aarch64/aarch64.opt (mlra): New option.
4022         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4023         (TARGET_LRA_P): Define.
4025 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4027         * expr.c (expand_assignment): Remove obsolete comment.
4029 2013-09-25  Jeff Law  <law@redhat.com>
4031         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4032         tree-ssa-threadupdate.h.
4033         (register_jump_thread): Similarly.
4034         * tree-ssa-threadupdate.h: New header file.
4035         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4036         * tree-vrp.c: Likewise.
4037         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4038         (thread_around_empty_blocks): Change type of path vector argument to
4039         an edge,type pair from just an edge.  Initialize both elements when
4040         appending to a jump threading path.  Tweak references to elements
4041         appropriately.
4042         (thread_across_edge): Similarly.  Release memory for the elements
4043         as needed.
4044         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4045         (dump_jump_thread_path): New function broken out from
4046         register_jump_thread.
4047         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4048         path vector entries.  Search the path for NULL edges and dump
4049         the path if one is found.  Tweak the conversion of path to 3-edge
4050         form to use the block copy type information embedded in the path.
4052 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4054         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4056 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4057             Vladimir Makarov  <vmakarov@redhat.com>
4059         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4060         from the least significant bit.
4061         (strip_address_mutations): Add bitfield operations handling.
4062         (must_be_index_p): Add shifting and rotate operations handling.
4063         (set_address_base): Use must_be_base_p predicate.
4064         (set_address_index): Use must_be_index_p predicate.
4066 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4067             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4068             Sergey Lega  <sergey.s.lega@intel.com>
4069             Anna Tikhonova  <anna.tikhonova@intel.com>
4070             Ilya Tocar  <ilya.tocar@intel.com>
4071             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4072             Ilya Verbin  <ilya.verbin@intel.com>
4073             Kirill Yukhin  <kirill.yukhin@intel.com>
4074             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4076         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4077         Use new names.
4078         (ix86_expand_vector_move_misalign): Support new unaligned load and
4079         stores and use new names.
4080         (CODE_FOR_sse2_storedqu): Rename to ...
4081         (CODE_FOR_sse2_storedquv16qi): ... this.
4082         (CODE_FOR_sse2_loaddqu): Rename to ...
4083         (CODE_FOR_sse2_loaddquv16qi): ... this.
4084         (CODE_FOR_avx_loaddqu256): Rename to ...
4085         (CODE_FOR_avx_loaddquv32qi): ... this.
4086         (CODE_FOR_avx_storedqu256): Rename to ...
4087         (CODE_FOR_avx_storedquv32qi): ... this.
4088         * config/i386/i386.md (fpint_logic): New.
4089         * config/i386/sse.md (VMOVE): Extend for AVX512.
4090         (VF): Ditto.
4091         (VF_128_256): New.
4092         (VF_512): Ditto.
4093         (VI_UNALIGNED_LOADSTORE): Ditto.
4094         (sse2_avx_avx512f): Ditto.
4095         (sse2_avx2): Extend for AVX512.
4096         (sse4_1_avx2): Ditto.
4097         (avx2_avx512f): New.
4098         (sse): Extend for AVX512.
4099         (sse2): Ditto.
4100         (sse4_1): Ditto.
4101         (avxsizesuffix): Ditto.
4102         (sseintvecmode): Ditto.
4103         (ssePSmode): Ditto.
4104         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4105         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4106         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4107         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4108         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4109         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4110         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4111         (STORENT_MODE): Extend for AVX512.
4112         (*absneg<mode>2): Replace constraint "x" with "v".
4113         (*mul<mode>3): Ditto.
4114         (*ieee_smin<mode>3): Ditto.
4115         (*ieee_smax<mode>3): Ditto.
4116         (avx_cmp<mode>3): Replace VF with VF_128_256.
4117         (*<sse>_maskcmp<mode>3_comm): Ditto.
4118         (<sse>_maskcmp<mode>3): Ditto.
4119         (<sse>_andnot<mode>3): Extend for AVX512.
4120         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4121         (<code><mode>3, fpint_logic): New.
4122         (*<code><mode>3): Extend for AVX512.
4123         (avx512flogicsuff): New.
4124         (avx512f_<logic><mode>): Ditto.
4125         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4126         VF_128_256.
4127         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4128         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4129         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4130         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4131         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4132         (xop_vpermil2<mode>3): Ditto.
4133         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4134         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4135         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4136         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4138 2013-09-25  Tom Tromey  <tromey@redhat.com>
4140         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4141         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4142         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4143         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4144         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4145         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4146         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4147         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4148         (GRAPHITE_HTAB_H): Remove.
4150 2013-09-25  Tom Tromey  <tromey@redhat.com>
4152         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4154 2013-09-25  Tom Tromey  <tromey@redhat.com>
4156         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4158 2013-09-25  Tom Tromey  <tromey@redhat.com>
4160         * config/i386/t-i386 (i386.o): Remove.
4161         (i386-c.o): Use COMPILE and POSTCOMPILE.
4163 2013-09-25  Tom Tromey  <tromey@redhat.com>
4165         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4167 2013-09-25  Tom Tromey  <tromey@redhat.com>
4169         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4170         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4171         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4172         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4173         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4174         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4175         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4176         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4177         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4178         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4179         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4180         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4181         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4182         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4183         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4184         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4185         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4186         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4187         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4188         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4189         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4190         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4191         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4192         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4193         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4194         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4195         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4196         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4197         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4198         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4199         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4200         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4201         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4202         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4203         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4204         (graphite-clast-to-gimple.o, graphite-dependences.o)
4205         (graphite-interchange.o, graphite-poly.o)
4206         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4207         (graphite-optimize-isl.o, tree-vect-loop.o)
4208         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4209         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4210         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4211         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4212         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4213         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4214         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4215         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4216         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4217         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4218         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4219         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4220         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4221         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4222         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4223         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4224         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4225         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4226         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4227         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4228         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4229         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4230         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4231         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4232         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4233         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4234         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4235         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4236         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4237         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4238         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4239         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4240         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4241         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4242         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4243         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4244         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4245         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4246         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4247         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4248         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4249         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4250         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4251         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4252         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4253         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4254         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4255         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4256         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4257         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4258         (gcov-dump.o): Remove.
4259         (default-c.o): Use COMPILE and POSTCOMPILE.
4260         (CFLAGS-gcc.o): New variable.
4261         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4263 2013-09-25  Tom Tromey  <tromey@redhat.com>
4265         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4266         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4267         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4268         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4269         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4270         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4271         (c-family/array-notation-common.o, c-family/stub-objc.o)
4272         (c-family/c-ubsan.o): Remove.
4274 2013-09-25  Tom Tromey  <tromey@redhat.com>
4276         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4278 2013-09-25  Tom Tromey  <tromey@redhat.com>
4280         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4281         to add -DENABLE_SHARED_LIBGCC.
4282         (gcc.o): Don't use subshell.
4284 2013-09-25  Tom Tromey  <tromey@redhat.com>
4286         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4287         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4288         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4289         * configure, config.in: Rebuild.
4291 2013-09-25  Tom Tromey  <tromey@redhat.com>
4293         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4294         (COMPILE, POSTCOMPILE): New variables.
4295         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4296         (DEPFILES): New variable.
4297         Include ".Po" files.
4298         * configure.ac: Add checks for dependency checking.
4299         * configure, aclocal.m4: Regenerate.
4301 2013-09-25  Tom Tromey  <tromey@redhat.com>
4303         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4304         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4306 2013-09-25  Tom Tromey  <tromey@redhat.com>
4308         * Makefile.in (generated_files): Add options.h,
4309         target-hooks-def.h, insn-opinit.h,
4310         common/common-target-hooks-def.h, pass-instances.def,
4311         c-family/c-target-hooks-def.h.
4313 2013-09-25  Jeff Law  <law@redhat.com>
4315         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4316         than foo[foo.length () - 1] to access last member in a vec.
4317         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4319 2013-09-25  Richard Biener  <rguenther@suse.de>
4321         PR middle-end/58521
4322         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4324 2013-09-25  Jan Hubicka  <jh@suse.cz>
4326         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4327         test.
4329 2013-09-25  Marek Polacek  <polacek@redhat.com>
4331         PR sanitizer/58420
4332         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4333         when determining the type name.
4335 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4337         * config/sh/sh.md: Fix formatting.
4339 2013-09-24  Xinliang David Li  <davidxl@google.com>
4341         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4342         max peel iterations parameter.
4343         * param.def: New parameter.
4344         * doc/invoke.texi: Document New parameter.
4346 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4348         * gimple-pretty-print.c: Various whitespace tweaks.
4349         * tree-core.h: Likewise.
4350         * tree-pretty-print.c: Likewise.
4351         * tree-ssa-alias.c: Likewise.
4352         * tree-ssa-copy.c: Likewise.
4353         * tree-ssanames.c: Likewise.
4354         * tree-ssanames.h: Likewise.
4355         * tree-vrp.c: Likewise.
4357 2013-09-24  Alan Modra  <amodra@gmail.com>
4359         PR middle-end/57134
4360         PR middle-end/57586
4361         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4362         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4363         inout operands.
4365 2013-09-24  Richard Biener  <rguenther@suse.de>
4367         PR middle-end/58513
4368         * tree.c (reference_alias_ptr_type): Move ...
4369         * alias.c (reference_alias_ptr_type): ... here and implement
4370         in terms of the new reference_alias_ptr_type_1.
4371         (ref_all_alias_ptr_type_p): New helper.
4372         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4373         use ref_all_alias_ptr_type_p.
4374         (get_deref_alias_set): Add flag_strict_aliasing check here.
4375         (reference_alias_ptr_type_1): New function, split out from ...
4376         (get_alias_set): ... here.
4377         (alias_ptr_types_compatible_p): New function.
4378         * alias.h (reference_alias_ptr_type): Declare.
4379         (alias_ptr_types_compatible_p): Likewise.
4380         * tree.h (reference_alias_ptr_type): Remove.
4381         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4382         to compare MEM_REF alias types.
4384 2013-09-24  Richard Biener  <rguenther@suse.de>
4386         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4388 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4390         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4391         reload helper function arrays into a single array reg_addr.
4392         (reload_fpr_gpr): Likewise.
4393         (reload_gpr_vsx): Likewise.
4394         (reload_vsx_gpr): Likewise.
4395         (struct rs6000_reg_addr): Likewise.
4396         (reg_addr): Likewise.
4397         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4398         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4399         (rs6000_init_hard_regno_mode_ok): Likewise.
4400         (rs6000_secondary_reload_direct_move): Likewise.
4401         (rs6000_secondary_reload): Likewise.
4403         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4404         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4405         power8 changes.  Put wg constraint in alphabetical order.
4407         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4408         for future work to add ISA 2.07 VSX single precision support.
4409         (-mvsx-scalar-double): Change default from -1 to 1, update
4410         documentation comment.
4411         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4412         (-mupper-regs-df): New debug switch to control whether DF values
4413         can go in the traditional Altivec registers.
4414         (-mupper-regs-sf): New debug switch to control whether SF values
4415         can go in the traditional Altivec registers.
4417         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4418         and wy constraints.
4419         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4420         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4421         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4422         (rs6000_debug_legitimate_address_p): Print if we are running
4423         before, during, or after reload.
4424         (rs6000_secondary_reload): Add a comment.
4425         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4427         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4428         constraints.  Update documentation string.
4429         (wd constraint): Likewise.
4430         (wf constraint): Likewise.
4431         (wg constraint): Likewise.
4432         (wn constraint): Likewise.
4433         (ws constraint): Likewise.
4434         (wt constraint): Likewise.
4435         (wx constraint): Likewise.
4436         (wz constraint): Likewise.
4437         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4438         instructions.
4439         (ww constraint): Likewise.
4440         (wy constraint): Likewise.
4441         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4442         in the previous submissions.
4443         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4445 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4447         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4448         (REG_BR_PROB): Say that the probability is stored in an int_list.
4449         * reg-notes.def: Update commentary to mention INT_LIST.
4450         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4451         (INT_LIST): New rtx.
4452         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4453         * rtlanal.c (int_reg_note_p): New function.
4454         (alloc_reg_note): Assert that the note does not have an int argument.
4455         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4456         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4457         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4458         rather than an INSN_LIST.  Handle INT_LIST.
4459         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4460         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4461         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4462         Manipulate them as ints rather than rtxes.
4463         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4464         * regmove.c (copy_src_to_dest): Likewise.
4465         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4467         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4468         into the cases that need it.
4469         * config/arm/arm.c (arm_unwind_emit): Likewise.
4471         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4472         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4473         * loop-doloop.c (add_test, doloop_modify): Likewise.
4474         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4475         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4476         * predict.c (combine_predictions_for_insn): Likewise.
4477         * print-rtl.c (print_rtx): Handle INT_LIST.
4478         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4479         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4480         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4481         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4482         (ix86_split_fp_branch, predict_jump): Likewise.
4483         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4484         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4485         * config/spu/spu.c (ea_load_store_inline): Likewise.
4487         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4488         value of a REG_BR_PROB note.
4489         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4490         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4491         * emit-rtl.c (try_split): Likewise.
4492         * predict.c (br_prob_note_reliable_p): Likewise.
4493         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4494         * reorg.c (mostly_true_jump): Likewise.
4495         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4496         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4497         * config/i386/i386.c (ix86_print_operand): Likewise.
4498         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4499         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4500         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4501         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4502         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4503         * config/sparc/sparc.c (output_cbranch): Likewise.
4504         * config/spu/spu.c (get_branch_target): Likewise.
4505         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4506         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4508 2013-09-23  Jan Hubicka  <jh@suse.cz>
4510         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4511         for ipa-devirt.
4512         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4513         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4514         of external calls
4515         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4516         (gimple_fold_call): Dump inconsistent devirtualizations; add
4517         sanity check for type based devirtualizations.
4518         * ipa-prop.c: Include ipa-utils.h
4519         (ipa_intraprocedural_devirtualization): Add sanity check.
4520         (try_make_edge_direct_virtual_call): Likewise.
4522 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4524         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4525         assignment statements.
4527 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4529         * gimple-pretty-print.c (dump_ssaname_info): New function.
4530         (dump_gimple_phi): Call it.
4531         (pp_gimple_stmt_1): Likewise.
4532         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4533         (range_info_def): Declare.
4534         * tree-pretty-print.c (pp_double_int): New function.
4535         (dump_generic_node): Call it.
4536         * tree-pretty-print.h (pp_double_int): Declare.
4537         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4538         * tree-ssanames.h (range_info_def): New structure.
4539         (value_range_type): Move definition here.
4540         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4541         Declare.
4542         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4543         initialization.
4544         (set_range_info): New function.
4545         (get_range_info): Likewise.
4546         (duplicate_ssa_name_range_info): Likewise.
4547         (duplicate_ssa_name_fn): Check pointer type and call
4548         duplicate_ssa_name_range_info.
4549         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4550         * tree-vrp.c (value_range_type): Remove definition, now in
4551         tree-ssanames.h.
4552         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4553         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4554         (SSA_NAME_RANGE_INFO): New macro.
4556 2013-09-23  Richard Biener  <rguenther@suse.de>
4558         PR tree-optimization/58464
4559         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4560         (phi_trans_add): Change to add conditionally on being not present.
4561         (phi_translate_1): Remove recursion detection here.
4562         (phi_translate): Pre-seed the cache with NULL to catch
4563         recursion here in a more generic way.
4564         (bitmap_find_leader): Adjust comment.
4565         (get_representative_for): Dump value-numbers.
4566         (create_expression_by_pieces): Likewise.
4567         (insert_into_preds_of_block): Likewise.
4569 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4571         PR target/58475
4572         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4573         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4575 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4577         Revert r202780:
4578         2013-09-20  Renlin Li  <renlin.li@arm.com>
4580         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4581         plus_constant.
4582         (aarch64_expand_epilogue): Likewise.
4583         (aarch64_legitimize_reload_address): Likewise.
4585 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4587         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4588         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4590 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4592         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4593         comparison operators when -fno-trapping-math is in effect.
4594         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4595         comparison operators in TFmode and assert that unsupported operators
4596         cannot reach here.
4597         (ia64_print_operand): Likewise.
4599 2013-09-21  Jan Hubicka  <jh@suse.cz>
4601         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4602         (sse_typeless_stores): Enable for core
4603         (sse_load0_by_pxor): Likewise.
4604         (four_jump_limit): Disable for core.
4605         (pad_returns): Likewise.
4606         (avoid_vector_decode): Likewise.
4607         (fuse_cmp_and_branch): Enable for cores.
4608         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4610 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4612         PR middle-end/56791
4613         * config/pa/pa.c (pa_option_override): Disable auto increment and
4614         decrement instructions until reload is completed.
4616         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4617         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4618         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4620 2013-09-20  DJ Delorie  <dj@redhat.com>
4621             Nick Clifton  <nickc@redhat.com>
4623         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4624         (need_to_save): Save bank 0 on interrupts.
4625         (characterize_address): Strip far address wrappers.
4626         (rl78_as_legitimate_address): Likewise.
4627         (transcode_memory_rtx): Likewise.
4628         (rl78_peep_movhi_p): Disable this peephole after devirt.
4629         (rl78_propogate_register_origins): Forget all origins when a
4630         CLOBBER is seen.
4631         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4632         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4633         peephole2's.
4634         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4635         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4636         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4637         relocating.
4638         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4640 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4642         * config/pa/pa.md: In "scc" insn patterns, change output template to
4643         handle const0_rtx in reg_or_0_operand operands.
4645 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4647         PR target/56875
4648         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4649         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4651 2013-09-20  Richard Biener  <rguenther@suse.de>
4653         PR middle-end/58484
4654         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4655         remembering SSA name version and block index.
4656         (new_scev_info_str): Adjust.
4657         (hash_scev_info): Likewise.  Also hash the block index.
4658         (eq_scev_info): Adjust.
4659         (find_var_scev_info): Likewise.
4660         (struct instantiate_cache_entry): Remove.
4661         (struct instantiate_cache_type): Use a htab to map name, block
4662         to chrec.
4663         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4664         (get_instantiated_value_entry): Likewise.
4665         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4666         (instantiate_scev_name): Adjust.
4668 2013-09-20  Jeff Law  <law@redhat.com>
4670         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4672 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4674         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4675         Call aarch64_simd_expand_args to update op[argc].
4677 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4679         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4680         plugin argument.
4682 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4684         * gengtype.c (file_rules): Added rule for *.cc files.
4685         (get_output_file_with_visibility): Give fatal message when no
4686         rules found.
4688 2013-09-20  Renlin Li  <renlin.li@arm.com>
4690         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4691         (aarch64_expand_epilogue): Likewise.
4692         (aarch64_legitimize_reload_address): Likewise.
4694 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4696         PR middle-end/57748
4697         * expr.c (expand_assignment): Remove misalignp code path.
4699 2013-09-20  Marek Polacek  <polacek@redhat.com>
4701         PR sanitizer/58413
4702         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4703         TYPE_PRECISION.  Add asserts.
4705 2013-09-20  Richard Biener  <rguenther@suse.de>
4707         PR tree-optimization/58453
4708         * tree-loop-distribution.c (distribute_loop): Apply the cost
4709         model for -ftree-loop-distribute-patterns, too.
4711 2013-09-20  Richard Biener  <rguenther@suse.de>
4713         PR middle-end/58473
4714         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4715         make type comparison less strict.
4717 2013-09-20  Alan Modra  <amodra@gmail.com>
4719         * configure: Regenerate.
4720         * aclocal.m4: Regenerate.
4722 2013-09-20  Marek Polacek  <polacek@redhat.com>
4724         PR other/58467
4725         * doc/extend.texi: Document that attribute used is meant to be used
4726         on variables with static storage duration.
4728 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4730         PR tree-optimization/58472
4731         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4732         simd_lane_access set inv_p = false.
4733         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4734         the simduid magic VAR_DECL.
4736 2013-09-19  Jan Hubicka  <jh@suse.cz>
4738         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4740 2013-09-17  Jeff Law  <law@redhat.com>
4742         * tree-ssa-dom.c (record_temporary_equivalences): New function
4743         split out of dom_opt_dom_walker::after_dom_children.
4744         (dom_opt_dom_walker::thread_across_edge): Move common code
4745         in here from dom_opt_dom_walker::after_dom_children.
4746         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4748 2013-09-19  Jan Hubicka  <jh@suse.cz>
4750         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4751         (TARGET_GENERIC): Use PROCESOR_GENERIC
4752         (enum processor_type): Unify generic32 and 64.
4753         * i386.md (cpu): Likewise.
4754         * x86-tune.def (use_leave): Enable for generic32.
4755         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4756         * athlon.md: Change generic64 to generic in all occurences.
4757         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4758         (ix86_target_macros_internal): Likewise.
4759         * driver-i386.c (host_detect_local_cpu): Likewise.
4760         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4761         to ..
4762         (generic_memcpy, generic_memset, generic_cost): This one.
4763         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4764         (m_GENERIC32, m_GENERIC64): Remove.
4765         (m_GENERIC): Turn into one flag.
4766         (processor_target): Unify generic tunnings.
4767         (ix86_option_override_internal): Replace generic32/64 by generic.
4768         (ix86_issue_rate): Likewise.
4769         (ix86_adjust_cost): Likewise.
4771 2013-09-19  Jan Hubicka  <jh@suse.cz>
4773         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4774         of speculative flag.
4776 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4778         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4779         GOMP_sections_start, even if !exit_reachable.
4781 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4783         * lra-constraints.c (need_for_all_save_p): Use macro
4784         HARD_REGNO_CALL_PART_CLOBBERED.
4785         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4786         set up pseudo conflict hard regs.
4788 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4790         PR target/58452
4791         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4792         operands.
4794 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4796         PR rtl-optimization/58438
4797         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4798         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4799         from previous subpasses.
4801 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4803         * arm.c (arm_get_frame_offsets): Validate architecture supports
4804         LDRD/STRD before accepting the tuning preference.
4805         (arm_expand_prologue): Likewise.
4806         (arm_expand_epilogue): Likewise.
4808 2013-09-18  Richard Biener  <rguenther@suse.de>
4810         PR tree-optimization/58417
4811         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4812         have chrecs with symbols defined in the loop as operands.
4813         (chrec_fold_multiply): Likewise.
4814         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4815         parameters before folding binary operations.
4816         (struct instantiate_cache_entry_hasher): Remove.
4817         (struct instantiate_cache_type): Use a pointer-map.
4818         (instantiate_cache_type::instantiate_cache_type): New function.
4819         (instantiate_cache_type::get): Likewise.
4820         (instantiate_cache_type::set): Likewise.
4821         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4822         (get_instantiated_value_entry): Likewise.
4823         (global_cache): New global.
4824         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4825         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4826         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4827         (instantiate_scev_name): Adjust.
4828         (instantiate_scev): Construct global instead of local cache.
4829         (resolve_mixers): Likewise.
4831 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4832             Paolo Carlini  <paolo.carlini@oracle.com>
4834         PR c++/58458
4835         * doc/implement-cxx.texi: Fix references to the C++ standards.
4837 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4839         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4840         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4841         simd_lane_access drs, update also DR_ALIGNED_TO.
4843 2013-09-18  Marek Polacek  <polacek@redhat.com>
4845         PR sanitizer/58411
4846         * doc/extend.texi: Document no_sanitize_undefined attribute.
4847         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4848         no_sanitize_undefined attribute.
4850 2013-09-18  Nick Clifton  <nickc@redhat.com>
4852         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4853         (ASM_DECLARE_FUNCTION_NAME): Define.
4855 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4857         * compare-elim.c (find_comparison_dom_walker): New class
4858         (find_comparisons_in_bb): Rename to
4859         find_comparison_dom_walker::before_dom_children
4860         (find_comparisons): Adjust
4861         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4862         adjust.
4863         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4864         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4865         (init_walk_dominator_tree): Remove declaration.
4866         (fini_walk_dominator_tree): Remove declaration.
4867         * fwprop.c (single_def_use_dom_walker): New class
4868         (single_def_use_enter_block): Convert to
4869         single_def_use_dom_walker::before_dom_children.
4870         (single_def_use_leave_block): Convert to
4871         single_def_use_dom_walker::after_dom_children.
4872         (build_single_def_use_links): Adjust.
4873         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4874         class.
4875         (find_candidates_in_block): Convert to
4876         find_candidates_dom_walker::before_dom_children.
4877         (execute_strength_reduction): Adjust.
4878         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4879         (sese_dom_walker): New class.
4880         (sese_dom_walker::sese_dom_walker): New constructor.
4881         (sese_dom_walker::~sese_dom_walker): New destructor.
4882         (build_sese_conditions_before): Convert to
4883         sese_dom_walker::before_dom_children.
4884         (build_sese_conditions_after): Convert to
4885         sese_dom_walker::after_dom_children.
4886         (build_poly_scop): Adjust
4887         * tree-into-ssa.c (rewrite_dom_walker): New class
4888         (rewrite_enter_block): Convert to
4889         rewrite_dom_walker::before_dom_children.
4890         (rewrite_leave_block): Convert to
4891         rewrite_dom_walker::after_dom_children.
4892         (rewrite_update_dom_walker): New class.
4893         (rewrite_update_enter_block): Convert to
4894         rewrite_update_dom_walker::before_dom_children.
4895         (rewrite_update_leave_block): Convert to
4896         rewrite_update_dom_walker::after_dom_children.
4897         (rewrite_blocks, rewrite_into_ssa): Adjust.
4898         (mark_def_dom_walker): New class.
4899         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4900         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4901         (mark_def_sites_blocks): Convert to
4902         mark_def_dom_walker::before_dom_children.
4903         (mark_def_site_blocks): Remove.
4904         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4905         (tree_ssa_dominator_optimize): Adjust.
4906         (dom_thread_across_edge): Convert to method
4907         dom_opt_dom_walker::thread_across_edge.
4908         (dom_opt_enter_block): Convert to member function
4909         dom_opt_dom_walker::before_dom_children.
4910         (dom_opt_leave_block): Convert to member function
4911         dom_opt_dom_walker::after_dom_children.
4912         * tree-ssa-dse.c (dse_dom_walker): New class.
4913         (dse_enter_block): Convert to member function
4914         dse_dom_walker::before_dom_children.
4915         (tree_ssa_dse): Adjust.
4916         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4917         (determine_invariantness_stmt): Convert to method
4918         invariantness_dom_walker::before_dom_children.
4919         (determine_invariantness): Remove
4920         (move_computations_dom_walker): New class.
4921         (move_computations_stmt): Convert to method
4922         move_computations_dom_walker::before_dom_children.
4923         (move_computations, tree_ssa_lim): Adjust.
4924         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4925         (nt_init_block): Make method
4926         notrappping_dom_walker::before_dom_children.
4927         (nt_fini_block): Make
4928         method nontrapping_dom_walker::after_dom_children.
4929         (get_non_trapping): Adjust.
4930         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4931         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4932         (eliminate_leave_block): Make method.
4933         eliminate_dom_walker::after_dom_children.
4934         (eliminate): Adjust
4935         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4936         (strlen_enter_block): Make method
4937         strlen_dom_walker::before_dom_children.
4938         (strlen_leave_block): Make
4939         method strlen_dom_walker::after_dom_children.
4940         (tree_ssa_strlen): Adjust.
4941         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4942         (tree_ssa_uncprop): Adjust.
4943         (uncprop_leave_block): Make method
4944         uncprop_dom_walker::after_dom_children.
4945         (uncprop_leave_block): Make method
4946         uncprop_dom_walker::before_dom_children.
4948 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4950         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4951         branch in same basic block.  Check both src and dest of the move insn.
4953 2013-09-17  Nick Clifton  <nickc@redhat.com>
4955         * config/rl78/rl78-real.md (bf): New pattern.
4956         (bt): New pattern.
4957         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4958         (rl78_print_operand): Do not put a # before a %B.
4959         * config/rl78/rl78.opt: Tweak doc strings.
4961 2013-09-17  DJ Delorie  <dj@redhat.com>
4963         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4964         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4965         (rl78_option_override): Likewise, if -mallregs.
4966         (is_virtual_register): Likewise.
4967         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4968         (REGNO_OK_FOR_BASE_P): Likewise.
4969         * config/rl78/rl78.opt (-mallregs): New.
4971 2013-09-17  Nick Clifton  <nickc@redhat.com>
4973         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4974         For interrupt functions: save all call clobbered registers if the
4975         interrupt handler is not a leaf function.
4976         (rl78_expand_prologue): Always recompute the frame information.
4977         For interrupt functions: only select bank 0 if one of the bank 0
4978         registers is going to be psuhed.
4980 2013-09-17  DJ Delorie  <dj@redhat.com>
4982         * config/rl78/constraints.md: For each W* constraint, rename to C*
4983         and create a W* constraint that checks for an optional ES: prefix
4984         pattern also.
4985         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4986         (es_addr): New.  Used to wrap far addresses.
4987         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4988         (rl78_es_base): New.
4989         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4990         wrapped far addresses.
4991         (rl78_print_operand_1): Unwrap far addresses before processing.
4992         (rl78_lo16): Wrap far addresses in unspecs.
4993         (rl78_es_addr): New.
4994         (rl78_es_base): New.
4995         (insn_ok_now): Check for not-yet-wrapped far addresses.
4996         (transcode_memory_rtx): Properly re-wrap far addresses.
4998 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5000         * config/sparc/t-rtems: Add leon3 multilibs.
5002 2013-09-17  Cong Hou  <congh@google.com>
5004         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
5005         when checking the dot production pattern. The type of rhs operand
5006         of multiply is now checked correctly.
5008 2013-09-17  Jeff Law  <law@redhat.com>
5010         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5011         edge implied equivalences into successor phis.
5012         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
5013         here from tree-ssa-threadedge.c.
5014         (mark_threaded_blocks): When threading through a joiner, if both
5015         successors of the joiner's clone reach the same block, verify the
5016         PHI arguments are equal.  If not, cancel the jump threading request.
5017         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
5018         tree-ssa-threadupdate.c
5019         (thread_across_edge): Don't check PHI argument equality when
5020         threading through joiner block here.
5022 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5024         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5025         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5026         (warn_uninit, warn_uninitialized_vars,
5027         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5028         Move to tree-ssa-uninit.c.
5029         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5030         (has_undefined_value_p): New.  Pass dependant parts of
5031         ssa_undefined_value_p.
5032         (uninit_undefined_value_p): Use has_undefined_value_p.
5033         (warn_uninit, warn_uninitialized_vars,
5034         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5035         Move from tree-ssa.c.
5036         * tree-ssa.h: Adjust prototypes.
5038 2013-09-17  Jan Hubicka  <jh@suse.cz>
5040         PR middle-end/58332
5041         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5042         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5043         FUNCTION_NOT_OPTIMIZED.
5044         * ipa-inline-analysis.c (compute_inline_parameters): Function
5045         not optimized is not inlinable unless it is alwaysinline.
5046         (inline_analyze_function): Force calls in not optimized
5047         function not inlinable.
5049 2013-09-17  Jan Hubicka  <jh@suse.cz>
5051         PR middle-end/58329
5052         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5053         to return NULL.
5054         * ipa.c (function_and_variable_visibility): Likewise.
5055         * ipa-profile.c (ipa_profile): Likewise.
5057 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5059         PR ipa/58398
5060         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5061         attribute, and don't inline the resolver in this case.
5063 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5065         * coverage.c (get_coverage_counts): Add missing newline.
5067 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5069         PR tree-optimization/58088
5070         * fold-const.c (mask_with_trailing_zeros): New function.
5071         (fold_binary_loc): Make sure we don't recurse infinitely
5072         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5073         Use mask_with_trailing_zeros where appropriate.
5075 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5077         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5078         of 'prev' var to get better distance estimation.
5080 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5082         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5083         * tree-inline.c (is_parameter_of): New predicate.
5084         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5085         a parameter has been remapped.
5086         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5087         (optimize_inline_calls): Initialize transform_parameter.
5088         (copy_gimple_seq_and_replace_locals): Likewise.
5089         (tree_function_versioning): Likewise.
5090         (maybe_inline_call_in_expr): Likewise.
5092 2013-09-17  Nick Clifton  <nickc@redhat.com>
5094         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5095         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5096         interrupt handlers.
5097         (is_attr_func): New function.
5098         (msp430_is_interrupt_func): New function.
5099         (is_naked_func): New function.
5100         (is_reentrant_func): New function.
5101         (is_critical_func): New function.
5102         (msp430_start_function): Add annotations for function attributes.
5103         (msp430_attr): New function.
5104         (msp430_attribute_table): New.
5105         (msp430_function_section): New function.
5106         (TARGET_ASM_FUNCTION_SECTION): Define.
5107         (msp430_builtin): New enum.
5108         (msp430_init_builtins): New function.
5109         (msp430_builtin_devl): New function.
5110         (msp430_expand_builtin): New function.
5111         (TARGET_INIT_BUILTINS): Define.
5112         (TARGET_EXPAND_BUILTINS): Define.
5113         (TARGET_BUILTIN_DECL): Define.
5114         (msp430_expand_prologue): Add support for naked, interrupt,
5115         critical and reentrant functions.
5116         (msp430_expand_epilogue): Likewise.
5117         (msp430_print_operand): Handle 'O' character.
5118         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5119         NO_TRAMPOLINES.
5120         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5121         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5122         (pushm): Use a 'n' rather than an 'i' constraint.
5123         (msp_return): Add generation of the interrupt return instruction.
5124         (disable_interrupts): New pattern.
5125         (enable_interrupts): New pattern.
5126         (push_intr_state): New pattern.
5127         (pop_intr_state): New pattern.
5128         (bic_SR): New pattern.
5129         (bis_SR): New pattern.
5130         * doc/extend.texi: Document MSP430 function attributes and builtin
5131         functions.
5133 2013-09-17  Richard Biener  <rguenther@suse.de>
5135         PR tree-optimization/58432
5136         * tree-loop-distribution.c (tree_loop_distribution): Also
5137         scan PHIs for outside loop uses and seed a partition from them.
5139 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5141         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5142         (restructure_reference): Call backtrace_base_for_ref.
5144 2013-09-17  Alan Modra  <amodra@gmail.com>
5146         PR target/57589
5147         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5148         patch.
5150 2013-09-16  DJ Delorie  <dj@redhat.com>
5152         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5153         vregs location for RL78/G10.
5154         (rl78_expand_prologue): Avoid SEL on G10.
5155         (rl78_expand_epilogue): Likewise.
5156         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5157         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5158         __RL78_G10__ when appropriate.
5159         (ASM_SPEC): Pass -mg10 along to the assembler.
5160         * config/rl78/rl78.md (sel_rb): Disable for G10.
5161         * config/rl78/rl78.opt: Add -mg10 option.
5162         * config/rl78/t-rl78: Add -mg10 multilib.
5164 2013-09-16  Xinliang David Li  <davidxl@google.com>
5166         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5167         * omp-low.c (omp_max_vf): Ditto.
5168         (expand_omp_simd): Ditto.
5169         * tree-vectorizer.c (vectorize_loops): Ditto.
5170         (gate_vect_slp): Ditto.
5171         (gate_increase_alignment): Ditto.
5172         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5173         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5174         (gate_tree_vectorize): Name change.
5175         (tree_vectorize): Ditto.
5176         (pass_vectorize::gate): Call new function.
5177         (pass_vectorize::execute): Ditto.
5178         * opts.c: O3 default setting change.
5179         (finish_options): Check new flag.
5180         * doc/invoke.texi: Document new flags.
5181         * common.opt: New flags.
5183 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5185         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5186         (Condition Code): Don't reference it.
5187         * doc/tm.texi: Regenerate.
5189 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5191         PR middle-end/58418
5192         * lra-constraints.c (undo_optional_reloads): Consider all optional
5193         reload even if it did not get a hard reg.
5195 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5197         * dumpfile.c (dump_loc): Remove newline emission.
5198         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5199         emission to dump_printf_loc calls where missing.
5200         (vect_mark_for_runtime_alias_test): Ditto.
5201         (vect_analyze_data_ref_dependence): Ditto.
5202         (vect_analyze_data_ref_dependences): Ditto.
5203         (vect_slp_analyze_data_ref_dependence): Ditto.
5204         (vect_slp_analyze_data_ref_dependences): Ditto.
5205         (vect_compute_data_ref_alignment): Ditto.
5206         (vect_update_misalignment_for_peel): Ditto.
5207         (vect_verify_datarefs_alignment): Ditto.
5208         (vector_alignment_reachable_p): Ditto.
5209         (vect_get_data_access_cost): Ditto.
5210         (vect_enhance_data_refs_alignment): Ditto.
5211         (vect_find_same_alignment_drs): Ditto.
5212         (vect_analyze_data_refs_alignment): Ditto.
5213         (vect_analyze_group_access): Ditto.
5214         (vect_analyze_data_ref_access): Ditto.
5215         (vect_analyze_data_ref_accesses): Ditto.
5216         (vect_prune_runtime_alias_test_list): Ditto.
5217         (vect_analyze_data_refs): Ditto.
5218         (vect_create_addr_base_for_vector_ref): Ditto.
5219         (vect_create_data_ref_ptr): Ditto.
5220         (vect_grouped_store_supported): Ditto.
5221         (vect_grouped_load_supported): Ditto.
5222         * value-prof.c (check_counter): Ditto.
5223         (check_ic_target): Ditto.
5224         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5225         (vect_recog_widen_mult_pattern): Ditto.
5226         (vect_recog_widen_sum_pattern): Ditto.
5227         (vect_recog_over_widening_pattern): Ditto.
5228         (vect_recog_widen_shift_pattern): Ditto.
5229         (vect_recog_rotate_pattern): Ditto.
5230         (vect_recog_vector_vector_shift_pattern): Ditto.
5231         (vect_recog_divmod_pattern): Ditto.
5232         (vect_recog_mixed_size_cond_pattern): Ditto.
5233         (vect_recog_bool_pattern): Ditto.
5234         (vect_pattern_recog_1): Ditto.
5235         (vect_pattern_recog): Ditto.
5236         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5237         (vect_is_simple_iv_evolution): Ditto.
5238         (vect_analyze_scalar_cycles_1): Ditto.
5239         (vect_get_loop_niters): Ditto.
5240         (vect_analyze_loop_1): Ditto.
5241         (vect_analyze_loop_form): Ditto.
5242         (vect_analyze_loop_operations): Ditto.
5243         (vect_analyze_loop_2): Ditto.
5244         (vect_analyze_loop): Ditto.
5245         (report_vect_op): Ditto.
5246         (vect_is_slp_reduction): Ditto.
5247         (vect_is_simple_reduction_1): Ditto.
5248         (vect_get_known_peeling_cost): Ditto.
5249         (vect_estimate_min_profitable_iters): Ditto.
5250         (vect_model_reduction_cost): Ditto.
5251         (vect_model_induction_cost): Ditto.
5252         (get_initial_def_for_induction): Ditto.
5253         (vect_create_epilog_for_reduction): Ditto.
5254         (vectorizable_reduction): Ditto.
5255         (vectorizable_induction): Ditto.
5256         (vectorizable_live_operation): Ditto.
5257         (vect_loop_kill_debug_uses): Ditto.
5258         (vect_transform_loop): Ditto.
5259         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5260         (vect_stmt_relevant_p): Ditto.
5261         (process_use): Ditto.
5262         (vect_mark_stmts_to_be_vectorized): Ditto.
5263         (vect_model_simple_cost): Ditto.
5264         (vect_model_promotion_demotion_cost): Ditto.
5265         (vect_model_store_cost): Ditto.
5266         (vect_get_store_cost): Ditto.
5267         (vect_model_load_cost): Ditto.
5268         (vect_get_load_cost): Ditto.
5269         (vect_init_vector_1): Ditto.
5270         (vect_get_vec_def_for_operand): Ditto.
5271         (vect_finish_stmt_generation): Ditto.
5272         (vectorizable_call): Ditto.
5273         (vectorizable_conversion): Ditto.
5274         (vectorizable_assignment): Ditto.
5275         (vectorizable_shift): Ditto.
5276         (vectorizable_operation): Ditto.
5277         (vectorizable_store): Ditto.
5278         (vectorizable_load): Ditto.
5279         (vectorizable_condition): Ditto.
5280         (vect_analyze_stmt): Ditto.
5281         (vect_transform_stmt): Ditto.
5282         (vect_is_simple_use): Ditto.
5283         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5284         (vect_can_advance_ivs_p): Ditto.
5285         (vect_update_ivs_after_vectorizer): Ditto.
5286         (vect_do_peeling_for_loop_bound): Ditto.
5287         (vect_gen_niters_for_prolog_loop): Ditto.
5288         (vect_update_inits_of_drs): Ditto.
5289         (vect_create_cond_for_alias_checks): Ditto.
5290         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5291         (vect_build_slp_tree_1): Ditto.
5292         (vect_supported_load_permutation_p): Ditto.
5293         (vect_analyze_slp_instance): Ditto.
5294         (vect_analyze_slp): Ditto.
5295         (vect_make_slp_decision): Ditto.
5296         (vect_detect_hybrid_slp): Ditto.
5297         (vect_bb_vectorization_profitable_p): Ditto.
5298         (vect_slp_analyze_bb_1): Ditto.
5299         (vect_update_slp_costs_according_to_vf): Ditto.
5300         (vect_get_mask_element): Ditto.
5301         (vect_transform_slp_perm_load): Ditto.
5302         (vect_schedule_slp_instance): Ditto.
5303         (vect_schedule_slp): Ditto.
5304         (vect_slp_transform_bb): Ditto.
5305         * profile.c (read_profile_edge_counts): Ditto.
5306         (compute_branch_probabilities): Ditto.
5307         * coverage.c (get_coverage_counts): Ditto.
5309 2013-09-16  Diego Novillo  <dnovillo@google.com>
5311         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5313 2013-09-16  Jan Hubicka  <jh@suse.cz>
5315         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5316         abstract functions; for static functions check the presence of body.
5318 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5320         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5321         * config/aarch64/aarch64-simd.md
5322         (aarch64_mla_elt<mode>): New.
5323         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5324         (aarch64_mls_elt<mode>): Likewise.
5325         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5326         (aarch64_fma4_elt<mode>): Likewise.
5327         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5328         (aarch64_fma4_elt_to_128v2df): Likewise.
5329         (aarch64_fma4_elt_to_64df): Likewise.
5330         (fnma<mode>4): Likewise.
5331         (aarch64_fnma4_elt<mode>): Likewise.
5332         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5333         (aarch64_fnma4_elt_to_128v2df): Likewise.
5334         (aarch64_fnma4_elt_to_64df): Likewise.
5335         * config/aarch64/iterators.md (VDQSF): New.
5336         * config/aarch64/arm_neon.h
5337         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5338         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5340 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5342         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5343         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5344         (aarch64_mul3_elt_to_128df): Likewise.
5345         (aarch64_mul3_elt_to_64v2df): Likewise.
5346         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5347         (VMUL): New.
5348         (VMUL_CHANGE_NLANES) Likewise.
5349         (h_con): Likewise.
5350         (f): Likewise.
5351         * config/aarch64/arm_neon.h
5352         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5354 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5356         * config/aarch64/arm_neon.h
5357         (vcvtx_high_f32_f64): Fix parameters.
5359 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5360             Uros Bizjak  <ubizjak@gmail.com>
5362         * config/alpha.c: Include tree-ssanames.h.
5364 2013-09-16  Richard Biener  <rguenther@suse.de>
5366         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5367         (dot_rdg_1): Handle control_dd.
5368         (create_edge_for_control_dependence): New function.
5369         (create_rdg_edges): Add control dependences if asked for.
5370         (build_rdg): Likewise.
5371         (generate_loops_for_partition): If there are not necessary
5372         control stmts remove all their dependencies.
5373         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5374         (distribute_loop): Pass on control dependences.
5375         (tree_loop_distribution): Compute control dependences and remove
5376         restriction on number of loop nodes.
5378 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5380         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5381         for internal calls.
5383 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5385         * cse.c (try_const_anchors): Punt on CC modes.
5387 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5389         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5391 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5393         PR target/58382
5394         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5395         calls to word_mode.
5397 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5399         PR target/48094
5400         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5401         seen.
5402         (darwin_objc1_section): Likewise.
5403         (darwin_file_end): Emit Image Info section when required.
5405 2013-09-14  Jan Hubicka  <jh@suse.cz>
5407         * tree-into-ssa.c (gate_into_ssa): New.
5408         (pass_data_build_ssa): Use it.
5409         * cgraph.h (expand_thunk): Update prototype.
5410         * cgraphunit.c (analyze_function): Expand thunks early.
5411         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5412         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5413         set lowered flag; do not add new function.
5414         (assemble_thunks_and_aliases): Update.
5415         * tree-ssa.c (gate_init_datastructures): New gate.
5416         (pass_data_init_datastructures): Use it.
5418 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5420         PR target/58269
5421         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5422         xmm register set described in the psABI.
5424 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5426         * dwarf2out.c (should_emit_struct_debug): Add check
5427         for type_decl variable is not NULL.
5429 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5431         * config.gcc: Use new winnt-c.c target hooks
5432         * config/t-winnt: New file
5433         * config/winnt-c.c: New file
5434         * doc/tm.texi.in: Document new hook
5435         * doc/tm.texi: Regenerated
5437 2013-09-13  Jan Hubicka  <jh@suse.cz>
5439         PR middle-end/58094
5440         * ipa-inline.c (check_callers): New function.
5441         (check_caller_edge): Remove.
5442         (want_inline_function_to_all_callers_p): Also permit alises that are
5443         called dirrectly.
5444         (inline_to_all_callers): Terminate the walk when devirtualization
5445         introduce new calls.
5447 2013-09-13  Jan Hubicka  <jh@suse.cz>
5449         * ipa-inline-analysis.c (struct growth_data): Add node.
5450         (do_estimate_growth_1): Fix detection of recursion.
5452 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5454         PR tree-optimization/58392
5455         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5456         to avoid shadowing of outer loop variable.  If
5457         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5458         replace_by_duplicate_decl simduid of loops that have it set and
5459         set dest_cfun->has_simduid_loops and/or
5460         dest_cfun->has_force_vect_loops.
5461         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5462         instead of maybe_lookup_decl.
5463         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5464         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5465         call.  Copy over force_vect and copy and remap simduid.  Set
5466         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5467         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5468         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5469         caller.  Don't set cfun->has_simduid_loops and/or
5470         cfun->has_force_vect_loops here.
5471         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5472         caller.
5473         (expand_call_inline, tree_function_versioning): Adjust copy_body
5474         callers.
5476 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5478         PR bootstrap/58388
5479         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5480         the assert if the edge was a speculative one.
5482 2013-09-13  Richard Biener  <rguenther@suse.de>
5484         * tree-data-ref.h (known_dependences_p): Move here ...
5485         * tree-loop-distribution.c (known_dependences_p): ... from here.
5486         (dump_rdg_component, debug_rdg_component): Remove.
5487         (dump_rdg): Adjust.
5488         (generate_loops_for_partition): Use gimple_uid instead of
5489         relying on matching stmt visit order.
5490         (rdg_build_partitions): Take starting stmt vector.
5491         (ldist_gen): Merge into ...
5492         (distribute_loop): ... this function.  Do not compute starting
5493         vertices vector.
5494         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5496 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5498         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5499         Set type attribute correctly. Set predicable_short_it attribute.
5500         (cmpsi_shiftsi): Remove %? from output template.
5502 2013-09-13  Richard Biener  <rguenther@suse.de>
5504         * tree-loop-distribution.c (struct rdg_component,
5505         rdg_defs_used_in_other_loops_p, free_rdg_components,
5506         rdg_build_components): Remove.
5507         (stmts_from_loop): Do not record virtual PHIs.
5508         (generate_loops_for_partition): Skip virtual PHIs.
5509         (build_rdg_partition_for_component): Rename to ...
5510         (build_rdg_partition_for_vertex): ... this and adjust.
5511         (rdg_build_partitions): Take a vector of starting vertices
5512         instead of components.  Remove unnecessary leftover handling.
5513         (ldist_gen): Do not build components or record other stores.
5514         (distribute_loop): Do not distribute loops containing stmts
5515         with side-effects.
5517 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5519         PR target/58314
5520         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5522 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5524         * config.gcc: Separate cases for mingw and cygwin targets,
5525         and add 64-bit cygwin target case.
5527         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5528         dll-export inline-functions.
5529         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5532 2013-09-13  Jeff Law  <law@redhat.com>
5534         PR middle-end/58387
5535         Revert:
5536         2013-09-06  Jeff Law  <law@redhat.com>
5538         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5539         edge implied equivalences into successor phis.
5541 2013-09-12  DJ Delorie  <dj@redhat.com>
5543         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5544         separators.
5546 2013-09-12  Brooks Moses  <bmoses@google.com>
5548         PR driver/42955
5549         * Makefile.in: Do not install driver binaries in $(target)/bin.
5551 2013-09-12  DJ Delorie  <dj@redhat.com>
5553         * config/rl78/rl78.opt (mrelax): New.
5554         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5555         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5557         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5558         between SP and FP.
5559         (rl78_expand_epilogue): Likewise.
5561 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5563         PR middle-end/58335
5564         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5565         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5566         pointer elimination with using remove_reg_equal_offset_note.
5568 2013-09-12  DJ Delorie  <dj@redhat.com>
5570         * config/msp430/: New port.
5571         * config.gcc (msp430): Added.
5572         * doc/invoke.texi: Document MSP430 options.
5573         * doc/install.texi: Document msp430-elf
5574         * doc/md.texi: Document msp430-elf
5575         * doc/contrib.texi: Document msp430-elf
5577         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5578         PSImode.
5580 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5582         PR ipa/58389
5583         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5584         reference descriptor is NULL.
5585         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5586         edge to NULL.
5588 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5590         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5591         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5592         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5593         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5594         make_temp_ssa_name): move to tree-ssanames.h
5595         * tree-ssa-alias.h: Move prototype.
5596         * tree-ssa.h: Include tree-ssanames.h.
5597         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5598         * tree-ssanames.h: New.  Move items from tree-flow*.h
5599         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5601 2013-09-12  Richard Biener  <rguenther@suse.de>
5603         PR tree-optimization/58404
5604         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5605         propagate non-invariant addresses into dereferences wrapped
5606         in component references.
5608 2013-09-12  Richard Biener  <rguenther@suse.de>
5610         PR tree-optimization/58402
5611         * passes.def: Move pass_late_warn_uninitialized later.
5613 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5615         * tree-ssa.h: New.  Move content from tree-flow.h and
5616         tree-flow-inline.h.
5617         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5618         Move prototypes belonging to tree-ssa.c.
5619         * tree-flow-inline.h (redirect_edge_var_map_def,
5620         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5621         tree-ssa.h.
5622         * gimple.h: Adjust prototypes.
5623         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5624         to...
5625         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5626         * tree.h: Move prototype to tree-ssa.h.
5627         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5628         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5629         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5630         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5631         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5632         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5633         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5634         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5635         graphite-blocking.c, graphite-clast-to-gimple.c,
5636         graphite-dependences.c, graphite-interchange.c,
5637         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5638         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5639         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5640         ipa-reference.c, ipa-split.c, ipa-utils.c,
5641         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5642         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5643         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5644         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5645         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5646         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5647         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5648         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5649         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5650         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5651         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5652         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5653         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5654         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5655         value-prof.c, var-tracking.c,
5656         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5658 2013-09-12  Richard Biener  <rguenther@suse.de>
5660         PR tree-optimization/58396
5661         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5662         (build_rdg): Take a loop-nest parameter, fix memleaks.
5663         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5665 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5667         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5668         for SLM.
5670 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5672         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5673         The FMA instruction names should have a 'v' prefix.
5675 2013-09-12  Richard Biener  <rguenther@suse.de>
5677         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5678         (dot_rdg): Use popen instead of system in optional code.
5679         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5680         (already_processed_vertex_p): Adjust.
5681         (has_anti_or_output_dependence, predecessor_has_mem_write,
5682         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5683         rdg_flag_uses): Remove.
5684         (rdg_flag_vertex): Simplify.
5685         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5686         remove recursion.
5687         (build_rdg_partition_for_component): Process the first vertex
5688         of a component only.
5689         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5691 2013-09-12  Alan Modra  <amodra@gmail.com>
5693         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5695 2013-09-11  DJ Delorie  <dj@redhat.com>
5696             Nick Clifton  <nickc@redhat.com>
5698         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5699         (rl78_stack_based_mem): New.
5700         * config/rl78/constraints.md (Iv08): New.
5701         (Iv16): New.
5702         (Iv24): New.
5703         (Is09): New.
5704         (Is17): New.
5705         (Is25): New.
5706         (ISsi): New.
5707         (IShi): New.
5708         (ISqi): New.
5709         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5710         (movhi): Likewise.
5711         (movsi): Change from expand to insn-and-split.
5712         (ashrsi3): Clobber AX.
5713         (lshrsi3): New.
5714         (ashlsi3): New.
5715         (cbranchsi4): New.
5716         * config/rl78/rl78.md (CC_REG): Fix.
5717         (addsi3): Allow memory and immediate operands.
5718         (addsi3_internal): Split into...
5719         (addsi3_internal_virt): ...new, and ...
5720         (addsi3_internal_real): ...new.
5721         (subsi): New.
5722         (subsi3_internal_virt): New.
5723         (subsi3_internal_real): New.
5724         (mulsi3): Add memory operand.
5725         (mulsi3_rl78): Likewise.
5726         (mulsi3_g13): Likewise.
5727         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5728         (cbranchqi4_real): Add more constraint options.
5729         (cbranchhi4_real): Expand pattern.
5730         (cbranchhi4_real_signed): New.
5731         (cbranchhi4_real_inverted): New.
5732         (cbranchsi4_real_lt): New.
5733         (cbranchsi4_real_ge): New.
5734         (cbranchsi4_real_signed): New.
5735         (cbranchsi4_real): New.
5736         (peephole2): New.
5737         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5738         constant shifts.
5739         (lshrsi3_virt): Likewise.
5740         (ashlsi3_virt): Likewise.
5741         (cbranchqi4_virt_signed): New.
5742         (cbranchhi4_virt_signed): New.
5743         (cbranchsi4_virt): New.
5744         * config/rl78/rl78.c: Whitespace fixes throughout.
5745         (move_elim_pass): New.
5746         (pass_data_rl78_move_elim): New.
5747         (pass_rl78_move_elim): New.
5748         (make_pass_rl78_move_elim): New.
5749         (rl78_devirt_info): Run devirt earlier.
5750         (rl78_move_elim_info): New.
5751         (rl78_asm_file_start): Register it.
5752         (rl78_split_movsi): New.
5753         (rl78_as_legitimate_address): Allow virtual base registers when
5754         appropriate.
5755         (rl78_addr_space_convert): Remove spurious debug stuff.
5756         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5757         (rl78_print_operand): More cases for not printing '#'.
5758         (rl78_expand_compare): Remove most of the logic.
5759         (content_memory): New.
5760         (clear_content_memory): New.
5761         (get_content_index): New.
5762         (get_content_name): New.
5763         (display_content_memory): New.
5764         (update_content): New.
5765         (record_content): New.
5766         (already_contains): New.
5767         (insn_ok_now): Re-recog insns with virtual registers.
5768         (add_postponed_content_update): New.
5769         (process_postponed_content_update): New.
5770         (gen_and_emit_move): New.
5771         (transcode_memory_rtx): Record new location content.
5772         Use gen_and_emit_move.
5773         (force_into_acc): New.
5774         (move_to_acc): Use gen_and_emit_move.
5775         (move_from_acc): Likewise.
5776         (move_acc_to_reg): Likewise.
5777         (move_to_x): Likewise.
5778         (move_to_hl): Likewise.
5779         (move_to_de): Likewise.
5780         (rl78_alloc_physical_registers_op1): Record location content.
5781         (has_constraint): New.
5782         (rl78_alloc_physical_registers_op2): Record location content.
5783         Optimize use of HL.
5784         (rl78_alloc_physical_registers_ro1): Likewise.
5785         (rl78_alloc_physical_registers_cmp): Likewise.
5786         (rl78_alloc_physical_registers_umul): Likewise.
5787         (rl78_alloc_address_registers_macax): New.
5788         (rl78_alloc_physical_registers): Initialize and set location
5789         content memory as needed.
5790         (rl78_reorg): Make sure split2 is called.
5791         (rl78_rtx_costs): New.
5793 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5795         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5796         for (not (neg ...)) and (neg (not ...)) cases.
5798 2013-09-11  Richard Biener  <rguenther@suse.de>
5800         PR middle-end/58377
5801         * passes.def: Split critical edges before late uninit warning passes.
5802         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5804 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5806         PR tree-optimization/58385
5807         * fold-const.c (build_range_check): If both low and high are NULL,
5808         use omit_one_operand_loc to preserve exp side-effects.
5810 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5812         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5814 2013-09-11  Richard Biener  <rguenther@suse.de>
5816         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5817         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5818         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5819         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5820         stmts_from_loop, known_dependences_p, build_empty_rdg,
5821         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5822         * tree-loop-distribution.c: ... here.
5823         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5824         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5825         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5826         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5827         * tree-loop-distribution.c: ... here.
5828         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5829         (struct partition_s): Add loops member.
5830         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5831         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5832         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5834 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5835             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5836             Sergey Lega  <sergey.s.lega@intel.com>
5837             Anna Tikhonova  <anna.tikhonova@intel.com>
5838             Ilya Tocar  <ilya.tocar@intel.com>
5839             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5840             Ilya Verbin  <ilya.verbin@intel.com>
5841             Kirill Yukhin  <kirill.yukhin@intel.com>
5842             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5844         * config/i386/constraints.md (k): New.
5845         (Yk): Ditto.
5846         * config/i386/i386.c (const regclass_map): Add new mask registers.
5847         (dbx_register_map): Ditto.
5848         (dbx64_register_map): Ditto.
5849         (svr4_dbx_register_map): Ditto.
5850         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5851         disabled.
5852         (ix86_preferred_reload_class): Disable constants for mask registers.
5853         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5854         (ix86_hard_regno_mode_ok): Support new mask registers.
5855         (x86_order_regs_for_local_alloc): Ditto.
5856         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5857         (FIXED_REGISTERS): Add new mask registers.
5858         (CALL_USED_REGISTERS): Ditto.
5859         (REG_ALLOC_ORDER): Ditto.
5860         (VALID_MASK_REG_MODE): New.
5861         (FIRST_MASK_REG): Ditto.
5862         (LAST_MASK_REG): Ditto.
5863         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5864         (MAYBE_MASK_CLASS_P): New.
5865         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5866         (REG_CLASS_CONTENTS): Ditto.
5867         (MASK_REGNO_P): New.
5868         (ANY_MASK_REG_P): Ditto.
5869         (HI_REGISTER_NAMES): Add new mask registers.
5870         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5871         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5872         mask registers.
5873         (attribute "type"): Add mskmov, msklog.
5874         (attribute "length_immediate"): Support them.
5875         (attribute "memory"): Ditto.
5876         (attribute "prefix_0f"): Ditto.
5877         (*movhi_internal): Support new mask registers.
5878         (*movqi_internal): Ditto.
5879         (define_split): Split out clobber pattern is a logic
5880         insn on mask registers.
5881         (*k<logic><mode>): New.
5882         (*andhi_1): Extend to support mask regs.
5883         (*andqi_1): Extend to support mask regs.
5884         (kandn<mode>): New.
5885         (define_split): Split and-not to and and not if operands
5886         are not mask regs.
5887         (*<code><mode>_1): Separate HI mode to new pattern...
5888         (*<code>hi_1): This.
5889         (*<code>qi_1): Extend to support mask regs.
5890         (kxnor<mode>): New.
5891         (kortestzhi): Ditto.
5892         (kortestchi): Ditto.
5893         (kunpckhi): Ditto.
5894         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5895         (*one_cmplhi2_1): ...Here, now with mask registers support.
5896         (*one_cmplqi2_1): Support new mask registers.
5897         (HI/QImode arithmetics splitter): Don't split if mask registers
5898         are used.
5899         (HI/QImode not splitter): Ditto.
5900         * config/i386/predicated.md (mask_reg_operand): New.
5901         (general_reg_operand): Ditto.
5903 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5905         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5906         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5908 2013-09-10  Jeff Law  <law@redhat.com>
5910         PR tree-optimization/58380
5911         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5912         of threading through a buried loop header.
5914         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5915         return value for single successor case.
5917 2013-09-10  Jan Hubicka  <jh@suse.cz>
5919         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5921 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5923         PR target/58361
5924         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5925         support conditional execution.
5926         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5928 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5930         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5931         constraint pass.
5932         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5933         Check destination too to check move insn.
5934         (undo_optional_reloads): Add check that the original peudo did not
5935         changed its allocation and the optional reload was inherited on last
5936         inheritance pass.  Break loop after deciding to keep optional reload.
5937         (lra_undo_inheritance): Add check that inherited pseudo still in
5938         memory.
5940 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5942         * config/aarch64/aarch64.md (generic_sched): New.
5943         * config/aarch64/aarch64-generic.md (load): Make conditional
5944         on generic_sched attribute.
5945         (nonload): Likewise.
5947 2013-09-10  Jan Hubicka  <jh@suse.cz>
5949         * lto-cgraph.c: Include ipa-utils.h.
5950         (compute_ltrans_boundary): Also add possible targets into the boundary.
5952 2013-09-10  Jan Hubicka  <jh@suse.cz>
5954         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5955         VAR_DECL of vtable rather than full expression.
5957 2013-09-10  Jan Hubicka  <jh@suse.cz>
5958             Paolo Carlini  <paolo.carlini@oracle.com>
5960         * cgraphunit.c (analyze_functions): Save input_location, set it
5961         to UNKNOWN_LOCATION and restore it at the end.
5963 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5965         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5966         represented by a thunk.
5968 2013-09-10  Jeff Law  <law@redhat.com>
5970         PR tree-optimization/58343
5971         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5972         jump threading requests through loop headers buried in the middle
5973         of a jump threading path.
5975         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5976         in return value/type.
5978 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5980         PR rtl-optimization/58365
5981         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5982         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5983         it differs.
5985 2013-09-10  Richard Biener  <rguenther@suse.de>
5987         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5988         * tree-data-ref.c (create_rdg_vertices): Collect all data
5989         references, signal failure to the caller, use data-ref API.
5990         (build_rdg): Compute data references only once.  Maintain lifetime
5991         of data references and data dependences from within RDG.
5992         (free_rdg): Free dependence relations.
5993         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5994         inventing extra dependences.
5995         (distribute_loop): Update for RDG API changes.
5997 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5999         * doc/invoke.texi (fms-extensions): Document changed
6000         behavior for ms-abi targets.
6001         * config/i386/i386.c (ix86_option_override_internal):
6002         Set default value of option -fms-extension for ms-abi targets.
6004 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6006         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
6008 2013-09-10  Alan Modra  <amodra@gmail.com>
6010         PR target/58330
6011         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
6013 2013-09-10  Alan Modra  <amodra@gmail.com>
6015         * config/rs6000/predicates.md (add_cint_operand): New.
6016         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
6017         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
6018         using add_cint_operand.
6019         (largetoc_high_plus_aix): Likewise.
6021 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6023         PR tree-optimization/58364
6024         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6025         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6026         the current range can't be an unconditional true or false.
6028 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6030         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6032 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6034         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6036 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6038         PR c++/43452
6039         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6041 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6043         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6044         NO_REGS for immediate that can't be moved directly into FP_REGS.
6046 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6048         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6049         comparison with negated operand.
6050         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6051         RTL form.
6053 2013-09-09  Richard Biener  <rguenther@suse.de>
6055         PR middle-end/58326
6056         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6057         of a subloop record all its block as affecting loop-closed SSA form.
6059 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6061         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6062         of an rtx/bitpos pair.
6063         (store_fixed_bit_field): Update accordingly.
6065 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6067         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6068         GEN_INT.
6069         * builtins.c (expand_errno_check): Likewise.
6070         * dwarf2cfi.c (init_return_column_size): Likewise.
6071         * except.c (sjlj_mark_call_sites): Likewise.
6072         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6073         * lra-constraints.c (emit_inc): Likewise.
6074         * ree.c (combine_set_extension): Likewise.
6075         * regmove.c (fixup_match_2): Likewise.
6076         * reload1.c (inc_for_reload): Likewise.
6078 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6080         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6081         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6082         Use gen_int_mode with the mode of the associated simplify_* call.
6083         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6084         * expmed.c (expand_shift_1): Likewise.
6085         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6086         * loop-iv.c (iv_number_of_iterations): Likewise.
6087         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6088         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6090 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6092         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6093         of the associated expand_* call.
6094         (asan_emit_stack_protection): Likewise.
6095         * builtins.c (round_trampoline_addr): Likewise.
6096         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6097         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6098         (emit_store_flag): Likewise.
6099         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6100         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6101         Likewise.
6102         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6103         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6104         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6105         * modulo-sched.c (generate_prolog_epilog): Likewise.
6106         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6107         (expand_ctz, expand_ffs, expand_unop): Likewise.
6109 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6111         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6112         of the associated gen_rtx_* call.
6113         * caller-save.c (init_caller_save): Likewise.
6114         * combine.c (find_split_point, make_extraction): Likewise.
6115         (make_compound_operation): Likewise.
6116         * dwarf2out.c (mem_loc_descriptor): Likewise.
6117         * explow.c (plus_constant, probe_stack_range): Likewise.
6118         * expmed.c (expand_mult_const): Likewise.
6119         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6120         * reload1.c (init_reload): Likewise.
6121         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6122         * var-tracking.c (adjust_mems): Likewise.
6123         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6124         rather than gen_rtx_fmt_ee.
6126 2013-09-09  Jan Hubicka  <jh@suse.cz>
6128         PR middle-end/58294
6129         * value-prof.c (gimple_ic): Copy also abnormal edges.
6131 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6133         * asan.c (asan_shadow_cst): Use gen_int_mode.
6135 2013-09-08  Jan Hubicka  <jh@suse.cz>
6137         * ipa-profile.c: Add toplevel comment.
6138         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6139         (contains_hot_call_p): New function.
6140         (ipa_propagate_frequency): Set frequencies based on counts when
6141         profile is read.
6142         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6143         profile; do not tamper with profile after inlining if it is read.
6145 2013-09-08  Jan Hubicka  <jh@suse.cz>
6147         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6148         speculative edges.
6150 2013-09-08  Jan Hubicka  <jh@suse.cz>
6152         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6153         summary generation.
6155 2013-09-08  Jeff Law  <law@redhat.com>
6157         PR bootstrap/58340
6158         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6159         of 'found'.
6161 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6163         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6165 2013-09-08  Jan Hubicka  <jh@suse.cz>
6167         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6169 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6171         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6172         for non-debug insns.
6173         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6174         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6175         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6176         accordingly.
6178 2013-09-08  Jan Hubicka  <jh@suse.cz>
6180         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6181         targets and devirtualize to BUILT_IN_UNREACHABLE.
6182         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6183         * ipa.c (walk_polymorphic_call_targets): New function.
6184         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6185         functions; use the new timevar.
6186         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6187         was removed from the program.
6188         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6189         not consider it in the walk when its vtable is dead.
6190         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6191         record_binfo.
6192         (devirt_variable_node_removal_hook): New function.
6193         (possible_polymorphic_call_targets): Also register
6194         devirt_variable_node_removal_hook.
6195         (ipa_devirt): Do not do non-speculative devirtualization.
6196         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6198 2013-09-08  Jan Hubicka  <jh@suse.cz>
6200         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6201         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6202         varpool_remove_variable_insertion_hook): Declare.
6203         * varpool.c (varpool_node_hook_list): New structure.
6204         (first_varpool_node_removal_hook,
6205         first_varpool_variable_insertion_hook): New variables.
6206         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6207         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6208         varpool_remove_variable_insertion_hook,
6209         varpool_call_variable_insertion_hooks): New functions.
6210         (varpool_remove_node): Use it.
6212 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6214         PR c++/54941
6215         * diagnostic.c (diagnostic_build_prefix): When s.file is
6216         "<built-in>" don't output line and column numbers.
6218 2013-09-06  Jan Hubicka  <jh@suse.cz>
6220         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6221         * lto-streamer-out.c: Stream thunks, too.
6222         * lto-streamer-in.c (input_function): Pop cfun here
6223         (lto_read_body): Instead of here.
6225 2013-09-06  Caroline Tice  <cmtice@google.com>
6227         * doc/install.texi: Add documentation for the --enable-vtable-verify
6228         and the --disable-libvtv configure options.
6230 2013-09-06  Jeff Law  <law@redhat.com>
6232         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6233         edge implied equivalences into successor phis.
6235 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6237         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6239 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6241         * resource.c (mark_target_live_regs): Compute resources taking
6242         into account if a call is predicated or not.
6244 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6246         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6247         the compiler in the function names.
6249 2013-09-06  Jan Hubicka  <jh@suse.cz>
6251         PR middle-end/58094
6252         * ipa-inline.c (has_caller_p): New function.
6253         (want_inline_function_to_all_callers_p): Use it.
6254         (sum_callers, inline_to_all_callers): Break out from ...
6255         (ipa_inline): ... here.
6257 2013-09-06  Jan Hubicka  <jh@suse.cz>
6259         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6260         only when AVX is enabled.
6262 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6264         * config/aarch64/aarch64.md
6265         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6266         is fpsimd_<load/store>2.
6267         (load_pair<mode>): Likewise.
6268         (store_pair<mode>): Likewise.
6270 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6272         * config/arm/types.md (type): Add "mrs" type.
6273         * config/aarch64/aarch64.md
6274         (aarch64_load_tp_hard): Make type "mrs".
6275         * config/arm/arm.md
6276         (load_tp_hard): Make type "mrs".
6277         * config/arm/cortex-a15.md: Update with new attributes.
6278         * config/arm/cortex-a5.md: Update with new attributes.
6279         * config/arm/cortex-a53.md: Update with new attributes.
6280         * config/arm/cortex-a7.md: Update with new attributes.
6281         * config/arm/cortex-a8.md: Update with new attributes.
6282         * config/arm/cortex-a9.md: Update with new attributes.
6283         * config/arm/cortex-m4.md: Update with new attributes.
6284         * config/arm/cortex-r4.md: Update with new attributes.
6285         * config/arm/fa526.md: Update with new attributes.
6286         * config/arm/fa606te.md: Update with new attributes.
6287         * config/arm/fa626te.md: Update with new attributes.
6288         * config/arm/fa726te.md: Update with new attributes.
6290 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6292         * config/aarch64/aarch64.md
6293         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6294         (*movtf_aarch64): Likewise.
6295         * config/arm/arm.md
6296         (thumb1_movdi_insn): Use "multiple" for type where more than one
6297         instruction is used for a move.
6298         (*arm32_movhf): Likewise.
6299         (*thumb_movdf_insn): Likewise.
6301 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6303         * config/arm/types.md (type): Rename fcpys to fmov.
6304         * config/arm/vfp.md
6305         (*arm_movsi_vfp): Rename type fcpys as fmov.
6306         (*thumb2_movsi_vfp): Likewise
6307         (*movhf_vfp_neon): Likewise
6308         (*movhf_vfp): Likewise
6309         (*movsf_vfp): Likewise
6310         (*thumb2_movsf_vfp): Likewise
6311         (*movsfcc_vfp): Likewise
6312         (*thumb2_movsfcc_vfp): Likewise
6313         * config/aarch64/aarch64-simd.md
6314         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6315         * config/aarch64/aarch64.md
6316         (*movsi_aarch64): Replace type mov_reg with fmovs.
6317         (*movdi_aarch64): Likewise
6318         (*movsf_aarch64): Likewise
6319         (*movdf_aarch64): Likewise
6320         * config/arm/arm.c
6321         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6322         * config/arm/iwmmxt.md
6323         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6324         * config/arm/arm1020e.md: Update with new attributes.
6325         * config/arm/cortex-a15-neon.md: Update with new attributes.
6326         * config/arm/cortex-a5.md: Update with new attributes.
6327         * config/arm/cortex-a53.md: Update with new attributes.
6328         * config/arm/cortex-a7.md: Update with new attributes.
6329         * config/arm/cortex-a8-neon.md: Update with new attributes.
6330         * config/arm/cortex-a9.md: Update with new attributes.
6331         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6332         * config/arm/cortex-r4f.md: Update with new attributes.
6333         * config/arm/marvell-pj4.md: Update with new attributes.
6334         * config/arm/vfp11.md: Update with new attributes.
6336 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6338         * config/aarch64/aarch64.md
6339         (*madd<mode>): Fix type attribute.
6340         (*maddsi_uxtw): Likewise.
6341         (*msub<mode>): Likewise.
6342         (*msubsi_uxtw): Likewise.
6343         (<su_optab>maddsidi4): Likewise.
6344         (<su_optab>msubsidi4): Likewise.
6346 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6348         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6349         * config/arm/arm.md (core_cycles): Remove fdiv.
6350         * config/arm/vfp.md:
6351         (*sqrtsf2_vfp): Update for attribute changes.
6352         (*sqrtdf2_vfp): Likewise.
6353         * config/aarch64/aarch64.md:
6354         (sqrt<mode>2): Update for attribute changes.
6355         * config/arm/arm1020e.md: Update with new attributes.
6356         * config/arm/cortex-a15-neon.md: Update with new attributes.
6357         * config/arm/cortex-a5.md: Update with new attributes.
6358         * config/arm/cortex-a53.md: Update with new attributes.
6359         * config/arm/cortex-a7.md: Update with new attributes.
6360         * config/arm/cortex-a8-neon.md: Update with new attributes.
6361         * config/arm/cortex-a9.md: Update with new attributes.
6362         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6363         * config/arm/cortex-r4f.md: Update with new attributes.
6364         * config/arm/marvell-pj4.md: Update with new attributes.
6365         * config/arm/vfp11.md: Update with new attributes.
6367 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6369         * config/arm/types.md
6370         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6371         * config/aarch64/aarch64.md
6372         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6373         new attributes.
6374         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6375         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6376         (float<GPI:mode><GPF:mode>2): Likewise.
6377         * config/arm/vfp.md
6378         (*truncsisf2_vfp): Update with new attributes.
6379         (*truncsidf2_vfp): Likewise.
6380         (fixuns_truncsfsi2): Likewise.
6381         (fixuns_truncdfsi2): Likewise.
6382         (*floatsisf2_vfp): Likewise.
6383         (*floatsidf2_vfp): Likewise.
6384         (floatunssisf2): Likewise.
6385         (floatunssidf2): Likewise.
6386         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6387         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6388         * config/arm/arm1020e.md: Update with new attributes.
6389         * config/arm/cortex-a15-neon.md: Update with new attributes.
6390         * config/arm/cortex-a5.md: Update with new attributes.
6391         * config/arm/cortex-a53.md: Update with new attributes.
6392         * config/arm/cortex-a7.md: Update with new attributes.
6393         * config/arm/cortex-a8-neon.md: Update with new attributes.
6394         * config/arm/cortex-a9.md: Update with new attributes.
6395         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6396         * config/arm/cortex-r4f.md: Update with new attributes.
6397         * config/arm/marvell-pj4.md: Update with new attributes.
6398         * config/arm/vfp11.md: Update with new attributes.
6400 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6402         * config/aarch64/arm_neon.h
6403         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6404         (vqtbx<1,2,3,4><q>_s8): Likewise.
6406 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6408         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6409         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6410         (add<mode>3): Add type attribute.
6411         (add<mode>3): Likewise.
6412         (usadd<mode>3): Likewise.
6413         (ssadd<mode>3): Likewise.
6414         (sub<mode>3): Likewise.
6415         (sub<mode>3): Likewise.
6416         (ussub<mode>3): Likewise.
6417         (sssub<mode>3): Likewise.
6418         (ssmulsa3): Likewise.
6419         (usmulusa3): Likewise.
6420         (arm_usatsihi): Likewise.
6421         * config/arm/vfp.md
6422         (*movdi_vfp): Add types for all instructions.
6423         (*movdi_vfp_cortexa8): Likewise.
6424         (*movhf_vfp_neon): Likewise.
6425         (*movhf_vfp): Likewise.
6426         (*movdf_vfp): Likewise.
6427         (*thumb2_movdf_vfp): Likewise.
6428         (*thumb2_movdfcc_vfp): Likewise.
6429         * config/arm/arm.md: Add type attribute to all insn patterns.
6430         (*thumb1_adddi3): Add type attribute.
6431         (*arm_adddi3): Likewise.
6432         (*adddi_sesidi_di): Likewise.
6433         (*adddi_zesidi_di): Likewise.
6434         (*thumb1_addsi3): Likewise.
6435         (addsi3_compare0): Likewise.
6436         (*addsi3_compare0_scratch): Likewise.
6437         (*compare_negsi_si): Likewise.
6438         (cmpsi2_addneg): Likewise.
6439         (*addsi3_carryin_<optab>): Likewise.
6440         (*addsi3_carryin_alt2_<optab>): Likewise.
6441         (*addsi3_carryin_clobercc_<optab>): Likewise.
6442         (*subsi3_carryin): Likewise.
6443         (*subsi3_carryin_const): Likewise.
6444         (*subsi3_carryin_compare): Likewise.
6445         (*subsi3_carryin_compare_const): Likewise.
6446         (*arm_subdi3): Likewise.
6447         (*thumb_subdi3): Likewise.
6448         (*subdi_di_zesidi): Likewise.
6449         (*subdi_di_sesidi): Likewise.
6450         (*subdi_zesidi_di): Likewise.
6451         (*subdi_sesidi_di): Likewise.
6452         (*subdi_zesidi_ze): Likewise.
6453         (thumb1_subsi3_insn): Likewise.
6454         (*arm_subsi3_insn): Likewise.
6455         (*anddi3_insn): Likewise.
6456         (*anddi_zesidi_di): Likewise.
6457         (*anddi_sesdi_di): Likewise.
6458         (*ne_zeroextracts): Likewise.
6459         (*ne_zeroextracts): Likewise.
6460         (*ite_ne_zeroextr): Likewise.
6461         (*ite_ne_zeroextr): Likewise.
6462         (*anddi_notdi_di): Likewise.
6463         (*anddi_notzesidi): Likewise.
6464         (*anddi_notsesidi): Likewise.
6465         (andsi_notsi_si): Likewise.
6466         (thumb1_bicsi3): Likewise.
6467         (*iordi3_insn): Likewise.
6468         (*iordi_zesidi_di): Likewise.
6469         (*iordi_sesidi_di): Likewise.
6470         (*thumb1_iorsi3_insn): Likewise.
6471         (*xordi3_insn): Likewise.
6472         (*xordi_zesidi_di): Likewise.
6473         (*xordi_sesidi_di): Likewise.
6474         (*arm_xorsi3): Likewise.
6475         (*andsi_iorsi3_no): Likewise.
6476         (*smax_0): Likewise.
6477         (*smax_m1): Likewise.
6478         (*arm_smax_insn): Likewise.
6479         (*smin_0): Likewise.
6480         (*arm_smin_insn): Likewise.
6481         (*arm_umaxsi3): Likewise.
6482         (*arm_uminsi3): Likewise.
6483         (*minmax_arithsi): Likewise.
6484         (*minmax_arithsi_): Likewise.
6485         (*satsi_<SAT:code>): Likewise.
6486         (arm_ashldi3_1bit): Likewise.
6487         (arm_ashrdi3_1bit): Likewise.
6488         (arm_lshrdi3_1bit): Likewise.
6489         (*arm_negdi2): Likewise.
6490         (*thumb1_negdi2): Likewise.
6491         (*arm_negsi2): Likewise.
6492         (*thumb1_negsi2): Likewise.
6493         (*negdi_extendsid): Likewise.
6494         (*negdi_zero_extend): Likewise.
6495         (*arm_abssi2): Likewise.
6496         (*thumb1_abssi2): Likewise.
6497         (*arm_neg_abssi2): Likewise.
6498         (*thumb1_neg_abss): Likewise.
6499         (one_cmpldi2): Likewise.
6500         (extend<mode>di2): Likewise.
6501         (*compareqi_eq0): Likewise.
6502         (*arm_extendhisi2addsi): Likewise.
6503         (*arm_movdi): Likewise.
6504         (*thumb1_movdi_insn): Likewise.
6505         (*arm_movt): Likewise.
6506         (*thumb1_movsi_insn): Likewise.
6507         (pic_add_dot_plus_four): Likewise.
6508         (pic_add_dot_plus_eight): Likewise.
6509         (tls_load_dot_plus_eight): Likewise.
6510         (*thumb1_movhi_insn): Likewise.
6511         (*thumb1_movsf_insn): Likewise.
6512         (*movdf_soft_insn): Likewise.
6513         (*thumb_movdf_insn): Likewise.
6514         (cbranchsi4_insn): Likewise.
6515         (cbranchsi4_scratch): Likewise.
6516         (*negated_cbranchsi4): Likewise.
6517         (*tbit_cbranch): Likewise.
6518         (*tlobits_cbranch): Likewise.
6519         (*tstsi3_cbranch): Likewise.
6520         (*cbranchne_decr1): Likewise.
6521         (*addsi3_cbranch): Likewise.
6522         (*addsi3_cbranch_scratch): Likewise.
6523         (*arm_cmpdi_insn): Likewise.
6524         (*arm_cmpdi_unsig): Likewise.
6525         (*arm_cmpdi_zero): Likewise.
6526         (*thumb_cmpdi_zero): Likewise.
6527         (*deleted_compare): Likewise.
6528         (*mov_scc): Likewise.
6529         (*mov_negscc): Likewise.
6530         (*mov_notscc): Likewise.
6531         (*cstoresi_eq0_thumb1_insn): Likewise.
6532         (cstoresi_nltu_thumb1): Likewise.
6533         (cstoresi_ltu_thu): Likewise.
6534         (thumb1_addsi3_addgeu): Likewise.
6535         (*arm_jump): Likewise.
6536         (*thumb_jump): Likewise.
6537         (*check_arch2): Likewise.
6538         (arm_casesi_internal): Likewise.
6539         (thumb1_casesi_dispatch): Likewise.
6540         (*arm_indirect_jump): Likewise.
6541         (*thumb1_indirect_jump): Likewise.
6542         (nop): Likewise.
6543         (*and_scc): Likewise.
6544         (*ior_scc): Likewise.
6545         (*compare_scc): Likewise.
6546         (*cond_move): Likewise.
6547         (*cond_arith): Likewise.
6548         (*cond_sub): Likewise.
6549         (*cmp_ite0): Likewise.
6550         (*cmp_ite1): Likewise.
6551         (*cmp_and): Likewise.
6552         (*cmp_ior): Likewise.
6553         (*ior_scc_scc): Likewise.
6554         (*ior_scc_scc_cmp): Likewise.
6555         (*and_scc_scc): Likewise.
6556         (*and_scc_scc_cmp): Likewise.
6557         (*and_scc_scc_nod): Likewise.
6558         (*negscc): Likewise.
6559         (movcond_addsi): Likewise.
6560         (movcond): Likewise.
6561         (*ifcompare_plus_move): Likewise.
6562         (*if_plus_move): Likewise.
6563         (*ifcompare_move_plus): Likewise.
6564         (*if_move_plus): Likewise.
6565         (*ifcompare_arith_arith): Likewise.
6566         (*if_arith_arith): Likewise.
6567         (*ifcompare_arith_move): Likewise.
6568         (*if_arith_move): Likewise.
6569         (*ifcompare_move_arith): Likewise.
6570         (*if_move_arith): Likewise.
6571         (*ifcompare_move_not): Likewise.
6572         (*if_move_not): Likewise.
6573         (*ifcompare_not_move): Likewise.
6574         (*if_not_move): Likewise.
6575         (*ifcompare_shift_move): Likewise.
6576         (*if_shift_move): Likewise.
6577         (*ifcompare_move_shift): Likewise.
6578         (*if_move_shift): Likewise.
6579         (*ifcompare_shift_shift): Likewise.
6580         (*ifcompare_not_arith): Likewise.
6581         (*ifcompare_arith_not): Likewise.
6582         (*if_arith_not): Likewise.
6583         (*ifcompare_neg_move): Likewise.
6584         (*if_neg_move): Likewise.
6585         (*ifcompare_move_neg): Likewise.
6586         (*if_move_neg): Likewise.
6587         (prologue_thumb1_interwork): Likewise.
6588         (*cond_move_not): Likewise.
6589         (*sign_extract_onebit): Likewise.
6590         (*not_signextract_onebit): Likewise.
6591         (stack_tie): Likewise.
6592         (align_4): Likewise.
6593         (align_8): Likewise.
6594         (consttable_end): Likewise.
6595         (consttable_1): Likewise.
6596         (consttable_2): Likewise.
6597         (consttable_4): Likewise.
6598         (consttable_8): Likewise.
6599         (consttable_16): Likewise.
6600         (*thumb1_tablejump): Likewise.
6601         (prefetch): Likewise.
6602         (force_register_use): Likewise.
6603         (thumb_eh_return): Likewise.
6604         (load_tp_hard): Likewise.
6605         (load_tp_soft): Likewise.
6606         (tlscall): Likewise.
6607         (*arm_movtas_ze): Likewise.
6608         (*arm_rev): Likewise.
6609         (*arm_revsh): Likewise.
6610         (*arm_rev16): Likewise.
6611         * config/arm/thumb2.md
6612         (*thumb2_smaxsi3): Likewise.
6613         (*thumb2_sminsi3): Likewise.
6614         (*thumb32_umaxsi3): Likewise.
6615         (*thumb2_uminsi3): Likewise.
6616         (*thumb2_negdi2): Likewise.
6617         (*thumb2_abssi2): Likewise.
6618         (*thumb2_neg_abss): Likewise.
6619         (*thumb2_movsi_insn): Likewise.
6620         (tls_load_dot_plus_four): Likewise.
6621         (*thumb2_movhi_insn): Likewise.
6622         (*thumb2_mov_scc): Likewise.
6623         (*thumb2_mov_negs): Likewise.
6624         (*thumb2_mov_negs): Likewise.
6625         (*thumb2_mov_nots): Likewise.
6626         (*thumb2_mov_nots): Likewise.
6627         (*thumb2_movsicc_): Likewise.
6628         (*thumb2_movsfcc_soft_insn): Likewise.
6629         (*thumb2_indirect_jump): Likewise.
6630         (*thumb2_and_scc): Likewise.
6631         (*thumb2_ior_scc): Likewise.
6632         (*thumb2_ior_scc_strict_it): Likewise.
6633         (*thumb2_cond_move): Likewise.
6634         (*thumb2_cond_arith): Likewise.
6635         (*thumb2_cond_ari): Likewise.
6636         (*thumb2_cond_sub): Likewise.
6637         (*thumb2_negscc): Likewise.
6638         (*thumb2_movcond): Likewise.
6639         (thumb2_casesi_internal): Likewise.
6640         (thumb2_casesi_internal_pic): Likewise.
6641         (*thumb2_alusi3_short): Likewise.
6642         (*thumb2_mov<mode>_shortim): Likewise.
6643         (*thumb2_addsi_short): Likewise.
6644         (*thumb2_subsi_short): Likewise.
6645         (thumb2_addsi3_compare0): Likewise.
6646         (*thumb2_cbz): Likewise.
6647         (*thumb2_cbnz): Likewise.
6648         (*thumb2_one_cmplsi2_short): Likewise.
6649         (*thumb2_negsi2_short): Likewise.
6650         (*orsi_notsi_si): Likewise.
6651         * config/arm/arm1020e.md: Update with new attributes.
6652         * config/arm/arm1026ejs.md: Update with new attributes.
6653         * config/arm/arm1136jfs.md: Update with new attributes.
6654         * config/arm/arm926ejs.md: Update with new attributes.
6655         * config/arm/cortex-a15.md: Update with new attributes.
6656         * config/arm/cortex-a5.md: Update with new attributes.
6657         * config/arm/cortex-a53.md: Update with new attributes.
6658         * config/arm/cortex-a7.md: Update with new attributes.
6659         * config/arm/cortex-a8.md: Update with new attributes.
6660         * config/arm/cortex-a9.md: Update with new attributes.
6661         * config/arm/cortex-m4.md: Update with new attributes.
6662         * config/arm/cortex-r4.md: Update with new attributes.
6663         * config/arm/fa526.md: Update with new attributes.
6664         * config/arm/fa606te.md: Update with new attributes.
6665         * config/arm/fa626te.md: Update with new attributes.
6666         * config/arm/fa726te.md: Update with new attributes.
6668 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6670         * config/aarch64/aarch64-simd.md
6671         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6672         <vwx> iterator to ensure correct register choice.
6673         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6674         (aarch64_sqdmull_n<mode>): Likewise.
6675         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6676         * config/aarch64/arm_neon.h
6677         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6678         (vml<as><q>_n_<su>16): Likewise.
6679         (vml<as>l_high_lane<q>_<su>16): Likewise.
6680         (vml<as>l_high_n_<su>16): Likewise.
6681         (vml<as>l_lane<q>_<su>16): Likewise.
6682         (vml<as>l_n_<su>16): Likewise.
6683         (vmul<q>_lane<q>_<su>16): Likewise.
6684         (vmul<q>_n_<su>16): Likewise.
6685         (vmull_lane<q>_<su>16): Likewise.
6686         (vmull_n_<su>16): Likewise.
6687         (vmull_high_lane<q>_<su>16): Likewise.
6688         (vmull_high_n_<su>16): Likewise.
6689         (vqrdmulh<q>_n_s16): Likewise.
6691 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6693         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6694         have the correct lane parameter.
6696 2013-09-06  Richard Biener <rguenther@suse.de>
6698         * cfganal.c (control_dependences::~control_dependences):
6699         Properly free all of the vector.
6701 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6703         PR target/58269
6704         * config/i386/i386.c (ix86_conditional_register_usage):
6705         Proper initialize extended SSE registers.
6707 2013-09-06  Jan Hubicka  <jh@suse.cz>
6709         PR tree-optimization/58311
6710         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6712 2013-09-06  Jan Hubicka  <jh@suse.cz>
6714         * Makefile.in (tree-sra.o): Update dependencies.
6715         * tree-sra.c: Include ipa-utils.h
6716         (scan_function): Use recursive_call_p.
6717         (has_caller_p): New function.
6718         (cgraph_for_node_and_aliases): Count also callers of aliases.
6720 2013-09-06  Jan Hubicka  <jh@suse.cz>
6722         PR middle-end/58094
6723         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6724         * tree-tailcall.c: Include ipa-utils.h.
6725         (find_tail_calls): Use it.
6726         * ipa-pure-const.c (check_call): Likewise.
6727         * ipa-utils.c (recursive_call_p): New function.
6728         * ipa-utils.h (recursive_call_p): Dclare.
6729         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6730         (symtab_semantically_equivalent_p): New function.
6731         * Makefile.in (tree-tailcall.o): Update dependencies.
6733 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6735         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6736         non-inlinable part.
6738 2013-09-06  Richard Biener  <rguenther@suse.de>
6740         * lto-streamer.h (lto_global_var_decls): Remove.
6741         * Makefile.in (OBJS): Remove lto-symtab.o.
6742         (lto-symtab.o): Remove.
6743         (GTFILES): Remove lto-symtab.c
6744         * lto-symtab.c: Move to lto/
6746 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6748         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6749         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6750         (UNSPEC_FPINT_RINT): New constant definitions.
6751         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6752         definition with 2 attributes.
6753         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6754         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6755         definitions.
6757 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6759         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6760         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6761         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6762         attribute to "z196_alone".
6763         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6764         "zEC12_simple".
6766 2013-09-06  Richard Biener  <rguenther@suse.de>
6768         * basic-block.h (class control_dependences): New.
6769         * tree-ssa-dce.c (control_dependence_map): Remove.
6770         (cd): New global.
6771         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6772         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6773         find_pdom, find_control_dependence, find_all_control_dependences):
6774         Move to cfganal.c.
6775         (mark_control_dependent_edges_necessary,
6776         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6777         tree_dce_done, perform_tree_ssa_dce): Adjust.
6778         * cfganal.c (set_control_dependence_map_bit,
6779         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6780         find_all_control_dependences): Move from tree-ssa-dce.c and
6781         implement as methods of control_dependences class.
6782         (control_dependences::control_dependences): New.
6783         (control_dependences::~control_dependences): Likewise.
6784         (control_dependences::get_edges_dependent_on): Likewise.
6785         (control_dependences::get_edge): Likewise.
6787 2013-09-04  Jan Hubicka  <jh@suse.cz>
6789         * tree.c (types_same_for_odr): Drop overactive check.
6790         * ipa-devirt.c (hash_type_name): Likewise.
6792 2013-09-04  Jan Hubicka  <jh@suse.cz>
6794         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6795         (analyze_functions): ... here.
6797 2013-09-04  Jan Hubicka  <jh@suse.cz>
6799         PR middle-end/58201
6800         * cgraphunit.c (analyze_functions): Clear AUX fields
6801         after processing; initialize assembler name has.
6803 2013-09-05  Jeff Law  <law@redhat.com>
6805         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6806         from thread_around_empty_block.  Record threading path into PATH.
6807         Recurse if threading through the initial block is successful.
6808         (thread_across_edge): Corresponding changes to slightly simplify.
6810 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6812         * config/aarch64/aarch64.md
6813         (type): Remove frecpe, frecps, frecpx.
6814         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6815         fix to be a TARGET_SIMD instruction.
6816         (aarch64_frecps): Remove.
6817         * config/aarch64/aarch64-simd.md
6818         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6819         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6821 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6822             Sofiane Naci  <sofiane.naci@arm.com>
6824         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6825         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6826         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6827         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6828         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6829         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6830         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6831         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6832         into "clz, "rbit".  Rename "shift" to "shift_imm".
6833         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6834         changes.  Update for attribute changes all occurrences of arlo_* and
6835         shift* types.
6836         * config/arm/arm-fixed.md: Update for attribute changes
6837         all occurrences of arlo_* types.
6838         * config/arm/thumb2.md: Update for attribute changes all occurrences
6839         of arlo_* types.
6840         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6841         (cortexa7_older_only): Likewise.
6842         (cortexa7_younger):  Likewise.
6843         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6844         (1020alu_shift_op): Likewise.
6845         (1020alu_shift_reg_op): Likewise.
6846         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6847         (alu_shift_op): Likewise.
6848         (alu_shift_reg_op): Likewise.
6849         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6850         (11_alu_shift_op): Likewise.
6851         (11_alu_shift_reg_op): Likewise.
6852         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6853         (9_alu_shift_reg_op): Likewise.
6854         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6855         attribute changes.
6856         (cortex_a15_alu_shift): Likewise.
6857         (cortex_a15_alu_shift_reg): Likewise.
6858         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6859         attribute changes.
6860         (cortex_a5_alu_shift): Likewise.
6861         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6862         attribute changes.
6863         (cortex_a53_alu_shift): Likewise.
6864         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6865         attribute changes.
6866         (cortex_a7_alu_reg): Likewise.
6867         (cortex_a7_alu_shift): Likewise.
6868         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6869         attribute changes.
6870         (cortex_a8_alu_shift): Likewise.
6871         (cortex_a8_alu_shift_reg): Likewise.
6872         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6873         (cortex_a9_dp_shift): Likewise.
6874         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6875         attribute changes.
6876         * config/arm/cortex-r4.md
6877         (cortex_r4_alu): Update for attribute changes.
6878         (cortex_r4_mov): Likewise.
6879         (cortex_r4_alu_shift_reg): Likewise.
6880         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6881         (526_alu_shift_op): Likewise.
6882         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6883         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6884         (626te_alu_shift_op): Likewise.
6885         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6886         (726te_alu_shift_op): Likewise.
6887         (726te_alu_shift_reg_op): Likewise.
6888         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6889         (mp626_alu_shift_op): Likewise.
6890         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6891         (pj4_alu_conds): Likewise.
6892         (pj4_shift): Likewise.
6893         (pj4_shift_conds): Likewise.
6894         (pj4_alu_shift): Likewise.
6895         (pj4_alu_shift_conds): Likewise.
6896         * config/aarch64/aarch64.md: Update for attribute change
6897         all occurrences of arlo_* and shift* types.
6899 2013-09-05  Mike Stump  <mikestump@comcast.net>
6901         * tree.h: Move documentation for tree_function_decl to tree-core.h
6902         with the declaration.
6904 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6906         PR target/58139
6907         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6908         looking for widest mode.
6910 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6912         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6914 2013-09-05  Richard Biener  <rguenther@suse.de>
6916         PR tree-optimization/58137
6917         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6918         Do not create vectors of pointers.
6919         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6920         types for the components of the vector initializer.
6921         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6922         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6924 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6926         * ipa-prop.c (remove_described_reference): Accept missing references,
6927         return false if that hppens, otherwise return true.
6928         (cgraph_node_for_jfunc): New function.
6929         (try_decrement_rdesc_refcount): Likewise.
6930         (try_make_edge_direct_simple_call): Use them.
6931         (ipa_edge_removal_hook): Remove references from rdescs.
6932         (ipa_edge_duplication_hook): Clone rdescs and their references
6933         when the new edge has the same caller as the old one.
6934         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6935         reference before removing any edges.
6937 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6939         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6940         initial store.
6941         * thumb2.md (thumb2_storewb_parisi): New pattern.
6943 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6945         * config/aarch64/aarch64-option-extensions.def: Add
6946         AARCH64_OPT_EXTENSION of 'crc'.
6947         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6948         (AARCH64_ISA_CRC): Ditto.
6949         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6950         description of the CRC extension.
6952 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6954         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6955         * config/rs6000/linux.h: Ditto.
6956         * alpha/linux.h: Ditto.
6957         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6958         no_c99_libc_has_function.
6959         * config/c6x/uclinux-elf.h: Ditto.
6960         * config/lm32/uclinux-elf.h: Ditto.
6961         * config/m68k/uclinux.h: Ditto.
6962         * config/moxie/uclinux.h: Ditto.
6963         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6964         (crisv32-*-linux*, cris-*-linux*): Ditto.
6965         * config/bfin/bfin.c: Include "tm_p.h".
6967 2013-09-05  Richard Biener  <rguenther@suse.de>
6969         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6970         check for a definition without a basic-block.
6972 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6973             Sofiane Naci <sofiane.naci@arm.com>
6975         * config/aarch64/aarch64.md
6976         (*movti_aarch64): Rename r_2_f and f_2_r.
6977         (*movsf_aarch64): Likewise.
6978         (*movdf_aarch64): Likewise.
6979         (*movtf_aarch64): Likewise.
6980         (aarch64_movdi_<mode>low): Likewise.
6981         (aarch64_movdi_<mode>high): Likewise.
6982         (aarch64_mov<mode>high_di): Likewise.
6983         (aarch64_mov<mode>low_di): Likewise.
6984         (aarch64_movtilow_tilow): Likewise.
6985         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6986         values to config/arm/types.md
6987         (attribute "conds"): Update for attribute change.
6988         (anddi3_insn): Likewise.
6989         (iordi3_insn): Likewise.
6990         (xordi3_insn): Likewise.
6991         (one_cmpldi2): Likewise.
6992         * config/arm/types.md (type): Add Neon types.
6993         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6994         use "type" attribute.
6995         (movmisalign<mode>_neon_store): Likewise.
6996         (movmisalign<mode>_neon_load): Likewise.
6997         (vec_set<mode>_internal): Likewise.
6998         (vec_setv2di_internal): Likewise.
6999         (vec_extract<mode>): Likewise.
7000         (vec_extractv2di): Likewise.
7001         (add<mode>3_neon): Likewise.
7002         (adddi3_neon): Likewise.
7003         (sub<mode>3_neon): Likewise.
7004         (subdi3_neon): Likewise.
7005         (mul<mode>3_neon): Likewise.
7006         (mul<mode>3add<mode>_neon): Likewise.
7007         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
7008         (fma<VCVTF:mode>4)): Likewise.
7009         (fma<VCVTF:mode>4_intrinsic): Likewise.
7010         (fmsub<VCVTF:mode>4)): Likewise.
7011         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
7012         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
7013         (ior<mode>3): Likewise.
7014         (and<mode>3): Likewise.
7015         (anddi3_neon): Likewise.
7016         (orn<mode>3_neon): Likewise.
7017         (orndi3_neon): Likewise.
7018         (bic<mode>3_neon): Likewise.
7019         (bicdi3_neon): Likewise.
7020         (xor<mode>3): Likewise.
7021         (one_cmpl<mode>2): Likewise.
7022         (abs<mode>2): Likewise.
7023         (neg<mode>2): Likewise.
7024         (umin<mode>3_neon): Likewise.
7025         (umax<mode>3_neon): Likewise.
7026         (smin<mode>3_neon): Likewise.
7027         (smax<mode>3_neon): Likewise.
7028         (vashl<mode>3): Likewise.
7029         (vashr<mode>3_imm): Likewise.
7030         (vlshr<mode>3_imm): Likewise.
7031         (ashl<mode>3_signed): Likewise.
7032         (ashl<mode>3_unsigned): Likewise.
7033         (neon_load_count): Likewise.
7034         (ashldi3_neon_noclobber): Likewise.
7035         (signed_shift_di3_neon): Likewise.
7036         (unsigned_shift_di3_neon): Likewise.
7037         (ashrdi3_neon_imm_noclobber): Likewise.
7038         (lshrdi3_neon_imm_noclobber): Likewise.
7039         (widen_ssum<mode>3): Likewise.
7040         (widen_usum<mode>3): Likewise.
7041         (quad_halves_<code>v4si): Likewise.
7042         (quad_halves_<code>v4sf): Likewise.
7043         (quad_halves_<code>v8hi): Likewise.
7044         (quad_halves_<code>v16qi): Likewise.
7045         (reduc_splus_v2di): Likewise.
7046         (neon_vpadd_internal<mode>): Likewise.
7047         (neon_vpsmin<mode>): Likewise.
7048         (neon_vpsmax<mode>): Likewise.
7049         (neon_vpumin<mode>): Likewise.
7050         (neon_vpumax<mode>): Likewise.
7051         (ss_add<mode>_neon): Likewise.
7052         (us_add<mode>_neon): Likewise.
7053         (ss_sub<mode>_neon): Likewise.
7054         (us_sub<mode>_neon): Likewise.
7055         (neon_vadd<mode>_unspec): Likewise.
7056         (neon_vaddl<mode>): Likewise.
7057         (neon_vaddw<mode>): Likewise.
7058         (neon_vhadd<mode>): Likewise.
7059         (neon_vqadd<mode>): Likewise.
7060         (neon_vaddhn<mode>): Likewise.
7061         (neon_vmul<mode>): Likewise.
7062         (neon_vmla<mode>): Likewise.
7063         (neon_vmlal<mode>): Likewise.
7064         (neon_vmls<mode>): Likewise.
7065         (neon_vmlsl<mode>): Likewise.
7066         (neon_vqdmulh<mode>): Likewise.
7067         (neon_vqdmlal<mode>): Likewise.
7068         (neon_vqdmlsl<mode>): Likewise.
7069         (neon_vmull<mode>): Likewise.
7070         (neon_vqdmull<mode>): Likewise.
7071         (neon_vsub<mode>_unspec): Likewise.
7072         (neon_vsubl<mode>): Likewise.
7073         (neon_vsubw<mode>): Likewise.
7074         (neon_vqsub<mode>): Likewise.
7075         (neon_vhsub<mode>): Likewise.
7076         (neon_vsubhn<mode>): Likewise.
7077         (neon_vceq<mode>): Likewise.
7078         (neon_vcge<mode>): Likewise.
7079         (neon_vcgeu<mode>): Likewise.
7080         (neon_vcgt<mode>): Likewise.
7081         (neon_vcgtu<mode>): Likewise.
7082         (neon_vcle<mode>): Likewise.
7083         (neon_vclt<mode>): Likewise.
7084         (neon_vcage<mode>): Likewise.
7085         (neon_vcagt<mode>): Likewise.
7086         (neon_vtst<mode>): Likewise.
7087         (neon_vabd<mode>): Likewise.
7088         (neon_vabdl<mode>): Likewise.
7089         (neon_vaba<mode>): Likewise.
7090         (neon_vabal<mode>): Likewise.
7091         (neon_vmax<mode>): Likewise.
7092         (neon_vmin<mode>): Likewise.
7093         (neon_vpaddl<mode>): Likewise.
7094         (neon_vpadal<mode>): Likewise.
7095         (neon_vpmax<mode>): Likewise.
7096         (neon_vpmin<mode>): Likewise.
7097         (neon_vrecps<mode>): Likewise.
7098         (neon_vrsqrts<mode>): Likewise.
7099         (neon_vqabs<mode>): Likewise.
7100         (neon_vqneg<mode>): Likewise.
7101         (neon_vcls<mode>): Likewise.
7102         (clz<mode>2): Likewise.
7103         (popcount<mode>2): Likewise.
7104         (neon_vrecpe): Likewise.
7105         (neon_vrsqrte): Likewise.
7106         (neon_vget_lane<mode>_sext_internal): Likewise.
7107         (neon_vget_lane<mode>_zext_internal): Likewise.
7108         (neon_vdup_n<mode>): Likewise.
7109         (neon_vdup_nv2di): Likewise.
7110         (neon_vdpu_lane<mode>_internal): Likewise.
7111         (neon_vswp<mode>): Likewise.
7112         (float<mode><V_cvtto>2): Likewise.
7113         (floatuns<mode><V_cvtto>2): Likewise.
7114         (fix_trunc<mode><V_cvtto>)2): Likewise
7115         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7116         (neon_vcvt<mode>): Likewise.
7117         (neon_vcvtv4sfv4hf): Likewise.
7118         (neon_vcvtv4hfv4sf): Likewise.
7119         (neon_vcvt_n<mode>): Likewise.
7120         (neon_vmovn<mode>): Likewise.
7121         (neon_vqmovn<mode>): Likewise.
7122         (neon_vqmovun<mode>): Likewise.
7123         (neon_vmovl<mode>): Likewise.
7124         (neon_vmul_lane<mode>): Likewise.
7125         (neon_vmull_lane<mode>): Likewise.
7126         (neon_vqdmull_lane<mode>): Likewise.
7127         (neon_vqdmulh_lane<mode>): Likewise.
7128         (neon_vmla_lane<mode>): Likewise.
7129         (neon_vmlal_lane<mode>): Likewise.
7130         (neon_vqdmlal_lane<mode>): Likewise.
7131         (neon_vmls_lane<mode>): Likewise.
7132         (neon_vmlsl_lane<mode>): Likewise.
7133         (neon_vqdmlsl_lane<mode>): Likewise.
7134         (neon_vext<mode>): Likewise.
7135         (neon_vrev64<mode>): Likewise.
7136         (neon_vrev32<mode>): Likewise.
7137         (neon_vrev16<mode>): Likewise.
7138         (neon_vbsl<mode>_internal): Likewise.
7139         (neon_vshl<mode>): Likewise.
7140         (neon_vqshl<mode>): Likewise.
7141         (neon_vshr_n<mode>): Likewise.
7142         (neon_vshrn_n<mode>): Likewise.
7143         (neon_vqshrn_n<mode>): Likewise.
7144         (neon_vqshrun_n<mode>): Likewise.
7145         (neon_vshl_n<mode>): Likewise.
7146         (neon_vqshl_n<mode>): Likewise.
7147         (neon_vqshlu_n<mode>): Likewise.
7148         (neon_vshll_n<mode>): Likewise.
7149         (neon_vsra_n<mode>): Likewise.
7150         (neon_vsri_n<mode>): Likewise.
7151         (neon_vsli_n<mode>): Likewise.
7152         (neon_vtbl1v8qi): Likewise.
7153         (neon_vtbl2v8qi): Likewise.
7154         (neon_vtbl3v8qi): Likewise.
7155         (neon_vtbl4v8qi): Likewise.
7156         (neon_vtbx1v8qi): Likewise.
7157         (neon_vtbx2v8qi): Likewise.
7158         (neon_vtbx3v8qi): Likewise.
7159         (neon_vtbx4v8qi): Likewise.
7160         (neon_vtrn<mode>_internal): Likewise.
7161         (neon_vzip<mode>_internal): Likewise.
7162         (neon_vuzp<mode>_internal): Likewise.
7163         (neon_vld1<mode>): Likewise.
7164         (neon_vld1_lane<mode>): Likewise.
7165         (neon_vld1_dup<mode>): Likewise.
7166         (neon_vld1_dupv2di): Likewise.
7167         (neon_vst1<mode>): Likewise.
7168         (neon_vst1_lane<mode>): Likewise.
7169         (neon_vld2<mode>): Likewise.
7170         (neon_vld2_lane<mode>): Likewise.
7171         (neon_vld2_dup<mode>): Likewise.
7172         (neon_vst2<mode>): Likewise.
7173         (neon_vst2_lane<mode>): Likewise.
7174         (neon_vld3<mode>): Likewise.
7175         (neon_vld3qa<mode>): Likewise.
7176         (neon_vld3qb<mode>): Likewise.
7177         (neon_vld3_lane<mode>): Likewise.
7178         (neon_vld3_dup<mode>): Likewise.
7179         (neon_vst3<mode>): Likewise.
7180         (neon_vst3qa<mode>): Likewise.
7181         (neon_vst3qb<mode>): Likewise.
7182         (neon_vst3_lane<mode>): Likewise.
7183         (neon_vld4<mode>): Likewise.
7184         (neon_vld4qa<mode>): Likewise.
7185         (neon_vld4qb<mode>): Likewise.
7186         (neon_vld4_lane<mode>): Likewise.
7187         (neon_vld4_dup<mode>): Likewise.
7188         (neon_vst4<mode>): Likewise.
7189         (neon_vst4qa<mode>): Likewise.
7190         (neon_vst4qb<mode>): Likewise.
7191         (neon_vst4_lane<mode>): Likewise.
7192         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7193         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7194         (neon_vec_<US>mult_lo_<mode>): Likewise.
7195         (neon_vec_<US>mult_hi_<mode>): Likewise.
7196         (neon_vec_<US>shiftl_<mode>): Likewise.
7197         (neon_unpack<US>_<mode>): Likewise.
7198         (neon_vec_<US>mult_<mode>): Likewise.
7199         (vec_pack_trunc_<mode>): Likewise.
7200         (neon_vec_pack_trunk_<mode>): Likewise.
7201         (neon_vabd<mode>_2): Likewise.
7202         (neon_vabd<mode>_3): Likewise.
7203         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7204         (thumb2_movsi_vfp): Likewise.
7205         (movdi_vfp): Likewise.
7206         (movdi_vfp_cortexa8): Likewise.
7207         (movhf_vfp_neon): Likewise.
7208         (movhf_vfp): Likewiwse.
7209         (movsf_vfp): Likewiwse.
7210         (thumb2_movsf_vfp): Likewiwse.
7211         (movdf_vfp): Likewise.
7212         (thumb2_movdf_vfp): Likewise.
7213         (movsfcc_vfp): Likewise.
7214         (thumb2_movsfcc_vfp): Likewise.
7215         (movdfcc_vfp): Likewise.
7216         (thumb2_movdfcc_vfp): Likewise.
7217         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7218         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7219         (v10_v2c): Likewise.
7220         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7221         attribute change.
7222         (cortex_a15_neon_int_2): Likewise.
7223         (cortex_a15_neon_int_3): Likewise.
7224         (cortex_a15_neon_int_4): Likewise.
7225         (cortex_a15_neon_int_5): Likewise.
7226         (cortex_a15_neon_vqneg_vqabs): Likewise.
7227         (cortex_a15_neon_vmov): Likewise.
7228         (cortex_a15_neon_vaba): Likewise.
7229         (cortex_a15_neon_vaba_qqq): Likewise.
7230         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7231         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7232         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7233         scalar_64_32_long_scalar): Likewise.
7234         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7235         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7236         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7237         lotype_qdd_64_32_long): Likewise.
7238         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7239         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7240         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7241         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7242         (cortex_a15_neon_shift_1): Likewise.
7243         (cortex_a15_neon_shift_2): Likewise.
7244         (cortex_a15_neon_shift_3): Likewise.
7245         (cortex_a15_neon_vshl_ddd): Likewise.
7246         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7247         (cortex_a15_neon_vsra_vrsra): Likewise.
7248         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7249         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7250         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7251         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7252         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7253         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7254         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7255         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7256         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7257         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7258         (cortex_a15_neon_bp_simple): Likewise.
7259         (cortex_a15_neon_bp_2cycle): Likewise.
7260         (cortex_a15_neon_bp_3cycle): Likewise.
7261         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7262         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7263         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7264         (cortex_a15_neon_vld2_4_regs): Likewise.
7265         (cortex_a15_neon_vld3_vld4): Likewise.
7266         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7267         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7268         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7269         (cortex_a15_neon_vst3_vst4): Likewise.
7270         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7271         (cortex_a15_neon_vld3_vld4_lane" 10
7272         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7273         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7274         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7275         (cortex_a15_neon_ldm_2): Likewise.0
7276         (cortex_a15_neon_stm_2): Likewise.
7277         (cortex_a15_neon_mcr): Likewise.
7278         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7279         (cortex_a15_neon_mrc): Likewise.
7280         (cortex_a15_neon_mrrc): Likewise.
7281         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7282         change.
7283         (cortex_a15_alu_shift): Likewise.
7284         (cortex_a15_alu_shift_reg): Likewise.
7285         (cortex_a15_mult32): Likewise.
7286         (cortex_a15_mult64): Likewise.
7287         (cortex_a15_block): Likewise.
7288         (cortex_a15_branch): Likewise.
7289         (cortex_a15_load1): Likewise.
7290         (cortex_a15_load3): Likewise.
7291         (cortex_a15_store1): Likewise.
7292         (cortex_a15_store3): Likewise.
7293         (cortex_a15_call): Likewise.
7294         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7295         (cortex_a5_f2r): Likewise.
7296         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7297         change.
7298         (cortex_a53_f2r): Likewise.
7299         * config/arm/cortex-a7.md
7300         (cortex_a7_branch): Update for attribute change.
7301         (cortex_a7_call): Likewise.
7302         (cortex_a7_alu_imm): Likewise.
7303         (cortex_a7_alu_reg): Likewise.
7304         (cortex_a7_alu_shift): Likewise.
7305         (cortex_a7_mul): Likewise.
7306         (cortex_a7_load1): Likewise.
7307         (cortex_a7_store1): Likewise.
7308         (cortex_a7_load2): Likewise.
7309         (cortex_a7_store2): Likewise.
7310         (cortex_a7_load3): Likewise.
7311         (cortex_a7_store3): Likewise.
7312         (cortex_a7_load4): Likewise.
7313         (cortex_a7_store4): Likewise.
7314         (cortex_a7_fpalu): Likewise.
7315         (cortex_a7_fconst): Likewise.
7316         (cortex_a7_fpmuls): Likewise.
7317         (cortex_a7_neon_mul): Likewise.
7318         (cortex_a7_fpmacs): Likewise.
7319         (cortex_a7_neon_mla: Likewise.
7320         (cortex_a7_fpmuld: Likewise.
7321         (cortex_a7_fpmacd: Likewise.
7322         (cortex_a7_fpfmad: Likewise.
7323         (cortex_a7_fdivs: Likewise.
7324         (cortex_a7_fdivd: Likewise.
7325         (cortex_a7_r2f: Likewise.
7326         (cortex_a7_f2r: Likewise.
7327         (cortex_a7_f_flags: Likewise.
7328         (cortex_a7_f_loads: Likewise.
7329         (cortex_a7_f_loadd: Likewise.
7330         (cortex_a7_f_stores: Likewise.
7331         (cortex_a7_f_stored: Likewise.
7332         (cortex_a7_neon): Likewise.
7333         * config/arm/cortex-a8-neon.md
7334         (cortex_a8_neon_mrc): Update for attribute change.
7335         (cortex_a8_neon_mrrc): Likewise.
7336         (cortex_a8_neon_int_1): Likewise.
7337         (cortex_a8_neon_int_2): Likewise.
7338         (cortex_a8_neon_int_3): Likewise.
7339         (cortex_a8_neon_int_4): Likewise.
7340         (cortex_a8_neon_int_5): Likewise.
7341         (cortex_a8_neon_vqneg_vqabs): Likewise.
7342         (cortex_a8_neon_vmov): Likewise.
7343         (cortex_a8_neon_vaba): Likewise.
7344         (cortex_a8_neon_vaba_qqq): Likewise.
7345         (cortex_a8_neon_vsma): Likewise.
7346         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7347         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7348         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7349         long_scalar): Likewise.
7350         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7351         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7352         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7353         long_scalar_qdd_64_32_long): Likewise.
7354         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7355         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7356         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7357         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7358         (cortex_a8_neon_shift_1): Likewise.
7359         (cortex_a8_neon_shift_2): Likewise.
7360         (cortex_a8_neon_shift_3): Likewise.
7361         (cortex_a8_neon_vshl_ddd): Likewise.
7362         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7363         (cortex_a8_neon_vsra_vrsra): Likewise.
7364         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7365         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7366         (cortex_a8_neon_fp_vsum): Likewise.
7367         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7368         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7369         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7370         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7371         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7372         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7373         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7374         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7375         (cortex_a8_neon_bp_simple): Likewise.
7376         (cortex_a8_neon_bp_2cycle): Likewise.
7377         (cortex_a8_neon_bp_3cycle): Likewise.
7378         (cortex_a8_neon_ldr): Likewise.
7379         (cortex_a8_neon_str): Likewise.
7380         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7381         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7382         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7383         (cortex_a8_neon_vld2_4_regs): Likewise.
7384         (cortex_a8_neon_vld3_vld4): Likewise.
7385         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7386         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7387         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7388         (cortex_a8_neon_vst3_vst4): Likewise.
7389         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7390         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7391         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7392         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7393         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7394         (cortex_a8_neon_mcr): Likewise.
7395         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7396         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7397         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7398         change.
7399         (ca9_neon_mrrc): Likewise.
7400         (cortex_a9_neon_int_1): Likewise.
7401         (cortex_a9_neon_int_2): Likewise.
7402         (cortex_a9_neon_int_3): Likewise.
7403         (cortex_a9_neon_int_4): Likewise.
7404         (cortex_a9_neon_int_5): Likewise.
7405         (cortex_a9_neon_vqneg_vqabs): Likewise.
7406         (cortex_a9_neon_vmov): Likewise.
7407         (cortex_a9_neon_vaba): Likewise.
7408         (cortex_a9_neon_vaba_qqq): Likewise.
7409         (cortex_a9_neon_vsma): Likewise.
7410         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7411         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7412         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7413         long_scalar): Likewise.
7414         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7415         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7416         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7417         long_scalar_qdd_64_32_long): Likewise.
7418         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7419         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7420         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7421         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7422         (cortex_a9_neon_shift_1): Likewise.
7423         (cortex_a9_neon_shift_2): Likewise.
7424         (cortex_a9_neon_shift_3): Likewise.
7425         (cortex_a9_neon_vshl_ddd): Likewise.
7426         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7427         (cortex_a9_neon_vsra_vrsra): Likewise.
7428         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7429         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7430         (cortex_a9_neon_fp_vsum): Likewise.
7431         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7432         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7433         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7434         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7435         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7436         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7437         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7438         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7439         (cortex_a9_neon_bp_simple): Likewise.
7440         (cortex_a9_neon_bp_2cycle): Likewise.
7441         (cortex_a9_neon_bp_3cycle): Likewise.
7442         (cortex_a9_neon_ldr): Likewise.
7443         (cortex_a9_neon_str): Likewise.
7444         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7445         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7446         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7447         (cortex_a9_neon_vld2_4_regs): Likewise.
7448         (cortex_a9_neon_vld3_vld4): Likewise.
7449         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7450         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7451         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7452         (cortex_a9_neon_vst3_vst4): Likewise.
7453         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7454         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7455         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7456         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7457         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7458         (cortex_a9_neon_mcr): Likewise.
7459         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7460         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7461         (cortex_a9_fps): Likewise.
7462         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7463         change.
7464         (cortex_m4_fmuls): Likewise.
7465         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7466         change.
7467         (cortex_r4_mrc): Likewise.
7468         * config/arm/iterators.md: Update comment referring to neon_type.
7469         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7470         (iwmmxt_movsi_insn): Likewise.
7471         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7472         attribute change.
7473         (pj4_core_to_vfp): Likewise.
7474         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7475         attribute change.
7476         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7477         (vfp_fstore): Likewise.
7478         * doc/md.texi: Change references to neon_type to refer to type.
7480 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7482         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7484 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7486         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7487         lglobl if not weak.
7489 2013-09-04  Easwaran Raman  <eraman@google.com>
7491         PR middle-end/57370
7492         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7493         (build_and_add_sum): Use it.
7494         (appears_later_in_bb): Simplify code.
7496 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7498         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7500 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7502         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7504 2013-09-04  Jan Hubicka  <jh@suse.cz>
7506         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7507         * ipa-devirt.c: Include diganostic.h
7508         (odr_type_d): Add types and types_set.
7509         (hash_type_name): Work for types with vtables during LTO.
7510         (odr_hasher::remove): Fix comment; destroy types_set.
7511         (add_type_duplicate): New function,
7512         (get_odr_type): Use it.
7513         (dump_type_inheritance_graph): Dump type duplicates.
7514         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7515         graph.
7516         * tree.c (types_same_for_odr): Give exact answers on types with
7517         virtual tables.
7519 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7521         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7522         explaining their differences.
7524 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7526         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7528 2013-09-03  Jeff Law  <law@redhat.com>
7530         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7531         when not threading through a joiner block.  Pass joiner/no joiner
7532         state to register_jump_thread.
7533         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7534         state from argument rather than implying on path length.
7535         Dump the entire jump thread path into debugging dump.
7536         * tree-flow.h (register_jump_thread): Update prototype.
7538 2013-08-29  Xinliang David Li  <davidxl@google.com>
7540         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7541         Remove a trivial gcc_assert.
7543 2013-08-29  Xinliang David Li  <davidxl@google.com>
7545         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7546         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7547         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7548         Delay base decl alignment adjustment.
7549         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7550         * tree-vectorizer.h: New data structure.
7551         (set_dr_misalignment): New function.
7552         (dr_misalignment): Ditto.
7553         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7554         (vectorizable_load): Ditto.
7555         (ensure_base_align): New function.
7556         (vectorize_loops): Add dbg_cnt support.
7557         (execute_vect_slp): Ditto.
7558         * dbgcnt.def: New debug counter.
7559         * Makefile: New dependency.
7561 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7563         Revert:
7565         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7567         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7569 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7571         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7572         function descriptor.
7574 2013-09-03  Richard Biener  <rguenther@suse.de>
7576         * tree-affine.c (add_elt_to_tree): Fix association issue,
7577         avoid useless converts and make sure to always return a
7578         properly typed result.
7580 2013-09-03  Richard Biener  <rguenther@suse.de>
7582         PR middle-end/57656
7583         * fold-const.c (negate_expr_p): Fix division case.
7584         (negate_expr): Likewise.
7586 2013-09-03  Richard Biener  <rguenther@suse.de>
7588         PR lto/58285
7589         * tree-streamer-out.c: Include tm.h.
7590         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7592 2013-09-03  Jan Hubicka  <jh@suse.cz>
7594         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7596 2013-09-03  Alan Modra  <amodra@gmail.com>
7598         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7599         multilibs to big-endian target and vice versa.
7600         * config/rs6000/t-linux64: Use := assignment on all vars.
7601         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7602         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7603         * config/rs6000/t-linux64le: New file.
7604         * config/rs6000/t-linux64bele: New file.
7605         * config/rs6000/t-linux64lebe: New file.
7607 2013-09-02  Jan Hubicka  <jh@suse.cz>
7609         * ipa-inline-transform.c (inline_transform): Do not
7610         optimize_inline_calls when not optimizing.
7612 2013-09-02  Jan Hubicka  <jh@suse.cz>
7614         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7615         duplicated nodes for assembler names.
7616         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7617         hard registers from assembler name hash.
7619 2013-09-02  Jan Hubicka  <jh@suse.cz>
7621         * ipa-split.c (execute_split_functions): Split externally visible
7622         functions called once.
7624 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7626         PR ipa/58106
7627         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7628         linked list.  When finding the correct duplicate, also consider also
7629         the caller in additon to its inlined_to node.
7631 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7633         * config/aarch64/aarch64-simd-builtins.def
7634         (dup_lane_scalar): Remove.
7635         * config/aarch64/aarch64-simd.md
7636         (aarch64_simd_dup): Add 'w->w' alternative.
7637         (aarch64_dup_lane<mode>): Allow for VALL.
7638         (aarch64_dup_lane_scalar<mode>): Remove.
7639         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7640         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7641         (aarch64_get_lane_unsigned<mode>): Likewise.
7642         (aarch64_get_lane<mode>): Likewise.
7643         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7644         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7645         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7646         (VCON): Change container of V2SF.
7647         (vswap_width_name): Likewise.
7648         * config/aarch64/arm_neon.h
7649         (__aarch64_vdup_lane_any): New.
7650         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7651         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7652         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7654 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7656         PR middle-end/56382
7657         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7658         the source or the destination is a single hard register.
7660 2013-09-02  Richard Biener  <rguenther@suse.de>
7662         PR middle-end/57511
7663         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7664         non-linear SCEVs.
7666 2013-09-02  Richard Biener  <rguenther@suse.de>
7668         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7669         arithmetic to sizetype.
7671 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7673         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7674         Find auto-increment use both before and after candidate.
7676 2013-09-02  Marek Polacek  <polacek@redhat.com>
7678         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7680 2013-09-01  Jan Hubicka  <jh@suse.cz>
7682         * Makefile.in: Add ipa-profile.o
7683         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7684         * cgraph.c (struct cgraph_propagate_frequency_data,
7685         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7686         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7687         * cgraph.h (cgraph_propagate_frequency): Remove.
7688         * ipa-inline-analysis.c: Include ipa-utils.h;
7689         drop duplicated cfgloop.h.
7690         (inline_update_callee_summaries): Update.
7691         * ipa-profile.c: New file.
7692         * ipa-utils.h (ipa_propagate_frequency): Declare.
7693         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7694         data-streamer.h, value-prof.h.
7695         (symtab_remove_unreachable_nodes): Update profile.
7696         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7697         account_time_size, cmp_counts, dump_histogram,
7698         ipa_profile_generate_summary, ipa_profile_write_summary,
7699         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7700         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7701         Move to ipa-profile.c.
7703 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7705         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7707 2013-09-01  Jan Hubicka  <jh@suse.cz>
7709         * common.opt (fdevirtualize-speculatively): New function.
7710         * invoke.texi (fdevirtualize-speculatively): Document.
7711         * ipa-devirt.c: Include ipa-inline.h
7712         (likely_target_p): New function.
7713         (ipa_devirt): New function.
7714         (gate_ipa_devirt): New function.
7715         (pass_data_ipa_devirt): New static var.
7716         (pass_ipa_devirt): Likewise.
7717         (make_pass_ipa_devirt): New function.
7718         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7719         (common_handle_option): Disable devirtualization when
7720         value range profiling is available.
7721         * passes.def (pass_ipa_devirt): Add.
7722         * timever.def (TV_IPA_DEVIRT): New timevar.
7723         * tree-pass.h (make_pass_ipa_devirt):
7725 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7727         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7728         include sanitize(undefined).
7730 2013-08-31  Diego Novillo  <dnovillo@google.com>
7732         * Makefile.in (TREE_CORE_H): Define.
7733         (TREE_H): Use.
7734         (GTFILES): Add tree-core.h.
7735         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7736         size the array.
7737         * tree-core.h: New file.
7738         Move all data structures, enum, typedefs, global
7739         declarations and constants from ...
7740         * tree.h: ... here.
7742 2013-08-31  Jan Hubicka  <jh@suse.cz>
7744         * bulitins.c (expand_builtin): Do not early exit for gcov
7745         instrumented functions.
7747 2013-08-31  Marek Polacek  <polacek@redhat.com>
7749         * ubsan.c: Include tm_p.h.
7751 2013-08-31  Jan Hubicka  <jh@suse.cz>
7753         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7754         warning.
7756         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7757         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7758         * tree-cfg.c (verify_gimple_label): ... here.
7759         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7760         (ipa_merge_profiles): New function.
7761         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7762         (lto_input_function_body): Likewise.
7763         * ipa-utils.h (ipa_merge_profiles): Declare.
7764         * lto-streamer.h (lto_input_function_body): Update prototype.
7765         (emit_label_in_global_context_p): Remove.
7766         * lto-symtab.c: Include ipa-utils.h
7767         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7769 2013-08-31  Jan Hubicka  <jh@suse.cz>
7771         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7773 2013-08-31  Jan Hubicka  <jh@suse.cz>
7775         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7777 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7779         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7780         "cmp" RTX before signed_comparison_operator check to account
7781         for "code" changes.
7783 2013-08-30  Jan Hubicka  <jh@suse.cz>
7785         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7786         (detect_type_change_1): Rename to ...
7787         (detect_type_change): ... this one; early return on non-polymorphic
7788         types.
7789         (detect_type_change_ssa): Add comp_type parameter; update
7790         use of detect_type_change.
7791         (compute_complex_assign_jump_func): Add param_type parameter;
7792         update use of detect_type_change_ssa.
7793         (compute_complex_ancestor_jump_func): Likewise.
7794         (ipa_get_callee_param_type): New function.
7795         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7796         update calls to the jump function computation functions.
7798 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7799             Steven Bosscher  <steven@gcc.gnu.org>
7801         * cfgrtl.c (fixup_new_cold_bb): New routine.
7802         (commit_edge_insertions): Invoke fixup_partitions.
7803         (find_partition_fixes): New routine.
7804         (fixup_partitions): Ditto.
7805         (verify_hot_cold_block_grouping): Update comments.
7806         (rtl_verify_edges): Invoke find_partition_fixes.
7807         (rtl_verify_bb_pointers): Update comments.
7808         (rtl_verify_bb_layout): Ditto.
7809         * basic-block.h (probably_never_executed_edge_p): Declare.
7810         (fixup_partitions): Ditto.
7811         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7812         * bb-reorder.c (sanitize_hot_paths): New function.
7813         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7814         sanitize_hot_paths.
7815         * predict.c (probably_never_executed_edge_p): New routine.
7816         * cfg.c (check_bb_profile): Add partition insanity warnings.
7818 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7820         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7822 2013-08-30  Marek Polacek  <polacek@redhat.com>
7824         * Makefile.in (ubsan.o): Add.
7825         (c-family/c-ubsan.o): Add.
7826         (builtins.o): Add ubsan.h dependency.
7827         * ubsan.h: New file.
7828         * ubsan.c: New file.
7829         * common.opt: Add -fsanitize=undefined option.
7830         (flag_sanitize): Add variable.
7831         (fsanitize=): Add option.  Add Driver.
7832         (fsanitize=thread): Remove option.
7833         (fsanitize=address): Likewise.
7834         (static-libubsan): New option.
7835         * doc/invoke.texi: Document the new flag and -static-libubsan.
7836         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7837         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7838         * builtin-attrs.def (ATTR_COLD): Define.
7839         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7840         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7841         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7842         * flag-types.h (sanitize_code): New enum.
7843         * opts.c (common_handle_option): Parse command line arguments
7844         of -fsanitize=.  Add -fsanitize=unreachable option.
7845         * varasm.c (get_variable_section): Adjust.
7846         (assemble_noswitch_variable): Likewise.
7847         (assemble_variable): Likewise.
7848         (output_constant_def_contents): Likewise.
7849         (categorize_decl_for_section): Likewise.
7850         (place_block_symbol): Likewise.
7851         (output_object_block): Likewise.
7852         * builtins.def: Likewise.
7853         * toplev.c (compile_file): Likewise.
7854         (process_options): Likewise.
7855         * cppbuiltin.c: Likewise.
7856         * tsan.c (tsan_pass): Likewise.
7857         (tsan_gate): Likewise.
7858         (tsan_gate_O0): Likewise.
7859         * cfgexpand.c (partition_stack_vars): Likewise.
7860         (expand_stack_vars): Likewise.
7861         (defer_stack_allocation): Likewise.
7862         (expand_used_vars): Likewise.
7863         * cfgcleanup.c (old_insns_match_p): Likewise.
7864         * asan.c (asan_finish_file): Likewise.
7865         (asan_instrument): Likewise.
7866         (gate_asan): Likewise.
7867         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7868         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7869         (asan_global_struct): Use pointer_sized_int_node instead
7870         calling build_nonstandard_integer_type.
7871         (initialize_sanitizer_builtins): Likewise.
7872         (asan_finish_file): Likewise.
7873         * gcc.c: Document %{%:function(args):X}.
7874         (static_spec_functions): Add sanitize.
7875         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7876         store funcval != NULL there.
7877         (do_spec_1): Adjust handle_spec_function caller.
7878         (handle_braces): Allow %:function(args) as condition.
7879         (sanitize_spec_function): New function.
7880         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7881         (LIBUBSAN_SPEC): Likewise.
7882         (LIBUBSAN_EARLY_SPEC): Likewise.
7883         (SANITIZER_SPEC): Handle libubsan.
7884         (SANITIZER_EARLY_SPEC): Likewise.
7885         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7886         instead of fsanitize=address.
7887         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7888         instead of fsanitize=address*.
7889         * builtins.c: Include ubsan.h.
7890         (fold_builtin_0): Instrument __builtin_unreachable.
7891         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7892         instead of flag_asan.
7893         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7894         (pointer_sized_int_node): Define.
7895         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7897 2013-08-30  Mike Stump  <mikestump@comcast.net>
7899         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7900         with RE patterns.
7902 2013-08-29  Jan Hubicka  <jh@suse.cz>
7904         * cgraph.c (cgraph_function_body_availability): Handle weakref
7905         correctly.
7906         * passes.def: Remove pass_fixup_cfg.
7907         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7908         track when we need to remove functions.
7909         (gate_ipa_inline): Execute inlining always; add comment why.
7910         (pass_data_ipa_inline): Remove TODO_remove_functions.
7911         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7912         do not produce summaries.
7913         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7914         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7915         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7916         constant pool and vtable.
7918 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7920         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7921         New arm_neon.h's internal macros to specify 64-bit constants.
7922         Avoid using stdint.h's macros.
7924 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7926         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7927         before applying REG_P.
7929 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7931         PR tree-optimization/58277
7932         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7933         after seeing too many stmts with vdef in between dombb and current
7934         bb, invalidate everything.
7936 2013-08-30  Richard Biener  <rguenther@suse.de>
7938         * fold-const.c (fold_single_bit_test): Fix overflow test.
7940 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7942         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7943         and which can live in a register, always retrieve the value on entry.
7944         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7945         passed by invisible reference specially.
7946         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7947         (vt_add_function_parameter): Correctly deal with a parameter passed by
7948         invisible reference.
7950 2013-08-30  Jan Hubicka  <jh@suse.cz>
7952         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7954 2013-08-30  Richard Biener  <rguenther@suse.de>
7956         PR tree-optimization/58228
7957         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7958         allow invariant loads in nested loop vectorization.
7960 2013-08-30  Richard Biener  <rguenther@suse.de>
7962         PR tree-optimization/58223
7963         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7964         (has_anti_or_output_dependence): ... this and adjust to also
7965         look for output dependences.
7966         (mark_nodes_having_upstream_mem_writes): Adjust.
7967         (rdg_flag_uses): Likewise.
7969 2013-08-30  Richard Biener  <rguenther@suse.de>
7971         PR tree-optimization/58010
7972         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7973         assert that we have a loop-closed PHI.
7975 2013-08-29  Jan Hubicka  <jh@suse.cz>
7977         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7978         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7979         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7980         (lto_free_function_in_decl_state_for_node): New function.
7982 2013-08-29  Xinliang David Li  <davidxl@google.com>
7984         * loop-unroll.c (report_unroll_peel): Minor message change.
7985         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7986         Emit alignment peeling message with default -fopt-info.
7987         (vect_loop_versioning): Emit loop version info message.
7988         * tree-vectorizer.c (vectorize_loops): Minor message change.
7989         (execute_vect_slp): Ditto.
7991 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7993         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7994         of the clone from the DECL_NAME of the original function.
7996 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7998         * passes.c (register_pass): Add overload.
7999         * tree-pass.h (register_pass): Forward declare it.  Add comment.
8001 2013-08-29  Jan Hubicka  <jh@suse.cz>
8003         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
8004         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
8005         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
8006         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8007         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
8008         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
8009         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8010         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
8012 2013-08-29  Teresa Johnson  <tejohnson@google.com>
8014         * dumpfile.c (dump_loc): Output column number.
8015         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
8016         * doc/invoke.texi: Document optall -fopt-info flag.
8017         * profile.c (read_profile_edge_counts): Use new dump framework.
8018         (compute_branch_probabilities): Ditto.
8019         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8020         when pass not in any opt group.
8021         * pass_manager.h (pass_manager::get_pass_profile): New method.
8022         * value-prof.c (check_counter): Use new dump framework.
8023         (check_ic_target): Ditto.
8024         * coverage.c (get_coverage_counts): Ditto.
8025         (coverage_init): Setup new dump framework.
8027 2013-08-29  Richard Biener  <rguenther@suse.de>
8029         PR tree-optimization/58246
8030         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8031         handle the dominance check inside a basic-block.
8033 2013-08-29  Richard Biener  <rguenther@suse.de>
8035         PR middle-end/57287
8036         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8037         of default defs that appear in abnormal PHI nodes.
8039 2013-08-29  Richard Biener  <rguenther@suse.de>
8041         PR tree-optimization/57685
8042         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8043         single-use operands to avoid exponential complexity.
8045 2013-08-28  Dehao Chen  <dehao@google.com>
8047         * ipa-inline.c (edge_badness): Fix integer underflow.
8049 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8051         * gtm-builtins.def (_ITM_free): Declare leaf.
8053 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8055         PR target/58067
8056         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8057         (*tls_local_dynamic_base_64_largepic): Likewise.
8058         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8059         Remove predicate from call operand.
8060         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8061         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8063 2013-08-28  Jeff Law  <law@redhat.com>
8065         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8066         checks for the number of predecessors and successors allowed.
8067         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8068         which require copying a joiner block if there is a request which
8069         is a subpath that requires no joiner block copying.
8071 2013-08-28  Jan Hubicka  <jh@suse.cz>
8073         * lto-streamer-out.c (DFS_write_tree_body): Drop
8074         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8075         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8076         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8077         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8078         Do not read DECL_ERROR_ISSUED.
8079         (unpack_ts_decl_with_vis_value_fields): Do not read
8080         DECL_DEFER_OUTPUT.
8081         (lto_input_ts_binfo_tree_pointers): Do not read
8082         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8083         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8084         write DECL_ERROR_ISSUED..
8085         (pack_ts_decl_with_vis_value_fields): Do not write
8086         DECL_DEFER_OUTPUT.
8087         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8088         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8089         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8090         * tree.h (tree_decl_common): Update comment.
8091         (DECL_ERROR_ISSUED): Remove.
8093 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8095         PR middle-end/58257
8096         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8098 2013-08-28  Jan Hubicka  <jh@suse.cz>
8100         * builtins.def (free): Declare leaf.
8102 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8104         * gdbhooks.py: New.
8105         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8106         * configure: Regenerate.
8108 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8110         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8111         (ipa_ancestor_jf_data): Likewise.
8112         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8113         (ipa_get_jf_pass_through_type_preserved): New function.
8114         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8115         (ipa_get_jf_ancestor_type_preserved): New function.
8116         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8117         (ipa_get_jf_ancestor_result): Likewise.
8118         (propagate_vals_accross_pass_through): Use
8119         ipa_get_jf_pass_through_result to do all the value mappings.
8120         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8121         type_preserved flag.
8122         (ipa_set_jf_cst_copy): New function.
8123         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8124         (ipa_set_jf_arith_pass_through): Likewise.
8125         (ipa_set_ancestor_jf): Likewise.
8126         (compute_complex_assign_jump_func): Set type_preserved instead of
8127         punting.
8128         (ipa_compute_jump_functions_for_edge): Likewise.
8129         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8130         (update_jump_functions_after_inlining): Update type_preserved.
8131         Explicitely create jump functions when combining one with pass_through.
8132         (ipa_write_jump_function): Stream the type_preserved flags.
8133         (ipa_read_jump_function): Likewise.
8135 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8136             Aldy Hernandez  <aldyh@redhat.com>
8138         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8139         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8140         * function.h (struct function): Add has_force_vect_loops and
8141         has_simduid_loops.
8142         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8143         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8144         handle it.
8145         * gimple.def: Update CLAUSES comments.
8146         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8147         (gimple_build_omp_for): Add argument to prototype.
8148         (gimple_omp_for_kind): New.
8149         (gimple_omp_for_set_kind): New.
8150         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8151         GOVD_DATA_SHARE_CLASS.
8152         (enum omp_region_type): Add ORT_SIMD.
8153         (gimple_add_tmp_var): Handle ORT_SIMD.
8154         (gimplify_var_or_parm_decl): Same.
8155         (is_gimple_stmt): Same.
8156         (omp_firstprivatize_variable): Same.
8157         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8158         (omp_notice_variable): Handle ORT_SIMD.
8159         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8160         (omp_check_private): Handle ORT_SIMD.
8161         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8162         OMP_CLAUSE_SAFELEN.
8163         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8164         Handle OMP_CLAUSE_LASTPRIVATE.
8165         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8166         OMP_CLAUSE_SAFELEN.
8167         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8168         (gimplify_expr): Handle OMP_SIMD.
8169         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8170         (expand_GOMP_SIMD_VF): New.
8171         (expand_GOMP_SIMD_LAST_LANE): New.
8172         * internal-fn.def (GOMP_SIMD_LANE): New.
8173         (GOMP_SIMD_VF): New.
8174         (GOMP_SIMD_LAST_LANE): New.
8175         * omp-low.c: Include target.h.
8176         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8177         OMP_CLAUSE_SAFELEN.
8178         (check_omp_nesting_restrictions): Same.
8179         (omp_max_vf): New.
8180         (lower_rec_simd_input_clauses): New.
8181         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8182         OMP_CLAUSE_LINEAR.
8183         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8184         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8185         (expand_omp_build_assign): New.
8186         (expand_omp_for_init_counts): New.
8187         (expand_omp_for_init_vars): New.
8188         (extract_omp_for_update_vars): New.
8189         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8190         and rewrite accordingly.
8191         (expand_omp_simd): New.
8192         (expand_omp_for): Use expand_omp_simd.
8193         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8194         (lower_omp_for): Do not lower the body.
8195         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8196         in their own loops.
8197         * tree-flow.h (find_omp_clause): Remove prototype.
8198         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8199         forcing vectorization of the loop, or if flag_tree_vectorize.
8200         (gate_tree_if_conversion): Similarly.
8201         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8202         gimple_build_omp_for.
8203         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8204         * tree-parloops (create_parallel_loop): Pass kind argument to
8205         gimple_build_omp_for.
8206         * tree-pretty-print.c (dump_omp_clause): Add cases for
8207         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8208         OMP_CLAUSE__SIMDUID_.
8209         (dump_generic_node): Handle OMP_SIMD.
8210         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8211         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8212         unroll OMP_SIMD loops here.
8213         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8214         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8215         loop->safelen.
8216         (vect_analyze_data_refs): Handle simd loops.
8217         * tree-vect-loop.c (vectorizable_live_operation): Handle
8218         IFN_GOMP_SIMD*.
8219         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8220         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8221         (vectorizable_load): Same.
8222         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8223         (struct simduid_to_vf): New.
8224         (simduid_to_vf::hash): New.
8225         (simduid_to-vf::equal): New.
8226         (struct simd_array_to_simduid): New.
8227         (simd_array_to_simduid::hash): New.
8228         (simd_array_to_simduid::equal): New.
8229         (adjust_simduid_builtins): New.
8230         (struct note_simd_array_uses_struct): New.
8231         (note_simd_array_uses_cb): New.
8232         (note_simd_array_uses): New.
8233         (vectorize_loops): Handle simd hints and adjust simd builtins
8234         accordingly.
8235         * tree-vectorizer.h (struct _stmt_vec_info): Add
8236         simd_lane_access_p field.
8237         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8238         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8239         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8240         (omp_clause_code_name): Same.
8241         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8242         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8243         * tree.def (OMP_SIMD): New entry.
8244         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8245         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8246         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8247         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8248         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8249         (OMP_CLAUSE_LINEAR_STEP): New.
8250         (OMP_CLAUSE_SAFELEN_EXPR): New.
8251         (OMP_CLAUSE__SIMDUID__DECL): New.
8252         (find_omp_clause): New prototype.
8254 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8256         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8257         Haswell processor detection.
8259 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8261         PR target/57927
8262         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8263         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8264         AVX2 capable processors.
8266 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8268         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8269         of vget_low_* with implementations in terms of other intrinsics.
8271 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8273         PR middle-end/57219
8274         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8275         values to -1, 0 and 1.
8277 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8279         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8280         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8281         (<optab><mode>3_insn): Remove.
8282         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8283         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8284         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8285         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8286         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8287         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8288         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8289         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8290         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8291         (ror<mode>3_insn): Likewise.
8292         * config/aarch64/predicates.md (aarch64_simd_register): New.
8294 2013-08-27  Richard Biener  <rguenther@suse.de>
8296         PR tree-optimization/57521
8297         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8298         one edge is non-critical.
8299         (find_phi_replacement_condition): Make sure to use a non-critical
8300         edge.  Cleanup and remove old bug workarounds.
8301         (bb_postdominates_preds): Remove.
8302         (if_convertible_loop_p_1): Do not compute post-dominators.
8303         (combine_blocks): Do not free post-dominators.
8304         (main_tree_if_conversion): Likewise.
8305         (pass_data_if_conversion): Add TODO_verify_ssa.
8307 2013-08-27  DJ Delorie  <dj@redhat.com>
8309         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8311 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8313         * function.c (assign_parm_find_data_types): Set passed_mode and
8314         nominal_mode to the TYPE_MODE of nominal_type for the built
8315         pointer type in case of the struct-pass-by-reference.
8317 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8319         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8320         (UINT16_TYPE): Change default to "unsigned int".
8322         * config/avr/avr.opt (mfract-convert-truncate): New option.
8323         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8324         is set, round negative fractional integers according to n1169
8325         when converting to integer types.
8327 2013-08-26  Jan Hubicka  <jh@suse.cz>
8329         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8330         methods can not be called indirectly when their address is not taken.
8332 2013-08-26  Jan Hubicka  <jh@suse.cz>
8334         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8335         ctor_for_folding.
8337 2013-08-26  Jan Hubicka  <jh@suse.cz>
8339         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8340         can be unshared.
8342 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8344         * reload.c (find_valid_class): Allow classes that do not include
8345         FIRST_PSEUDO_REGISTER - 1.
8347 2013-08-26  Jan Hubicka  <jh@suse.cz>
8349         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8350         fix edge count/frequency when speculation failed; fix type check
8351         for the direct call.
8353 2013-08-26  Jan Hubicka  <jh@suse.cz>
8355         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8357 2013-08-26  Jan Hubicka  <jh@suse.cz>
8359         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8360         to be changed by edge redirection.
8362 2013-08-26  Jan Hubicka  <jh@suse.cz>
8364         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8365         formating; add sanity check.
8366         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8367         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8368         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8369         (resolve_noninline_speculation): Update callee keys, too.
8371 2013-08-26  Jan Hubicka  <jh@suse.cz>
8373         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8374         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8376 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8378         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8379         into proper place.
8381 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8383         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8384         (x86_64_elf_select_section): Ditto.
8385         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8386         arguments.
8387         (ix86_pass_by_reference): Ditto.
8388         (output_set_got): Ditto.
8389         (ix86_unary_operator_ok): Ditto.
8390         (ix86_expand_builtin): Ditto.
8392 2013-08-23  Jan Hubicka  <jh@suse.cz>
8394         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8396 2013-08-23  Jan Hubicka  <jh@suse.cz>
8398         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8399         (tree_decl_with_vis): Add FINAL field.
8401 2013-08-23  Jeff Law  <law@redhat.com>
8403         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8404         the debugging dump when the expression is fully redundant.
8406 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8408         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8409         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8410         * pretty-print.c (pp_formatted_text_data): Likewise.
8411         (pp_write_text_to_stream): Likewise.
8412         (pp_write_text_as_dot_label_to_stream): Likewise.
8413         (pp_append_r): Likewise.
8414         (pp_format): Likewise.
8415         (pp_flush): Likewise.
8416         (pp_clear_output_area): Likewise.
8417         (pp_append_text): Likewise.
8418         (pp_formatted_text): Likewise.
8419         (pp_remaining_character_count_for_line): Likewise.
8420         (pp_newline): Likewise.
8421         (pp_character): Likewise.
8422         (output_buffer::~output_buffer): Define.
8423         (pretty_printer::~pretty_printer): Destruct output buffer.
8424         * pretty-print.h (output_buffer::~output_buffer): Declare.
8425         (pretty_printer::~pretty_printer): Declare virtual.
8427 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8429         PR other/57324
8430         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8431         HOST_WIDE_INT_M1U): New macros.
8432         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8433         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8434         unsigned -1 for lshift.
8435         * cse.c (cse_insn): Likewise.
8436         * double-int.c (rshift_double, lshift_double): Likewise.
8437         * builtins.c (fold_builtin_bitop): Likewise.
8438         * combine.c (force_to_mode): Likewise.
8439         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8440         * simplify-rtx.c (simplify_const_unary_operation,
8441         simplify_const_binary_operation): Likewise.
8442         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8443         check_va_list_escapes): Likewise.
8444         * rtlanal.c (nonzero_bits1): Likewise.
8445         * expmed.c (expand_smod_pow2): Likewise.
8446         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8448 2013-08-23  Jan Hubicka  <jh@suse.cz>
8450         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8451         as having address taken.
8453 2013-08-23  Jan Hubicka  <jh@suse.cz>
8455         * ipa-utils.h (method_class_type): Declare.
8456         * ipa-devirt.c (method_class_type): Export.
8458         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8459         do not walk base classes of anonymous types.
8461 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8463         PR rtl-optimization/58220
8464         PR regression/58221
8465         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8466         handle SEQUENCE insns properly.
8468 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8470         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8471         definition.
8472         (pp_newline_and_indent): Likewise.
8473         (pp_separate_with): Likewise.
8474         * pretty-print.c (pp_newline_and_flush): Define.
8475         (pp_newline_and_indent): Likewise.
8476         (pp_separate_with): Likewise.
8478 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8480         PR target/58218
8481         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8482         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8484 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8486         * config/i386/predicates.md (ext_sse_reg_operand): New.
8487         * config/i386/i386.md (*movti_internal): Use
8488         predicate to determine if EVEX is needed.
8489         (*movsi_internal): Ditto.
8490         (*movdf_internal): Ditto.
8491         (*movsf_internal): Ditto.
8492         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8494 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8496         PR tree-optimization/58209
8497         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8498         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8499         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8500         emit POINTER_PLUS_EXPR.
8501         (create_tailcall_accumulator): For pointer result type accumulate in
8502         sizetype type.
8504 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8506         * configure.ac: Add backslashes missing from the last change.
8507         * configure: Regenerate.
8509 2013-08-22  Jan Hubicka  <jh@susue.cz>
8511         * ipa.c (function_and_variable_visibility): First remember function
8512         was global and then make it local.
8514 2013-08-22  Julian Brown  <julian@codesourcery.com>
8516         * configure.ac: Add aarch64 to list of arches which use "nop" in
8517         debug_line test.
8518         * configure: Regenerate.
8520 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8522         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8523         gnu_libc_has_function.
8524         * config/s390/tpf.h: Likewise.
8526 2013-08-22  Jan Hubicka  <jh@susue.cz>
8528         * timevar.c (validate_phases): Add cast.
8530 2013-08-22  Jan Hubicka  <jh@susue.cz>
8532         * timevar.c (validate_phases): Use size_t for memory.
8533         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8535 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8537         * pretty-print.h (output_buffer::output_buffer): Declare.
8538         (pretty_printer::pretty_printer): Likewise.
8539         (pp_construct): Remove.
8540         * pretty-print.c (output_buffer::output_buffer): Define.
8541         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8542         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8543         (print_gimple_expr): Likewise.
8544         (print_gimple_seq): Likewise.
8545         (gimple_dump_bb): Likewise.
8546         * sched-vis.c (dump_value_slim): Likewise.
8547         (dump_insn_slim): Likewise.
8548         (dump_rtl_slim): Likewise.
8549         (str_pattern_slim): Likewise.
8550         * tree-mudflap.c (mf_varname_tree): Likewise.
8551         * graph.c (print_graph_cfg): Likewise.
8552         (start_graph_dump): Likewise.
8553         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8554         placement-new.
8555         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8556         pretty printer initialization.
8557         * coretypes.h (diagnostic_context): Remove superflous type alias
8558         declaration.
8559         (pretty_printer): Likewise.  Declare directly as a class.
8560         (pretty_print_info): Remove declaration as class.
8561         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8562         and pp_clear_output_area.
8563         (asan_add_global): Likewise.
8565 2013-08-22  Jan Hubicka  <jh@suse.cz>
8567         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8568         * ipa-utils.h (update_type_inheritance_graph): Declare.
8569         (possible_polymorphic_call_target_p): Declare.
8570         (possible_polymorphic_call_target_p): New.
8571         * ipa-devirt.c: Update toplevel comments.
8572         (cached_polymorphic_call_targets): Move up.
8573         (odr_type_d): Move ID down.
8574         (polymorphic_type_binfo_p): Update comment.
8575         (odr_hasher::remove): Likewise;
8576         (get_odr_type): Set anonymous_namespace.
8577         (dump_odr_type): Dump it.
8578         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8579         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8580         (record_binfo): Add comment.
8581         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8582         built.
8583         (devirt_node_removal_hook): Do not iCE when cache is freed.
8584         (possible_polymorphic_call_target_p): New predicate.
8585         (update_type_inheritance_graph): New function.
8587 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8588             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8589             Sergey Lega  <sergey.s.lega@intel.com>
8590             Anna Tikhonova  <anna.tikhonova@intel.com>
8591             Ilya Tocar  <ilya.tocar@intel.com>
8592             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8593             Ilya Verbin  <ilya.verbin@intel.com>
8594             Kirill Yukhin  <kirill.yukhin@intel.com>
8595             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8597         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8598         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8599         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8600         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8601         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8602         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8603         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8604         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8605         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8606         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8607         OPT_mavx512pf, OPT_mavx512er cases.
8608         * config/i386/constraints.md (v): New constraint.
8609         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8610         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8611         (bit_AVX512CD): New.
8612         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8613         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8614         * config/i386/i386-c.c (ix86_target_macros_internal):
8615         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8616         __AVX512PF__.
8617         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8618         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8619         * config/i386/i386.c (regclass_map, dbx_register_map)
8620         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8621         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8622         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8623         -mavx512pf options.
8624         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8625         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8626         -mavx512pf options.  Fix formatting.
8627         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8628         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8629         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8630         -mavx512cd, -mavx512pf options.
8631         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8632         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8633         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8634         ALL_SSE_REGS.
8635         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8636         (ix86_set_reg_reg_cost): Ditto.
8637         (x86_order_regs_for_local_alloc): Ditto.
8638         (MAX_VECT_LEN): Extend to 64-byte.
8639         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8640         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8641         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8642         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8643         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8644         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8645         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8646         (SSE_REG_MODE_P): Support new modes.
8647         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8648         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8649         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8650         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8651         (REG_CLASS_CONTENTS): Add new registers.
8652         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8653         (EXT_REX_SSE_REGNO_P): New.
8654         (HI_REGISTER_NAMES): Add new registers.
8655         * config/i386/i386.md: Define constants for new registers.
8656         (mode): Add new 512-bit modes.
8657         (prefix): Support evex prefix.
8658         (isa): Support avx512f, noavx512f, fma_avx512f.
8659         (ssemodesuffix): Add new 512-bit modes.
8660         (movxi): New.
8661         (*movxi_internal_avx512f): Ditto.
8662         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8663         Support MODE_XI.
8664         (*movsi_internal): Likewise.
8665         (*movdf_internal): Likewise.
8666         (*movsf_internal): Likewise.
8667         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8668         (<code><mode>3): Likewise.
8669         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8670         New.
8671         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8672         with the new constraint "v".
8673         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8674         modes.
8675         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8676         with the new constraint "v".
8677         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8678         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8679         (*<plusminus_insn><mode>3): Likewise.
8680         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8681         (*mul<mode>3): Likewise.
8682         (<sse>_vmmul<mode>3): Likewise.
8683         (<sse>_div<mode>3): Likewise.
8684         (<sse>_vmdiv<mode>3): Likewise.
8685         (<sse>_sqrt<mode>2): Likewise.
8686         (<sse>_vmsqrt<mode>2): Likewise.
8687         (*<code><mode>3_finite): Likewise.
8688         (*<code><mode>3) <smaxmin>: Likewise.
8689         (<sse>_vm<code><mode>3): Likewise.
8690         (*<code><mode>3) <any_logic>: Likewise.
8691         (*fma_fmadd_<mode>): Likewise.
8692         (*fma_fmsub_<mode>): Likewise.
8693         (*fma_fnmadd_<mode>): Likewise.
8694         (*fma_fnmsub_<mode>): Likewise.
8695         (*fma_fmaddsub_<mode>): Likewise.
8696         (*fma_fmsubadd_<mode>): Likewise.
8697         (*fmai_fmadd_<mode>): Likewise.
8698         (*fmai_fmsub_<mode>): Likewise.
8699         (*fmai_fnmadd_<mode>): Likewise.
8700         (*fmai_fnmsub_<mode>): Likewise.
8701         (sse_cvtsi2ss): Likewise.
8702         (sse_cvtsi2ssq): Likewise.
8703         (sse_cvtss2si): Likewise.
8704         (sse_cvtss2si_2): Likewise.
8705         (sse_cvtss2siq): Likewise.
8706         (sse_cvtss2siq_2): Likewise.
8707         (sse_cvttss2si): Likewise.
8708         (sse_cvtss2siq_2): Likewise.
8709         (float<sseintvecmodelower><mode>2): Likewise.
8710         (sse2_cvtsd2si_2): Likewise.
8711         (sse2_cvtsd2siq_2): Likewise.
8712         (*<plusminus_insn><mode>3): Likewise.
8713         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8714         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8715         (ashr<mode>3): Likewise.
8716         (<shift_insn><mode>3): Likewise.
8717         (avx2_<code><mode>3): Likewise.
8718         (*avx2_<code><mode>3): Likewise.
8719         (*andnot<mode>3): Likewise.
8720         (*<code><mode>3) <any_logic>: Likewise.
8721         (abs<mode>2): Likewise.
8722         (avx2_permvar<mode>): Likewise.
8723         (avx2_perm<mode>_1): Likewise.
8724         (*avx_vpermilp<mode>): Likewise.
8725         (avx_vpermilvar<mode>3): Likewise.
8726         (avx2_ashrv<mode>): Likewise.
8727         (avx2_<shift_insn>v<mode>): Likewise.
8728         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8729         -mavx512cd.
8730         * doc/rtl.texi: Document XImode.
8732 2013-08-21  Jeff Law  <law@redhat.com>
8734         * tree-flow.h (register_jump_thread): Pass vector of edges
8735         instead of each important edge.
8736         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8737         thread path into a vector and pass that to register_jump_thread.
8738         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8739         passed in edge vector to the current 3-edge form.
8741         Revert:
8742         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8744         * dce.c (fini_dce): Call df_analyze again just in case
8745         delete_unmarked_insns removed anything.
8747 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8749         * reload.h (struct reg_equivs): Rename to ..
8750         (struct reg_equivs_s): .. this.
8752 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8754         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8756 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8758         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8760 2013-08-21  Jeff Law  <law@redhat.com>
8762         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8763         simplify assignments too.  If the RHS collapses to a singleton
8764         range, then return the value for the range.
8766 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8768         * config/i386/sse.md (V16): Rename to...
8769         (VMOVE): this.
8770         (mov<mode>): Update iterator name.
8771         (*mov<mode>_internal): Ditto.
8772         (push<mode>1): Ditto.
8773         (movmisalign<mode>): Ditto.
8775 2013-08-20  Jan Hubicka  <jh@suse.cz>
8777         PR bootstrap/58186
8778         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8779         entry for direct edges.
8780         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8782 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8784         Revert my last two changes, r201865 and r201864:
8786         Revert r201865:
8787         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8789         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8790         instances can own GC refs.
8792         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8793         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8794         (gcc::context::gt_pch_nx): Likewise.
8795         (gcc::context::gt_pch_nx):  Likewise.
8796         * ggc.h (gt_ggc_mx <T>): New.
8797         (gt_pch_nx_with_op <T>): New.
8798         (gt_pch_nx <T>): New.
8799         * passes.c (opt_pass::gt_ggc_mx): New.
8800         (opt_pass::gt_pch_nx): New.
8801         (opt_pass::gt_pch_nx_with_op): New.
8802         (pass_manager::gt_ggc_mx): New.
8803         (pass_manager::gt_pch_nx): New.
8804         (pass_manager::gt_pch_nx_with_op): New.
8805         (pass_manager::operator new): Use
8806         ggc_internal_cleared_alloc_stat rather than xcalloc.
8807         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8808         (pass_manager::gt_ggc_mx): New.
8809         (pass_manager::gt_pch_nx): New.
8810         (pass_manager::gt_pch_nx_with_op): New.
8811         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8812         (opt_pass::operator new): New.
8813         (opt_pass::gt_ggc_mx): New.
8814         (opt_pass::gt_pch_nx): New.
8815         (opt_pass::gt_pch_nx_with_op): New.
8817         Revert r201864:
8818         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8820         * Makefile.in (GTFILES): Add context.h.
8821         * context.c (gcc::context::operator new): New.
8822         (gcc::context::gt_ggc_mx): New.
8823         (gcc::context::gt_pch_nx): New.
8824         (gcc::context::gt_pch_nx): New.
8825         * context.h (gcc::context): Add GTY((user)) marking.
8826         (gcc::context::operator new): New.
8827         (gcc::context::gt_ggc_mx): New.
8828         (gcc::context::gt_pch_nx): New.
8829         (gcc::context::gt_pch_nx): New.
8830         (g): Add GTY marking.
8831         (gt_ggc_mx (gcc::context *)): New.
8832         (gt_pch_nx (gcc::context *)): New.
8833         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8834         void *cookie)): New.
8835         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8837 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8839         * dce.c (fini_dce): Call df_analyze again just in case
8840         delete_unmarked_insns removed anything.
8842 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8844         PR rtl-optimizations/57451
8845         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8846         from crossing split section boundaries.
8848 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8850         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8851         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8852         with MULTLIB_DEFAULTS.
8854 2013-08-20  Nick Clifton  <nickc@redhat.com>
8856         * target.def (narrow_volatile_bitfield): Note that the default
8857         value is false, not !TARGET_STRICT_ALIGN.
8858         * doc/tm.texi: Regenerate.
8860 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8862         Fix LIB_SPEC for systems without libpthread.
8864         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8865         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8866         for Android.
8867         * config/i386/linux-common.h: Likewise.
8868         * config/mips/linux-common.h: Likewise.
8870 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8872         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8873         checks.
8875 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8877         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8878         instances can own GC refs.
8880         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8881         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8882         (gcc::context::gt_pch_nx): Likewise.
8883         (gcc::context::gt_pch_nx):  Likewise.
8884         * ggc.h (gt_ggc_mx <T>): New.
8885         (gt_pch_nx_with_op <T>): New.
8886         (gt_pch_nx <T>): New.
8887         * passes.c (opt_pass::gt_ggc_mx): New.
8888         (opt_pass::gt_pch_nx): New.
8889         (opt_pass::gt_pch_nx_with_op): New.
8890         (pass_manager::gt_ggc_mx): New.
8891         (pass_manager::gt_pch_nx): New.
8892         (pass_manager::gt_pch_nx_with_op): New.
8893         (pass_manager::operator new): Use
8894         ggc_internal_cleared_alloc_stat rather than xcalloc.
8895         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8896         (pass_manager::gt_ggc_mx): New.
8897         (pass_manager::gt_pch_nx): New.
8898         (pass_manager::gt_pch_nx_with_op): New.
8899         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8900         (opt_pass::operator new): New.
8901         (opt_pass::gt_ggc_mx): New.
8902         (opt_pass::gt_pch_nx): New.
8903         (opt_pass::gt_pch_nx_with_op): New.
8905 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8907         * Makefile.in (GTFILES): Add context.h.
8908         * context.c (gcc::context::operator new): New.
8909         (gcc::context::gt_ggc_mx): New.
8910         (gcc::context::gt_pch_nx): New.
8911         (gcc::context::gt_pch_nx): New.
8912         * context.h (gcc::context): Add GTY((user)) marking.
8913         (gcc::context::operator new): New.
8914         (gcc::context::gt_ggc_mx): New.
8915         (gcc::context::gt_pch_nx): New.
8916         (gcc::context::gt_pch_nx): New.
8917         (g): Add GTY marking.
8918         (gt_ggc_mx (gcc::context *)): New.
8919         (gt_pch_nx (gcc::context *)): New.
8920         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8921         void *cookie)): New.
8922         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8924 2013-08-20  Alan Modra  <amodra@gmail.com>
8926         PR target/57865
8927         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8928         (rs6000_emit_epilogue): Likewise.
8930 2013-08-19  Dehao Chen  <dehao@google.com>
8932         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8934 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8935             Jakub Jelinek  <jakub@redhat.com>
8937         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8938         (BUILT_IN_FABSD64): Likewise.
8939         (BUILT_IN_FABSD128): Likewise.
8940         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8941         (fold_builtin_1): Likewise.
8942         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8943         destination and source operands.
8944         (*abstd2_fpr): Likewise.
8945         (*nabstd2_fpr): Likewise.
8947 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8949         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8950         JUMP_P and INSN_P.
8952 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8954         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8955         incomplete.
8957 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8959         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8960         * builtins.c (default_libc_has_function): New.
8961         (gnu_libc_has_function): Ditto.
8962         (no_c99_libc_has_function): Ditto.
8963         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8964         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8965         (fold_builtin_sincos): Likewise.
8966         (fold_builtin_cexp): Likewise.
8967         * builtins.def (DEF_C94_BUILTIN): Likewise.
8968         (DEF_C99_BUILTIN): Likewise.
8969         (DEF_C99_C90RES_BUILTIN): Likewise.
8970         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8971         definitions to using this define.
8972         * config/darwin-protos.h (darwin_libc_has_function): New.
8973         * config/darwin.c (darwin_libc_has_function): Ditto.
8974         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8975         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8976         * config/darwin.h: Ditto.
8977         * config/elfos.h: Ditto.
8978         * config/freebsd.h: Ditto.
8979         * config/i386/cygming.h: Ditto.
8980         * config/i386/djgpp.h: Ditto.
8981         * config/i386/i386-interix.h: Ditto.
8982         * config/microblaze/microblaze.h: Ditto.
8983         * config/mmix/mmix.h: Ditto.
8984         * config/gnu-user.h: Ditto.
8985         * config/ia64/hpux.h: Ditto.
8986         * config/pa/pa-hpux.h: Ditto.
8987         * config/pdp11/pdp11.h: Ditto.
8988         * config/picochip/picochip.h: Ditto.
8989         * config/linux.h: Ditto.
8990         * config/netbsd.h: Ditto.
8991         * config/openbsd.h: Ditto.
8992         * config/rs6000/aix43.h: Ditto.
8993         * config/rs6000/aix51.h: Ditto.
8994         * config/rs6000/aix52.h: Ditto.
8995         * config/rs6000/aix53.h: Ditto.
8996         * config/rs6000/aix61.h: Ditto.
8997         * config/rs6000/darwin.h: Ditto.
8998         * config/rs6000/linux.h: Ditto.
8999         * config/rs6000/linux64.h: Ditto.
9000         * config/s390/tpf.h: Ditto.
9001         * config/sol2-10.h: Ditto.
9002         * config/sol2.h: Ditto.
9003         * config/vms/vms.h: Ditto.
9004         * config/vxworks.h: Ditto.
9005         * config/linux-android.c (linux_android_libc_has_function):
9006         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
9007         * config/linux-protos.h (linux_android_libc_has_function):
9008         New declaration.
9009         * config/i386/i386.c (ix86_libc_has_function): New.
9010         * config/i386/i386-protos.h
9011         (ix86_libc_has_function): New declaration.
9012         * config/i386/i386.md
9013         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
9014         ("isinf<mode>2): Likewise.
9015         * convert.c (convert_to_integer): Using new target hook
9016         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
9017         TARGET_C99_FUNCTIONS.
9018         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9019         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9020         * coretypes.h (function_class): New enum for different
9021         classes of functions.
9022         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9023         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9024         (TARGET_HAS_SINCOS): Likewise.
9025         (TARGET_LIBC_HAS_FUNCTION): New.
9026         * doc/tm.texi: Regenerated.
9027         * targhooks.h (default_libc_has_function): New declaration.
9028         (no_c99_libc_has_function): Ditto.
9029         (gnu_libc_has_function): Ditto.
9030         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9031         and TARGET_HAS_SINCOS.
9033 2013-08-18  Jan Hubicka  <jh@suse.cz>
9035         * Makeifle-in (ipa-devirt.o): New.
9036         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9037         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9038         (analyze_functions): Look into possible targets of polymorphic call.
9039         * dumpfile.c (dump_files): Add type-inheritance dump.
9040         * dumpfile.h (TDI_inheritance): New.
9041         * ipa-devirt.c: New file.
9042         * ipa-utils.h (odr_type_d): Forward declare.
9043         (odr_type): New type.
9044         (build_type_inheritance_graph): Declare.
9045         (possible_polymorphic_call_targets): Declare and introduce inline
9046         variant when only edge is pased.
9047         (dump_possible_polymorphic_call_targets): Likewise.
9048         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9049         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9050         (types_same_for_odr): ... here.
9051         * tree.h (type_in_anonymous_namespace_p): Declare.
9053 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9055         PR tree-optimization/58006
9056         * tree-parloops.c (take_address_of): Don't ICE if get_name
9057         returns NULL.
9058         (eliminate_local_variables_stmt): Remove clobber stmts.
9060 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9062         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9063         error message is issued for an alias to undefined symbol.
9065 2013-08-18  Jan Hubicka  <jh@suse.cz>
9067         * cgraph.c (cgraph_create_indirect_edge): Discover
9068         polymorphic calls and record basic info into indirect_info.
9069         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9070         devirtualization, ignore objc function calls.
9071         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9072         call with no parm index info.
9073         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9074         * tree.c (virtual_method_call_p): New function.
9075         * tree.h (virtual_method_call_p): Declare.
9077 2013-08-16  Jan Hubicka  <jh@suse.cz>
9079         PR middle-end/58179
9080         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9082 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9084         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9085         attribute.
9087 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9089         * gengtype.c (type_for_name): Add special-case support for
9090         locating types within the "gcc::" namespace.
9091         (open_base_files): Emit a "using namespace gcc" directive.
9093 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9095         PR target/58160
9096         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9097         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9099         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9100         array instead of each individual operand as a separate argument.
9101         (emit_fusion_gpr_load): Likewise.
9102         (expand_fusion_gpr_load): Add new function declaration.
9104         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9105         signature to have the operands passed as an array, instead of as
9106         separate arguments.  Allow ZERO_EXTEND to be in the memory
9107         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9108         depend on the register live/dead flags when peepholes are run.
9109         (expand_fusion_gpr_load): New function to be called from the
9110         peephole2 pass, to change the register that addis sets to be the
9111         target register.
9112         (emit_fusion_gpr_load): Change the calling signature to have the
9113         operands passed as an array, instead of as separate arguments.
9114         Allow ZERO_EXTEND to be in the memory address, and also
9115         SIGN_EXTEND if -mpower8-fusion-sign.
9117         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9118         unspec enumeration.
9119         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9120         adjust the register addis loads up in the peephole2 pass.  Do not
9121         depend on the register live/dead state when the peephole pass is done.
9123 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9125         * gengtype.c (create_user_defined_type): Ensure that the kind
9126         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9127         declaration is seen before the GTY((user)) marking.
9129 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9131         PR target/58105
9132         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9134 2013-08-16  Jan Hubicka  <jh@suse.cz>
9136         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9137         arugment expected_type.
9138         (gimple_fold_call): Use it.
9139         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9140         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9141         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9142         (try_make_edge_direct_virtual_call): Likewise.
9143         * tree.c (obj_type_ref_class): New.
9144         * tree.h (obj_type_ref_class): Use it.
9146 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9148         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9149         (rtl_slim_pp): Likewise.
9150         (init_rtl_slim_pretty_print): Likewise.
9151         (dump_value_slim):  Don't call it.  Use local pretty printer.
9152         (dump_insn_slim): Likewise.
9153         (dump_rtl_slim): Likewise.
9154         (str_pattern_slim): Likewise.
9155         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9156         Simplify.
9158 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9160         PR tree-optimization/58164
9161         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9162         walk gimple_goto_dest of GIMPLE_GOTO.
9164         PR tree-optimization/58165
9165         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9166         bi_call must be the last stmt in a bb, don't split_block, instead
9167         use fallthru edge from it and give up if there is none.
9168         Release conds vector when returning early.
9170 2013-08-14  Xinliang David Li  <davidxl@google.com>
9172         * config/i386/i386.c (ix86_option_override_internal):
9173         Remove unused variable and field.
9175 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9177         PR target/57949
9178         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9179         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9180         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9181         and Linux, correct BLKmode alignment when 128-bit alignment is
9182         required and compatibility flag is not set.
9183         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9184         for zero-size arguments when compatibility flag is not set.
9186 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9188         PR tree-optimization/58145
9189         * tree-sra.c (build_ref_for_offset): If prev_base has
9190         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9192 2013-08-14  Xinliang David Li  <davidxl@google.com>
9194         * config/i386/i386.c (ix86_option_override_internal):
9195         Fix uninitialized variable error.
9197 2013-08-14  Xinliang David Li  <davidxl@google.com>
9199         * config/i386/i386.opt: Define two new options.
9200         * config/i386/x86-tune.def: Add arch selector field in macros.
9201         * config/i386/i386.h: Adjust macro definition.
9202         * config/i386/i386.c (ix86_option_override_internal):
9203         Refactor the code.
9204         (parse_mtune_ctrl_str): New function.
9205         (set_ix86_tune_features): New function.
9206         (ix86_function_specific_restore): Call the new helper function.
9208 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9210         PR rtl-optimization/57662
9211         * sel-sched.c (code_motion_process_successors): When the current insn
9212         is removed after the recursive traversal, break from the loop.
9213         Add comments and debug printouts.
9215 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9216             Alexandre Oliva  <aoliva@redhat.com>
9218         PR target/58067
9219         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9220         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9221         there also UNSPEC_PLTOFF.
9223 2013-08-14  Marek Polacek  <polacek@redhat.com>
9225         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9226         NUM_CONDITIONS bit positions.
9228 2013-08-13  Cary Coutant  <ccoutant@google.com>
9230         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9231         (attr_checksum): Hash vector contents instead of pointer.
9232         (attr_checksum_ordered): Likewise.
9234 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9236         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9237         when Pmode != word_mode.  Add length_address attribute.
9238         (sse3_monitor_<mode>): Merge from sse3_monitor and
9239         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9240         Pmode != word_mode.  Update insn length attribute.
9241         * config/i386/i386.c (ix86_option_override_internal): Update
9242         ix86_gen_monitor selection for merged sse3_monitor insn.
9244 2013-08-13  Julian Brown  <julian@codesourcery.com>
9246         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9247         perform invalid legitimization on greater-than-word-size modes for
9248         TARGET_E500_DOUBLE.
9250 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9252         * ira.c (setup_class_translate_array): Use aclass instead of cl
9253         for classes not fully covered by allocno classes.
9255 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9257         PR tree-optimization/57661
9258         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9259         * tree-inline.c (tree_function_versioning): Initialize it.
9260         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9261         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9262         that is not being copied.
9264         PR sanitizer/56417
9265         * asan.c (instrument_strlen_call): Fix typo in comment.
9266         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9268 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9270         * config/mips/mips.md (prefetch): Use lw instead of ld on
9271         loongson in 32bit mode.
9273 2013-08-13  Nick Clifton  <nickc@redhat.com>
9275         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9277 2013-08-13  Jan Hubicka  <jh@suse.cz>
9279         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9280         introduced edge; fix typo in sanity check.
9281         (cgraph_resolve_speculation): Export; improve diagnostic.
9282         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9283         speculation at type mismatch.
9284         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9285         (cgraph_resolve_speculation): Declare.
9286         (symtab_can_be_discarded): New function.
9287         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9288         * ipa-inline-transform.c (speculation_removed): New global var.
9289         (clone_inlined_nodes): See if speculation can be removed.
9290         (inline_call): If speculations was removed, we growths may not match.
9291         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9292         (speculation_useful_p): New function.
9293         (resolve_noninline_speculation): New function.
9294         (inline_small_functions): Resolve useless speculations.
9295         * ipa-inline.h (speculation_useful_p): Declare
9296         * ipa.c (can_replace_by_local_alias): Simplify.
9297         (ipa_profile): Produce speculative calls in non-lto, too;
9298         add simple cost model; produce local aliases.
9300 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9302         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9303         PASS_MANAGER_H.
9305 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9307         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9309 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9311         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9312         iterator instead of X87MODEF.
9314 2013-08-12  Perez Read  <netfirewall@gmail.com>
9316         PR target/58132
9317         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9318         operand 0 for intel asm alternative.
9319         (*movabs<mode>_2): Ditto for operand 1.
9321 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9323         * config/aarch64/arm_none.h
9324         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9326 2013-08-12  Nick Clifton  <nickc@redhat.com>
9328         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9330 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9332         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9333         expand for QI/HImode operand to produce more effictive code for
9334         unsigned char(short) --> float(double) conversion.
9336 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9338         * doc/invoke.texi: Mention that -ftls-model does not force the final
9339         model.
9341 2013-08-12  Marek Polacek  <polacek@redhat.com>
9342             Marc Glisse  <marc.glisse@inria.fr>
9344         PR tree-optimization/57980
9345         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9346         when creating -1 constant.
9348 2013-08-10  Jan Hubicka  <jh@suse.cz>
9350         Workaround binutils PR14342.
9351         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9352         (gimple_init_edge_profiler): Likewise.
9353         (gimple_gen_ic_func_profiler): Likewise.
9355 2013-08-09  Jan Hubicka  <jh@suse.cz>
9357         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9359 2013-08-09  Xinliang David Li  <davidxl@google.com>
9361         * config/i386/stringop.def: New file.
9362         * config/i386/stringop.opt: New file.
9363         * config/i386/i386-opts.h: Include stringopt.def.
9364         * config/i386/i386.opt: Include stringopt.opt.
9365         * config/i386/i386.c (ix86_option_override_internal):
9366         Override default size based stringop inline strategies with options.
9367         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9368         New function.
9370 2013-08-09  Jan Hubicka  <jh@suse.cz>
9372         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9374 2013-08-09  Jan Hubicka  <jh@suse.cz>
9376         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9377         CGRAPH_FREQ_MAX.
9378         (dump_cgraph_node): Dump profile-id.
9379         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9380         and common_target_probability.
9381         * lto-cgraph.c (lto_output_edge): Stream common targets.
9382         (lto_output_node): Stream profile ids.
9383         (input_node): Stream profile ids.
9384         (input_edge): Stream common targets.
9385         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9386         * ipa.c: Include value-prof.h
9387         (ipa_profile_generate_summary): Turn indirect call statement histograms
9388         into common targets.
9389         (ipa_profile): Turn common targets into speculative edges.
9391 2013-08-09  Jan Hubicka  <jh@suse.cz>
9393         * cgraph.h (cgraph_node): Add profile_id.
9394         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9395         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9396         (del_node_map): Update.
9397         (find_func_by_funcdef_no): Replace by ...
9398         (find_func_by_profile_id): ... this one.
9399         (gimple_ic_transform): Do not remove useful histograms when
9400         speculation is not done; dump info when indirect call removal
9401         can happen at LTO.
9402         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9403         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9404         (__gcov_indirect_call_profiler_v2): .. this one.
9405         * profile.h (init_node_map): Update.
9406         * coverage.c (coverage_compute_profile_id): New function.
9407         * coverage.h (coverage_compute_profile_id): Declare.
9408         * tree-profile.c (init_ic_make_global_vars): Make
9409         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9410         (gimple_init_edge_profiler): Update prototype of
9411         __gcov_indirect_call_profiler.
9412         (gimple_gen_ic_func_profiler): Simplify.
9413         (tree_profiling): Use init_node_map
9415 2013-08-09  Jan Hubicka  <jh@suse.cz>
9417         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9418         non-speculative refs.
9419         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9420         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9421         (cgraph_set_call_stmt): Likewise.
9422         (cgraph_create_edge_1): Fix release checking compilatoin;
9423         clear lto_stmt_uid.
9424         (cgraph_free_edge): Free indirect info.
9425         (cgraph_turn_edge_to_speculative): New function.
9426         (cgraph_speculative_call_info): New function.
9427         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9428         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9429         (dump_cgraph_node): Dump speculation.
9430         (verify_edge_count_and_frequency): Accept speculative edges.
9431         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9432         (verify_cgraph_node): Handle speculation.
9433         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9434         (cgraph_set_call_stmt): Update prototype.
9435         (cgraph_make_edge_direct): Update prototype.
9436         (cgraph_speculative_call_info): Declare.
9437         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9438         to change; update call of ipa_find_references.
9439         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9440         lto_stmt_uid and speculative flags.
9441         (ipa_dump_references): Dump speculation.
9442         (ipa_clone_references): Clone speculative flag.
9443         (ipa_clone_referring): Likewise.
9444         (ipa_clone_ref): New function.
9445         (ipa_find_reference): Look into lto_stmt_uids
9446         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9447         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9448         (ipa_find_reference): Update declaration.
9449         (ipa_clone_ref): Declare.
9450         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9451         stream speculative flag.
9452         (lto_output_ref): Stream statements uids and speculation.
9453         (input_ref): Likewise.
9454         (input_edge): Stream speuclation.
9455         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9456         (cgraph_set_call_stmt_including_clones): Handle speculation.
9457         * ipa-inline.c (heap_edge_removal_hook): New function.
9458         (inline_small_functions): Register it.
9459         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9460         also initialize refs.
9461         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9462         edge to change.
9463         (try_make_edge_direct_simple_call): Likewise.
9464         (try_make_edge_direct_simple_call): Likewise.
9465         (update_indirect_edges_after_inlining): Likewise.
9466         (remove_described_reference): Look proper lto_stmt_uid.
9467         (propagate_controlled_uses): Likewise.
9468         (propagate_controlled_uses): Liekwise.
9469         * tree-inline.c (copy_bb): Copy speculative edges.
9470         (redirect_all_calls): New function.
9471         (copy_cfg_body): Do redirection after loop info is updated.
9472         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9474 2013-08-09  Jan Hubicka  <jh@suse.cz>
9476         * lto-streamer-out.c (output_function): Renumber PHIs.
9477         * lto-streamer-in.c (input_function): Likewise.
9479 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9481         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9482         (get_lane_unsigned): Likewise.
9483         (dup_lane_scalar): Likewise.
9484         (get_lane): enable for VALL.
9485         * config/aarch64/aarch64-simd.md
9486         (aarch64_dup_lane_scalar<mode>): Remove.
9487         (aarch64_get_lane_signed<mode>): Likewise.
9488         (aarch64_get_lane_unsigned<mode>): Likewise.
9489         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9490         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9491         (aarch64_get_lane<mode>): Enable for all vector modes.
9492         (aarch64_get_lanedi): Remove misleading constraints.
9493         * config/aarch64/arm_neon.h
9494         (__aarch64_vget_lane_any): Define.
9495         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9496         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9497         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9498         * config/aarch64/iterators.md (VDQQH): New.
9499         (VDQQHS): Likewise.
9500         (vwcore): Likewise.
9502 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9504         * configure.ac: Add GAS check for LEON instructions on SPARC.
9505         * configure: Regenerate.
9506         * config.in: Likewise.
9507         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9508         sparc*-*-* block.
9509         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9510         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9511         for LEON or LEON3.
9512         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9513         (AS_LEON_FLAG): New macro.
9514         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9515         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9516         Deal with LEON and LEON3 for the memory model.
9517         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9518         (atomic_compare_and_swap<mode>_1): Likewise.
9519         (*atomic_compare_and_swap<mode>_1): Likewise.
9521 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9523         * config/arm/neon.md (vcond): Fix floating-point vector
9524         comparisons against 0.
9526 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9528         * lra-constraints.c (emit_spill_move): Remove assert.
9529         (process_alt_operands): Add more debugging
9530         output.  Increase reject for spilling into memory.  Decrease
9531         reject for reloading scratch.
9532         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9534 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9536         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9537         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9538         micromips incompatible.  Add nan2008.
9539         (MULTILIB_DIRNAMES): Add nan2008.
9540         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9541         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9542         and micromips incompatible.  Add nan2008.
9543         (MULTILIB_DIRNAMES): Add nan2008.
9544         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9546 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9548         PR rtl-optimization/58079
9549         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9550         simplify_comparison has widened a comparison with an integer.
9552 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9554         * config/arm/neon.md (movmisalign<mode>): Disable when we
9555         don't allow unaligned accesses.
9556         (*movmisalign<mode>_neon_store): Likewise.
9557         (*movmisalign<mode>_neon_load): Likewise.
9558         (*movmisalign<mode>_neon_store): Likewise.
9559         (*movmisalign<mode>_neon_load): Likewise.
9561 2013-08-08  Jan Hubicka  <jh@suse.cz>
9563         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9564         (make_pass_rebuild_cgraph_edges): Also clear references.
9565         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9566         * ipa-inline-transform.c (inline_transform): Remove all references
9567         after inlining.
9568         * cgraphunit.c (expand_function): Remove all references after
9569         expansion.
9570         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9571         (ipa_find_reference): Rewrite to iterator.
9572         (remove_stmt_references): Likewise.
9573         (ipa_clear_stmts_in_references): New function.
9574         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9575         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9576         clear references.
9577         * ipa-split.c (split_function): Remove references in split function.
9579 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9581         PR target/57431
9582         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9583         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9585 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9587         PR target/56979
9588         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9589         suggested mode for the assignment isn't compatible with the
9590         registers required.
9592 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9594         PR target/58065
9595         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9597 2013-08-07  Xinliang David Li  <davidxl@google.com>
9599         * config/i386/i386.opt: New option -mtune-ctrl=.
9600         * config/i386/x86-tune.def: New file.
9601         * config/i386/i386.h: include x86-tune.def.
9602         * config/i386/i386.c (ix86_option_override_internal):
9603         Parsing -mtune-ctrl= option and set tune features.
9605 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9607         PR other/12081
9608         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9609         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9610         to gen_2arg_fn_t.
9612 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9614         * rtl.h (update_alignments): Declare.
9615         * final.c (grow_label_align): New function extracted from...
9616         (shorten_branches): ...here.  Call it.
9617         (update_alignments): New function.
9618         * reorg.c (sibling_labels): New variable.
9619         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9620         the new label along with it onto the sibling_labels vector.
9621         (fill_simple_delay_slots): Adjust call to get_label_before.
9622         (fill_slots_from_thread): Likewise.
9623         (relax_delay_slots): Likewise.
9624         (make_return_insns): Likewise.
9625         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9627 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9629         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9630         document its semantics.
9631         (diagnostic_report_diagnostic): Adjust accordingly.
9633 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9635         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9636         (sparc_option_override): ...and port to new C++ pass API.
9637         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9639 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9641         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9643 2013-08-06  Caroline Tice  <cmtice@google.com>
9645         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9646         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9647         * tree-pass.h: Add pass_vtable_verify.
9648         * varasm.c (assemble_variable): Add code to properly set the comdat
9649         section and name for the .vtable_map_vars section.
9650         (assemble_vtyv_preinit_initializer): New function.
9651         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9652         LINK_ONCE flag.
9653         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9654         * vtable-verify.c: New file.
9655         * vtable-verify.h: New file.
9656         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9657         initialiation levels.
9658         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9659         * passes.def: Insert pass_vtable_verify.
9660         * aclocal.m4: Reorder includes.
9661         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9662         -fvtv-counts options.
9663         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9664         as appropriate, if -fvtable-verify=... is used.
9665         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9666         -fvtable-verify=... is used.
9667         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9668         (vtable-verify.o): Add new build rule.
9669         (GTFILES): Add vtable-verify.c to list.
9670         * common.opt (fvtable-verify=): New flag.
9671         (vtv_priority): Values for fvtable-verify= flag.
9672         (fvtv-counts): New flag.
9673         (fvtv-debug): New flag.
9674         * tree.h (save_vtable_map_decl): New extern function decl.
9676 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9678         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9679         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9680         (pass_data_rl78_devirt): ...new pass_data instance and...
9681         (make_pass_rl78_devirt): ...new function.
9682         (rl78_asm_file_start): Port pass registration to new C++ API.
9684 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9686         * coretypes.h (rtl_opt_pass): Add.
9687         (gcc::context): Add.
9688         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9689         (epiphany_init): Port to new C++ pass API.
9690         (epiphany_optimize_mode_switching): Likewise.
9691         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9692         (pass_manager::get_pass_mode_switching): New.
9693         (pass_manager::get_pass_peephole2): New.
9694         * mode-switching.c (pass_mode_switching): Add clone method.
9695         * recog.c (pass_peephole2): Add clone method.
9696         (pass_split_all_insns): Add clone method.
9698 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9700         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9701         (mips_option_override): ...here, porting to new C++ API for passes.
9703 2013-08-06  Jan Hubicka  <jh@suse.cz>
9705         * cgraph.c (cgraph_get_body): New function based on lto.c
9706         implementation.
9707         * cgraph.h (cgraph_get_body): Declare.
9708         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9709         LTO paths.
9710         * cgraphunit.c (expand_function): Get body prior expanding.
9711         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9712         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9713         really need.
9714         * passes.c (do_per_function_toporder): Get body.
9715         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9716         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9718 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9720         PR fortran/57987
9721         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9722         is not re-finalized.  Rename second parameter to no_collect.
9724 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9726         PR middle-end/58041
9727         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9728         MEM_REF has proper alignment information.
9730 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9732         PR other/12081
9733         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9734         class insn_gen_fn.
9735         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9736         rtx (*) (rtx, ...) with insn_gen_fn.
9737         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9738         insn_gen_fn::stored_funcptr.  Add initializer braces.
9740 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9742         Rewrite how instances of passes are cloned to remove assumptions
9743         about their sizes (thus allowing pass subclasses to have
9744         additional data fields, albeit non-GC-managed ones at this point).
9746         * passes.c (make_pass_instance): Now that passes have clone
9747         methods, rewrite this function to eliminate XNEW and memcpy
9748         calls that used hardcoded sizes.  Since this function no longer
9749         creates pass instances, rename it to...
9750         (add_pass_instance): ...this.  Document the old way that passes were
9751         numbered and flagged, and rework this function to continue using it.
9752         (next_pass_1): Add an initial_pass argument for use by
9753         add_pass_instance.
9754         (position_pass): When adding multiple instances of a pass, use
9755         the pass's clone method, rather than relying on the XNEW/memcpy
9756         within the former make_pass_instance (now add_pass_instance).
9757         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9758         the initial instance of the current pass within the pass manager.
9760 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9762         This is the automated part of the conversion of passes from C
9763         structs to C++ classes.
9765         Patch autogenerated by refactor_passes.py from
9766         https://github.com/davidmalcolm/gcc-refactoring-scripts
9767         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9769         * asan.c (pass_asan): Convert from a global struct to a subclass of
9770         gimple_opt_pass along with...
9771         (pass_data_asan): ...new pass_data instance and...
9772         (make_pass_asan): ...new function.
9773         (pass_asan_O0): Convert from a global struct to a subclass of
9774         gimple_opt_pass along with...
9775         (pass_data_asan_O0): ...new pass_data instance and...
9776         (make_pass_asan_O0): ...new function.
9777         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9778         subclass of rtl_opt_pass along with...
9779         (pass_data_inc_dec): ...new pass_data instance and...
9780         (make_pass_inc_dec): ...new function.
9781         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9782         a subclass of rtl_opt_pass along with...
9783         (pass_data_reorder_blocks): ...new pass_data instance and...
9784         (make_pass_reorder_blocks): ...new function.
9785         (pass_duplicate_computed_gotos): Convert from a global struct to a
9786         subclass of rtl_opt_pass along with...
9787         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9788         (make_pass_duplicate_computed_gotos): ...new function.
9789         (pass_partition_blocks): Convert from a global struct to a subclass of
9790         rtl_opt_pass along with...
9791         (pass_data_partition_blocks): ...new pass_data instance and...
9792         (make_pass_partition_blocks): ...new function.
9793         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9794         struct to a subclass of rtl_opt_pass along with...
9795         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9796         and...
9797         (make_pass_branch_target_load_optimize1): ...new function.
9798         (pass_branch_target_load_optimize2): Convert from a global struct to a
9799         subclass of rtl_opt_pass along with...
9800         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9801         and...
9802         (make_pass_branch_target_load_optimize2): ...new function.
9803         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9804         of rtl_opt_pass along with...
9805         (pass_data_jump): ...new pass_data instance and...
9806         (make_pass_jump): ...new function.
9807         (pass_jump2): Convert from a global struct to a subclass of
9808         rtl_opt_pass along with...
9809         (pass_data_jump2): ...new pass_data instance and...
9810         (make_pass_jump2): ...new function.
9811         * cfgexpand.c (pass_expand): Convert from a global struct to a
9812         subclass of rtl_opt_pass along with...
9813         (pass_data_expand): ...new pass_data instance and...
9814         (make_pass_expand): ...new function.
9815         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9816         of rtl_opt_pass along with...
9817         (pass_data_free_cfg): ...new pass_data instance and...
9818         (make_pass_free_cfg): ...new function.
9819         (pass_into_cfg_layout_mode): Convert from a global struct to a
9820         subclass of rtl_opt_pass along with...
9821         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9822         (make_pass_into_cfg_layout_mode): ...new function.
9823         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9824         subclass of rtl_opt_pass along with...
9825         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9826         (make_pass_outof_cfg_layout_mode): ...new function.
9827         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9828         struct to a subclass of gimple_opt_pass along with...
9829         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9830         (make_pass_build_cgraph_edges): ...new function.
9831         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9832         subclass of gimple_opt_pass along with...
9833         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9834         (make_pass_rebuild_cgraph_edges): ...new function.
9835         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9836         subclass of gimple_opt_pass along with...
9837         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9838         and...
9839         (make_pass_remove_cgraph_callee_edges): ...new function.
9840         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9841         struct to a subclass of rtl_opt_pass along with...
9842         (pass_data_stack_adjustments): ...new pass_data instance and...
9843         (make_pass_stack_adjustments): ...new function.
9844         * combine.c (pass_combine): Convert from a global struct to a subclass
9845         of rtl_opt_pass along with...
9846         (pass_data_combine): ...new pass_data instance and...
9847         (make_pass_combine): ...new function.
9848         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9849         global struct to a subclass of rtl_opt_pass along with...
9850         (pass_data_compare_elim_after_reload): ...new pass_data instance
9851         and...
9852         (make_pass_compare_elim_after_reload): ...new function.
9853         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9854         of rtl_opt_pass along with...
9855         (pass_data_rtl_cprop): ...new pass_data instance and...
9856         (make_pass_rtl_cprop): ...new function.
9857         * cse.c (pass_cse): Convert from a global struct to a subclass of
9858         rtl_opt_pass along with...
9859         (pass_data_cse): ...new pass_data instance and...
9860         (make_pass_cse): ...new function.
9861         (pass_cse2): Convert from a global struct to a subclass of
9862         rtl_opt_pass along with...
9863         (pass_data_cse2): ...new pass_data instance and...
9864         (make_pass_cse2): ...new function.
9865         (pass_cse_after_global_opts): Convert from a global struct to a
9866         subclass of rtl_opt_pass along with...
9867         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9868         (make_pass_cse_after_global_opts): ...new function.
9869         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9870         of rtl_opt_pass along with...
9871         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9872         (make_pass_ud_rtl_dce): ...new function.
9873         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9874         rtl_opt_pass along with...
9875         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9876         (make_pass_fast_rtl_dce): ...new function.
9877         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9878         a subclass of rtl_opt_pass along with...
9879         (pass_data_df_initialize_opt): ...new pass_data instance and...
9880         (make_pass_df_initialize_opt): ...new function.
9881         (pass_df_initialize_no_opt): Convert from a global struct to a
9882         subclass of rtl_opt_pass along with...
9883         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9884         (make_pass_df_initialize_no_opt): ...new function.
9885         (pass_df_finish): Convert from a global struct to a subclass of
9886         rtl_opt_pass along with...
9887         (pass_data_df_finish): ...new pass_data instance and...
9888         (make_pass_df_finish): ...new function.
9889         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9890         rtl_opt_pass along with...
9891         (pass_data_rtl_dse1): ...new pass_data instance and...
9892         (make_pass_rtl_dse1): ...new function.
9893         (pass_rtl_dse2): Convert from a global struct to a subclass of
9894         rtl_opt_pass along with...
9895         (pass_data_rtl_dse2): ...new pass_data instance and...
9896         (make_pass_rtl_dse2): ...new function.
9897         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9898         subclass of rtl_opt_pass along with...
9899         (pass_data_dwarf2_frame): ...new pass_data instance and...
9900         (make_pass_dwarf2_frame): ...new function.
9901         * except.c (pass_set_nothrow_function_flags): Convert from a global
9902         struct to a subclass of rtl_opt_pass along with...
9903         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9904         and...
9905         (make_pass_set_nothrow_function_flags): ...new function.
9906         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9907         subclass of rtl_opt_pass along with...
9908         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9909         and...
9910         (make_pass_convert_to_eh_region_ranges): ...new function.
9911         * final.c (pass_compute_alignments): Convert from a global struct to a
9912         subclass of rtl_opt_pass along with...
9913         (pass_data_compute_alignments): ...new pass_data instance and...
9914         (make_pass_compute_alignments): ...new function.
9915         (pass_final): Convert from a global struct to a subclass of
9916         rtl_opt_pass along with...
9917         (pass_data_final): ...new pass_data instance and...
9918         (make_pass_final): ...new function.
9919         (pass_shorten_branches): Convert from a global struct to a subclass of
9920         rtl_opt_pass along with...
9921         (pass_data_shorten_branches): ...new pass_data instance and...
9922         (make_pass_shorten_branches): ...new function.
9923         (pass_clean_state): Convert from a global struct to a subclass of
9924         rtl_opt_pass along with...
9925         (pass_data_clean_state): ...new pass_data instance and...
9926         (make_pass_clean_state): ...new function.
9927         * function.c (pass_instantiate_virtual_regs): Convert from a global
9928         struct to a subclass of rtl_opt_pass along with...
9929         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9930         (make_pass_instantiate_virtual_regs): ...new function.
9931         (pass_leaf_regs): Convert from a global struct to a subclass of
9932         rtl_opt_pass along with...
9933         (pass_data_leaf_regs): ...new pass_data instance and...
9934         (make_pass_leaf_regs): ...new function.
9935         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9936         subclass of rtl_opt_pass along with...
9937         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9938         and...
9939         (make_pass_thread_prologue_and_epilogue): ...new function.
9940         (pass_match_asm_constraints): Convert from a global struct to a
9941         subclass of rtl_opt_pass along with...
9942         (pass_data_match_asm_constraints): ...new pass_data instance and...
9943         (make_pass_match_asm_constraints): ...new function.
9944         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9945         subclass of rtl_opt_pass along with...
9946         (pass_data_rtl_fwprop): ...new pass_data instance and...
9947         (make_pass_rtl_fwprop): ...new function.
9948         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9949         rtl_opt_pass along with...
9950         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9951         (make_pass_rtl_fwprop_addr): ...new function.
9952         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9953         rtl_opt_pass along with...
9954         (pass_data_rtl_pre): ...new pass_data instance and...
9955         (make_pass_rtl_pre): ...new function.
9956         (pass_rtl_hoist): Convert from a global struct to a subclass of
9957         rtl_opt_pass along with...
9958         (pass_data_rtl_hoist): ...new pass_data instance and...
9959         (make_pass_rtl_hoist): ...new function.
9960         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9961         subclass of gimple_opt_pass along with...
9962         (pass_data_lower_cf): ...new pass_data instance and...
9963         (make_pass_lower_cf): ...new function.
9964         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9965         from a global struct to a subclass of gimple_opt_pass along with...
9966         (pass_data_strength_reduction): ...new pass_data instance and...
9967         (make_pass_strength_reduction): ...new function.
9968         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9969         of rtl_opt_pass along with...
9970         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9971         (make_pass_rtl_ifcvt): ...new function.
9972         (pass_if_after_combine): Convert from a global struct to a subclass of
9973         rtl_opt_pass along with...
9974         (pass_data_if_after_combine): ...new pass_data instance and...
9975         (make_pass_if_after_combine): ...new function.
9976         (pass_if_after_reload): Convert from a global struct to a subclass of
9977         rtl_opt_pass along with...
9978         (pass_data_if_after_reload): ...new pass_data instance and...
9979         (make_pass_if_after_reload): ...new function.
9980         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9981         a subclass of rtl_opt_pass along with...
9982         (pass_data_initialize_regs): ...new pass_data instance and...
9983         (make_pass_initialize_regs): ...new function.
9984         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9985         of ipa_opt_pass_d along with...
9986         (pass_data_ipa_cp): ...new pass_data instance and...
9987         (make_pass_ipa_cp): ...new function.
9988         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9989         global struct to a subclass of gimple_opt_pass along with...
9990         (pass_data_inline_parameters): ...new pass_data instance and...
9991         (make_pass_inline_parameters): ...new function.
9992         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9993         subclass of gimple_opt_pass along with...
9994         (pass_data_early_inline): ...new pass_data instance and...
9995         (make_pass_early_inline): ...new function.
9996         (pass_ipa_inline): Convert from a global struct to a subclass of
9997         ipa_opt_pass_d along with...
9998         (pass_data_ipa_inline): ...new pass_data instance and...
9999         (make_pass_ipa_inline): ...new function.
10000         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
10001         struct to a subclass of gimple_opt_pass along with...
10002         (pass_data_local_pure_const): ...new pass_data instance and...
10003         (make_pass_local_pure_const): ...new function.
10004         (pass_ipa_pure_const): Convert from a global struct to a subclass of
10005         ipa_opt_pass_d along with...
10006         (pass_data_ipa_pure_const): ...new pass_data instance and...
10007         (make_pass_ipa_pure_const): ...new function.
10008         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
10009         to a subclass of ipa_opt_pass_d along with...
10010         (pass_data_ipa_reference): ...new pass_data instance and...
10011         (make_pass_ipa_reference): ...new function.
10012         * ipa-split.c (pass_split_functions): Convert from a global struct to
10013         a subclass of gimple_opt_pass along with...
10014         (pass_data_split_functions): ...new pass_data instance and...
10015         (make_pass_split_functions): ...new function.
10016         (pass_feedback_split_functions): Convert from a global struct to a
10017         subclass of gimple_opt_pass along with...
10018         (pass_data_feedback_split_functions): ...new pass_data instance and...
10019         (make_pass_feedback_split_functions): ...new function.
10020         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10021         global struct to a subclass of simple_ipa_opt_pass along with...
10022         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10023         instance and...
10024         (make_pass_ipa_function_and_variable_visibility): ...new function.
10025         (pass_ipa_free_inline_summary): Convert from a global struct to a
10026         subclass of simple_ipa_opt_pass along with...
10027         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10028         (make_pass_ipa_free_inline_summary): ...new function.
10029         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10030         subclass of ipa_opt_pass_d along with...
10031         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10032         and...
10033         (make_pass_ipa_whole_program_visibility): ...new function.
10034         (pass_ipa_profile): Convert from a global struct to a subclass of
10035         ipa_opt_pass_d along with...
10036         (pass_data_ipa_profile): ...new pass_data instance and...
10037         (make_pass_ipa_profile): ...new function.
10038         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10039         ipa_opt_pass_d along with...
10040         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10041         (make_pass_ipa_cdtor_merge): ...new function.
10042         * ira.c (pass_ira): Convert from a global struct to a subclass of
10043         rtl_opt_pass along with...
10044         (pass_data_ira): ...new pass_data instance and...
10045         (make_pass_ira): ...new function.
10046         (pass_reload): Convert from a global struct to a subclass of
10047         rtl_opt_pass along with...
10048         (pass_data_reload): ...new pass_data instance and...
10049         (make_pass_reload): ...new function.
10050         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10051         subclass of rtl_opt_pass along with...
10052         (pass_data_cleanup_barriers): ...new pass_data instance and...
10053         (make_pass_cleanup_barriers): ...new function.
10054         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10055         of rtl_opt_pass along with...
10056         (pass_data_loop2): ...new pass_data instance and...
10057         (make_pass_loop2): ...new function.
10058         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10059         rtl_opt_pass along with...
10060         (pass_data_rtl_loop_init): ...new pass_data instance and...
10061         (make_pass_rtl_loop_init): ...new function.
10062         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10063         rtl_opt_pass along with...
10064         (pass_data_rtl_loop_done): ...new pass_data instance and...
10065         (make_pass_rtl_loop_done): ...new function.
10066         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10067         subclass of rtl_opt_pass along with...
10068         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10069         (make_pass_rtl_move_loop_invariants): ...new function.
10070         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10071         rtl_opt_pass along with...
10072         (pass_data_rtl_unswitch): ...new pass_data instance and...
10073         (make_pass_rtl_unswitch): ...new function.
10074         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10075         subclass of rtl_opt_pass along with...
10076         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10077         and...
10078         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10079         (pass_rtl_doloop): Convert from a global struct to a subclass of
10080         rtl_opt_pass along with...
10081         (pass_data_rtl_doloop): ...new pass_data instance and...
10082         (make_pass_rtl_doloop): ...new function.
10083         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10084         a subclass of rtl_opt_pass along with...
10085         (pass_data_lower_subreg): ...new pass_data instance and...
10086         (make_pass_lower_subreg): ...new function.
10087         (pass_lower_subreg2): Convert from a global struct to a subclass of
10088         rtl_opt_pass along with...
10089         (pass_data_lower_subreg2): ...new pass_data instance and...
10090         (make_pass_lower_subreg2): ...new function.
10091         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10092         struct to a subclass of ipa_opt_pass_d along with...
10093         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10094         (make_pass_ipa_lto_gimple_out): ...new function.
10095         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10096         of ipa_opt_pass_d along with...
10097         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10098         (make_pass_ipa_lto_finish_out): ...new function.
10099         * mode-switching.c (pass_mode_switching): Convert from a global struct
10100         to a subclass of rtl_opt_pass along with...
10101         (pass_data_mode_switching): ...new pass_data instance and...
10102         (make_pass_mode_switching): ...new function.
10103         * modulo-sched.c (pass_sms): Convert from a global struct to a
10104         subclass of rtl_opt_pass along with...
10105         (pass_data_sms): ...new pass_data instance and...
10106         (make_pass_sms): ...new function.
10107         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10108         subclass of gimple_opt_pass along with...
10109         (pass_data_expand_omp): ...new pass_data instance and...
10110         (make_pass_expand_omp): ...new function.
10111         (pass_lower_omp): Convert from a global struct to a subclass of
10112         gimple_opt_pass along with...
10113         (pass_data_lower_omp): ...new pass_data instance and...
10114         (make_pass_lower_omp): ...new function.
10115         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10116         of gimple_opt_pass along with...
10117         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10118         (make_pass_diagnose_omp_blocks): ...new function.
10119         * passes.c (pass_early_local_passes): Convert from a global struct to
10120         a subclass of simple_ipa_opt_pass along with...
10121         (pass_data_early_local_passes): ...new pass_data instance and...
10122         (make_pass_early_local_passes): ...new function.
10123         (pass_all_early_optimizations): Convert from a global struct to a
10124         subclass of gimple_opt_pass along with...
10125         (pass_data_all_early_optimizations): ...new pass_data instance and...
10126         (make_pass_all_early_optimizations): ...new function.
10127         (pass_all_optimizations): Convert from a global struct to a subclass
10128         of gimple_opt_pass along with...
10129         (pass_data_all_optimizations): ...new pass_data instance and...
10130         (make_pass_all_optimizations): ...new function.
10131         (pass_all_optimizations_g): Convert from a global struct to a subclass
10132         of gimple_opt_pass along with...
10133         (pass_data_all_optimizations_g): ...new pass_data instance and...
10134         (make_pass_all_optimizations_g): ...new function.
10135         (pass_rest_of_compilation): Convert from a global struct to a subclass
10136         of rtl_opt_pass along with...
10137         (pass_data_rest_of_compilation): ...new pass_data instance and...
10138         (make_pass_rest_of_compilation): ...new function.
10139         (pass_postreload): Convert from a global struct to a subclass of
10140         rtl_opt_pass along with...
10141         (pass_data_postreload): ...new pass_data instance and...
10142         (make_pass_postreload): ...new function.
10143         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10144         subclass of rtl_opt_pass along with...
10145         (pass_data_gcse2): ...new pass_data instance and...
10146         (make_pass_gcse2): ...new function.
10147         * postreload.c (pass_postreload_cse): Convert from a global struct to
10148         a subclass of rtl_opt_pass along with...
10149         (pass_data_postreload_cse): ...new pass_data instance and...
10150         (make_pass_postreload_cse): ...new function.
10151         * predict.c (pass_profile): Convert from a global struct to a subclass
10152         of gimple_opt_pass along with...
10153         (pass_data_profile): ...new pass_data instance and...
10154         (make_pass_profile): ...new function.
10155         (pass_strip_predict_hints): Convert from a global struct to a subclass
10156         of gimple_opt_pass along with...
10157         (pass_data_strip_predict_hints): ...new pass_data instance and...
10158         (make_pass_strip_predict_hints): ...new function.
10159         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10160         of rtl_opt_pass along with...
10161         (pass_data_peephole2): ...new pass_data instance and...
10162         (make_pass_peephole2): ...new function.
10163         (pass_split_all_insns): Convert from a global struct to a subclass of
10164         rtl_opt_pass along with...
10165         (pass_data_split_all_insns): ...new pass_data instance and...
10166         (make_pass_split_all_insns): ...new function.
10167         (pass_split_after_reload): Convert from a global struct to a subclass
10168         of rtl_opt_pass along with...
10169         (pass_data_split_after_reload): ...new pass_data instance and...
10170         (make_pass_split_after_reload): ...new function.
10171         (pass_split_before_regstack): Convert from a global struct to a
10172         subclass of rtl_opt_pass along with...
10173         (pass_data_split_before_regstack): ...new pass_data instance and...
10174         (make_pass_split_before_regstack): ...new function.
10175         (pass_split_before_sched2): Convert from a global struct to a subclass
10176         of rtl_opt_pass along with...
10177         (pass_data_split_before_sched2): ...new pass_data instance and...
10178         (make_pass_split_before_sched2): ...new function.
10179         (pass_split_for_shorten_branches): Convert from a global struct to a
10180         subclass of rtl_opt_pass along with...
10181         (pass_data_split_for_shorten_branches): ...new pass_data instance
10182         and...
10183         (make_pass_split_for_shorten_branches): ...new function.
10184         * ree.c (pass_ree): Convert from a global struct to a subclass of
10185         rtl_opt_pass along with...
10186         (pass_data_ree): ...new pass_data instance and...
10187         (make_pass_ree): ...new function.
10188         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10189         subclass of rtl_opt_pass along with...
10190         (pass_data_stack_regs): ...new pass_data instance and...
10191         (make_pass_stack_regs): ...new function.
10192         (pass_stack_regs_run): Convert from a global struct to a subclass of
10193         rtl_opt_pass along with...
10194         (pass_data_stack_regs_run): ...new pass_data instance and...
10195         (make_pass_stack_regs_run): ...new function.
10196         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10197         subclass of rtl_opt_pass along with...
10198         (pass_data_cprop_hardreg): ...new pass_data instance and...
10199         (make_pass_cprop_hardreg): ...new function.
10200         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10201         subclass of rtl_opt_pass along with...
10202         (pass_data_reginfo_init): ...new pass_data instance and...
10203         (make_pass_reginfo_init): ...new function.
10204         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10205         of rtl_opt_pass along with...
10206         (pass_data_regmove): ...new pass_data instance and...
10207         (make_pass_regmove): ...new function.
10208         * regrename.c (pass_regrename): Convert from a global struct to a
10209         subclass of rtl_opt_pass along with...
10210         (pass_data_regrename): ...new pass_data instance and...
10211         (make_pass_regrename): ...new function.
10212         * reorg.c (pass_delay_slots): Convert from a global struct to a
10213         subclass of rtl_opt_pass along with...
10214         (pass_data_delay_slots): ...new pass_data instance and...
10215         (make_pass_delay_slots): ...new function.
10216         (pass_machine_reorg): Convert from a global struct to a subclass of
10217         rtl_opt_pass along with...
10218         (pass_data_machine_reorg): ...new pass_data instance and...
10219         (make_pass_machine_reorg): ...new function.
10220         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10221         of rtl_opt_pass along with...
10222         (pass_data_sched): ...new pass_data instance and...
10223         (make_pass_sched): ...new function.
10224         (pass_sched2): Convert from a global struct to a subclass of
10225         rtl_opt_pass along with...
10226         (pass_data_sched2): ...new pass_data instance and...
10227         (make_pass_sched2): ...new function.
10228         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10229         to a subclass of rtl_opt_pass along with...
10230         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10231         (make_pass_stack_ptr_mod): ...new function.
10232         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10233         to a subclass of rtl_opt_pass along with...
10234         (pass_data_rtl_store_motion): ...new pass_data instance and...
10235         (make_pass_rtl_store_motion): ...new function.
10236         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10237         of gimple_opt_pass along with...
10238         (pass_data_tracer): ...new pass_data instance and...
10239         (make_pass_tracer): ...new function.
10240         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10241         to a subclass of gimple_opt_pass along with...
10242         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10243         (make_pass_diagnose_tm_blocks): ...new function.
10244         (pass_lower_tm): Convert from a global struct to a subclass of
10245         gimple_opt_pass along with...
10246         (pass_data_lower_tm): ...new pass_data instance and...
10247         (make_pass_lower_tm): ...new function.
10248         (pass_tm_init): Convert from a global struct to a subclass of
10249         gimple_opt_pass along with...
10250         (pass_data_tm_init): ...new pass_data instance and...
10251         (make_pass_tm_init): ...new function.
10252         (pass_tm_mark): Convert from a global struct to a subclass of
10253         gimple_opt_pass along with...
10254         (pass_data_tm_mark): ...new pass_data instance and...
10255         (make_pass_tm_mark): ...new function.
10256         (pass_tm_edges): Convert from a global struct to a subclass of
10257         gimple_opt_pass along with...
10258         (pass_data_tm_edges): ...new pass_data instance and...
10259         (make_pass_tm_edges): ...new function.
10260         (pass_tm_memopt): Convert from a global struct to a subclass of
10261         gimple_opt_pass along with...
10262         (pass_data_tm_memopt): ...new pass_data instance and...
10263         (make_pass_tm_memopt): ...new function.
10264         (pass_ipa_tm): Convert from a global struct to a subclass of
10265         simple_ipa_opt_pass along with...
10266         (pass_data_ipa_tm): ...new pass_data instance and...
10267         (make_pass_ipa_tm): ...new function.
10268         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10269         subclass of gimple_opt_pass along with...
10270         (pass_data_call_cdce): ...new pass_data instance and...
10271         (make_pass_call_cdce): ...new function.
10272         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10273         subclass of gimple_opt_pass along with...
10274         (pass_data_build_cfg): ...new pass_data instance and...
10275         (make_pass_build_cfg): ...new function.
10276         (pass_split_crit_edges): Convert from a global struct to a subclass of
10277         gimple_opt_pass along with...
10278         (pass_data_split_crit_edges): ...new pass_data instance and...
10279         (make_pass_split_crit_edges): ...new function.
10280         (pass_warn_function_return): Convert from a global struct to a
10281         subclass of gimple_opt_pass along with...
10282         (pass_data_warn_function_return): ...new pass_data instance and...
10283         (make_pass_warn_function_return): ...new function.
10284         (pass_warn_function_noreturn): Convert from a global struct to a
10285         subclass of gimple_opt_pass along with...
10286         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10287         (make_pass_warn_function_noreturn): ...new function.
10288         (pass_warn_unused_result): Convert from a global struct to a subclass
10289         of gimple_opt_pass along with...
10290         (pass_data_warn_unused_result): ...new pass_data instance and...
10291         (make_pass_warn_unused_result): ...new function.
10292         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10293         a subclass of gimple_opt_pass along with...
10294         (pass_data_merge_phi): ...new pass_data instance and...
10295         (make_pass_merge_phi): ...new function.
10296         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10297         a subclass of gimple_opt_pass along with...
10298         (pass_data_lower_complex): ...new pass_data instance and...
10299         (make_pass_lower_complex): ...new function.
10300         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10301         gimple_opt_pass along with...
10302         (pass_data_lower_complex_O0): ...new pass_data instance and...
10303         (make_pass_lower_complex_O0): ...new function.
10304         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10305         subclass of gimple_opt_pass along with...
10306         (pass_data_lower_eh): ...new pass_data instance and...
10307         (make_pass_lower_eh): ...new function.
10308         (pass_refactor_eh): Convert from a global struct to a subclass of
10309         gimple_opt_pass along with...
10310         (pass_data_refactor_eh): ...new pass_data instance and...
10311         (make_pass_refactor_eh): ...new function.
10312         (pass_lower_resx): Convert from a global struct to a subclass of
10313         gimple_opt_pass along with...
10314         (pass_data_lower_resx): ...new pass_data instance and...
10315         (make_pass_lower_resx): ...new function.
10316         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10317         of gimple_opt_pass along with...
10318         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10319         (make_pass_lower_eh_dispatch): ...new function.
10320         (pass_cleanup_eh): Convert from a global struct to a subclass of
10321         gimple_opt_pass along with...
10322         (pass_data_cleanup_eh): ...new pass_data instance and...
10323         (make_pass_cleanup_eh): ...new function.
10324         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10325         to a subclass of simple_ipa_opt_pass along with...
10326         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10327         (make_pass_ipa_lower_emutls): ...new function.
10328         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10329         a subclass of gimple_opt_pass along with...
10330         (pass_data_if_conversion): ...new pass_data instance and...
10331         (make_pass_if_conversion): ...new function.
10332         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10333         subclass of gimple_opt_pass along with...
10334         (pass_data_build_ssa): ...new pass_data instance and...
10335         (make_pass_build_ssa): ...new function.
10336         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10337         global struct to a subclass of gimple_opt_pass along with...
10338         (pass_data_loop_distribution): ...new pass_data instance and...
10339         (make_pass_loop_distribution): ...new function.
10340         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10341         subclass of gimple_opt_pass along with...
10342         (pass_data_mudflap_1): ...new pass_data instance and...
10343         (make_pass_mudflap_1): ...new function.
10344         (pass_mudflap_2): Convert from a global struct to a subclass of
10345         gimple_opt_pass along with...
10346         (pass_data_mudflap_2): ...new pass_data instance and...
10347         (make_pass_mudflap_2): ...new function.
10348         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10349         subclass of gimple_opt_pass along with...
10350         (pass_data_mudflap_1): ...new pass_data instance and...
10351         (make_pass_mudflap_1): ...new function.
10352         (pass_mudflap_2): Convert from a global struct to a subclass of
10353         gimple_opt_pass along with...
10354         (pass_data_mudflap_2): ...new pass_data instance and...
10355         (make_pass_mudflap_2): ...new function.
10356         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10357         gimple_opt_pass along with...
10358         (pass_data_nrv): ...new pass_data instance and...
10359         (make_pass_nrv): ...new function.
10360         (pass_return_slot): Convert from a global struct to a subclass of
10361         gimple_opt_pass along with...
10362         (pass_data_return_slot): ...new pass_data instance and...
10363         (make_pass_return_slot): ...new function.
10364         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10365         to a subclass of gimple_opt_pass along with...
10366         (pass_data_object_sizes): ...new pass_data instance and...
10367         (make_pass_object_sizes): ...new function.
10368         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10369         global struct to a subclass of gimple_opt_pass along with...
10370         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10371         and...
10372         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10373         (pass_fixup_cfg): Convert from a global struct to a subclass of
10374         gimple_opt_pass along with...
10375         (pass_data_fixup_cfg): ...new pass_data instance and...
10376         (make_pass_fixup_cfg): ...new function.
10377         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10378         (make_pass_mudflap_1): ...new function.
10379         (pass_mudflap_2): Replace declaration with that of...
10380         (make_pass_mudflap_2): ...new function.
10381         (pass_asan): Replace declaration with that of...
10382         (make_pass_asan): ...new function.
10383         (pass_asan_O0): Replace declaration with that of...
10384         (make_pass_asan_O0): ...new function.
10385         (pass_tsan): Replace declaration with that of...
10386         (make_pass_tsan): ...new function.
10387         (pass_tsan_O0): Replace declaration with that of...
10388         (make_pass_tsan_O0): ...new function.
10389         (pass_lower_cf): Replace declaration with that of...
10390         (make_pass_lower_cf): ...new function.
10391         (pass_refactor_eh): Replace declaration with that of...
10392         (make_pass_refactor_eh): ...new function.
10393         (pass_lower_eh): Replace declaration with that of...
10394         (make_pass_lower_eh): ...new function.
10395         (pass_lower_eh_dispatch): Replace declaration with that of...
10396         (make_pass_lower_eh_dispatch): ...new function.
10397         (pass_lower_resx): Replace declaration with that of...
10398         (make_pass_lower_resx): ...new function.
10399         (pass_build_cfg): Replace declaration with that of...
10400         (make_pass_build_cfg): ...new function.
10401         (pass_early_tree_profile): Replace declaration with that of...
10402         (make_pass_early_tree_profile): ...new function.
10403         (pass_cleanup_eh): Replace declaration with that of...
10404         (make_pass_cleanup_eh): ...new function.
10405         (pass_sra): Replace declaration with that of...
10406         (make_pass_sra): ...new function.
10407         (pass_sra_early): Replace declaration with that of...
10408         (make_pass_sra_early): ...new function.
10409         (pass_early_ipa_sra): Replace declaration with that of...
10410         (make_pass_early_ipa_sra): ...new function.
10411         (pass_tail_recursion): Replace declaration with that of...
10412         (make_pass_tail_recursion): ...new function.
10413         (pass_tail_calls): Replace declaration with that of...
10414         (make_pass_tail_calls): ...new function.
10415         (pass_tree_loop): Replace declaration with that of...
10416         (make_pass_tree_loop): ...new function.
10417         (pass_tree_loop_init): Replace declaration with that of...
10418         (make_pass_tree_loop_init): ...new function.
10419         (pass_lim): Replace declaration with that of...
10420         (make_pass_lim): ...new function.
10421         (pass_tree_unswitch): Replace declaration with that of...
10422         (make_pass_tree_unswitch): ...new function.
10423         (pass_predcom): Replace declaration with that of...
10424         (make_pass_predcom): ...new function.
10425         (pass_iv_canon): Replace declaration with that of...
10426         (make_pass_iv_canon): ...new function.
10427         (pass_scev_cprop): Replace declaration with that of...
10428         (make_pass_scev_cprop): ...new function.
10429         (pass_empty_loop): Replace declaration with that of...
10430         (make_pass_empty_loop): ...new function.
10431         (pass_record_bounds): Replace declaration with that of...
10432         (make_pass_record_bounds): ...new function.
10433         (pass_graphite): Replace declaration with that of...
10434         (make_pass_graphite): ...new function.
10435         (pass_graphite_transforms): Replace declaration with that of...
10436         (make_pass_graphite_transforms): ...new function.
10437         (pass_if_conversion): Replace declaration with that of...
10438         (make_pass_if_conversion): ...new function.
10439         (pass_loop_distribution): Replace declaration with that of...
10440         (make_pass_loop_distribution): ...new function.
10441         (pass_vectorize): Replace declaration with that of...
10442         (make_pass_vectorize): ...new function.
10443         (pass_slp_vectorize): Replace declaration with that of...
10444         (make_pass_slp_vectorize): ...new function.
10445         (pass_complete_unroll): Replace declaration with that of...
10446         (make_pass_complete_unroll): ...new function.
10447         (pass_complete_unrolli): Replace declaration with that of...
10448         (make_pass_complete_unrolli): ...new function.
10449         (pass_parallelize_loops): Replace declaration with that of...
10450         (make_pass_parallelize_loops): ...new function.
10451         (pass_loop_prefetch): Replace declaration with that of...
10452         (make_pass_loop_prefetch): ...new function.
10453         (pass_iv_optimize): Replace declaration with that of...
10454         (make_pass_iv_optimize): ...new function.
10455         (pass_tree_loop_done): Replace declaration with that of...
10456         (make_pass_tree_loop_done): ...new function.
10457         (pass_ch): Replace declaration with that of...
10458         (make_pass_ch): ...new function.
10459         (pass_ccp): Replace declaration with that of...
10460         (make_pass_ccp): ...new function.
10461         (pass_phi_only_cprop): Replace declaration with that of...
10462         (make_pass_phi_only_cprop): ...new function.
10463         (pass_build_ssa): Replace declaration with that of...
10464         (make_pass_build_ssa): ...new function.
10465         (pass_build_alias): Replace declaration with that of...
10466         (make_pass_build_alias): ...new function.
10467         (pass_build_ealias): Replace declaration with that of...
10468         (make_pass_build_ealias): ...new function.
10469         (pass_dominator): Replace declaration with that of...
10470         (make_pass_dominator): ...new function.
10471         (pass_dce): Replace declaration with that of...
10472         (make_pass_dce): ...new function.
10473         (pass_dce_loop): Replace declaration with that of...
10474         (make_pass_dce_loop): ...new function.
10475         (pass_cd_dce): Replace declaration with that of...
10476         (make_pass_cd_dce): ...new function.
10477         (pass_call_cdce): Replace declaration with that of...
10478         (make_pass_call_cdce): ...new function.
10479         (pass_merge_phi): Replace declaration with that of...
10480         (make_pass_merge_phi): ...new function.
10481         (pass_split_crit_edges): Replace declaration with that of...
10482         (make_pass_split_crit_edges): ...new function.
10483         (pass_pre): Replace declaration with that of...
10484         (make_pass_pre): ...new function.
10485         (pass_profile): Replace declaration with that of...
10486         (make_pass_profile): ...new function.
10487         (pass_strip_predict_hints): Replace declaration with that of...
10488         (make_pass_strip_predict_hints): ...new function.
10489         (pass_lower_complex_O0): Replace declaration with that of...
10490         (make_pass_lower_complex_O0): ...new function.
10491         (pass_lower_complex): Replace declaration with that of...
10492         (make_pass_lower_complex): ...new function.
10493         (pass_lower_vector): Replace declaration with that of...
10494         (make_pass_lower_vector): ...new function.
10495         (pass_lower_vector_ssa): Replace declaration with that of...
10496         (make_pass_lower_vector_ssa): ...new function.
10497         (pass_lower_omp): Replace declaration with that of...
10498         (make_pass_lower_omp): ...new function.
10499         (pass_diagnose_omp_blocks): Replace declaration with that of...
10500         (make_pass_diagnose_omp_blocks): ...new function.
10501         (pass_expand_omp): Replace declaration with that of...
10502         (make_pass_expand_omp): ...new function.
10503         (pass_expand_omp_ssa): Replace declaration with that of...
10504         (make_pass_expand_omp_ssa): ...new function.
10505         (pass_object_sizes): Replace declaration with that of...
10506         (make_pass_object_sizes): ...new function.
10507         (pass_strlen): Replace declaration with that of...
10508         (make_pass_strlen): ...new function.
10509         (pass_fold_builtins): Replace declaration with that of...
10510         (make_pass_fold_builtins): ...new function.
10511         (pass_stdarg): Replace declaration with that of...
10512         (make_pass_stdarg): ...new function.
10513         (pass_early_warn_uninitialized): Replace declaration with that of...
10514         (make_pass_early_warn_uninitialized): ...new function.
10515         (pass_late_warn_uninitialized): Replace declaration with that of...
10516         (make_pass_late_warn_uninitialized): ...new function.
10517         (pass_cse_reciprocals): Replace declaration with that of...
10518         (make_pass_cse_reciprocals): ...new function.
10519         (pass_cse_sincos): Replace declaration with that of...
10520         (make_pass_cse_sincos): ...new function.
10521         (pass_optimize_bswap): Replace declaration with that of...
10522         (make_pass_optimize_bswap): ...new function.
10523         (pass_optimize_widening_mul): Replace declaration with that of...
10524         (make_pass_optimize_widening_mul): ...new function.
10525         (pass_warn_function_return): Replace declaration with that of...
10526         (make_pass_warn_function_return): ...new function.
10527         (pass_warn_function_noreturn): Replace declaration with that of...
10528         (make_pass_warn_function_noreturn): ...new function.
10529         (pass_cselim): Replace declaration with that of...
10530         (make_pass_cselim): ...new function.
10531         (pass_phiopt): Replace declaration with that of...
10532         (make_pass_phiopt): ...new function.
10533         (pass_forwprop): Replace declaration with that of...
10534         (make_pass_forwprop): ...new function.
10535         (pass_phiprop): Replace declaration with that of...
10536         (make_pass_phiprop): ...new function.
10537         (pass_tree_ifcombine): Replace declaration with that of...
10538         (make_pass_tree_ifcombine): ...new function.
10539         (pass_dse): Replace declaration with that of...
10540         (make_pass_dse): ...new function.
10541         (pass_nrv): Replace declaration with that of...
10542         (make_pass_nrv): ...new function.
10543         (pass_rename_ssa_copies): Replace declaration with that of...
10544         (make_pass_rename_ssa_copies): ...new function.
10545         (pass_sink_code): Replace declaration with that of...
10546         (make_pass_sink_code): ...new function.
10547         (pass_fre): Replace declaration with that of...
10548         (make_pass_fre): ...new function.
10549         (pass_check_data_deps): Replace declaration with that of...
10550         (make_pass_check_data_deps): ...new function.
10551         (pass_copy_prop): Replace declaration with that of...
10552         (make_pass_copy_prop): ...new function.
10553         (pass_vrp): Replace declaration with that of...
10554         (make_pass_vrp): ...new function.
10555         (pass_uncprop): Replace declaration with that of...
10556         (make_pass_uncprop): ...new function.
10557         (pass_return_slot): Replace declaration with that of...
10558         (make_pass_return_slot): ...new function.
10559         (pass_reassoc): Replace declaration with that of...
10560         (make_pass_reassoc): ...new function.
10561         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10562         (make_pass_rebuild_cgraph_edges): ...new function.
10563         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10564         (make_pass_remove_cgraph_callee_edges): ...new function.
10565         (pass_build_cgraph_edges): Replace declaration with that of...
10566         (make_pass_build_cgraph_edges): ...new function.
10567         (pass_local_pure_const): Replace declaration with that of...
10568         (make_pass_local_pure_const): ...new function.
10569         (pass_tracer): Replace declaration with that of...
10570         (make_pass_tracer): ...new function.
10571         (pass_warn_unused_result): Replace declaration with that of...
10572         (make_pass_warn_unused_result): ...new function.
10573         (pass_diagnose_tm_blocks): Replace declaration with that of...
10574         (make_pass_diagnose_tm_blocks): ...new function.
10575         (pass_lower_tm): Replace declaration with that of...
10576         (make_pass_lower_tm): ...new function.
10577         (pass_tm_init): Replace declaration with that of...
10578         (make_pass_tm_init): ...new function.
10579         (pass_tm_mark): Replace declaration with that of...
10580         (make_pass_tm_mark): ...new function.
10581         (pass_tm_memopt): Replace declaration with that of...
10582         (make_pass_tm_memopt): ...new function.
10583         (pass_tm_edges): Replace declaration with that of...
10584         (make_pass_tm_edges): ...new function.
10585         (pass_split_functions): Replace declaration with that of...
10586         (make_pass_split_functions): ...new function.
10587         (pass_feedback_split_functions): Replace declaration with that of...
10588         (make_pass_feedback_split_functions): ...new function.
10589         (pass_strength_reduction): Replace declaration with that of...
10590         (make_pass_strength_reduction): ...new function.
10591         (pass_ipa_lower_emutls): Replace declaration with that of...
10592         (make_pass_ipa_lower_emutls): ...new function.
10593         (pass_ipa_function_and_variable_visibility): Replace declaration with
10594         that of...
10595         (make_pass_ipa_function_and_variable_visibility): ...new function.
10596         (pass_ipa_tree_profile): Replace declaration with that of...
10597         (make_pass_ipa_tree_profile): ...new function.
10598         (pass_early_local_passes): Replace declaration with that of...
10599         (make_pass_early_local_passes): ...new function.
10600         (pass_ipa_whole_program_visibility): Replace declaration with that
10601         of...
10602         (make_pass_ipa_whole_program_visibility): ...new function.
10603         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10604         (make_pass_ipa_lto_gimple_out): ...new function.
10605         (pass_ipa_increase_alignment): Replace declaration with that of...
10606         (make_pass_ipa_increase_alignment): ...new function.
10607         (pass_ipa_inline): Replace declaration with that of...
10608         (make_pass_ipa_inline): ...new function.
10609         (pass_ipa_free_lang_data): Replace declaration with that of...
10610         (make_pass_ipa_free_lang_data): ...new function.
10611         (pass_ipa_free_inline_summary): Replace declaration with that of...
10612         (make_pass_ipa_free_inline_summary): ...new function.
10613         (pass_ipa_cp): Replace declaration with that of...
10614         (make_pass_ipa_cp): ...new function.
10615         (pass_ipa_reference): Replace declaration with that of...
10616         (make_pass_ipa_reference): ...new function.
10617         (pass_ipa_pure_const): Replace declaration with that of...
10618         (make_pass_ipa_pure_const): ...new function.
10619         (pass_ipa_pta): Replace declaration with that of...
10620         (make_pass_ipa_pta): ...new function.
10621         (pass_ipa_lto_finish_out): Replace declaration with that of...
10622         (make_pass_ipa_lto_finish_out): ...new function.
10623         (pass_ipa_tm): Replace declaration with that of...
10624         (make_pass_ipa_tm): ...new function.
10625         (pass_ipa_profile): Replace declaration with that of...
10626         (make_pass_ipa_profile): ...new function.
10627         (pass_ipa_cdtor_merge): Replace declaration with that of...
10628         (make_pass_ipa_cdtor_merge): ...new function.
10629         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10630         of...
10631         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10632         (pass_init_datastructures): Replace declaration with that of...
10633         (make_pass_init_datastructures): ...new function.
10634         (pass_fixup_cfg): Replace declaration with that of...
10635         (make_pass_fixup_cfg): ...new function.
10636         (pass_expand): Replace declaration with that of...
10637         (make_pass_expand): ...new function.
10638         (pass_instantiate_virtual_regs): Replace declaration with that of...
10639         (make_pass_instantiate_virtual_regs): ...new function.
10640         (pass_rtl_fwprop): Replace declaration with that of...
10641         (make_pass_rtl_fwprop): ...new function.
10642         (pass_rtl_fwprop_addr): Replace declaration with that of...
10643         (make_pass_rtl_fwprop_addr): ...new function.
10644         (pass_jump): Replace declaration with that of...
10645         (make_pass_jump): ...new function.
10646         (pass_jump2): Replace declaration with that of...
10647         (make_pass_jump2): ...new function.
10648         (pass_lower_subreg): Replace declaration with that of...
10649         (make_pass_lower_subreg): ...new function.
10650         (pass_cse): Replace declaration with that of...
10651         (make_pass_cse): ...new function.
10652         (pass_fast_rtl_dce): Replace declaration with that of...
10653         (make_pass_fast_rtl_dce): ...new function.
10654         (pass_ud_rtl_dce): Replace declaration with that of...
10655         (make_pass_ud_rtl_dce): ...new function.
10656         (pass_rtl_dce): Replace declaration with that of...
10657         (make_pass_rtl_dce): ...new function.
10658         (pass_rtl_dse1): Replace declaration with that of...
10659         (make_pass_rtl_dse1): ...new function.
10660         (pass_rtl_dse2): Replace declaration with that of...
10661         (make_pass_rtl_dse2): ...new function.
10662         (pass_rtl_dse3): Replace declaration with that of...
10663         (make_pass_rtl_dse3): ...new function.
10664         (pass_rtl_cprop): Replace declaration with that of...
10665         (make_pass_rtl_cprop): ...new function.
10666         (pass_rtl_pre): Replace declaration with that of...
10667         (make_pass_rtl_pre): ...new function.
10668         (pass_rtl_hoist): Replace declaration with that of...
10669         (make_pass_rtl_hoist): ...new function.
10670         (pass_rtl_store_motion): Replace declaration with that of...
10671         (make_pass_rtl_store_motion): ...new function.
10672         (pass_cse_after_global_opts): Replace declaration with that of...
10673         (make_pass_cse_after_global_opts): ...new function.
10674         (pass_rtl_ifcvt): Replace declaration with that of...
10675         (make_pass_rtl_ifcvt): ...new function.
10676         (pass_into_cfg_layout_mode): Replace declaration with that of...
10677         (make_pass_into_cfg_layout_mode): ...new function.
10678         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10679         (make_pass_outof_cfg_layout_mode): ...new function.
10680         (pass_loop2): Replace declaration with that of...
10681         (make_pass_loop2): ...new function.
10682         (pass_rtl_loop_init): Replace declaration with that of...
10683         (make_pass_rtl_loop_init): ...new function.
10684         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10685         (make_pass_rtl_move_loop_invariants): ...new function.
10686         (pass_rtl_unswitch): Replace declaration with that of...
10687         (make_pass_rtl_unswitch): ...new function.
10688         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10689         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10690         (pass_rtl_doloop): Replace declaration with that of...
10691         (make_pass_rtl_doloop): ...new function.
10692         (pass_rtl_loop_done): Replace declaration with that of...
10693         (make_pass_rtl_loop_done): ...new function.
10694         (pass_web): Replace declaration with that of...
10695         (make_pass_web): ...new function.
10696         (pass_cse2): Replace declaration with that of...
10697         (make_pass_cse2): ...new function.
10698         (pass_df_initialize_opt): Replace declaration with that of...
10699         (make_pass_df_initialize_opt): ...new function.
10700         (pass_df_initialize_no_opt): Replace declaration with that of...
10701         (make_pass_df_initialize_no_opt): ...new function.
10702         (pass_reginfo_init): Replace declaration with that of...
10703         (make_pass_reginfo_init): ...new function.
10704         (pass_inc_dec): Replace declaration with that of...
10705         (make_pass_inc_dec): ...new function.
10706         (pass_stack_ptr_mod): Replace declaration with that of...
10707         (make_pass_stack_ptr_mod): ...new function.
10708         (pass_initialize_regs): Replace declaration with that of...
10709         (make_pass_initialize_regs): ...new function.
10710         (pass_combine): Replace declaration with that of...
10711         (make_pass_combine): ...new function.
10712         (pass_if_after_combine): Replace declaration with that of...
10713         (make_pass_if_after_combine): ...new function.
10714         (pass_ree): Replace declaration with that of...
10715         (make_pass_ree): ...new function.
10716         (pass_partition_blocks): Replace declaration with that of...
10717         (make_pass_partition_blocks): ...new function.
10718         (pass_match_asm_constraints): Replace declaration with that of...
10719         (make_pass_match_asm_constraints): ...new function.
10720         (pass_regmove): Replace declaration with that of...
10721         (make_pass_regmove): ...new function.
10722         (pass_split_all_insns): Replace declaration with that of...
10723         (make_pass_split_all_insns): ...new function.
10724         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10725         (make_pass_fast_rtl_byte_dce): ...new function.
10726         (pass_lower_subreg2): Replace declaration with that of...
10727         (make_pass_lower_subreg2): ...new function.
10728         (pass_mode_switching): Replace declaration with that of...
10729         (make_pass_mode_switching): ...new function.
10730         (pass_sms): Replace declaration with that of...
10731         (make_pass_sms): ...new function.
10732         (pass_sched): Replace declaration with that of...
10733         (make_pass_sched): ...new function.
10734         (pass_ira): Replace declaration with that of...
10735         (make_pass_ira): ...new function.
10736         (pass_reload): Replace declaration with that of...
10737         (make_pass_reload): ...new function.
10738         (pass_clean_state): Replace declaration with that of...
10739         (make_pass_clean_state): ...new function.
10740         (pass_branch_prob): Replace declaration with that of...
10741         (make_pass_branch_prob): ...new function.
10742         (pass_value_profile_transformations): Replace declaration with that
10743         of...
10744         (make_pass_value_profile_transformations): ...new function.
10745         (pass_postreload_cse): Replace declaration with that of...
10746         (make_pass_postreload_cse): ...new function.
10747         (pass_gcse2): Replace declaration with that of...
10748         (make_pass_gcse2): ...new function.
10749         (pass_split_after_reload): Replace declaration with that of...
10750         (make_pass_split_after_reload): ...new function.
10751         (pass_branch_target_load_optimize1): Replace declaration with that
10752         of...
10753         (make_pass_branch_target_load_optimize1): ...new function.
10754         (pass_thread_prologue_and_epilogue): Replace declaration with that
10755         of...
10756         (make_pass_thread_prologue_and_epilogue): ...new function.
10757         (pass_stack_adjustments): Replace declaration with that of...
10758         (make_pass_stack_adjustments): ...new function.
10759         (pass_peephole2): Replace declaration with that of...
10760         (make_pass_peephole2): ...new function.
10761         (pass_if_after_reload): Replace declaration with that of...
10762         (make_pass_if_after_reload): ...new function.
10763         (pass_regrename): Replace declaration with that of...
10764         (make_pass_regrename): ...new function.
10765         (pass_cprop_hardreg): Replace declaration with that of...
10766         (make_pass_cprop_hardreg): ...new function.
10767         (pass_reorder_blocks): Replace declaration with that of...
10768         (make_pass_reorder_blocks): ...new function.
10769         (pass_branch_target_load_optimize2): Replace declaration with that
10770         of...
10771         (make_pass_branch_target_load_optimize2): ...new function.
10772         (pass_leaf_regs): Replace declaration with that of...
10773         (make_pass_leaf_regs): ...new function.
10774         (pass_split_before_sched2): Replace declaration with that of...
10775         (make_pass_split_before_sched2): ...new function.
10776         (pass_compare_elim_after_reload): Replace declaration with that of...
10777         (make_pass_compare_elim_after_reload): ...new function.
10778         (pass_sched2): Replace declaration with that of...
10779         (make_pass_sched2): ...new function.
10780         (pass_stack_regs): Replace declaration with that of...
10781         (make_pass_stack_regs): ...new function.
10782         (pass_stack_regs_run): Replace declaration with that of...
10783         (make_pass_stack_regs_run): ...new function.
10784         (pass_df_finish): Replace declaration with that of...
10785         (make_pass_df_finish): ...new function.
10786         (pass_compute_alignments): Replace declaration with that of...
10787         (make_pass_compute_alignments): ...new function.
10788         (pass_duplicate_computed_gotos): Replace declaration with that of...
10789         (make_pass_duplicate_computed_gotos): ...new function.
10790         (pass_variable_tracking): Replace declaration with that of...
10791         (make_pass_variable_tracking): ...new function.
10792         (pass_free_cfg): Replace declaration with that of...
10793         (make_pass_free_cfg): ...new function.
10794         (pass_machine_reorg): Replace declaration with that of...
10795         (make_pass_machine_reorg): ...new function.
10796         (pass_cleanup_barriers): Replace declaration with that of...
10797         (make_pass_cleanup_barriers): ...new function.
10798         (pass_delay_slots): Replace declaration with that of...
10799         (make_pass_delay_slots): ...new function.
10800         (pass_split_for_shorten_branches): Replace declaration with that of...
10801         (make_pass_split_for_shorten_branches): ...new function.
10802         (pass_split_before_regstack): Replace declaration with that of...
10803         (make_pass_split_before_regstack): ...new function.
10804         (pass_convert_to_eh_region_ranges): Replace declaration with that
10805         of...
10806         (make_pass_convert_to_eh_region_ranges): ...new function.
10807         (pass_shorten_branches): Replace declaration with that of...
10808         (make_pass_shorten_branches): ...new function.
10809         (pass_set_nothrow_function_flags): Replace declaration with that of...
10810         (make_pass_set_nothrow_function_flags): ...new function.
10811         (pass_dwarf2_frame): Replace declaration with that of...
10812         (make_pass_dwarf2_frame): ...new function.
10813         (pass_final): Replace declaration with that of...
10814         (make_pass_final): ...new function.
10815         (pass_rtl_seqabstr): Replace declaration with that of...
10816         (make_pass_rtl_seqabstr): ...new function.
10817         (pass_release_ssa_names): Replace declaration with that of...
10818         (make_pass_release_ssa_names): ...new function.
10819         (pass_early_inline): Replace declaration with that of...
10820         (make_pass_early_inline): ...new function.
10821         (pass_inline_parameters): Replace declaration with that of...
10822         (make_pass_inline_parameters): ...new function.
10823         (pass_update_address_taken): Replace declaration with that of...
10824         (make_pass_update_address_taken): ...new function.
10825         (pass_convert_switch): Replace declaration with that of...
10826         (make_pass_convert_switch): ...new function.
10827         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10828         to a subclass of simple_ipa_opt_pass along with...
10829         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10830         (make_pass_ipa_tree_profile): ...new function.
10831         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10832         subclass of gimple_opt_pass along with...
10833         (pass_data_sra_early): ...new pass_data instance and...
10834         (make_pass_sra_early): ...new function.
10835         (pass_sra): Convert from a global struct to a subclass of
10836         gimple_opt_pass along with...
10837         (pass_data_sra): ...new pass_data instance and...
10838         (make_pass_sra): ...new function.
10839         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10840         gimple_opt_pass along with...
10841         (pass_data_early_ipa_sra): ...new pass_data instance and...
10842         (make_pass_early_ipa_sra): ...new function.
10843         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10844         subclass of gimple_opt_pass along with...
10845         (pass_data_ccp): ...new pass_data instance and...
10846         (make_pass_ccp): ...new function.
10847         (pass_fold_builtins): Convert from a global struct to a subclass of
10848         gimple_opt_pass along with...
10849         (pass_data_fold_builtins): ...new pass_data instance and...
10850         (make_pass_fold_builtins): ...new function.
10851         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10852         subclass of gimple_opt_pass along with...
10853         (pass_data_copy_prop): ...new pass_data instance and...
10854         (make_pass_copy_prop): ...new function.
10855         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10856         global struct to a subclass of gimple_opt_pass along with...
10857         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10858         (make_pass_rename_ssa_copies): ...new function.
10859         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10860         subclass of gimple_opt_pass along with...
10861         (pass_data_dce): ...new pass_data instance and...
10862         (make_pass_dce): ...new function.
10863         (pass_dce_loop): Convert from a global struct to a subclass of
10864         gimple_opt_pass along with...
10865         (pass_data_dce_loop): ...new pass_data instance and...
10866         (make_pass_dce_loop): ...new function.
10867         (pass_cd_dce): Convert from a global struct to a subclass of
10868         gimple_opt_pass along with...
10869         (pass_data_cd_dce): ...new pass_data instance and...
10870         (make_pass_cd_dce): ...new function.
10871         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10872         subclass of gimple_opt_pass along with...
10873         (pass_data_dominator): ...new pass_data instance and...
10874         (make_pass_dominator): ...new function.
10875         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10876         gimple_opt_pass along with...
10877         (pass_data_phi_only_cprop): ...new pass_data instance and...
10878         (make_pass_phi_only_cprop): ...new function.
10879         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10880         subclass of gimple_opt_pass along with...
10881         (pass_data_dse): ...new pass_data instance and...
10882         (make_pass_dse): ...new function.
10883         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10884         a subclass of gimple_opt_pass along with...
10885         (pass_data_forwprop): ...new pass_data instance and...
10886         (make_pass_forwprop): ...new function.
10887         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10888         struct to a subclass of gimple_opt_pass along with...
10889         (pass_data_tree_ifcombine): ...new pass_data instance and...
10890         (make_pass_tree_ifcombine): ...new function.
10891         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10892         subclass of gimple_opt_pass along with...
10893         (pass_data_ch): ...new pass_data instance and...
10894         (make_pass_ch): ...new function.
10895         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10896         subclass of gimple_opt_pass along with...
10897         (pass_data_tree_loop): ...new pass_data instance and...
10898         (make_pass_tree_loop): ...new function.
10899         (pass_tree_loop_init): Convert from a global struct to a subclass of
10900         gimple_opt_pass along with...
10901         (pass_data_tree_loop_init): ...new pass_data instance and...
10902         (make_pass_tree_loop_init): ...new function.
10903         (pass_lim): Convert from a global struct to a subclass of
10904         gimple_opt_pass along with...
10905         (pass_data_lim): ...new pass_data instance and...
10906         (make_pass_lim): ...new function.
10907         (pass_tree_unswitch): Convert from a global struct to a subclass of
10908         gimple_opt_pass along with...
10909         (pass_data_tree_unswitch): ...new pass_data instance and...
10910         (make_pass_tree_unswitch): ...new function.
10911         (pass_predcom): Convert from a global struct to a subclass of
10912         gimple_opt_pass along with...
10913         (pass_data_predcom): ...new pass_data instance and...
10914         (make_pass_predcom): ...new function.
10915         (pass_vectorize): Convert from a global struct to a subclass of
10916         gimple_opt_pass along with...
10917         (pass_data_vectorize): ...new pass_data instance and...
10918         (make_pass_vectorize): ...new function.
10919         (pass_graphite): Convert from a global struct to a subclass of
10920         gimple_opt_pass along with...
10921         (pass_data_graphite): ...new pass_data instance and...
10922         (make_pass_graphite): ...new function.
10923         (pass_graphite_transforms): Convert from a global struct to a subclass
10924         of gimple_opt_pass along with...
10925         (pass_data_graphite_transforms): ...new pass_data instance and...
10926         (make_pass_graphite_transforms): ...new function.
10927         (pass_check_data_deps): Convert from a global struct to a subclass of
10928         gimple_opt_pass along with...
10929         (pass_data_check_data_deps): ...new pass_data instance and...
10930         (make_pass_check_data_deps): ...new function.
10931         (pass_iv_canon): Convert from a global struct to a subclass of
10932         gimple_opt_pass along with...
10933         (pass_data_iv_canon): ...new pass_data instance and...
10934         (make_pass_iv_canon): ...new function.
10935         (pass_scev_cprop): Convert from a global struct to a subclass of
10936         gimple_opt_pass along with...
10937         (pass_data_scev_cprop): ...new pass_data instance and...
10938         (make_pass_scev_cprop): ...new function.
10939         (pass_record_bounds): Convert from a global struct to a subclass of
10940         gimple_opt_pass along with...
10941         (pass_data_record_bounds): ...new pass_data instance and...
10942         (make_pass_record_bounds): ...new function.
10943         (pass_complete_unroll): Convert from a global struct to a subclass of
10944         gimple_opt_pass along with...
10945         (pass_data_complete_unroll): ...new pass_data instance and...
10946         (make_pass_complete_unroll): ...new function.
10947         (pass_complete_unrolli): Convert from a global struct to a subclass of
10948         gimple_opt_pass along with...
10949         (pass_data_complete_unrolli): ...new pass_data instance and...
10950         (make_pass_complete_unrolli): ...new function.
10951         (pass_parallelize_loops): Convert from a global struct to a subclass
10952         of gimple_opt_pass along with...
10953         (pass_data_parallelize_loops): ...new pass_data instance and...
10954         (make_pass_parallelize_loops): ...new function.
10955         (pass_loop_prefetch): Convert from a global struct to a subclass of
10956         gimple_opt_pass along with...
10957         (pass_data_loop_prefetch): ...new pass_data instance and...
10958         (make_pass_loop_prefetch): ...new function.
10959         (pass_iv_optimize): Convert from a global struct to a subclass of
10960         gimple_opt_pass along with...
10961         (pass_data_iv_optimize): ...new pass_data instance and...
10962         (make_pass_iv_optimize): ...new function.
10963         (pass_tree_loop_done): Convert from a global struct to a subclass of
10964         gimple_opt_pass along with...
10965         (pass_data_tree_loop_done): ...new pass_data instance and...
10966         (make_pass_tree_loop_done): ...new function.
10967         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10968         struct to a subclass of gimple_opt_pass along with...
10969         (pass_data_cse_reciprocals): ...new pass_data instance and...
10970         (make_pass_cse_reciprocals): ...new function.
10971         (pass_cse_sincos): Convert from a global struct to a subclass of
10972         gimple_opt_pass along with...
10973         (pass_data_cse_sincos): ...new pass_data instance and...
10974         (make_pass_cse_sincos): ...new function.
10975         (pass_optimize_bswap): Convert from a global struct to a subclass of
10976         gimple_opt_pass along with...
10977         (pass_data_optimize_bswap): ...new pass_data instance and...
10978         (make_pass_optimize_bswap): ...new function.
10979         (pass_optimize_widening_mul): Convert from a global struct to a
10980         subclass of gimple_opt_pass along with...
10981         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10982         (make_pass_optimize_widening_mul): ...new function.
10983         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10984         subclass of gimple_opt_pass along with...
10985         (pass_data_phiopt): ...new pass_data instance and...
10986         (make_pass_phiopt): ...new function.
10987         (pass_cselim): Convert from a global struct to a subclass of
10988         gimple_opt_pass along with...
10989         (pass_data_cselim): ...new pass_data instance and...
10990         (make_pass_cselim): ...new function.
10991         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10992         subclass of gimple_opt_pass along with...
10993         (pass_data_phiprop): ...new pass_data instance and...
10994         (make_pass_phiprop): ...new function.
10995         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10996         subclass of gimple_opt_pass along with...
10997         (pass_data_pre): ...new pass_data instance and...
10998         (make_pass_pre): ...new function.
10999         (pass_fre): Convert from a global struct to a subclass of
11000         gimple_opt_pass along with...
11001         (pass_data_fre): ...new pass_data instance and...
11002         (make_pass_fre): ...new function.
11003         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
11004         subclass of gimple_opt_pass along with...
11005         (pass_data_reassoc): ...new pass_data instance and...
11006         (make_pass_reassoc): ...new function.
11007         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
11008         subclass of gimple_opt_pass along with...
11009         (pass_data_sink_code): ...new pass_data instance and...
11010         (make_pass_sink_code): ...new function.
11011         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
11012         subclass of gimple_opt_pass along with...
11013         (pass_data_strlen): ...new pass_data instance and...
11014         (make_pass_strlen): ...new function.
11015         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
11016         struct to a subclass of gimple_opt_pass along with...
11017         (pass_data_build_alias): ...new pass_data instance and...
11018         (make_pass_build_alias): ...new function.
11019         (pass_build_ealias): Convert from a global struct to a subclass of
11020         gimple_opt_pass along with...
11021         (pass_data_build_ealias): ...new pass_data instance and...
11022         (make_pass_build_ealias): ...new function.
11023         (pass_ipa_pta): Convert from a global struct to a subclass of
11024         simple_ipa_opt_pass along with...
11025         (pass_data_ipa_pta): ...new pass_data instance and...
11026         (make_pass_ipa_pta): ...new function.
11027         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11028         subclass of gimple_opt_pass along with...
11029         (pass_data_uncprop): ...new pass_data instance and...
11030         (make_pass_uncprop): ...new function.
11031         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11032         global struct to a subclass of gimple_opt_pass along with...
11033         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11034         (make_pass_late_warn_uninitialized): ...new function.
11035         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11036         to a subclass of gimple_opt_pass along with...
11037         (pass_data_init_datastructures): ...new pass_data instance and...
11038         (make_pass_init_datastructures): ...new function.
11039         (pass_early_warn_uninitialized): Convert from a global struct to a
11040         subclass of gimple_opt_pass along with...
11041         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11042         (make_pass_early_warn_uninitialized): ...new function.
11043         (pass_update_address_taken): Convert from a global struct to a
11044         subclass of gimple_opt_pass along with...
11045         (pass_data_update_address_taken): ...new pass_data instance and...
11046         (make_pass_update_address_taken): ...new function.
11047         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11048         struct to a subclass of gimple_opt_pass along with...
11049         (pass_data_release_ssa_names): ...new pass_data instance and...
11050         (make_pass_release_ssa_names): ...new function.
11051         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11052         subclass of gimple_opt_pass along with...
11053         (pass_data_stdarg): ...new pass_data instance and...
11054         (make_pass_stdarg): ...new function.
11055         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11056         global struct to a subclass of gimple_opt_pass along with...
11057         (pass_data_convert_switch): ...new pass_data instance and...
11058         (make_pass_convert_switch): ...new function.
11059         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11060         to a subclass of gimple_opt_pass along with...
11061         (pass_data_tail_recursion): ...new pass_data instance and...
11062         (make_pass_tail_recursion): ...new function.
11063         (pass_tail_calls): Convert from a global struct to a subclass of
11064         gimple_opt_pass along with...
11065         (pass_data_tail_calls): ...new pass_data instance and...
11066         (make_pass_tail_calls): ...new function.
11067         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11068         struct to a subclass of gimple_opt_pass along with...
11069         (pass_data_lower_vector): ...new pass_data instance and...
11070         (make_pass_lower_vector): ...new function.
11071         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11072         gimple_opt_pass along with...
11073         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11074         (make_pass_lower_vector_ssa): ...new function.
11075         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11076         to a subclass of gimple_opt_pass along with...
11077         (pass_data_slp_vectorize): ...new pass_data instance and...
11078         (make_pass_slp_vectorize): ...new function.
11079         (pass_ipa_increase_alignment): Convert from a global struct to a
11080         subclass of simple_ipa_opt_pass along with...
11081         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11082         (make_pass_ipa_increase_alignment): ...new function.
11083         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11084         gimple_opt_pass along with...
11085         (pass_data_vrp): ...new pass_data instance and...
11086         (make_pass_vrp): ...new function.
11087         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11088         subclass of simple_ipa_opt_pass along with...
11089         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11090         (make_pass_ipa_free_lang_data): ...new function.
11091         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11092         gimple_opt_pass along with...
11093         (pass_data_tsan): ...new pass_data instance and...
11094         (make_pass_tsan): ...new function.
11095         (pass_tsan_O0): Convert from a global struct to a subclass of
11096         gimple_opt_pass along with...
11097         (pass_data_tsan_O0): ...new pass_data instance and...
11098         (make_pass_tsan_O0): ...new function.
11099         * var-tracking.c (pass_variable_tracking): Convert from a global
11100         struct to a subclass of rtl_opt_pass along with...
11101         (pass_data_variable_tracking): ...new pass_data instance and...
11102         (make_pass_variable_tracking): ...new function.
11103         * web.c (pass_web): Convert from a global struct to a subclass of
11104         rtl_opt_pass along with...
11105         (pass_data_web): ...new pass_data instance and...
11106         (make_pass_web): ...new function.
11107         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11108         declaration with that of...
11109         (make_pass_mode_switch_use): ...new function.
11110         (pass_resolve_sw_modes): Replace declaration with that of...
11111         (make_pass_resolve_sw_modes): ...new function.
11112         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11113         from a global struct to a subclass of rtl_opt_pass along with...
11114         (pass_data_mode_switch_use): ...new pass_data instance and...
11115         (make_pass_mode_switch_use): ...new function.
11116         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11117         from a global struct to a subclass of rtl_opt_pass along with...
11118         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11119         (make_pass_resolve_sw_modes): ...new function.
11120         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11121         struct to a subclass of rtl_opt_pass along with...
11122         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11123         (make_pass_insert_vzeroupper): ...new function.
11124         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11125         global struct to a subclass of rtl_opt_pass along with...
11126         (pass_data_work_around_errata): ...new pass_data instance and...
11127         (make_pass_work_around_errata): ...new function.
11128         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11129         struct to a subclass of rtl_opt_pass along with...
11130         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11131         (make_pass_mips_machine_reorg2): ...new function.
11133 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11135         * passes.c (pass_manager::operator new): New.
11137 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11139         Handwritten part of conversion of passes to C++ classes.
11141         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11142         (toplev.o): Add dep on PASS_MANAGER_H.
11143         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11144         of early local pases to reflect this moving from a global to a
11145         member of gcc::pass_manager.
11146         (cgraph_add_new_function): Likewise.
11147         * lto-cgraph.c (lto_output_node): Update for conversion of
11148         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11149         * passes.c (opt_pass::clone): New.
11150         (opt_pass::gate): New.
11151         (opt_pass::execute): New.
11152         (opt_pass::opt_pass): New.
11153         (pass_manager::execute_early_local_passes): New.
11154         (pass_manager::execute_pass_mode_switching): new.
11155         (finish_optimization_passes): Convert to...
11156         (pass_manager::finish_optimization_passes): ...this.
11157         (finish_optimization_passes): Update for conversion of passes to
11158         C++ classes.
11159         (register_dump_files_1): Use has_gate since we cannot portably
11160         check a vtable entry against NULL.
11161         (dump_one_pass): Likewise.
11162         (ipa_write_summaries_2): Likewise.
11163         (ipa_write_optimization_summaries_1): Likewise.
11164         (ipa_read_summaries_1): Likewise.
11165         (ipa_read_optimization_summaries_1): Likewise.
11166         (execute_ipa_stmt_fixups): Likewise.
11167         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11168         pass-creation functions rather than wiring up globals, and
11169         storing the results in fields of pass_manager generated using
11170         pass-instances.def.
11171         (pass_manager::dump_profile_report): Update for conversion of
11172         passes to C++ classes.
11173         (pass_manager::execute_ipa_summary_passes): Likewise.
11174         (execute_one_ipa_transform_pass): Likewise.
11175         (execute_one_pass): Use has_gate and has_execute since we cannot
11176         portably check a vtable entry against NULL.
11177         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11178         (pass_manager): Use pass-instances.def to add fields for the
11179         various pass instances.
11180         * toplev.c (finalize): Update for move of
11181         finish_optimization_passes to a method of gcc::pass_manager.
11182         * toplev.h (finish_optimization_passes): Move to method of class
11183         pass_manager.
11184         * tree-pass.h (struct pass_data): New.
11185         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11186         (opt_pass::gate): Convert to virtual function.
11187         (opt_pass::~opt_pass): New.
11188         (opt_pass::clone): New.
11189         (opt_pass::execute): Convert to virtual function.
11190         (opt_pass::opt_pass): New.
11191         (opt_pass::ctxt_): new.
11192         (gimple_opt_pass): Convert to subclass of opt_pass.
11193         (gimple_opt_pass::gimple_opt_pass): New.
11194         (rtl_opt_pass): Convert to subclass of opt_pass.
11195         (rtl_opt_pass::rtl_opt_pass): New.
11196         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11197         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11198         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11199         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11200         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11201         invocation of pass_mode_switching to reflect this moving from a
11202         global to a member of gcc::pass_manager.
11203         (ix86_option_override): Rework how pass_insert_vzeroupper is
11204         added to the pass_manager to reflect autogenerated changes.
11205         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11207 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11209         PR rtl-optimization/57708
11210         * recog.c (peep2_find_free_register): Validate all regs in a
11211         multi-reg mode.
11213 2013-08-05  Jan Hubicka  <jh@suse.cz>
11215         PR lto/57602
11216         * cgraph.c (verify_cgraph_node): Accept local flags from other
11217         partitions.
11218         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11219         (function_and_variable_visibility): Likewise.
11220         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11222 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11224         * graph.c (init_graph_slim_pretty_print): Remove.
11225         (print_graph_cfg): Do not call it.  Use local pretty printer.
11226         (start_graph_dump): Likewise.
11228 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11230         * gimple-pretty-print.c (buffer): Remove.
11231         (initialized): Likewise.
11232         (maybe_init_pretty_print): Likewise.
11233         (print_gimple_stmt): Do not call it.  Use non-static local
11234         pretty_printer variable.
11235         (print_gimple_expr): Likewise.
11236         (print_gimple_seq): Likewise.
11237         (gimple_dump_bb): Likewise.
11239 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11241         * asan.c (asan_pp): Remove.
11242         (asan_pp_initialized): Likewise.
11243         (asan_pp_initialize): Likewise.
11244         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11245         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11246         (asan_add_global): Likewise.
11248 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11250         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11251         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11252         * pretty-print.c (pp_formatted_text_data): Likewise.
11253         (pp_indent): Rename from pp_base_indent.
11254         (pp_format): Rename from pp_base_format.
11255         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11256         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11257         (pp_flush): Rename from pp_base_flush.
11258         (pp_set_line_maximum_length): Rename from
11259         pp_base_set_line_maximum_length.
11260         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11261         (pp_set_prefix): Rename from pp_base_set_prefix.
11262         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11263         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11264         (pp_append_text): Rename from pp_base_append_text.
11265         (pp_formatted_text): Rename from pp_base_formatted_text.
11266         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11267         (pp_remaining_character_count_for_line): Rename from
11268         pp_base_remaining_character_count_for_line.
11269         (pp_newline): Rename from pp_base_newline.
11270         (pp_character): Rename from pp_base_character.
11271         (pp_string): Rename from pp_base_string.
11272         (pp_maybe_space): Rename from pp_base_maybe_space.
11273         * asan.c (asan_pp_string): Adjust.
11274         (asan_emit_stack_protection): Likewise.
11275         (asan_add_global): Likewise.
11276         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11277         * tree-mudflap.c (mf_varname_tree): Likewise.
11278         * tree-pretty-print.c (pp_tree_identifier): Rename from
11279         pp_base_tree_identifier.
11280         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11281         Declare as function.
11283 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11285         * pretty-print.h (pp_bar_bar): New.
11286         (pp_ampersand_ampersand): Likewise.
11287         (pp_less_equal): Likewise.
11288         (pp_greater_equal): Likewise.
11289         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11290         printer functions instead of pp_string or operators and punctuators.
11291         (dump_gimple_call): Likewise.
11292         (dump_gimple_omp_for): Likewise.
11293         (dump_gimple_transaction): Likewise.
11294         (dump_gimple_phi): Likewise.
11295         (pp_gimple_stmt_1): Likewise.
11296         * sched-vis.c (print_insn): Likewise.
11297         * tree-mudflap.c (mf_varname_tree): Likewise.
11298         * tree-pretty-print.c (dump_block_node): Likewise.
11299         (dump_generic_node): Likewise.
11301 2013-08-02  Jan Hubicka  <jh@suse.cz>
11303         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11304         boundaries.
11305         * lto-streamer-out.c (tree_is_indexable): Results decls and
11306         parm decls are not indexable.
11307         (DFS_write_tree_body): Do not follow args and results.
11308         (hash_tree): Likewise.
11309         (output_functions): Rearrange so struct function is needed
11310         only when real body is output; be able to also ouptut abstract
11311         functions; output DECL_ARGUMENTS and DECL_RESULT.
11312         (lto_output): When not in WPA, ale store abstract functions.
11313         (write_symbol): Do not care about RESULT_DECL.
11314         (output_symbol_p): Handle correctly sbtract decls.
11315         * lto-streamer-in.c (input_function): Rearrange so struct
11316         function can be NULL at entry; allow streaming of
11317         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11318         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11319         sanity check during LTO.
11320         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11321         RESULT_DECl and DECL_ARGUMENTS.
11322         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11323         Likewise.
11325 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11327         * pretty-print.h (pp_underscore): New.
11328         (pp_comma): Tidy.
11329         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11330         printer functions instead of pp_character.
11331         (dump_binary_rhs): Likewise.
11332         (dump_ternary_rhs): Likewise.
11333         (dump_gimple_call_args): Likewise.
11334         (pp_points_to_solution): Likewise.
11335         (dump_gimple_call): Likewise.
11336         (dump_gimple_switch): Likewise.
11337         (dump_gimple_cond): Likewise.
11338         (dump_gimple_bind): Likewise.
11339         (dump_gimple_try): Likewise.
11340         (dump_gimple_omp_for): Likewise.
11341         (dump_gimple_omp_continue): Likewise.
11342         (dump_gimple_omp_single): Likewise.
11343         (dump_gimple_omp_sections): Likewise.
11344         (dump_gimple_omp_block): Likewise.
11345         (dump_gimple_omp_critical): Likewise.
11346         (dump_gimple_transaction): Likewise.
11347         (dump_gimple_asm): Likewise.
11348         (dump_gimple_phi): Likewise.
11349         (dump_gimple_omp_parallel): Likewise.
11350         (dump_gimple_omp_task): Likewise.
11351         (dump_gimple_omp_atomic_load): Likewise.
11352         (dump_gimple_omp_atomic_store): Likewise.
11353         (dump_gimple_mem_ops): Likewise.
11354         (pp_gimple_stmt_1): Likewise.
11355         (pp_cfg_jump): Likewise.
11356         (dump_implicit_edges): Likewise.
11357         (gimple_dump_bb_for_graph): Likewise.
11358         * graph.c (draw_cfg_node): Likewise.
11359         * langhooks.c (lhd_print_error_function): Likewise.
11360         * sched-vis.c (print_exp): Likewise.
11361         (print_value): Likewise.
11362         (print_pattern): Likewise.
11363         (print_insn): Likewise.
11364         (rtl_dump_bb_for_graph): Likewise.
11365         * tree-pretty-print.c (dump_function_declaration): Likewise.
11366         (dump_array_domain): Likewise.
11367         (dump_omp_clause): Likewise.
11368         (dump_location): Likewise.
11369         (dump_generic_node): Likewise.
11370         (print_struct_decl): Likewise.
11371         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11373 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11375         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11376         candidate table when replacing a candidate statement.
11377         (replace_rhs_if_not_dup): Likewise.
11378         (replace_one_candidate): Likewise.
11380 2013-08-02  Jan Hubicka  <jh@suse.cz>
11381             Martin Liska  <marxin.liska@gmail.com>
11383         * cgraphunit.c (add_new_function): Fix logic when adding from
11384         late IPA pass.
11385         (assemble_thunk): Rename to ...
11386         (expand_thunk); .. this one; export; get it working with
11387         general functions; make produced gimple valid.
11388         * cgraph.h (expand_thunk): Declare.
11390 2013-08-02  Jan Hubicka  <jh@suse.cz>
11392         * ipa-cp.c (gather_context_independent_values): Use
11393         ipa_get_param_move_cost.
11394         (get_replacement_map): Remove PARAM; move parameter folding
11395         into tree-inline.c
11396         (create_specialized_node): Update.
11397         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11398         assert that we have gimple body; update move_cost.
11399         (count_formal_params): Assert that we have gimple body.
11400         (ipa_dump_param): New function.
11401         (ipa_alloc_node_params): Break out from ...
11402         (ipa_initialize_node_params): ... here.
11403         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11404         (ipa_write_node_info): Stream move costs.
11405         (ipa_read_node_info): Read move costs.
11406         (ipa_update_after_lto_read): Do not recompute node params.
11407         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11408         (ipa_get_param): Check we are not in WPA.
11409         (ipa_get_param_move_cost): New.
11410         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11411         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11412         parm numbers to be present.
11414 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11416         PR rtl-optimization/58048
11417         * lra-constraints.c (process_alt_operands): Don't check asm
11418         operand on register.
11420 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11422         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11423         the implied StoreLoad barrier for atomic operations if before.
11425 2013-08-02  Jan Hubicka  <jh@suse.cz>
11426             Martin Liska  <marxin.liska@gmail.com>
11428         * cgraph.c (cgraph_function_body_availability): Do not check
11429         cgraph flags.
11430         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11431         symtab_node_availability): Declare.
11432         * ipa.c (can_replace_by_local_alias): New.
11433         (function_and_variable_visibility): Use it.
11434         * symtab.c (symtab_for_node_and_aliases,
11435         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11437 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11439         PR rtl-optimization/57963
11440         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11441         (lra_constraints): Use them.
11443 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11445         * config/arm/types.md (define_attr "type"): Add "load_acq"
11446         and "store_rel".
11447         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11448         changes.
11449         (cortex_a53_store1): Likewise.
11451 2013-08-01  Jan Hubicka  <jh@suse.cz>
11453         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11454         partitions are not needed.
11456 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11458         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11459         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11460         MAYBE_NON_Q_CLASS_P where appropriate.
11462 2013-08-01  Jan Hubicka  <jh@suse.cz>
11464         * cgraph.h (release_function_body): Declare.
11465         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11466         of unused delcarations.
11468 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11470         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11471         RTL form when subtracting a constant.
11473 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11475         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11476         Generate canonical plus rtx with negated immediate instead of minus
11477         where appropriate.
11478         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11480 2013-08-01  Jan Hubicka  <jh@suse.cz>
11482         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11483         (cgraph_release_function_body): Likewise.
11484         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11485         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11486         to used_as_abstract_origin.
11487         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11488         symbols used as abstract origins.
11489         * cgraphunit.c (analyze_functions): Update.
11490         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11491         used_as_abstract_origin.
11492         * tree-inline.c (tree_function_versioning): Update
11493         used_as_abstract_origin; be ready for DECL_RESULT and
11494         DECL_ARGUMENTS to be NULL.
11496         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11497         for abstract functions.
11498         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11499         real symbols.
11501 2013-08-01  Jan Hubicka  <jh@suse.cz>
11503         * profile.c (compute_value_histograms): Fix thinko.
11505 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11507         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11508         aarch-common-protos.h to extra_headers.
11509         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11510         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11511         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11513 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11515         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11516         Include "../arm/types.md".  Define "type" attribute for all patterns.
11517         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11518         attribute changes.
11520 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11522         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11523         to support power8 load fusion.
11524         (fusion_gpr_mem_load): Likewise.
11526         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11528         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11529         declarations for power8 load fusion.
11530         (emit_fusion_gpr_load): Likewise.
11532         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11533         tuning for power8, turn on fusion mode by default.  Turn on sign
11534         extending fusion mode if normal fusion mode is on, and we are at
11535         -O2 or -O3.
11536         (fusion_gpr_load_p): New function, return true if we can fuse an
11537         addis instruction with a dependent load to a GPR.
11538         (emit_fusion_gpr_load): Emit the instructions for power8 load
11539         fusion to GPRs.
11541         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11542         (VSX load fusion peepholes): New peepholes to fuse together an
11543         addi instruction with a VSX load instruction.
11545         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11546         peepholes to fuse an addis instruction with a load to a GPR base
11547         register.  If we are supporting sign extending fusions, convert
11548         sign extending loads to zero extending loads and add an explicit
11549         sign extension.
11551 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11553         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11554         aarch-common-protos.h to extra_headers.
11555         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11556         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11557         (arm_early_store_addr_dep): Likewise.
11558         (arm_no_early_alu_shift_dep): Likewise.
11559         (arm_no_early_alu_shift_value_dep): Likewise.
11560         (arm_no_early_mul_dep): Likewise.
11561         (arm_no_early_store_addr_dep): Likewise.
11562         (arm_mac_accumulator_is_mul_result): Likewise.
11563         (arm_mac_accumulator_is_result): Likewise.
11564         * config/arm/aarch-common.c: ... here.  New file.
11565         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11566         here to ...
11567         (arm_early_store_addr_dep): Likewise.
11568         (arm_no_early_alu_shift_dep): Likewise.
11569         (arm_no_early_alu_shift_value_dep): Likewise.
11570         (arm_no_early_mul_dep): Likewise.
11571         (arm_no_early_store_addr_dep): Likewise.
11572         (arm_mac_accumulator_is_mul_result): Likewise.
11573         (arm_mac_accumulator_is_result): Likewise.
11574         * config/arm/aarch-common-protos.h: ... here.  New file.
11575         * config/arm/t-arm (aarch-common.o): Define.
11577 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11579         * config/arm/arm.md: Include new file "types.md".
11580         (define_attr "type"): Move from here to ...
11581         (define_attr "mul32"): Likewise.
11582         (define_attr "mul64"): Likewise.
11583         * config/arm/types.md: ... here.  New file.
11585 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11587         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11588         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11590 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11592         * gen-pass-instances.awk: Fix offset of substr().
11594 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11596         * Makefile.in (pass-instances.def): New.
11597         (passes.o): Replace dependency on passes.def with one on
11598         pass-instances.def
11600         * gen-pass-instances.awk: New.
11602         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11603         rather than passes.def, updating local definition of NEXT_PASS
11604         macro to add an extra NUM parameter (currently unused).
11606 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11608         * Makefile.in (PASS_MANAGER_H): New.
11609         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11610         (passes.o): Likewise.
11611         (statistics.o): Likewise.
11612         (cgraphunit.o): Likewise.
11613         (context.o): Depend on PASS_MANAGER_H.
11615         * pass_manager.h: New.
11617         * cgraphunit.c (cgraph_add_new_function): Update for moves
11618         of globals to fields of pass_manager.
11619         (analyze_function): Likewise.
11620         (expand_function): Likewise.
11621         (ipa_passes): Likewise.
11622         (compile): Likewise.
11624         * context.c (context::context): New.
11625         * context.h  (context::context): New.
11626         (context::get_passes): New.
11627         (context::passes_): New.
11629         * lto-cgraph.c (input_node): Update for moves of globals to
11630         fields of pass_manager.
11632         * passes.c (all_passes): Remove, in favor of a field of the
11633         same name within the new class pass_manager.
11634         (all_small_ipa_passes): Likewise.
11635         (all_lowering_passes): Likewise.
11636         (all_regular_ipa_passes): Likewise.
11637         (all_late_ipa_passes): Likewise.
11638         (all_lto_gen_passes): Likewise.
11639         (passes_by_id): Likewise.
11640         (passes_by_id_size): Likewise.
11641         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11642         the new class pass_manager.
11643         (set_pass_for_id): Convert to...
11644         (pass_manager::set_pass_for_id): ...method.
11645         (get_pass_for_id): Convert to...
11646         (pass_manager::get_pass_for_id): ...method.
11647         (register_one_dump_file): Move body of implementation into...
11648         (pass_manager::register_one_dump_file): ...here.
11649         (register_dump_files_1): Convert to...
11650         (pass_manager::register_dump_files_1): ...method.
11651         (register_dump_files): Convert to...
11652         (pass_manager::register_dump_files): ...method.
11653         (create_pass_tab): Update for moves of globals to fields of
11654         pass_manager.
11655         (dump_passes): Move body of implementation into...
11656         (pass_manager::dump_passes): ...here.
11657         (register_pass): Move body of implementation into...
11658         (pass_manager::register_pass): ...here.
11659         (init_optimization_passes): Convert into...
11660         (pass_manager::pass_manager): ...constructor for new
11661         pass_manager class, and initialize the pass_lists array.
11662         (check_profile_consistency): Update for moves of globals to
11663         fields of pass_manager.
11664         (dump_profile_report): Move body of implementation into...
11665         (pass_manager::dump_profile_report): ...here.
11666         (ipa_write_summaries_1): Update for moves of pass lists from
11667         being globals to fields of pass_manager.
11668         (ipa_write_optimization_summaries): Likewise.
11669         (ipa_read_summaries):  Likewise.
11670         (ipa_read_optimization_summaries): Likewise.
11671         (execute_all_ipa_stmt_fixups): Likewise.
11673         * statistics.c (statistics_fini): Update for moves of globals to
11674         fields of pass_manager.
11676         * toplev.c (general_init): Replace call to
11677         init_optimization_passes with construction of the pass_manager
11678         instance.
11680         * tree-pass.h (all_passes): Remove, in favor of a field of the
11681         same name within the new class pass_manager.
11682         (all_small_ipa_passes): Likewise.
11683         (all_lowering_passes): Likewise.
11684         (all_regular_ipa_passes): Likewise.
11685         (all_lto_gen_passes): Likewise.
11686         (all_late_ipa_passes): Likewise.
11687         (passes_by_id): Likewise.
11688         (passes_by_id_size): Likewise.
11689         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11690         the new class pass_manager.
11691         (get_pass_for_id): Remove.
11693 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11695         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11696         configs.
11698 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11700         * arm.md (mulhi3): New expand pattern.
11702 2013-07-30  Jan Hubicka  <jh@suse.cz>
11703             Martin Liska  <marxin.liska@gmail.com>
11705         * profile.c (compute_value_histograms): Do not ICE when
11706         there is mismatch only on some counters.
11708 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11710         PR rtl-optimization/57637
11711         * function.c (move_insn_for_shrink_wrap): Also check the
11712         GEN set of the LIVE problem for the liveness analysis
11713         if it exists, otherwise give up.
11715 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11717         PR tree-optimization/57993
11718         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11719         replaced statement in the candidate table.
11720         (phi_add_costs): Return infinite cost when the hidden basis does
11721         not dominate all phis on which the candidate is dependent.
11722         (replace_one_candidate): Record replaced statement in the
11723         candidate table.
11725 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11727         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11728         (ashlv2si3): New expander.
11729         (*ashlv2si3_i): New define_insn_and_split.
11730         * predicates.md (float_operation): Allow patterns with three
11731         basic sub-patterns.
11733         PR rtl-optimization/58021
11734         * mode-switching.c (create_pre_exit): Always split off preceding
11735         insns if we are not at the basic block head.
11737 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11739         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11740         (UCLIBC_DYNAMIC_LINKER): New macro.
11741         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11742         `-mnan=2008'.
11743         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11744         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11745         `-mnan=2008'.
11746         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11747         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11748         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11749         for SF and DF modes.  Use ieee_quad_format for TF mode.
11750         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11751         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11752         (mips_option_override): Handle `-mnan=legacy'.
11753         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11754         `-mabs=2008' and `-mnan=2008'.
11755         (OPTION_DEFAULT_SPECS): Add "nan" default.
11756         (ASM_SPEC): Handle `-mnan='.
11757         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11758         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11759         comment accordingly.
11760         (neg<mode>2): Likewise.
11761         * config/mips/mips.opt (mabs, mnan): New options.
11762         * doc/install.texi (Configuration): Document `--with-nan=' option.
11763         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11764         `-mnan=' options.
11765         (MIPS Options): Document them.
11766         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11767         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11768         * configure: Regenerate.
11769         * config.in: Regenerate.
11771 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11773         * config/i386/i386.md (float post-reload splitters): Do not check
11774         for subregs of SSE registers.
11776 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11777             H.J. Lu  <hongjiu.lu@intel.com>
11779         PR target/57954
11780         PR target/57988
11781         * config/i386/i386.md (post-reload splitter
11782         to avoid partial SSE reg dependency stalls): New pattern.
11784 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11786         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11788 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11790         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11791         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11793 2013-07-26  Cary Coutant  <ccoutant@google.com>
11795         * dwarf2out.c (die_checksum_ordered): Don't include template
11796         instantiations in signature.
11797         (is_template_parameter): New function.
11798         (is_template_instantiation): New function.
11799         (generate_skeleton_bottom_up): Don't include template instantiations
11800         in type unit DIE.
11801         (generate_skeleton): Likewise.
11802         (break_out_comdat_types): Move recursive call to break out nested
11803         types earlier.
11804         (prune_unused_types_mark_generic_parms_dies): Call
11805         is_template_parameter.
11807 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11809         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11810         uses vector registers.
11811         * config/aarch64/iterators.md: Add attributes rtn and vas.
11813 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11814             Richard Earnshaw  <richard.earnshaw@arm.com>
11816         * combine.c (simplify_comparison): Re-canonicalize operands
11817         where appropriate.
11818         * config/arm/arm.md (movcond_addsi): New splitter.
11820 2013-07-25  Sterling Augustine  <saugustine@google.com>
11822         * dwarf2out.c (size_of_pubnames): Move code to...
11823         (include_pubname_in_output): ...here.  New.
11824         (want_pubnames): Rearrange.
11825         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11827 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11829         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11831 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11833         PR target/38836
11834         * doc/extend.texi: Remove obsolete builtins. Fix
11835         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11837 2013-07-25  Jan Hubicka  <jh@suse.cz>
11839         * cgraph.c (release_function_body): Break out from ...
11840         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11841         and DECL_ARGUMENTS.
11842         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11843         old_tree in the map.
11844         (create_specialized_node): Update.
11845         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11846         into index.
11847         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11848         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11849         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11850         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11851         DECL_RESULT.
11853 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11855         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11856         addsi3_carryin_alt2_<optab>): Correct output template.
11858 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11860         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11861         Adjust for arm_restrict_it.
11862         Remove trailing whitespace.
11864 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11866         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11867         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11869         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11871 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11873         PR rtl-optimization/57960
11874         * lra-constraints.c (process_alt_operands): Use the right mode
11875         when checking strict_low.
11877 2013-07-25  Jan Hubicka  <jh@suse.cz>
11879         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11880         * cgraph.c (cgraph_remove_node): Do not release function body
11881         when in cgraph streaming.
11882         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11883         in other partitions are not considered reachable; fix handling of
11884         clones.
11886 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11888         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11890 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11892         PR target/19599
11893         PR target/57731
11894         PR target/57837
11895         * config/arm/arm.md ("*sibcall_insn): Replace use of
11896         Ss with US. Adjust output for v5 and v4t.
11897         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11899         * config/arm/constraints.md ("Ss"): Rename to US.
11901 2013-07-25  Terry Guo  <terry.guo@arm.com>
11903         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11904         shift_add/shift_sub0/shift_sub1 RTXs.
11906 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11907             Anton Blanchard  <anton@au1.ibm.com>
11909         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11910         (altivec_vpks<VI_char>ss): Likewise.
11911         (altivec_vpks<VI_char>us): Likewise.
11912         (altivec_vpku<VI_char>us): Likewise.
11913         (altivec_vpku<VI_char>um): Likewise.
11915 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11917         Introduce context class.
11919         * Makefile.in (CONTEXT_H): New.
11920         (OBJS): Add context.o.
11921         (toplev.o): Add CONTEXT_H to dependencies.
11922         (context.o): New.
11924         * toplev.c (general_init): Create the singleton gcc::context instance.
11926         * context.c: New.
11928         * context.h: New.
11930 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11932         PR rtl-optimization/57968
11933         * mode-switching.c (create_pre_exit): Allow instructions that
11934         don't set a return register to need a non-exit mode.
11936 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11937             Anton Blanchard  <anton@au1.ibm.com>
11939         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11940         operands to vperm for little endian.
11941         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11942         of lvsl to create the control mask for a vperm for little endian.
11944 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11945             Anton Blanchard  <anton@au1.ibm.com>
11947         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11948         two operands for little-endian.
11950 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11952         * config/mips/mips.c (mips_case_values_threshold): New.
11953         (TARGET_CASE_VALUES_THRESHOLD): Define.
11955 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11956             Anton Blanchard  <anton@au1.ibm.com>
11958         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11959         selection of field for vector splat in little endian mode.
11961 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11963         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11964         expanders to rs6000.md.
11965         (ior<mode>3): Likewise.
11966         (and<mode>3): Likewise.
11967         (one_cmpl<mode>2): Likewise.
11968         (nor<mode>3): Likewise.
11969         (andc<mode>3): Likewise.
11970         (eqv<mode>3): Likewise.
11971         (nand<mode>3): Likewise.
11972         (orc<mode>3): Likewise.
11974         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11975         declaration.
11977         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11978         to split multi-word logical operations.
11979         (rs6000_split_logical_di): Likewise.
11980         (rs6000_split_logical): Likewise.
11982         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11983         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11984         and allow TImode operations in 32-bit.
11985         (vsx_and<mode>3_64bit): Likewise.
11986         (vsx_ior<mode>3_32bit): Likewise.
11987         (vsx_ior<mode>3_64bit): Likewise.
11988         (vsx_xor<mode>3_32bit): Likewise.
11989         (vsx_xor<mode>3_64bit): Likewise.
11990         (vsx_one_cmpl<mode>2_32bit): Likewise.
11991         (vsx_one_cmpl<mode>2_64bit): Likewise.
11992         (vsx_nor<mode>3_32bit): Likewise.
11993         (vsx_nor<mode>3_64bit): Likewise.
11994         (vsx_andc<mode>3_32bit): Likewise.
11995         (vsx_andc<mode>3_64bit): Likewise.
11996         (vsx_eqv<mode>3_32bit): Likewise.
11997         (vsx_eqv<mode>3_64bit): Likewise.
11998         (vsx_nand<mode>3_32bit): Likewise.
11999         (vsx_nand<mode>3_64bit): Likewise.
12000         (vsx_orc<mode>3_32bit): Likewise.
12001         (vsx_orc<mode>3_64bit): Likewise.
12003         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
12004         logical types in GPRs.
12006         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
12007         logical insns to rs6000.md, and allow TImode operations in
12008         32-bit.
12009         (altivec_ior<mode>3): Likewise.
12010         (altivec_xor<mode>3): Likewise.
12011         (altivec_one_cmpl<mode>2): Likewise.
12012         (altivec_nor<mode>3): Likewise.
12013         (altivec_andc<mode>3): Likewise.
12015         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
12016         attributes for moving the 128-bit logical operations into
12017         rs6000.md.
12018         (BOOL_REGS_OUTPUT): Likewise.
12019         (BOOL_REGS_OP1): Likewise.
12020         (BOOL_REGS_OP2): Likewise.
12021         (BOOL_REGS_UNARY): Likewise.
12022         (BOOL_REGS_AND_CR0): Likewise.
12023         (one_cmpl<mode>2): Add support for DI logical operations on
12024         32-bit, splitting the operations to 32-bit.
12025         (anddi3): Likewise.
12026         (iordi3): Likewise.
12027         (xordi3): Likewise.
12028         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12029         changes to combine the 32/64-bit code, allow logical operations on
12030         TI mode in 32-bit, and to use similar match_operator patterns like
12031         scalar mode uses.  Combine the Altivec and VSX code for logical
12032         operations, and move it here.
12033         (ior<mode>3, 128-bit types): Likewise.
12034         (xor<mode>3, 128-bit types): Likewise.
12035         (one_cmpl<mode>3, 128-bit types): Likewise.
12036         (nor<mode>3, 128-bit types): Likewise.
12037         (andc<mode>3, 128-bit types): Likewise.
12038         (eqv<mode>3, 128-bit types): Likewise.
12039         (nand<mode>3, 128-bit types): Likewise.
12040         (orc<mode>3, 128-bit types): Likewise.
12041         (and<mode>3_internal): Likewise.
12042         (bool<mode>3_internal): Likewise.
12043         (boolc<mode>3_internal1): Likewise.
12044         (boolc<mode>3_internal2): Likewise.
12045         (boolcc<mode>3_internal1): Likewise.
12046         (boolcc<mode>3_internal2): Likewise.
12047         (eqv<mode>3_internal1): Likewise.
12048         (eqv<mode>3_internal2): Likewise.
12049         (one_cmpl1<mode>3_internal): Likewise.
12051 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12053         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12054         Rename flag_stack_usage to flag_stack_usage_info.
12056 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12058         * config/microblaze/sync.md: New file.
12059         * config/microblaze/microblaze.md: Include sync.md
12060         * config/microblaze/microblaze.c: Add print_operand 'y'.
12061         * config/microblaze/constraints.md: Add memory_contraint
12062         'Q' which is a single register.
12064 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12066         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12068 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12070         * reload.c (find_reloads): Exit loop once we find this operand
12071         cannot be reloaded somehow for this alternative.
12073         * reload.c (find_reloads): Exit loop once we find a hard register.
12075         * rtlanal.c (computed_jump_p): Exit loop once we find label
12076         reference is used.
12078         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12080         * cfgloopmanip.c (remove_path): Exit loop after setting
12081         irred_invalidated.
12083         * gensupport.c (subst_dup): Avoid loop if code is not
12084         MATCH_DUP nor MATCH_OP_DUP.
12086 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12088         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12090 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12092         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12093         true for SP_REGNUM if mode == ptr_mode.
12094         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12095         with value R0_REGNUM + 31.
12097 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12099         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12100         pad pointer-typed argument downward.
12102 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12104         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12105         and __ILP32__ when the ILP32 model is in use.
12107 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12109         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12110         (aarch64_load_symref_appropriately): In the case of
12111         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12112         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12113         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12114         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12115         if 'mode' doesn't equal to 'ptr_mode'.
12116         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12117         'vcall_offset'; change to call aarch64_emit_move differently depending
12118         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12119         to calculate the upper bound of 'vcall_offset'.
12120         (aarch64_cannot_force_const_mem): Change to also return true if
12121         mode != ptr_mode.
12122         (aarch64_legitimize_reload_address): In the case of large
12123         displacements, add new local variable 'xmode' and an assertion
12124         based on it; change to use 'xmode' to generate the new rtx and
12125         reload.
12126         (aarch64_asm_trampoline_template): Change to generate the template
12127         differently depending on TARGET_ILP32 or not; change to use
12128         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12129         (aarch64_trampoline_size): Removed.
12130         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12131         and replace immediate literals with it.  Change to use 'ptr_mode'
12132         instead of 'DImode' and call convert_memory_address if the mode
12133         of 'fnaddr' doesn't equal to 'ptr_mode'.
12134         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12135         to output symbol.
12136         (aarch64_elf_asm_destructor): Likewise.
12137         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12138         on TARGET_ILP32 instead of aarch64_trampoline_size.
12139         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12140         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12141         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12142         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12143         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12144         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12145         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12146         depending on the value of 'mode'.
12147         (add_losym_<mode>): New.
12148         (ldr_got_small_<mode>): New, based on ldr_got_small.
12149         (ldr_got_small): Remove.
12150         (ldr_got_small_sidi): New.
12151         * config/aarch64/iterators.md (P): New.
12152         (PTR): Change to 'ptr_mode' in the condition.
12154 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12156         * config.gcc (aarch64*-*-*): Support --with-abi.
12157         (aarch64*-*-elf): Support --with-multilib-list.
12158         (aarch64*-*-linux*): Likewise.
12159         (supported_defaults): Add abi to aarch64*-*-*.
12160         * configure.ac: Mention AArch64 for --with-multilib-list.
12161         * configure: Re-generated.
12162         * config/aarch64/biarchilp32.h: New file.
12163         * config/aarch64/biarchlp64.h: New file.
12164         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12165         (ABI_SPEC): Ditto.
12166         (MULTILIB_DEFAULTS): Ditto.
12167         (DRIVER_SELF_SPECS): Ditto.
12168         (ASM_SPEC): Update to also substitute -mabi.
12169         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12170         file whose name depends on -mabi= and -mbig-endian.
12171         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12172         TARGET_ILP32.
12173         (POINTER_SIZE): New define.
12174         (POINTERS_EXTEND_UNSIGNED): Ditto.
12175         (enum aarch64_abi_type): New enumeration tag.
12176         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12177         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12178         (TARGET_ILP32): New define.
12179         * config/aarch64/aarch64.opt (mabi): New.
12180         (aarch64_abi): New.
12181         (ilp32, lp64): New values for -mabi.
12182         * config/aarch64/t-aarch64 (comma): New define.
12183         (MULTILIB_OPTIONS): Ditto.
12184         (MULTILIB_DIRNAMES): Ditto.
12185         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12186         * doc/invoke.texi: Document -mabi for AArch64.
12188 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12190         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12192 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12193             Anton Blanchard  <anton@au1.ibm.com>
12195         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12196         endianness when selecting field to splat.
12198 2013-07-22  Eric Christopher  <echristo@gmail.com>
12200         * dwarf2out.c (die_odr_checksum): New function to use
12201         CHECKSUM_ macros and ULEB128 for DIE tag.
12202         (generate_type_signature): Use.
12204 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12206         * config.gcc (sparc*-*-*): Accept leon3 processor.
12207         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12208         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12209         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12210         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12211         (mfix-ut699): Adjust comment.
12212         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12213         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12214         (CPP_CPU_SPEC): Likewise.
12215         (ASM_CPU_SPEC): Likewise.
12216         * config/sparc/sparc.c (leon3_cost): New constant.
12217         (sparc_option_override): Add leon3 support.
12218         (mem_ref): New function.
12219         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12220         (sparc_do_work_around_errata): Look into the instruction in the delay
12221         slot and adjust accordingly.  Add fix for the data cache nullify issues
12222         of the UT699.  Change insertion position for the NOP.
12223         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12224         (leon3_load): New reservation.
12225         (leon_store): Bump latency to 2.
12226         (grfpu): New automaton.
12227         (grfpu_alu): New unit.
12228         (grfpu_ds): Likewise.
12229         (leon_fp_alu): Adjust.
12230         (leon_fp_mult): Delete.
12231         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12232         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12233         * config/sparc/sparc.md (cpu): Add leon3.
12234         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12235         (swapsi): Likewise.
12236         (atomic_test_and_set): Likewise.
12237         (ldstub): Likewise.
12239 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12241         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12242         default for R5900 targets.
12243         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12244         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12245         * config/mips/mips.c (mips_option_override): Report an error for
12246         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12247         for -march=r5900 -mhard-float.
12249 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12251         * df-problems.c (can_move_insns_across): Exit loop once we
12252         find a non-fixed, non-global register.
12254         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12255         setting can_throw.
12257         * omega.c (omega_eliminate_red): Break after setting red_found.
12258         (omega_problem_has_red_equations): Similarly after setting found.
12259         (omega_query_variable): Similarly after setting coupled.
12261 2013-07-22  Marek Polacek  <polacek@redhat.com>
12263         * gimplify.c: Don't include gimple.h twice.
12265 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12267         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12268         instead of TARGET_THUMB1.
12269         (Pz): New constraint.
12270         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12271         encodings.
12272         (compare_negsi_si): Likewise.
12273         (compare_addsi2_op0): Likewise.
12274         (compare_addsi2_op1): Likewise.
12275         (addsi3_carryin_<optab>): Likewise.
12276         (addsi3_carryin_alt2_<optab>): Likewise.
12277         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12278         for arm_restrict_it.
12279         (subsi3_carryin): Likewise.
12280         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12281         (minmax_arithsi): Disable for arm_restrict_it.
12282         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12283         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12284         (satsi_<SAT:code>_shift): Likewise.
12285         (arm_shiftsi3): Add alternative for 16-bit encoding.
12286         (arm32_movhf): Disable for arm_restrict_it.
12287         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12288         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12290 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12292         * config/arm/arm.md (attribute "insn"): Delete.
12293         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12294         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12295         (not_shiftsi): Update for attribute change.
12296         (not_shiftsi_compare0): Likewise.
12297         (not_shiftsi_compare0_scratch): Likewise.
12298         (arm_one_cmplsi2): Likewise.
12299         (thumb1_one_cmplsi2): Likewise.
12300         (notsi_compare0): Likewise.
12301         (notsi_compare0_scratch): Likewise.
12302         (thumb1_movdi_insn): Likewise.
12303         (arm_movsi_insn): Likewise.
12304         (movhi_insn_arch4): Likewise.
12305         (movhi_bytes): Likewise.
12306         (arm_movqi_insn): Likewise.
12307         (thumb1_movqi_insn): Likewise.
12308         (arm32_movhf): Likewise.
12309         (thumb1_movhf): Likewise.
12310         (arm_movsf_soft_insn): Likewise.
12311         (thumb1_movsf_insn): Likewise.
12312         (thumb_movdf_insn): Likewise.
12313         (movsicc_insn): Likewise.
12314         (movsfcc_soft_insn): Likewise.
12315         (and_scc): Likewise.
12316         (cond_move): Likewise.
12317         (if_move_not): Likewise.
12318         (if_not_move): Likewise.
12319         (if_shift_move): Likewise.
12320         (if_move_shift): Likewise.
12321         (if_shift_shift): Likewise.
12322         (if_not_arith): Likewise.
12323         (if_arith_not): Likewise.
12324         (cond_move_not): Likewise.
12325         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12326         (neon_mov<mode>): Likewise.
12327         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12328         (thumb2_movsi_vfp): Likewise.
12329         (movsf_vfp): Likewise.
12330         (thumb2_movsf_vfp): Likewise.
12331         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12332         change.
12333         (cortexa7_older_only): Likewise.
12334         (cortexa7_younger): Likewise.
12335         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12336         (1020alu_shift_op): Likewise.
12337         (1020alu_shift_reg_op): Likewise.
12338         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12339         (alu_shift_op): Likewise.
12340         (alu_shift_reg_op): Likewise.
12341         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12342         (11_alu_shift_op): Likewise.
12343         (11_alu_shift_reg_op): Likewise.
12344         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12345         (9_alu_shift_reg_op): Likewise.
12346         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12347         change.
12348         (cortex_a15_alu_shift): Likewise.
12349         (cortex_a15_alu_shift_reg): Likewise.
12350         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12351         (cortex_a5_alu_shift): Likewise.
12352         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12353         change.
12354         (cortex_a53_alu_shift): Likewise.
12355         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12356         change.
12357         (cortex_a7_alu_reg): Likewise.
12358         (cortex_a7_alu_shift): Likewise.
12359         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12360         (cortex_a8_alu_shift): Likewise.
12361         (cortex_a8_alu_shift_reg): Likewise.
12362         (cortex_a8_mov): Likewise.
12363         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12364         (cortex_a9_dp_shift): Likewise.
12365         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12366         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12367         (cortex_r4_mov): Likewise.
12368         (cortex_r4_alu_shift): Likewise.
12369         (cortex_r4_alu_shift_reg): Likewise.
12370         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12371         (526_alu_shift_op): Likewise.
12372         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12373         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12374         (626te_alu_shift_op): Likewise.
12375         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12376         (726te_alu_op): Likewise.
12377         (726te_alu_shift_op): Likewise.
12378         (726te_alu_shift_reg_op): Likewise.
12379         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12380         (mp626_alu_shift_op): Likewise.
12381         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12382         (pj4_alu_e1_conds): Likewise.
12383         (pj4_alu): Likewise.
12384         (pj4_alu_conds): Likewise.
12385         (pj4_shift): Likewise.
12386         (pj4_shift_conds): Likewise.
12387         (pj4_alu_shift): Likewise.
12388         (pj4_alu_shift_conds): Likewise.
12390 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12392         * config/arm/predicates.md (shiftable_operator_strict_it):
12393         New predicate.
12394         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12395         Disable cond_exec version for arm_restrict_it.
12396         (thumb2_smaxsi3): Convert to generate cond_exec.
12397         (thumb2_sminsi3): Likewise.
12398         (thumb32_umaxsi3): Likewise.
12399         (thumb2_uminsi3): Likewise.
12400         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12401         (thumb2_neg_abssi2): Likewise.
12402         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12403         (thumb2_movsicc_insn): Adjust alternatives.
12404         (thumb2_mov_negscc): Disable for arm_restrict_it.
12405         (thumb2_mov_negscc_strict_it): New pattern.
12406         (thumb2_mov_notscc_strict_it): New pattern.
12407         (thumb2_mov_notscc): Disable for arm_restrict_it.
12408         (thumb2_ior_scc): Likewise.
12409         (thumb2_ior_scc_strict_it): New pattern.
12410         (thumb2_cond_move): Adjust for arm_restrict_it.
12411         (thumb2_cond_arith): Disable for arm_restrict_it.
12412         (thumb2_cond_arith_strict_it): New pattern.
12413         (thumb2_cond_sub): Adjust for arm_restrict_it.
12414         (thumb2_movcond): Likewise.
12415         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12416         (thumb2_zero_extendhisi2_v6): Likewise.
12417         (thumb2_zero_extendqisi2_v6): Likewise.
12418         (orsi_notsi_si): Likewise.
12419         (orsi_not_shiftsi_si): Likewise.
12421 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12423         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12424         instruction sequence is 1 byte shorter.
12426 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12428         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12429         it is not needed after split.
12431 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12433         PR target/51784
12434         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12435         second label for nonlocal goto receivers. Don't output pic base labels
12436         unless we're producing PIC; mark that action unreachable().
12437         (ix86_save_reg): If the function contains a nonlocal label, save the
12438         PIC base reg.
12439         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12440         * config/darwin.c (emitted_pic_label_num): New GTY.
12441         (update_pic_label_number_if_needed): New.
12442         (machopic_output_function_base_name): Adjust for nonlocal receiver
12443         case.
12444         (machopic_should_output_picbase_label): New.
12445         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12446         (nonlocal_goto_receiver): New insn and split.
12448 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12450         * config/aarch64/aarch64-builtins.c
12451         (aarch64_fold_builtin): Fold abs in all modes.
12452         * config/aarch64/aarch64-simd-builtins.def
12453         (abs): Enable for all modes.
12454         * config/aarch64/arm_neon.h
12455         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12456         (vabs_f64): Add missing intrinsic.
12458 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12460         * config/aarch64/arm_neon.h (vabs_s64): New function
12462 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12464         PR target/57516
12465         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12466         * config/avr/avr.md (adjust_len): Add `round'.
12467         * config/avr/avr-protos.h (avr_out_round): New prototype.
12468         (avr_out_plus): Add `out_label' argument.
12469         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12470         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12471         Handle the case where `insn' is just a pattern.
12472         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12473         (avr_out_round): New function.
12474         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12476 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12478         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12479         Add check for flag_stack_usage to handle -fstack-usage support
12481 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12483         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12484         interaction for new Power8 flags and VSX.
12486 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12488         PR middle-end/57698
12489         * tree-inline.c (expand_call_inline): Emit errors during
12490         early_inlining only if optimization is not turned on.
12492 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12494         * passes.def: New.
12496         * passes.c (init_optimization_passes): Move the construction of
12497         the pass hierarchy into a new passes.def file.
12499         * Makefile.in (passes.o): Add dependency on passes.def.
12501 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12503         * passes.c (init_optimization_passes): Introduce macros for
12504         constructing the tree of passes (INSERT_PASSES_AFTER,
12505         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12506         TERMINATE_PASS_LIST).
12508 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12509             Wei Mi  <wmi@google.com>
12511         PR rtl-optimization/57878
12512         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12513         top.
12514         (reload_pseudo_compare_func): Check nregs first for reload
12515         pseudos.
12517 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12519         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12521 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12523         * read-rtl.c (validate_const_int): Once an invalid character is
12524         seen, quit the loop.
12526         * gengtype.c (write_roots): Similarly once we find the "deletable"
12527         or "if_marked" option.
12529 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12531         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12532         "xtab" and "sat".  Move value "clz" from here to ...
12533         (attriubte "type"): ... here.
12534         (satsi_<SAT:code>): Delete "insn" attribute.
12535         (satsi_<SAT:code>_shift): Likewise.
12536         (arm_zero_extendqisi2addsi): Likewise.
12537         (arm_extendqisi2addsi): Likewise.
12538         (clzsi2): Update for attribute changes.
12539         (rbitsi2): Likewise.
12540         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12541         attribute.
12542         (arm_usatsihi): Likewise.
12543         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12545 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12547         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12548         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12549         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12550         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12551         in alphabetical order.
12552         (attribute "core_cycles"): Update for attribute changes.
12553         (arm_addsi3): Likewise.
12554         (addsi3_compare0): Likewise.
12555         (addsi3_compare0_scratch): Likewise.
12556         (addsi3_compare_op1): Likewise.
12557         (addsi3_compare_op2): Likewise.
12558         (compare_addsi2_op0): Likewise.
12559         (compare_addsi2_op1): Likewise.
12560         (addsi3_carryin_shift_<optab>): Likewise.
12561         (subsi3_carryin_shift): Likewise.
12562         (rsbsi3_carryin_shift): Likewise.
12563         (arm_subsi3_insn): Likewise.
12564         (subsi3_compare0): Likewise.
12565         (subsi3_compare): Likewise.
12566         (arm_andsi3_insn): Likewise.
12567         (thumb1_andsi3_insn): Likewise.
12568         (andsi3_compare0): Likewise.
12569         (andsi3_compare0_scratch): Likewise.
12570         (zeroextractsi_compare0_scratch
12571         (andsi_not_shiftsi_si): Likewise.
12572         (iorsi3_insn): Likewise.
12573         (iorsi3_compare0): Likewise.
12574         (iorsi3_compare0_scratch): Likewise.
12575         (arm_xorsi3): Likewise.
12576         (thumb1_xorsi3_insn): Likewise.
12577         (xorsi3_compare0): Likewise.
12578         (xorsi3_compare0_scratch): Likewise.
12579         (satsi_<SAT:code>_shift): Likewise.
12580         (rrx): Likewise.
12581         (arm_shiftsi3): Likewise.
12582         (shiftsi3_compare0): Likewise.
12583         (not_shiftsi): Likewise.
12584         (not_shiftsi_compare0): Likewise.
12585         (not_shiftsi_compare0_scratch): Likewise.
12586         (arm_one_cmplsi2): Likewise.
12587         (thumb_one_complsi2): Likewise.
12588         (notsi_compare0): Likewise.
12589         (notsi_compare0_scratch): Likewise.
12590         (thumb1_zero_extendhisi2): Likewise.
12591         (arm_zero_extendhisi2): Likewise.
12592         (arm_zero_extendhisi2_v6): Likewise.
12593         (arm_zero_extendhisi2addsi): Likewise.
12594         (thumb1_zero_extendqisi2): Likewise.
12595         (thumb1_zero_extendqisi2_v6): Likewise.
12596         (arm_zero_extendqisi2): Likewise.
12597         (arm_zero_extendqisi2_v6): Likewise.
12598         (arm_zero_extendqisi2addsi): Likewise.
12599         (thumb1_extendhisi2): Likewise.
12600         (arm_extendhisi2): Likewise.
12601         (arm_extendhisi2_v6): Likewise.
12602         (arm_extendqisi): Likewise.
12603         (arm_extendqisi_v6): Likewise.
12604         (arm_extendqisi2addsi): Likewise.
12605         (thumb1_extendqisi2): Likewise.
12606         (thumb1_movdi_insn): Likewise.
12607         (arm_movsi_insn): Likewise.
12608         (movsi_compare0): Likewise.
12609         (movhi_insn_arch4): Likewise.
12610         (movhi_bytes): Likewise.
12611         (arm_movqi_insn): Likewise.
12612         (thumb1_movqi_insn): Likewise.
12613         (arm32_movhf): Likewise.
12614         (thumb1_movhf): Likewise.
12615         (arm_movsf_soft_insn): Likewise.
12616         (thumb1_movsf_insn): Likewise.
12617         (movdf_soft_insn): Likewise.
12618         (thumb_movdf_insn): Likewise.
12619         (arm_cmpsi_insn): Likewise.
12620         (cmpsi_shiftsi): Likewise.
12621         (cmpsi_shiftsi_swp): Likewise.
12622         (arm_cmpsi_negshiftsi_si): Likewise.
12623         (movsicc_insn): Likewise.
12624         (movsfcc_soft_insn): Likewise.
12625         (arith_shiftsi): Likewise.
12626         (arith_shiftsi_compare0
12627         (arith_shiftsi_compare0_scratch
12628         (sub_shiftsi): Likewise.
12629         (sub_shiftsi_compare0
12630         (sub_shiftsi_compare0_scratch
12631         (and_scc): Likewise.
12632         (cond_move): Likewise.
12633         (if_plus_move): Likewise.
12634         (if_move_plus): Likewise.
12635         (if_move_not): Likewise.
12636         (if_not_move): Likewise.
12637         (if_shift_move): Likewise.
12638         (if_move_shift): Likewise.
12639         (if_shift_shift): Likewise.
12640         (if_not_arith): Likewise.
12641         (if_arith_not): Likewise.
12642         (cond_move_not): Likewise.
12643         (thumb1_ashlsi3): Set type attribute.
12644         (thumb1_ashrsi3): Likewise.
12645         (thumb1_lshrsi3): Likewise.
12646         (thumb1_rotrsi3): Likewise.
12647         (shiftsi3_compare0_scratch): Likewise.
12648         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12649         (neon_mov<mode>): Likewise.
12650         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12651         attribute changes.
12652         (thumb2_movsi_insn): Likewise.
12653         (thumb2_cmpsi_neg_shiftsi): Likewise.
12654         (thumb2_extendqisi_v6): Likewise.
12655         (thumb2_zero_extendhisi2_v6): Likewise.
12656         (thumb2_zero_extendqisi2_v6): Likewise.
12657         (thumb2_shiftsi3_short): Likewise.
12658         (thumb2_addsi3_compare0_scratch): Likewise.
12659         (orsi_not_shiftsi_si): Likewise.
12660         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12661         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12662         changes.
12663         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12664         (1020alu_shift_op): Likewise.
12665         (1020alu_shift_reg_op): Likewise.
12666         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12667         (alu_shift_op): Likewise.
12668         (alu_shift_reg_op): Likewise.
12669         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12670         (11_alu_shift_op): Likewise.
12671         (11_alu_shift_reg_op): Likewise.
12672         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12673         (9_alu_shift_reg_op): Likewise.
12674         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12675         changes.
12676         (cortex_a15_alu_shift): Likewise.
12677         (cortex_a15_alu_shift_reg): Likewise.
12678         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12679         changes.
12680         (cortex_a5_alu_shift): Likewise.
12681         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12682         changes.
12683         (cortex_a53_alu_shift): Likewise.
12684         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12685         changes.
12686         (cortex_a7_alu_reg): Likewise.
12687         (cortex_a7_alu_shift): Likewise.
12688         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12689         changes.
12690         (cortex_a8_alu_shift): Likewise.
12691         (cortex_a8_alu_shift_reg): Likewise.
12692         (cortex_a8_mov): Likewise.
12693         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12694         (cortex_a9_dp_shift): Likewise.
12695         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12696         changes.
12697         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12698         changes.
12699         (cortex_r4_mov): Likewise.
12700         (cortex_r4_alu_shift): Likewise.
12701         (cortex_r4_alu_shift_reg): Likewise.
12702         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12703         (526_alu_shift_op): Likewise.
12704         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12705         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12706         (626te_alu_shift_op): Likewise.
12707         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12708         (726te_alu_op): Likewise.
12709         (726te_alu_shift_op): Likewise.
12710         (726te_alu_shift_reg_op): Likewise.
12711         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12712         (mp626_alu_shift_op): Likewise.
12713         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12714         (pj4_alu_e1_conds): Likewise.
12715         (pj4_alu): Likewise.
12716         (pj4_alu_conds): Likewise.
12717         (pj4_shift): Likewise.
12718         (pj4_shift_conds): Likewise.
12719         (pj4_alu_shift): Likewise.
12720         (pj4_alu_shift_conds): Likewise.
12721         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12722         changes.
12723         (cortexa7_older_only): Likewise.
12724         (cortexa7_younger): Likewise.
12726 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12728         * ipa-pure-const.c (generate_summary): Rename to...
12729         (pure_const_generate_summary): ... this.
12731 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12733         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12735 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12737         PR target/57909
12738         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12739         usage in HI mode.
12741 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12743         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12744         enabled without -march=zEC12.
12745         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12746         flags to be set.
12748 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12750         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12751         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12752         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12753         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12754         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12755         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12756         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12757         ISA_HAS_FP4.
12758         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12759         and ISA_HAS_NMADD3_NMSUB3.
12760         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12761         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12762         (nmsub4<mode>, nmsub3<mode>): Likewise.
12763         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12765 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12767         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12768         TARGET_MIPS5400 checking.
12770 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12771             Peter Bergner  <bergner@vnet.ibm.com>
12773         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12774         registers in the comment.
12775         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12776         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12777         rather than FIRST_PSEUDO_REGISTERS.
12779 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12781         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12782         enable extra ISA flags with TARGET_HTM.
12784 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12786         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12787         Fix comment typos.
12789 2013-07-15  Cong Hou  <congh@google.com>
12791         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12792         in compare function for sorting.
12794 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12796         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12797         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12798         * config/rs6000/rs6000.opt: Add -mhtm option.
12799         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12800         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12801         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12802         __HTM__ if the HTM instructions are available.
12803         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12804         htm_spr_reg_operand): New define_predicates.
12805         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12806         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12807         Include htm.md.
12808         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12809         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12810         HTM builtin functions.
12811         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12812         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12813         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12814         (rs6000_builtin_mask_calculate): Likewise.
12815         (rs6000_option_override_internal): Likewise.
12816         (bdesc_htm): Add new HTM builtin support.
12817         (htm_spr_num): New function.
12818         (htm_spr_regno): Likewise.
12819         (rs6000_htm_spr_icode): Likewise.
12820         (htm_expand_builtin): Likewise.
12821         (htm_init_builtins): Likewise.
12822         (rs6000_expand_builtin): Add support for HTM builtin functions.
12823         (rs6000_init_builtins): Likewise.
12824         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12825         option.
12826         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12827         (TARGET_HTM, MASK_HTM): Define macros.
12828         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12829         (FIXED_REGISTERS): Likewise.
12830         (CALL_USED_REGISTERS): Likewise.
12831         (CALL_REALLY_USED_REGISTERS): Likewise.
12832         (REG_ALLOC_ORDER): Likewise.
12833         (enum reg_class): Likewise.
12834         (REG_CLASS_NAMES): Likewise.
12835         (REG_CLASS_CONTENTS): Likewise.
12836         (REGISTER_NAMES): Likewise.
12837         (ADDITIONAL_REGISTER_NAMES): Likewise.
12838         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12839         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12840         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12841         * config/rs6000/htm.md: New file.
12842         * config/rs6000/htmintrin.h: New file.
12843         * config/rs6000/htmxlintrin.h: New file.
12845 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12847         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12848         Define SYMBOL_TINY_GOT, update comment.
12849         * config/aarch64/aarch64.c
12850         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12851         (aarch64_expand_mov_immediate): Likewise.
12852         (aarch64_print_operand): Likewise.
12853         (aarch64_classify_symbol): Likewise.
12854         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12855         (ldr_got_tiny): Define.
12857 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12859         PR tree-optimization/54094
12860         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12861           scheduling dimension for the parallelism check from the polyhedral
12862           information in the AST.
12863         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12864           in 2D + 1 form.
12866 2013-07-13  Jason Merrill  <jason@redhat.com>
12868         * print-tree.c (debug_vec_tree): Use debug_raw.
12869         (debug_raw (vec<tree, va_gc> &)): New.
12870         (debug_raw (vec<tree, va_gc> *)): New.
12871         * tree.h: Declare them.
12873 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12875         * ifcvt.c (ifcvt_after_combine): New static variable.
12876         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12877         for size.
12878         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12879         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12880         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12882 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12884         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12886 2013-07-12  Michael Matz  <matz@suse.de>
12888         PR middle-end/55771
12889         * convert.c (convert_to_real): Reject non-float inner types.
12891 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12893         * config/aarch64/aarch64-protos.h
12894         (aarch64_simd_immediate_valid_for_move): Remove.
12895         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12896         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12897         cases.
12898         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12900 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12902         * config/mips/mips.c (mips_conditional_register_usage): Do not
12903         use t[0-7] registers in MIPS16 mode when optimizing for size.
12905 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12907         * config/i386/i386.c (dispatch_function_versions): Fix array
12908         indexing of function_version_info to match actual_versions.
12910 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12912         * vec.h (struct va_gc): Move release out-of-line.
12913         (va_gc::release): Call ggc_free on released vec.
12915 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12917         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12918         Require GOT register as additional operand in UNSPEC.
12919         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12920         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12921         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12922         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12923         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12924         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12925         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12927 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12929         PR target/57631
12930         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12931         name seen by assembler/linker rather if available.
12933 2013-07-11  Andreas Schwab  <schwab@suse.de>
12935         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12937 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12939         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12941 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12943         * doc/tm.texi.in: Move hook documentation to ....
12944         * target.def: ... here.
12946         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12947         text on @hook line.
12948         * doc/tm.texi: Regenerate.
12950 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12952         PR c++/57869
12953         * doc/invoke.texi: Document -Wconditionally-supported.
12955 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12957         PR target/57844
12958         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12959         of my_fp.
12961 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12963         PR target/57506
12964         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12965         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12966         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12967         Remove duplicate devices.
12968         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12969         * config/avr/t-multilib: Regenerate.
12970         * config/avr/avr-tables.opt: Regenerate.
12971         * doc/avr-mmcu.texi: Regenerate.
12973 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12975         PR target/56987
12976         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12978 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12980         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12981         the cost of MULT when optimizing for size.
12983 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12985         * config/cr16/cr16-protos.h: Don't include target.h.
12987 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12989         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12990         adjust register size for TDmode and TFmode for VSX registers.
12992 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12994         PR target/56892
12995         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12996         hook_bool_const_tree_true.
12998 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13000         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
13001         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
13002         * config/s390/s390.md: Define FPR*_REGNUM constants.
13003         Fix FPR2_REGNUM constant (18 -> 17).
13004         ("*trunc<BFP:mode><DFP_ALL:mode>2")
13005         ("*trunc<DFP_ALL:mode><BFP:mode>2")
13006         ("trunc<BFP:mode><DFP_ALL:mode>2")
13007         ("trunc<DFP_ALL:mode><BFP:mode>2")
13008         ("*extend<BFP:mode><DFP_ALL:mode>2")
13009         ("*extend<DFP_ALL:mode><BFP:mode>2")
13010         ("extend<BFP:mode><DFP_ALL:mode>2")
13011         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
13012         FPR4_REGNUM.
13014 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
13016         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
13018 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13020         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13021         and cfun_fpr_bit_p to cfun_fpr_save_p.
13022         (s390_frame_area, s390_register_info, s390_frame_info)
13023         (s390_emit_prologue, s390_emit_epilogue)
13024         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13025         register numbers.
13026         * config/s390/s390.h: Define *_REGNUM macros for floating point
13027         register numbers.
13029 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13031         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13033 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13035         PR rtl-optimization/57786
13036         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13037         and break out of the loop when it is set to false.
13039 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13041         PR target/57819
13042         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13043         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13044         (const_int 63)) 0)).
13045         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13046         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13047         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13049         PR rtl-optimization/57829
13050         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13051         mask bits outside of mode are just sign-extension from mode to HWI.
13053 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13055         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13056         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13057         adjust_address instead of change_address to keep info about alignment.
13058         (emit_strmov): Remove.
13059         (emit_memmov): New function.
13060         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13061         (expand_movmem_epilogue): Likewise and return updated rtx for
13062         destination.
13063         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13064         destination and source.
13065         (decide_alignment): Refactor, handle vector_loop.
13066         (ix86_expand_movmem): Likewise.
13067         (ix86_expand_setmem): Likewise.
13068         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13070 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13072         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13073         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13075 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13077         * config/i386/sse.md (sse_movlhps): Change alternative 3
13078         of operand 2 to "m".
13080 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13082         PR target/57807
13083         * config/i386/sse.md (iptr): New mode attribute.
13084         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13085         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13086         (<sse>_vmmul<mode>3): Ditto.
13087         (<sse>_vmdiv<mode>3): Ditto.
13088         (sse_vmrcpv4sf2): Ditto.
13089         (<sse>_vmsqrt<mode>2): Ditto.
13090         (sse_vmrsqrtv4sf2): Ditto.
13091         (<sse>_vm<code><mode>3): Ditto.
13092         (avx_vmcmp<mode>3): Ditto.
13093         (<sse>_vmmaskcmp<mode>3): Ditto.
13094         (<sse>_comi): Ditto.
13095         (<sse>_ucomi): Ditto.
13096         (*xop_vmfrcz_<mode>): Ditto.
13097         (*fmai_fmadd_<mode>): Ditto.
13098         (*fmai_fmsub_<mode>): Ditto.
13099         (*fmai_fnmadd_<mode>): Ditto.
13100         (*fmai_fnmsub_<mode>): Ditto.
13101         (*fma4i_vmfmadd_<mode>): Ditto.
13102         (*fma4i_vmfmsub_<mode>): Ditto.
13103         (*fma4i_vmfnmadd_<mode>): Ditto.
13104         (*fma4i_vmfnmsub_<mode>): Ditto.
13105         (*xop_vmfrcz_<mode>): Ditto.
13106         (sse_cvtps2pi): Ditto.
13107         (sse_cvttps2pi): Ditto.
13108         (sse_cvtss2si): Ditto.
13109         (sse_cvtss2si_2): Ditto.
13110         (sse_cvtss2siq_2): Ditto.
13111         (sse_cvttss2si): Ditto.
13112         (sse_cvttss2siq): Ditto.
13113         (sse_cvtsd2si): Ditto.
13114         (sse_cvtsd2si_2): Ditto.
13115         (sse_cvtsd2siq_2): Ditto.
13116         (sse_cvttsd2si): Ditto.
13117         (sse_cvttsd2siq): Ditto.
13118         (sse_cvtsd2ss): Ditto.
13119         (sse_cvtss2sd): Ditto.
13120         (avx2_pbroadcast<mode>): Ditto.
13121         (avx2_pbroadcast<mode>_1): Ditto.
13122         (*avx_vperm_broadcast_v4sf): Ditto.
13124         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13125         (sse_movlhps): Ditto.
13126         (sse_storehps): Ditto.
13127         (sse_loadhps): Ditto.
13128         (sse_storelps): Ditto.
13129         (sse_loadlps): Ditto.
13130         (*vec_concatv4sf): Ditto.
13131         (*vec_interleave_highv2df): Ditto.
13132         (*vec_interleave_lowv2df): Ditto.
13133         (*vec_extractv2df_1_sse): Ditto.
13134         (*vec_extractv2df_0_sse): Ditto.
13135         (sse2_storelpd): Ditto.
13136         (sse2_loadlpd): Ditto.
13137         (sse2_movsd): Ditto.
13138         (*vec_concatv4si): Ditto.
13139         (vec_concatv2di): Ditto.
13141         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13142         for Intel asm dialect.
13143         (mmx_punpcklwd): Ditto.
13144         (mmx_punpckldq): Ditto.
13146         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13147         for intel assembler dialect.
13149 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13151         PR target/29776
13152         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13153         for BUILT_IN_C{LZ,LRSB}*.
13154         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13155         * tree-vrp.c (extract_range_basic): Handle
13156         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13157         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13158         fall thru to code calling set_value*.
13159         * builtins.c (expand_builtin): Remove *IMAX cases.
13160         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13161         if width is bigger than 2*HWI.
13163 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13165         PR rtl-optimization/55342
13166         * lra-int.h (lra_subreg_reload_pseudos): New.
13167         * lra.c: Add undoing optional reloads to the block diagram.
13168         (lra_subreg_reload_pseudos): New.
13169         (lra_optional_reload_pseudos): Change comments.
13170         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13171         lra_optional_reload_pseudos after undo transformations.
13172         * lra-assigns.c (pseudo_prefix_title): New.
13173         (lra_setup_reg_renumber): Use it.
13174         (spill_for): Ditto.  Check subreg reload pseudos too.
13175         (assign_by_spills): Consider subreg reload pseudos too.
13176         * lra-constraints.c (simplify_operand_subreg): Use
13177         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13178         (curr_insn_transform): Recognize and do optional reloads.
13179         (undo_optional_reloads): New.
13180         (lra_undo_inheritance): Call undo_optional_reloads.
13182 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13184         * tree-complex.c (expand_complex_operations_1): Fix typo.
13186 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13188         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13189         (tune_params): New member 'const vec_costs'.
13190         * config/aarch64/aarch64.c (generic_vector_cost): New.
13191         (generic_tunings): New member 'generic_vector_cost'.
13192         (aarch64_builtin_vectorization_cost): New.
13193         (aarch64_add_stmt_cost): New.
13194         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13195         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13197 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13199         PR target/57777
13200         * config/i386/predicates.md (vsib_address_operand): Disallow
13201         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13203 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13205         PR middle-end/55030
13206         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13207         expand_builtin_setjmp_receiver.
13208         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13209         with NULL for the label parameter.
13210         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13211         the frame-pointer.  Adjust comments.
13212         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13213         only if LABEL is non-NULL.
13215 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13217         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13218         (ARM_ABI_AAPCS64): Ditto.
13219         (arm_abi): Ditto.
13220         (ARM_DEFAULT_ABI): Ditto.
13222 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13224         * config/aarch64/aarch64-builtins.c
13225         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13226         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13227         (st1): Likewise.
13228         * config/aarch64/aarch64-simd.md
13229         (aarch64_ld1<VALL:mode>): New.
13230         (aarch64_st1<VALL:mode>): Likewise.
13231         * config/aarch64/arm_neon.h
13232         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13234 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13236         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13237         target ISA is AVX.
13238         (ix86_option_override_internal):Turn on all -mavx target flags by
13239         default as they are dependent on AVX anyway.
13241 2013-07-02  Cary Coutant  <ccoutant@google.com>
13243         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13244         deterministic hash.
13245         (loc_checksum_ordered): Likewise.
13246         (hash_loc_operands): Remove inline keyword.
13248 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13250         PR tree-optimization/57741
13251         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13252         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13253         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13254         Allow REAL_CST step_exprs if flag_associative_math.
13255         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13257 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13259         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13261 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13263         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13265 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13267         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13268         encoding.
13269         (iorsi3_insn): Likewise.
13270         (arm_xorsi3): Likewise.
13272 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13274         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13275         to ...
13276         (attribute "type"): ... here, and prefix with "wmmx_".
13277         (attribute "core_cycles"): Update for attribute changes.
13278         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13279         (tbcstv4hi): Likewise.
13280         (tbcstv2si): Likewise.
13281         (iwmmxt_iordi3): Likewise.
13282         (iwmmxt_xordi3): Likewise.
13283         (iwmmxt_anddi3): Likewise.
13284         (iwmmxt_nanddi3): Likewise.
13285         (iwmmxt_arm_movdi): Likewise.
13286         (iwmmxt_movsi_insn): Likewise.
13287         (mov<mode>_internal): Likewise.
13288         (and<mode>3_iwmmxt): Likewise.
13289         (ior<mode>3_iwmmxt): Likewise.
13290         (xor<mode>3_iwmmxt): Likewise.
13291         (add<mode>3_iwmmxt): Likewise.
13292         (ssaddv8qi3): Likewise.
13293         (ssaddv4hi3): Likewise.
13294         (ssaddv2si3): Likewise.
13295         (usaddv8qi3): Likewise.
13296         (usaddv4hi3): Likewise.
13297         (usaddv2si3): Likewise.
13298         (sub<mode>3_iwmmxt): Likewise.
13299         (sssubv8qi3): Likewise.
13300         (sssubv4hi3): Likewise.
13301         (sssubv2si3): Likewise.
13302         (ussubv8qi3): Likewise.
13303         (ussubv4hi3): Likewise.
13304         (ussubv2si3): Likewise.
13305         (mulv4hi3_iwmmxt): Likewise.
13306         (smulv4hi3_highpart): Likewise.
13307         (umulv4hi3_highpart): Likewise.
13308         (iwmmxt_wmacs): Likewise.
13309         (iwmmxt_wmacsz): Likewise.
13310         (iwmmxt_wmacu): Likewise.
13311         (iwmmxt_wmacuz): Likewise.
13312         (iwmmxt_clrdi): Likewise.
13313         (iwmmxt_clrv8qi): Likewise.
13314         (iwmmxt_clr4hi): Likewise.
13315         (iwmmxt_clr2si): Likewise.
13316         (iwmmxt_uavgrndv8qi3): Likewise.
13317         (iwmmxt_uavgrndv4hi3): Likewise.
13318         (iwmmxt_uavgv8qi3): Likewise.
13319         (iwmmxt_uavgv4hi3): Likewise.
13320         (iwmmxt_tinsrb): Likewise.
13321         (iwmmxt_tinsrh): Likewise.
13322         (iwmmxt_tinsrw): Likewise.
13323         (iwmmxt_textrmub): Likewise.
13324         (iwmmxt_textrmsb): Likewise.
13325         (iwmmxt_textrmuh): Likewise.
13326         (iwmmxt_textrmsh): Likewise.
13327         (iwmmxt_textrmw): Likewise.
13328         (iwmxxt_wshufh): Likewise.
13329         (eqv8qi3): Likewise.
13330         (eqv4hi3): Likewise.
13331         (eqv2si3): Likewise.
13332         (gtuv8qi3): Likewise.
13333         (gtuv4hi3): Likewise.
13334         (gtuv2si3): Likewise.
13335         (gtv8qi3): Likewise.
13336         (gtv4hi3): Likewise.
13337         (gtv2si3): Likewise.
13338         (smax<mode>3_iwmmxt): Likewise.
13339         (umax<mode>3_iwmmxt): Likewise.
13340         (smin<mode>3_iwmmxt): Likewise.
13341         (umin<mode>3_iwmmxt): Likewise.
13342         (iwmmxt_wpackhss): Likewise.
13343         (iwmmxt_wpackwss): Likewise.
13344         (iwmmxt_wpackdss): Likewise.
13345         (iwmmxt_wpackhus): Likewise.
13346         (iwmmxt_wpackwus): Likewise.
13347         (iwmmxt_wpackdus): Likewise.
13348         (iwmmxt_wunpckihb): Likewise.
13349         (iwmmxt_wunpckihh): Likewise.
13350         (iwmmxt_wunpckihw): Likewise.
13351         (iwmmxt_wunpckilb): Likewise.
13352         (iwmmxt_wunpckilh): Likewise.
13353         (iwmmxt_wunpckilw): Likewise.
13354         (iwmmxt_wunpckehub): Likewise.
13355         (iwmmxt_wunpckehuh): Likewise.
13356         (iwmmxt_wunpckehuw): Likewise.
13357         (iwmmxt_wunpckehsb): Likewise.
13358         (iwmmxt_wunpckehsh): Likewise.
13359         (iwmmxt_wunpckehsw): Likewise.
13360         (iwmmxt_wunpckelub): Likewise.
13361         (iwmmxt_wunpckeluh): Likewise.
13362         (iwmmxt_wunpckeluw): Likewise.
13363         (iwmmxt_wunpckelsb): Likewise.
13364         (iwmmxt_wunpckelsh): Likewise.
13365         (iwmmxt_wunpckelsw): Likewise.
13366         (ror<mode>3): Likewise.
13367         (ashr<mode>3_iwmmxt): Likewise.
13368         (lshr<mode>3_iwmmxt): Likewise.
13369         (ashl<mode>3_iwmmxt): Likewise.
13370         (ror<mode>3_di): Likewise.
13371         (ashr<mode>3_di): Likewise.
13372         (lshr<mode>3_di): Likewise.
13373         (ashl<mode>3_di): Likewise.
13374         (iwmmxt_wmadds): Likewise.
13375         (iwmmxt_wmaddu): Likewise.
13376         (iwmmxt_tmia): Likewise.
13377         (iwmmxt_tmiaph): Likewise.
13378         (iwmmxt_tmiabb): Likewise.
13379         (iwmmxt_tmiatb): Likewise.
13380         (iwmmxt_tmiabt): Likewise.
13381         (iwmmxt_tmiatt): Likewise.
13382         (iwmmxt_tmovmskb): Likewise.
13383         (iwmmxt_tmovmskh): Likewise.
13384         (iwmmxt_tmovmskw): Likewise.
13385         (iwmmxt_waccb): Likewise.
13386         (iwmmxt_wacch): Likewise.
13387         (iwmmxt_waccw): Likewise.
13388         (iwmmxt_waligni): Likewise.
13389         (iwmmxt_walignr): Likewise.
13390         (iwmmxt_walignr0): Likewise.
13391         (iwmmxt_walignr1): Likewise.
13392         (iwmmxt_walignr2): Likewise.
13393         (iwmmxt_walignr3): Likewise.
13394         (iwmmxt_wsadb): Likewise.
13395         (iwmmxt_wsadh): Likewise.
13396         (iwmmxt_wsadbz): Likewise.
13397         (iwmmxt_wsadhz): Likewise.
13398         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13399         (iwmmxt_wabsdiffb): Likewise.
13400         (iwmmxt_wabsdiffh): Likewise.
13401         (iwmmxt_wabsdiffw): Likewise.
13402         (iwmmxt_waddsubhx): Likewise
13403         (iwmmxt_wsubaddhx): Likewise.
13404         (addc<mode>3): Likewise.
13405         (iwmmxt_avg4): Likewise.
13406         (iwmmxt_avg4r): Likewise.
13407         (iwmmxt_wmaddsx): Likewise.
13408         (iwmmxt_wmaddux): Likewise.
13409         (iwmmxt_wmaddsn): Likewise.
13410         (iwmmxt_wmaddun): Likewise.
13411         (iwmmxt_wmulwsm): Likewise.
13412         (iwmmxt_wmulwum): Likewise.
13413         (iwmmxt_wmulsmr): Likewise.
13414         (iwmmxt_wmulumr): Likewise.
13415         (iwmmxt_wmulwsmr): Likewise.
13416         (iwmmxt_wmulwumr): Likewise.
13417         (iwmmxt_wmulwl): Likewise.
13418         (iwmmxt_wqmulm): Likewise.
13419         (iwmmxt_wqmulwm): Likewise.
13420         (iwmmxt_wqmulmr): Likewise.
13421         (iwmmxt_wqmulwmr): Likewise.
13422         (iwmmxt_waddbhusm): Likewise.
13423         (iwmmxt_waddbhusl): Likewise.
13424         (iwmmxt_wqmiabb): Likewise.
13425         (iwmmxt_wqmiabt): Likewise.
13426         (iwmmxt_wqmiatb): Likewise.
13427         (iwmmxt_wqmiatt): Likewise.
13428         (iwmmxt_wqmiabbn): Likewise.
13429         (iwmmxt_wqmiabtn): Likewise.
13430         (iwmmxt_wqmiatbn): Likewise.
13431         (iwmmxt_wqmiattn): Likewise.
13432         (iwmmxt_wmiabb): Likewise.
13433         (iwmmxt_wmiabt): Likewise.
13434         (iwmmxt_wmiatb): Likewise.
13435         (iwmmxt_wmiatt): Likewise.
13436         (iwmmxt_wmiabbn): Likewise.
13437         (iwmmxt_wmiabtn): Likewise.
13438         (iwmmxt_wmiatbn): Likewise.
13439         (iwmmxt_wmiattn): Likewise.
13440         (iwmmxt_wmiawbb): Likewise.
13441         (iwmmxt_wmiawbt): Likewise.
13442         (iwmmxt_wmiawtb): Likewise.
13443         (iwmmxt_wmiawtt): Likewise.
13444         (iwmmxt_wmiawbbn): Likewise.
13445         (iwmmxt_wmiawbtn): Likewise.
13446         (iwmmxt_wmiawtbn): Likewise.
13447         (iwmmxt_wmiawttn): Likewise.
13448         (iwmmxt_wmerge): Likewise.
13449         (iwmmxt_tandc<mode>3): Likewise.
13450         (iwmmxt_torc<mode>3): Likewise.
13451         (iwmmxt_torvsc<mode>3): Likewise.
13452         (iwmmxt_textrc<mode>3): Likewise.
13453         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13454         (wmmxt_pack): Likewise.
13455         (wmmxt_mult_c1): Likewise.
13456         (wmmxt_mult_c2): Likewise.
13457         (wmmxt_alu_c1): Likewise.
13458         (wmmxt_alu_c2): Likewise.
13459         (wmmxt_alu_c3): Likewise.
13460         (wmmxt_transfer_c1): Likewise.
13461         (wmmxt_transfer_c2): Likewise.
13462         (wmmxt_transfer_c3): Likewise.
13463         (marvell_f_iwmmxt_wstr): Likewise.
13464         (marvell_f_iwmmxt_wldr): Likewise.
13466 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13468         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13470 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13472         Revert:
13473         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13474         * lra-constraints.c (need_for_split_p): Check call used hard regs
13475         living through calls.
13477         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13478         call used regs for call insn.
13480 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13482         PR target/57736
13483         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13484         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13486 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13488         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13490 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13492         * lra-constraints.c (need_for_split_p): Check call used hard regs
13493         living through calls.
13495 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13497         PR target/57744
13498         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13499         to tie with any other modes.  Eliminate Altivec vector mode tests,
13500         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13501         code, to return 0 if testing MODE2 for a condition, if we've
13502         already tested MODE1 for the same condition.
13504 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13506         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13507         layout.
13509 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13511         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13512         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13514 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13516         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13517         Define.
13518         (aarch64_symbolic_constant_p): Remove.
13519         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13520         static.  Fix line length and white space.
13521         (aarch64_symbolic_constant_p): Remove.
13522         * config/aarch64/predicates.md (aarch64_valid_symref):
13523         Use aarch64_classify_symbol_expression.
13525 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13527         * config/arm/constraints.md (Ts): New constraint.
13528         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13529         16-bit encodings.
13530         (compare_scc): Use "Ts" constraint for operand 0.
13531         (ior_scc_scc): Likewise.
13532         (and_scc_scc): Likewise.
13533         (and_scc_scc_nodom): Likewise.
13534         (ior_scc_scc_cmp): Likewise for operand 7.
13535         (and_scc_scc_cmp): Likewise.
13536         * config/arm/thumb2.md (thumb2_movsi_insn):
13537         Add alternatives for 16-bit encodings.
13538         (thumb2_movhi_insn): Likewise.
13539         (thumb2_movsicc_insn): Likewise.
13540         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13541         (thumb2_negscc): Use "Ts" constraint.
13542         Move mvn instruction outside cond_exec block.
13543         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13544         for 16-bit encodings.
13546 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13548         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13549         encoding.
13550         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13551         (mulsi3subsi): Likewise.
13552         (mulsidi3adddi): Likewise.
13553         (mulsidi3_v6): Likewise.
13554         (umulsidi3_v6): Likewise.
13555         (umulsidi3adddi_v6): Likewise.
13556         (smulsi3_highpart_v6): Likewise.
13557         (umulsi3_highpart_v6): Likewise.
13558         (mulhisi3tb): Likewise.
13559         (mulhisi3bt): Likewise.
13560         (mulhisi3tt): Likewise.
13561         (maddhisi4): Likewise.
13562         (maddhisi4tb): Likewise.
13563         (maddhisi4tt): Likewise.
13564         (maddhidi4): Likewise.
13565         (maddhidi4tb): Likewise.
13566         (maddhidi4tt): Likewise.
13567         (zeroextractsi_compare0_scratch): Likewise.
13568         (insv_zero): Likewise.
13569         (insv_t2): Likewise.
13570         (anddi_notzesidi_di): Likewise.
13571         (anddi_notsesidi_di): Likewise.
13572         (andsi_notsi_si): Likewise.
13573         (iordi_zesidi_di): Likewise.
13574         (xordi_zesidi_di): Likewise.
13575         (andsi_iorsi3_notsi): Likewise.
13576         (smax_0): Likewise.
13577         (smax_m1): Likewise.
13578         (smin_0): Likewise.
13579         (not_shiftsi): Likewise.
13580         (unaligned_loadsi): Likewise.
13581         (unaligned_loadhis): Likewise.
13582         (unaligned_loadhiu): Likewise.
13583         (unaligned_storesi): Likewise.
13584         (unaligned_storehi): Likewise.
13585         (extv_reg): Likewise.
13586         (extzv_t2): Likewise.
13587         (divsi3): Likewise.
13588         (udivsi3): Likewise.
13589         (arm_zero_extendhisi2addsi): Likewise.
13590         (arm_zero_extendqisi2addsi): Likewise.
13591         (compareqi_eq0): Likewise.
13592         (arm_extendhisi2_v6): Likewise.
13593         (arm_extendqisi2addsi): Likewise.
13594         (arm_movt): Likewise.
13595         (thumb2_ldrd): Likewise.
13596         (thumb2_ldrd_base): Likewise.
13597         (thumb2_ldrd_base_neg): Likewise.
13598         (thumb2_strd): Likewise.
13599         (thumb2_strd_base): Likewise.
13600         (thumb2_strd_base_neg): Likewise.
13601         (arm_negsi2): Add alternative for 16-bit encoding.
13602         (arm_one_cmplsi2): Likewise.
13604 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13606         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13607         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13608         (movdfcc): Likewise.
13609         * config/arm/vfp.md (*thumb2_movsf_vfp):
13610         Disable predication for arm_restrict_it.
13611         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13612         (*thumb2_movdfcc_vfp): Likewise.
13613         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13614         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13615         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13616         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13617         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13618         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13619         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13620         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13621         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13622         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13623         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13624         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13625         Disable predication for arm_restrict_it.
13627 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13629         * config/i386/bmiintrin.h (_bextr_u32): New.
13630         (_bextr_u64): Ditto.
13632 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13634         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13635         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13636         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13637         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13638         * config/mips/n32-elf.h: ...this new file.
13640 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13642         PR target/57224
13643         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13644         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13646 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13648         * config/mips/mips-tables.opt: Regenerate.
13649         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13650         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13651         * doc/invoke.texi: Add -m14kc.
13653 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13655         PR target/57623
13656         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13657         constraints of operand 1 and 2.
13659         PR target/57623
13660         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13661         to match RTL canonicalization.  Swap predicates and
13662         constraints of operand 1 and 2.
13664 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13666         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13667         Process OP_INOUT regs for splitting too.
13669 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13671         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13672         decl before the loop, initialize to NULL.
13673         (vectorizable_load): Initialize ptr_incr to NULL.
13675 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13677         PR lto/57208
13678         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13679         * ipa-ref.c (ipa_maybe_record_reference): New function.
13680         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13681         * ipa-cp.c (create_specialized_node): Record potential references from
13682         aggvals.
13683         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13685 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13687         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13688         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13689         'mode' to force_reg.
13690         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13691         (aarch64_expand_mov_immediate): Likewise.
13693 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13695         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13696         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13698 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13700         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13701         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13702         (struct machine_function): Add tbegin_p.
13703         (s390_canonicalize_comparison): Fold CC mode compares to
13704         conditional jump if possible.
13705         (s390_emit_jump): Return the emitted jump.
13706         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13707         Handle CCRAWmode compares.
13708         (s390_option_override): Default to -mhtm if available.
13709         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13710         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13711         FPRs instead of df_regs_ever_live_p.
13712         (s390_optimize_nonescaping_tx): New function.
13713         (s390_init_frame_layout): Extend clobbered_regs array to cover
13714         FPRs as well.
13715         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13716         (s390_expand_tbegin): New function.
13717         (enum s390_builtin): New enum definition.
13718         (code_for_builtin): New array definition.
13719         (s390_init_builtins): New function.
13720         (s390_expand_builtin): New function.
13721         (TARGET_INIT_BUILTINS): Define.
13722         (TARGET_EXPAND_BUILTIN): Define.
13723         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13724         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13725         (s390_alc_comparison): Likewise.
13726         * config/s390/s390-modes.def: Add CCRAWmode.
13727         * config/s390/s390.h (processor_flags): Add PF_TX.
13728         (TARGET_CPU_HTM): Define macro.
13729         (TARGET_HTM): Define macro.
13730         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13731         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13732         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13733         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13734         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13735         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13736         ("*cc_to_int"): Move up.
13737         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13738         constants other than 0.
13739         ("*ccraw_to_int"): New insn and splitter definition.
13740         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13741         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13742         ("tx_assist"): New expander.
13743         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13744         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13745         * config/s390/s390.opt: Add -mhtm option.
13746         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13747         * config/s390/htmxlintrin.h: New file.
13748         * config/s390/htmintrin.h: New file.
13749         * config/s390/s390intrin.h: New file.
13750         * doc/extend.texi: Document htm builtins.
13751         * config.gcc: Add the new header files to extra_headers.
13753 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13755         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13756         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13758 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13759             Pat Haugen  <pthaugen@us.ibm.com>
13760             Peter Bergner  <bergner@vnet.ibm.com>
13762         * config/rs6000/power8.md: New.
13763         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13764         setting for power8 entry.
13765         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13766         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13767         test for Power4/Power5 only.
13768         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13769         support.
13770         (force_new_group): Adjust comment.
13771         * config/rs6000/rs6000.md: Include power8.md.
13773 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13775         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13776         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13777         declaration.
13778         (tune_params): Update comment.
13779         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13780         (arm_max_conditional_execute): New function.
13781         (thumb2_final_prescan_insn): Use max_insn_skipped and
13782         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13784 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13786         PR tree-optimization/57705
13787         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13788         SSA_NAME step, provided that it is not defined inside the loop.
13789         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13790         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13792 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13794         PR middle-end/57670
13795         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13796         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13797         calls in the dump.
13798         (ipa_note_param_call): Initialize member_ptr flag.
13799         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13800         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13801         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13802         (ipa_read_indirect_edge_info): Likewise.
13804 2013-06-25  Richard Biener  <rguenther@suse.de>
13806         PR middle-end/56977
13807         * passes.c (init_optimization_passes): Move pass_fold_builtins
13808         and pass_dce earlier with -Og.
13810 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13812         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13813         <BIT_FIELD_REF>: Remove trailing TAB.
13814         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13815         remove blank line.
13817 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13819         PR tree-optimization/57358
13820         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13821         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13822         (ipa_analyze_params_uses): Generate pessimistic info when true.
13824 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13826         PR tree-optimization/57539
13827         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13828         global.inlined_to of the new node to it.  All callers changed.
13829         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13830         inlining_into, pass it to cgraph_clone_node.
13831         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13832         ipa_free_edge_args_substructures.
13833         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13834         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13835         Assert we have found an rdesc in the rdesc list.
13837 2013-06-24  Richard Biener  <rguenther@suse.de>
13839         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13840         (pointer_set_lookup): Declare.
13841         (class pointer_map): New template class implementing a
13842         generic pointer to T map.
13843         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13844         pointer_map<T>::contains, pointer_map<T>::insert,
13845         pointer_map<T>::traverse): New functions.
13846         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13847         (pointer_set_lookup): New function.
13848         (pointer_set_contains): Use pointer_set_lookup.
13849         (pointer_set_insert): Likewise.
13850         (insert_aux): Remove.
13851         (struct pointer_map_t): Embed a pointer_set_t.
13852         (pointer_map_create): Adjust.
13853         (pointer_map_destroy): Likewise.
13854         (pointer_map_contains): Likewise.
13855         (pointer_map_insert): Likewise.
13856         (pointer_map_traverse): Likewise.
13857         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13858         pointer_map<unsigned> instead of a pointer_map_t.
13859         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13860         (streamer_tree_cache_lookup): Likewise.
13861         (streamer_tree_cache_create): Likewise.
13862         (streamer_tree_cache_delete): Likewise.
13863         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13864         pointer_map<unsigned> instead of a pointer_map_t.
13865         (lto_init_tree_ref_encoder): Adjust.
13866         (lto_destroy_tree_ref_encoder): Likewise.
13867         * lto-section-out.c (lto_output_decl_index): Likewise.
13868         (lto_record_function_out_decl_state): Likewise.
13869         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13871 2013-06-24  Richard Biener  <rguenther@suse.de>
13873         PR tree-optimization/57488
13874         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13876 2013-06-24  Alan Modra  <amodra@gmail.com>
13878         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13879         (gen_easy_altivec_constant): Likewise.
13880         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13881         easy_vector_constant_msb): Likewise.
13883 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13885         PR target/57688
13886         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13887         add missing return true.
13889 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13891         PR target/52483
13892         * config/sh/predicates.md (general_extend_operand): Invoke
13893         general_movsrc_operand for memory operands.
13894         (general_movsrc_operand): Allow reg+reg addressing, do not use
13895         general_operand for memory operands.
13897 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13899         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13900         when current target options does not apply.
13901         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13902         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13903         * config/i386/bmiintrin.h: Pass appropriate target
13904         attributes to header.
13905         * config/i386/mmintrin.h: Ditto.
13906         * config/i386/nmmintrin.h: Ditto.
13907         * config/i386/avx2intrin.h: Ditto.
13908         * config/i386/fxsrintrin.h: Ditto.
13909         * config/i386/tbmintrin.h: Ditto.
13910         * config/i386/xsaveintrin.h: Ditto.
13911         * config/i386/f16cintrin.h: Ditto.
13912         * config/i386/xtestintrin.h: Ditto.
13913         * config/i386/xsaveoptintrin.h: Ditto.
13914         * config/i386/bmi2intrin.h: Ditto.
13915         * config/i386/lzcntintrin.h: Ditto.
13916         * config/i386/smmintrin.h: Ditto.
13917         * config/i386/wmmintrin.h: Ditto.
13918         * config/i386/x86intrin.h: Remove all header include guards.
13919         * config/i386/prfchwintrin.h: Ditto.
13920         * config/i386/pmmintrin.h: Ditto.
13921         * config/i386/tmmintrin.h: Ditto.
13922         * config/i386/xmmintrin.h: Ditto.
13923         * config/i386/popcntintrin.h: Ditto.
13924         * config/i386/rdseedintrin.h: Ditto.
13925         * config/i386/ammintrin.h: Ditto.
13926         * config/i386/emmintrin.h: Ditto.
13927         * config/i386/immintrin.h: Remove all header include guards.
13928         * config/i386/fma4intrin.h: Ditto.
13929         * config/i386/lwpintrin.h: Ditto.
13930         * config/i386/xopintrin.h: Ditto.
13931         * config/i386/ia32intrin.h: Ditto.
13932         * config/i386/avxintrin.h: Ditto.
13933         * config/i386/rtmintrin.h: Ditto.
13934         * config/i386/fmaintrin.h: Ditto.
13935         * config/i386/mm3dnow.h: Ditto.
13937 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13939         * common/config/i386/i386-common.c: Handle LZCNT.
13941 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13943         * doc/extend.texi: Use __atomic_store_n instead of
13944         __atomic_store in HLE example.
13946 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13948         * config/sh/sh.c: Remove <cstdlib> workaround.
13950 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13952         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13954 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13956         * doc/extend.texi: Document that __atomic_clear and
13957           __atomic_test_and_set should only be used with bool.
13959 2013-06-20  Jan Hubicka  <jh@suse.cz>
13961         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13962         types_same_for_odr.
13963         * tree.c (decls_same_for_odr): New function.
13964         (same_for_edr): New function.
13965         (types_same_for_odr): New function.
13966         (get_binfo_at_offset): Use it.
13967         * tree.h (types_same_for_odr): Declare.
13969 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13970             Jason Merrill  <jason@redhat.com>
13972         * system.h: Include <cstdlib> as well as <stdlib.h>.
13974 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13976         PR target/57655
13977         * config/i386/i386.c (construct_container): Report error if
13978         long double is used with disabled x87 float returns.
13980 2013-06-20  Jan Hubicka  <jh@suse.cz>
13982         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13984 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13986         PR rtl-optimization/57425
13987         PR rtl-optimization/57569
13988         * alias.c (write_dependence_p): Remove parameters mem_mode and
13989         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13990         Changed all callers.
13991         (canon_anti_dependence): Get comments and semantics in sync.
13992         Add parameter mem_canonicalized.  Changed all callers.
13993         * rtl.h (canon_anti_dependence): Update prototype.
13995 2013-06-20  Richard Biener  <rguenther@suse.de>
13997         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13998         byte case, inline streamer_read_uchar and defer section
13999         overrun check.
14001 2013-06-20  Richard Biener  <rguenther@suse.de>
14003         PR tree-optimization/57584
14004         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
14005         SSA names into the expanded expression that take part in
14006         abnormal coalescing.
14008 2013-06-19  Sharad Singhai  <singhai@google.com>
14010         * gcov.c (print_usage): Handle new option.
14011         (process_args): Ditto.
14012         (get_gcov_intermediate_filename): New function.
14013         (output_intermediate_file): New function.
14014         (output_gcov_file): New function
14015         (generate_results): Handle new option.
14016         (release_function): Relase demangled name.
14017         (read_graph_file): Handle demangled name.
14018         (output_lines): Ditto.
14019         * doc/gcov.texi: Document gcov intermediate format.
14021 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14023         PR bootstrap/57604
14024         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14025         (lra_emit_add): Use the functions.  Add comment about Y as an
14026         address segment.
14028 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14030         PR driver/57652
14031         * collect2.c (collect_atexit): New.
14032         (collect_exit): Delete.
14033         (main): Register collect_atexit with atexit.
14034         (collect_wait): Change collect_exit to exit.
14035         (do_wait): Same.
14036         * collect2.h (collect_exit): Delete.
14037         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14039 2013-06-19  Wei Mi  <wmi@google.com>
14041         PR rtl-optimization/57518
14042         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14043         if regno is used in paradoxical subreg.
14044         (update_equiv_regs): Check pdx_subregs[regno] before
14045         set a reg to be equivalent with a mem.
14047 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14049         PR driver/57651
14050         * file-find.h (find_a_file): Add a mode parameter.
14051         * file-find.c (find_a_file): Likewise.
14052         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14053         with X_OK for the executables.
14054         * collect2.c (main): Call find_a_file with X_OK.
14056 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14058         PR target/56942
14059         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14060         Use NEXT_INSN instead of next_real_insn.
14062 2013-06-19  Jan Hubicka  <jh@suse.cz>
14064         * cgraph.h (const_value_known_p): Replace by ...
14065         (ctor_for_folding): .. this one.
14066         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14067         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14068         * expr.c (expand_expr_real_1): Likewise.
14069         (string_constant): Likewise.
14070         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14071         * ipa.c (process_references): Likewise.
14072         (symtab_remove_unreachable_nodes): Likewise.
14073         * ipa-inline-analysis.c (param_change_prob): Likewise.
14074         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14075         (get_base_constructor): Likwise.
14076         * varpool.c (varpool_remove_node): Likewise.
14077         (varpool_remove_initializer): LIkewise.
14078         (dump_varpool_node): LIkwise.
14079         (const_value_known_p): Rewrite to ...
14080         (ctor_for_folding): ... this one.
14082 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14084         PR driver/57651
14085         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14086         PERSONALITY in $PATH derived prefixes.
14088 2013-06-19  Jeff Law  <law@redhat.com>
14090         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14091         in comment.
14093         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14094         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14095         booleans.
14097 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14099         * config/arm/vfp.md: Move VFP instruction classification documentation
14100         to ...
14101         * config/arm/arm.md: ... here.  Update instruction classification
14102         documentation.
14104 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14106         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14107         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14108         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14110 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14112         PR target/57609
14113         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14114         with NEXT_INSN.  Use tablejump_p to check for jump table data
14115         insns.
14117 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14119         PR c++/56544
14120         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14121         that now in C++ the value is correct per the C++ standards.
14123 2013-06-19  Richard Biener  <rguenther@suse.de>
14125         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14126         for global context.
14128 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14130         Revert:
14131         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14133         PR target/57609
14134         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14135         with next_active_insn.
14137 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14139         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14140         functions are inlined during failures to flag an error.
14141         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14142         in early inline pass.
14144 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14146         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14147         in comments.
14149 2013-06-18  Julian Brown  <julian@codesourcery.com>
14151         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14152         Permit virtual register pre-reload if !strict.
14153         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14154         change.
14155         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14156         prototype.
14157         * config/arm/neon.md (movmisalign<mode>): Use
14158         neon_perm_struct_or_reg_operand instead of
14159         neon_struct_or_register_operand.
14160         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14161         neon_permissive_struct_operand instead of neon_struct_operand.
14162         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14163         neon_vector_mem_operand.
14164         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14165         neon_vector_mem_operand.
14166         (neon_permissive_struct_operand): New.
14167         (neon_struct_or_register_operand): Rename to...
14168         (neon_perm_struct_or_reg_operand): This. Adjust call to
14169         neon_vector_mem_operand.
14171 2013-06-18  Richard Biener  <rguenther@suse.de>
14173         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14174         * lto-streamer.h: Include pointer-set.h.
14175         (struct lto_decl_slot): Remove.
14176         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14177         Remove next_index entry.
14178         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14179         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14180         (lto_init_tree_ref_encoder): Adjust.
14181         (lto_destroy_tree_ref_encoder): Likewise.
14182         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14183         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14184         (lto_output_decl_index): Adjust.
14185         (lto_new_out_decl_state): Likewise.
14186         (lto_record_function_out_decl_state): Likewise.
14187         * lto-streamer-out.c (copy_function): Likewise.
14189 2013-06-18  Richard Biener  <rguenther@suse.de>
14191         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14192         * cgraphunit.c: Include cfgloop.h.
14193         (init_lowered_empty_function): Initialize the loop tree.
14194         (assemble_thunk): Insert new BBs into loops.
14196 2013-06-18  Richard Biener  <rguenther@suse.de>
14198         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14199         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14200         the map from cache entry to cache index optional.
14201         (streamer_tree_cache_replace_tree): Adjust accordingly.
14202         (streamer_tree_cache_append): Likewise.
14203         (streamer_tree_cache_delete): Likewise.
14204         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14205         streamer cache map from cache entry to cache index.
14206         * lto-streamer-out.c (create_output_block): Adjust.
14208 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14210         * config/arm/arm.md (attribute "insn"): Move multiplication and
14211         division attributes to...
14212         (attribute "type"): ... here.  Remove mult.
14213         (attribute "mul32"): New attribute.
14214         (attribute "mul64"): Add umaal.
14215         (*arm_mulsi3): Update attributes.
14216         (*arm_mulsi3_v6): Likewise.
14217         (*thumb_mulsi3): Likewise.
14218         (*thumb_mulsi3_v6): Likewise.
14219         (*mulsi3_compare0): Likewise.
14220         (*mulsi3_compare0_v6): Likewise.
14221         (*mulsi_compare0_scratch): Likewise.
14222         (*mulsi_compare0_scratch_v6): Likewise.
14223         (*mulsi3addsi): Likewise.
14224         (*mulsi3addsi_v6): Likewise.
14225         (*mulsi3addsi_compare0): Likewise.
14226         (*mulsi3addsi_compare0_v6): Likewise.
14227         (*mulsi3addsi_compare0_scratch): Likewise.
14228         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14229         (*mulsi3subsi): Likewise.
14230         (*mulsidi3adddi): Likewise.
14231         (*mulsi3addsi_v6): Likewise.
14232         (*mulsidi3adddi_v6): Likewise.
14233         (*mulsidi3_nov6): Likewise.
14234         (*mulsidi3_v6): Likewise.
14235         (*umulsidi3_nov6): Likewise.
14236         (*umulsidi3_v6): Likewise.
14237         (*umulsidi3adddi): Likewise.
14238         (*umulsidi3adddi_v6): Likewise.
14239         (*smulsi3_highpart_nov6): Likewise.
14240         (*smulsi3_highpart_v6): Likewise.
14241         (*umulsi3_highpart_nov6): Likewise.
14242         (*umulsi3_highpart_v6): Likewise.
14243         (mulhisi3): Likewise.
14244         (*mulhisi3tb): Likewise.
14245         (*mulhisi3bt): Likewise.
14246         (*mulhisi3tt): Likewise.
14247         (maddhisi4): Likewise.
14248         (*maddhisi4tb): Likewise.
14249         (*maddhisi4tt): Likewise.
14250         (maddhidi4): Likewise.
14251         (*maddhidi4tb): Likewise.
14252         (*maddhidi4tt): Likewise.
14253         (divsi3): Likewise.
14254         (udivsi3): Likewise.
14255         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14256         (thumb2_mulsi_short_compare0): Likewise.
14257         (thumb2_mulsi_short_compare0_scratch): Likewise.
14258         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14259         (1020mult2): Likewise.
14260         (1020mult3): Likewise.
14261         (1020mult4): Likewise.
14262         (1020mult5): Likewise.
14263         (1020mult6): Likewise.
14264         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14265         change.
14266         (cortex_a15_mult64): Likewise.
14267         (cortex_a15_sdiv): Likewise.
14268         (cortex_a15_udiv): Likewise.
14269         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14270         (mult2): Likewise.
14271         (mult3): Likewise.
14272         (mult4): Likewise.
14273         (mult5): Likewise.
14274         (mult6): Likewise.
14275         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14276         (pj4_ir_div): Likewise.
14277         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14278         (11_mult2): Likewise.
14279         (11_mult3): Likewise.
14280         (11_mult4): Likewise.
14281         (11_mult5): Likewise.
14282         (11_mult6): Likewise.
14283         (11_mult7): Likewise.
14284         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14285         (cortex_a8_mla): Likewise.
14286         (cortex_a8_mull): Likewise.
14287         (cortex_a8_smulwy): Likewise.
14288         (cortex_a8_smlald): Likewise.
14289         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14290         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14291         (cortex_r4_mul_3): Likewise.
14292         (cortex_r4_mla_4): Likewise.
14293         (cortex_r4_mla_3): Likewise.
14294         (cortex_r4_smlald): Likewise.
14295         (cortex_r4_mull): Likewise.
14296         (cortex_r4_sdiv): Likewise.
14297         (cortex_r4_udiv): Likewise.
14298         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14299         (cortex_a7_idiv): Likewise.
14300         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14301         (9_mult2): Likewise.
14302         (9_mult3): Likewise.
14303         (9_mult4): Likewise.
14304         (9_mult5): Likewise.
14305         (9_mult6): Likewise.
14306         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14307         (cortex_a53_sdiv): Likewise.
14308         (cortex_a53_udiv): Likewise.
14309         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14310         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14311         (mp626_mult2): Likewise.
14312         (mp626_mult3): Likewise.
14313         (mp626_mult4): Likewise.
14314         * config/arm/fa526.md (526_mult1): Update attribute change.
14315         (526_mult2): Likewise.
14316         * config/arm/arm-generic.md (mult): Update attribute change.
14317         (mult_ldsched_strongarm): Likewise.
14318         (mult_ldsched): Likewise.
14319         (multi_cycle): Likewise.
14320         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14321         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14322         (606te_mult2): Likewise.
14323         (606te_mult3): Likewise.
14324         (606te_mult4): Likewise.
14325         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14326         (cortex_a9_mac16): Likewise.
14327         (cortex_a9_multiply): Likewise.
14328         (cortex_a9_mac): Likewise.
14329         (cortex_a9_multiply_long): Likewise.
14330         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14331         (626te_mult2): Likewise.
14332         (626te_mult3): Likewise.
14333         (626te_mult4): Likewise.
14335 2013-06-18  Richard Biener  <rguenther@suse.de>
14337         PR lto/57334
14338         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14340 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14342         PR target/57609
14343         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14344         with next_active_insn.
14346 2013-06-18  Alan Modra  <amodra@gmail.com>
14348         * config/rs6000/rs6000.h (enum data_align): New.
14349         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14350         (DATA_ABI_ALIGNMENT): Define.
14351         (CONSTANT_ALIGNMENT): Correct comment.
14352         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14353         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14355 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14357         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14358         ATTRIBUTE_UNUSED marking.
14360 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14362         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14363         alternative and update.
14364         (aarch64_dup_lanedi): Delete.
14365         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14366         * config/aarch64/aarch64-simd-builtins.def: Update.
14368 2013-06-17  Richard Biener  <rguenther@suse.de>
14370         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14371         (lto_input_scc): Declare.
14372         (lto_input_tree_1): Likewise.
14373         (struct lto_stats_d): Add num_tree_bodies_output and
14374         num_pickle_refs_output.
14375         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14376         (lto_read_tree_1): Split out from ...
14377         (lto_read_tree): ... this.
14378         (lto_input_scc): New function.
14379         (lto_input_tree_1): Split out from ...
14380         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14381         (lto_data_in_create): Create the streamer cache without hashes.
14382         * lto-streamer-out.c (create_output_block): Create the streamer
14383         cache with hashes when not doing WPA.
14384         (lto_write_tree_1): Split out from ...
14385         (lto_write_tree): ... this.
14386         (get_symbol_initial_value): New function.
14387         (lto_output_tree_1): Split out from ...
14388         (lto_output_tree): ... this.  Write trees as series of SCCs
14389         using a DFS walk via DFS_write_tree.
14390         (struct sccs, struct scc_entry): New types.
14391         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14392         (DFS_write_tree_body): New function.
14393         (DFS_write_tree): Likewise.
14394         (hash_tree): Likewise.
14395         (scc_entry_compare): Likewise.
14396         (hash_scc): Likewise.
14397         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14398         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14399         TREE_CHAIN as regular reference.
14400         (streamer_read_integer_cst): Remove.
14401         (streamer_get_pickled_tree): Adjust.
14402         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14403         of DECL_EXTERNALs in BLOCK_VARS for now.
14404         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14405         reference.
14406         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14407         Add hash value argument and record that if hashes are recorded
14408         in the cache.
14409         (streamer_tree_cache_insert_1): Adjust.
14410         (streamer_tree_cache_insert): Likewise.
14411         (streamer_tree_cache_insert_at): Rename to ...
14412         (streamer_tree_cache_replace_tree): ... this and adjust.
14413         (streamer_tree_cache_append): Adjust.
14414         (record_common_node): Likewise.
14415         (streamer_tree_cache_create): Add argument whether to
14416         record hash values together with trees.
14417         (streamer_tree_cache_delete): Adjust.
14418         * tree-streamer.h (struct streamer_tree_cache_d): Add
14419         vector of hashes.
14420         (streamer_read_integer_cst): Remove.
14421         (streamer_tree_cache_insert): Adjust.
14422         (streamer_tree_cache_append): Likewise.
14423         (streamer_tree_cache_insert_at): Rename to ...
14424         (streamer_tree_cache_replace_tree): ... this and adjust.
14425         (streamer_tree_cache_create): Add argument whether to record hashes.
14426         (streamer_tree_cache_get): Rename to ...
14427         (streamer_tree_cache_get_tree): ... this.
14428         (streamer_tree_cache_get_hash): New function.
14429         * tree.c (cache_integer_cst): New function.
14430         * tree.h (cache_integer_cst): Declare.
14431         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14432         * lto-symtab.c (lto_varpool_replace_node): Only release
14433         DECL_INITIAL of non-prevailing decls.
14434         * varpool.c (varpool_remove_initializer): Do not release
14435         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14437 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14439         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14440         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14441         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14442         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14443         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14444         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14445         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14446         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14447         instead of TARGET_64BIT.
14448         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14449         Require ISA_HAS_<D>DIV.
14451 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14453         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14454         (mips*-*-linux*): Move default with_llsc setting to where other
14455         defaults are set.
14456         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14457         with_arch block.
14458         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14459         Likewise.  Remove default with_tune setting.  Move default float
14460         setting to its own block.  Handle with_llsc in the same block as above.
14462 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14464         PR rtl-optimization/57425
14465         PR rtl-optimization/57569
14466         * alias.c (write_dependence_p): Add new parameters mem_mode,
14467         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14468         Changed all callers.
14469         (canon_anti_dependence): New function.
14470         * cse.c (check_dependence): Use canon_anti_dependence.
14471         * cselib.c (cselib_invalidate_mem): Likewise.
14472         * rtl.h (canon_anti_dependence): Declare.
14474 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14476         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14477         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14478         ".set mips3" for 64-bit targets.
14480 2013-06-15  Dehao Chen  <dehao@google.com>
14482         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14483         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14484         (gimple_check_call_args): Likewise.
14485         * value-prof.c (check_ic_target): Likewise.
14486         * ipa-inline.c (early_inliner): Likewise.
14487         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14488         * cgraph.c (cgraph_create_edge_1): Likewise.
14489         (cgraph_make_edge_direct): Likewise.
14491 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14493         PR target/57615
14494         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14495         rs6000_output_move_128bit to handle emitting quad memory
14496         operations.  Set attribute length to 8 bytes.
14498 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14500         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14501         New pattern.
14502         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14503         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14504         (aarch64_<su>mlsl<mode>): Likewise.
14506 2013-06-14  Mike Stump  <mikestump@comcast.net>
14508         * Makefile.in (TARGET_H): Add insn-codes.h.
14510 2013-06-14  Alan Modra  <amodra@gmail.com>
14512         PR middle-end/57134
14513         PR middle-end/57586
14514         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14515         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14516         bitfield expansion when EXPAND_MEMORY.
14517         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14519 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14521         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14522         test for clearing quad memory on 32-bit later.
14524 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14526         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14527         (fold_negate_expr): Likewise.
14528         (fold_real_zero_addition_p): Handle vectors.
14529         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14531 2013-06-14  Alan Modra  <amodra@gmail.com>
14533         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14535 2013-06-13  Jan Hubicka  <jh@suse.cz>
14537         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14538         Local comdats are not externally visible.
14539         * symtab.c (dump_symtab_base): Dump externally visible.
14540         (verify_symtab_base): Verify back links in the symtab hash.
14542 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14544         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14545         CONVERT_EXPR as equal nodes.
14547 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14549         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14551 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14553         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14554         Generalize to complex and vector.
14555         * tree.c (build_all_ones_cst): New function.
14556         * tree.h (build_all_ones_cst): Declare it.
14558 2013-06-13  Alan Modra  <amodra@gmail.com>
14560         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14561         * config/rs6000/rs6000.md (signbittf2): New insn.
14562         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14563         (abstf2_internal, cmptf_internal2): Likewise.
14564         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14566 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14567             Pat Haugen  <pthaugen@us.ibm.com>
14568             Peter Bergner  <bergner@vnet.ibm.com>
14570         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14571         power8 byte, half-word, and quad-word atomic instructions.
14572         (emit_store_conditional): Likewise.
14573         (rs6000_expand_atomic_compare_and_swap): Likewise.
14574         (rs6000_expand_atomic_op): Likewise.
14576         * config/rs6000/sync.md (larx): Add new modes for power8.
14577         (stcx): Likewise.
14578         (AINT): New mode iterator to include TImode as well as normal
14579         integer modes on power8.
14580         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14581         that VSX registers are not considered.  Use AINT mode iterator
14582         instead of INT1 to allow inclusion of quad word atomic operations
14583         on power8.
14584         (load_locked<mode>): Likewise.
14585         (store_conditional<mode>): Likewise.
14586         (atomic_compare_and_swap<mode>): Likewise.
14587         (atomic_exchange<mode>): Likewise.
14588         (atomic_nand<mode>): Likewise.
14589         (atomic_fetch_<fetchop_name><mode>): Likewise.
14590         (atomic_nand_fetch<mode>): Likewise.
14591         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14592         each type.
14593         (ATOMIC): On power8, add QImode, HImode modes.
14594         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14595         modes that promote to SImode.
14596         (load_lockedti): Convert TImode arguments to PTImode, so that we
14597         get a guaranteed even/odd register pair.
14598         (load_lockedpti): Likewise.
14599         (store_conditionalti): Likewise.
14600         (store_conditionalpti): Likewise.
14602         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14603         atomic load/store instructions.
14604         (HSI): Likewise.
14606 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14608         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14609         loads.
14610         (insn_count): New attribute, with most cases extracted from...
14611         (length): ...here.  Redefine most cases in terms of insn_count.
14612         (single_insn): Delete.
14613         (can_delay): Use insn_count to check for single instructions.
14614         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14615         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14616         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14617         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14618         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14619         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14620         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14621         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14622         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14623         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14624         rather than "length".
14625         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14626         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14627         Use "insn_count" rather than "length".
14628         * config/mips/mips-dsp.md
14629         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14630         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14631         length attributes.
14633 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14635         PR tree-optimization/57361
14636         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14638 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14640         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14641         to split.
14642         (aarch64_simd_combine<mode>): New instruction expansion.
14643         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14644         function prototype.
14645         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14646         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14648 2013-06-12  Jan Hubicka  <jh@suse.cz>
14650         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14651         decl has when in streaming stage.
14652         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14653         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14655 2013-06-12  Roland Stigge  <stigge@antcom.de>
14657         PR target/57578
14658         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14660 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14662         PR tree-optimization/57537
14663         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14664         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14666 2013-06-12  Richard Biener  <rguenther@suse.de>
14668         * data-streamer.h (streamer_write_char_stream): CSE
14669         obs->current_pointer.
14670         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14671         streamer_write_char_stream manually and optimize the resulting loop.
14672         (streamer_write_hwi_stream): Likewise.
14674 2013-06-12  Jan Hubicka  <jh@suse.cz>
14676         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14677         * cgraph.h (varpool_create_empty_node): Declare.
14678         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14679         duplicated nodes.
14680         * symtab.c (symtab_unregister_node): Be lax about missin entries
14681         in node hash.
14682         (symtab_get_node): Update comment.
14683         * varpool.c (varpool_create_empty_node): Break out from ...
14684         (varpool_node_for_decl): ... here.
14685         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14687 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14689         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14690         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14691         part.  Use straight-line flow at the end.
14692         <COMPONENT_REF>: Remove superfluous else.
14693         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14695 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14697         PR target/56564
14698         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14699         target hook even for !TREE_PUBLIC decls.  If no resolution info
14700         is available, return false for common and external decls.
14702 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14704         * config/rl78/constraints.md (U): New constraint.
14705         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14706         valloc attribute.
14708 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14710         PR target/57589
14711         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14712         to allow returning address to AT_PLATFORM name.
14714 2013-06-11  Jan Hubicka  <jh@suse.cz>
14716         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14717         * cgraph.h (symtab_node_base): Add weakref flag.
14718         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14719         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14720         (output_weakrefs): Use weakref flag.
14721         * fold-const.c (simple_operand_p): Handle WEAK.
14722         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14723         * ipa.c (varpool_externally_visible_p): Drop weakref.
14724         (function_and_variable_visibility): Update comment; fix weakref
14725         sanity checks; do not clear DECL_WEAK on them.
14726         * lto-cgraph.c (lto_output_node): update.
14727         (lto_output_varpool_node): Update.
14728         (input_overwrite_node): Update.
14729         (input_node): Update.
14730         (input_varpool_node): Update.
14731         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14732         (lto_symtab_merge_symbols): Add sanity check.
14733         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14734         * passes.c (rest_of_decl_compilation): Set static flag, too.
14735         * symtab.c (dump_symtab_base): Dump weakref.
14736         (verify_symtab_base): Sanity check weakrefs.
14737         (symtab_make_decl_local): Remove duplicated code.
14738         (symtab_alias_ultimate_target): Simplify.
14739         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14741 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14743         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14744         sequence_vect == NULL.
14746 2013-06-11  DJ Delorie  <dj@redhat.com>
14748         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14749         (rl78_unwind_word_mode): New.
14751 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14753         * final.c (debug_prefix_maps): Make static.
14755 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14757         * function.c (initial_trampoline): Remove stray copy.
14759 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14761         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14763 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14765         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14766         within bounds at the beginning of the function.
14768 2013-06-11  Alan Modra  <amodra@gmail.com>
14770         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14771         reporting.
14772         (get_named_section): Don't NULL !DECL_P decl.
14774 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14776         * doc/invoke.texi (core-avx2): Document.
14777         (slm): Likewise.
14778         (atom): Updated with MOVBE.
14780 2013-06-11  Richard Biener  <rguenther@suse.de>
14782         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14784 2013-06-11  Anton Blanchard  <anton@samba.org>
14786         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14787         correct shift value in little-endian mode.
14789 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14791         PR target/56564
14792         * varasm.c (get_variable_align): Move #endif to the right place.
14794 2013-06-10  Cary Coutant  <ccoutant@google.com>
14796         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14797         for hash so that hash table traversal order is deterministic.
14799 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14800             Pat Haugen  <pthaugen@us.ibm.com>
14801             Peter Bergner  <bergner@vnet.ibm.com>
14803         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14804         of vectors in GPRS if they are direct moves or quad word load or
14805         store moves.
14807         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14808         declaration.
14809         (direct_move_p): Likewise.
14810         (quad_load_store_p): Likewise.
14812         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14813         classes into bins based on the physical register type.
14814         (reg_class_to_reg_type): Likewise.
14815         (IS_STD_REG_TYPE): Likewise.
14816         (IS_FP_VECT_REG_TYPE): Likewise.
14817         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14818         use direct move instructions.
14819         (reload_gpr_vsx): Likewise.
14820         (reload_vsx_gpr): Likewise.
14821         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14822         information that is a simplification of register classes.  Also
14823         precalculate direct move reload helpers.
14824         (direct_move_p): New function to return true if the operation can
14825         be done as a direct move instruciton.
14826         (quad_load_store_p): New function to return true if the operation
14827         is a quad memory operation.
14828         (rs6000_legitimize_address): If quad memory, only allow register
14829         indirect for TImode addresses.
14830         (rs6000_legitimate_address_p): Likewise.
14831         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14832         (rs6000_reload_register_type): Likewise.
14833         (register_to_reg_type): Return register type.
14834         (rs6000_secondary_reload_simple_move): New helper function for
14835         secondary reload and secondary memory needed to identify anything
14836         that is a simple move, and does not need reloading.
14837         (rs6000_secondary_reload_direct_move): New helper function for
14838         secondary reload to identify cases that can be done with several
14839         instructions via the direct move instructions.
14840         (rs6000_secondary_reload_move): New helper function for secondary
14841         reload to identify moves between register types that can be done.
14842         (rs6000_secondary_reload): Add support for quad memory operations
14843         and for direct move.
14844         (rs6000_secondary_memory_needed): Likewise.
14845         (rs6000_debug_secondary_memory_needed): Change argument names.
14846         (rs6000_output_move_128bit): New function to return the move to
14847         use for 128-bit moves, including knowing about the various
14848         limitations of quad memory operations.
14850         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14851         memory operations.  call rs6000_output_move_128bit for the actual
14852         instruciton(s) to generate.
14853         (vsx_movti_64bit): Likewise.
14855         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14856         (UNSPEC_P8V_MTVSRWZ): Likewise.
14857         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14858         (UNSPEC_P8V_MTVSRD): Likewise.
14859         (UNSPEC_P8V_XXPERMDI): Likewise.
14860         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14861         (UNSPEC_FUSION_GPR): Likewise.
14862         (FMOVE128_GPR): New iterator for direct move.
14863         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14864         (f32_sv): Likewise.
14865         (f32_dm): Likewise.
14866         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14867         loads and direct move instructions.
14868         (zero_extendsidi2_lfiwzx): Likewise.
14869         (extendsidi2_lfiwax): Likewise.
14870         (extendsidi2_nocell): Likewise.
14871         (floatsi<mode>2_lfiwax): Likewise.
14872         (lfiwax): Likewise.
14873         (floatunssi<mode>2_lfiwzx): Likewise.
14874         (lfiwzx): Likewise.
14875         (fix_trunc<mode>_stfiwx): Likewise.
14876         (fixuns_trunc<mode>_stfiwx): Likewise.
14877         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14878         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14879         (parity<mode>2_cmpb): Set length/type attr.
14880         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14881         for 'mr.' to fast_compare.
14882         (bpermd_<mode>): Change type attr to popcnt.
14883         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14884         (p8_mtvsrwz_1): Likewise.
14885         (p8_mtvsrwz_2): Likewise.
14886         (reload_fpr_from_gpr<mode>): Likewise.
14887         (p8_mtvsrd_1): Likewise.
14888         (p8_mtvsrd_2): Likewise.
14889         (p8_xxpermdi_<mode>): Likewise.
14890         (reload_vsx_from_gpr<mode>): Likewise.
14891         (reload_vsx_from_gprsf): Likewise.
14892         (p8_mfvsrd_3_<mode>): LIkewise.
14893         (reload_gpr_from_vsx<mode>): Likewise.
14894         (reload_gpr_from_vsxsf): Likewise.
14895         (p8_mfvsrd_4_disf): Likewise.
14896         (multi-word GPR splits): Do not split direct moves or quad memory
14897         operations.
14899 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14901         * tree-into-ssa.c (interesting_blocks): Make static.
14903 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14905         PR target/56564
14906         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14907         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14908         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14909         (get_variable_align): New function.
14910         (get_variable_section, emit_bss, emit_common,
14911         assemble_variable_contents, place_block_symbol): Use
14912         get_variable_align instead of DECL_ALIGN.
14913         (assemble_noswitch_variable): Add align argument, use it
14914         instead of DECL_ALIGN.
14915         (assemble_variable): Adjust caller.  Use get_variable_align
14916         instead of DECL_ALIGN.
14917         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14918         caller.
14919         (DATA_ABI_ALIGNMENT): Define.
14920         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14921         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14922         opt is false, only return the psABI mandated alignment increase.
14923         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14924         (DATA_ABI_ALIGNMENT): ... this.
14925         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14926         (DATA_ABI_ALIGNMENT): ... this.
14927         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14928         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14929         (DATA_ABI_ALIGNMENT): ... this.
14930         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14931         * doc/tm.texi: Regenerated.
14933 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14935         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14936         cmp_code to construct REG_EQUAL note.
14938 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14940         PR target/57568
14941         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14942         that operands[2] doesn't overlap with operands[0].
14944 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14945             Jan Hubicka  <jh@suse.cz>
14947         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14948         hack to mark symbols as used.
14950 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14952         PR rtl-optimization/57559
14953         * lra-constraints.c (process_alt_operands): Don't discourage
14954         memory with known offset for offsetable memory constraint.
14955         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14957 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14959         * varasm.c (struct oc_local_state): Reorder fields.
14960         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14961         and adjust accordingly.
14962         (output_constructor): Reorder initialization code and adjust call to
14963         output_constructor_bitfield.
14965 2013-06-07  Jan Hubicka  <jh@suse.cz>
14967         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14969 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14971         * tree-object-size.c (unknown): Make const.
14973 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14975         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14976         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14977         for last alternative in the cpu_facility attribute.
14979 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14981         PR target/56315
14982         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14983         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14984         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14985         * config/arm/constraints.md (Dg): New constraint.
14986         * config/arm/neon.md (xordi3_neon): Remove.
14987         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14988         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14990 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14992         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14993         Clean up alternatives.
14995 2013-06-07  Alan Modra  <amodra@gmail.com>
14997         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14998         va_list_gpr_size.
15000 2013-06-07  Alan Modra  <amodra@gmail.com>
15002         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
15004 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15006         * config/arm/constraints.md (Df): New constraint.
15007         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
15008         Correct length attribute for last two alternatives.
15010 2013-06-07  Alan Modra  <amodra@gmail.com>
15012         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
15013         override user -mfp-in-toc.
15014         (offsettable_ok_by_alignment): Consider just the current access
15015         rather than the whole object, unless BLKmode.  Handle
15016         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
15017         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
15018         for -mcmodel=medium.
15019         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15020         override user -mfp-in-toc or -msum-in-toc.  Default to
15021         -mno-fp-in-toc for -mcmodel=medium.
15023 2013-06-06  DJ Delorie  <dj@redhat.com>
15025         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15026         TARGET_VALID_POINTER_MODE.
15028 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15029             Pat Haugen  <pthaugen@us.ibm.com>
15030             Peter Bergner  <bergner@vnet.ibm.com>
15032         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15033         Document new power8 builtins.
15035         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15036         condition code register, to allow 128-bit logical operations to be
15037         done in the VSX or GPR registers.
15038         (nor<mode>3): Use the canonical form for nor.
15039         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15040         vclz*, and vpopcnt* vector instructions.
15041         (nand<mode>3): Likewise.
15042         (orc<mode>3): Likewise.
15043         (clz<mode>2): LIkewise.
15044         (popcount<mode>2): Likewise.
15046         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15047         that only the GPRs are recognized.
15049         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15050         support for new power8 builtins.
15052         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15053         builtin functions.
15054         (xscvdpspn): Likewise.
15055         (vclz): Likewise.
15056         (vclzb): Likewise.
15057         (vclzh): Likewise.
15058         (vclzw): Likewise.
15059         (vclzd): Likewise.
15060         (vpopcnt): Likewise.
15061         (vpopcntb): Likewise.
15062         (vpopcnth): Likewise.
15063         (vpopcntw): Likewise.
15064         (vpopcntd): Likewise.
15065         (vgbbd): Likewise.
15066         (vmrgew): Likewise.
15067         (vmrgow): Likewise.
15068         (eqv): Likewise.
15069         (eqv_v16qi3): Likewise.
15070         (eqv_v8hi3): Likewise.
15071         (eqv_v4si3): Likewise.
15072         (eqv_v2di3): Likewise.
15073         (eqv_v4sf3): Likewise.
15074         (eqv_v2df3): Likewise.
15075         (nand): Likewise.
15076         (nand_v16qi3): Likewise.
15077         (nand_v8hi3): Likewise.
15078         (nand_v4si3): Likewise.
15079         (nand_v2di3): Likewise.
15080         (nand_v4sf3): Likewise.
15081         (nand_v2df3): Likewise.
15082         (orc): Likewise.
15083         (orc_v16qi3): Likewise.
15084         (orc_v8hi3): Likewise.
15085         (orc_v4si3): Likewise.
15086         (orc_v2di3): Likewise.
15087         (orc_v4sf3): Likewise.
15088         (orc_v2df3): Likewise.
15090         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15091         allow power8 quad mode in 64-bit.
15092         (rs6000_builtin_vectorized_function): Add support to vectorize
15093         ISA 2.07 count leading zeros, population count builtins.
15094         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15095         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15096         (builtin_function_type): Add vgbbd builtin function which takes an
15097         unsigned argument.
15098         (altivec_expand_vec_perm_const): Add support for new power8 merge
15099         instructions.
15101         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15102         that does not include TImdoe for use with 32-bit.
15103         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15104         instructions.
15105         (UNSPEC_VSX_CVDPSPN): Likewise.
15106         (vsx_xscvdpspn): Likewise.
15107         (vsx_xscvspdpn): Likewise.
15108         (vsx_xscvdpspn_scalar): Likewise.
15109         (vsx_xscvspdpn_directmove): Likewise.
15110         (vsx_and<mode>3): Split logical operations into 32-bit and
15111         64-bit. Add support to do logical operations on TImode as well as
15112         VSX vector types.  Allow logical operations to be done in either
15113         VSX registers or in general purpose registers in 64-bit mode.  Add
15114         splitters if GPRs were used. For AND, add clobber of CCmode to
15115         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15116         encoding.
15117         (vsx_and<mode>3_32bit): Likewise.
15118         (vsx_and<mode>3_64bit): Likewise.
15119         (vsx_ior<mode>3): Likewise.
15120         (vsx_ior<mode>3_32bit): Likewise.
15121         (vsx_ior<mode>3_64bit): Likewise.
15122         (vsx_xor<mode>3): Likewise.
15123         (vsx_xor<mode>3_32bit): Likewise.
15124         (vsx_xor<mode>3_64bit): Likewise.
15125         (vsx_one_cmpl<mode>2): Likewise.
15126         (vsx_one_cmpl<mode>2_32bit): Likewise.
15127         (vsx_one_cmpl<mode>2_64bit): Likewise.
15128         (vsx_nor<mode>3): Likewise.
15129         (vsx_nor<mode>3_32bit): Likewise.
15130         (vsx_nor<mode>3_64bit): Likewise.
15131         (vsx_andc<mode>3): Likewise.
15132         (vsx_andc<mode>3_32bit): Likewise.
15133         (vsx_andc<mode>3_64bit): Likewise.
15134         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15135         and xxlorc instructions.
15136         (vsx_eqv<mode>3_64bit): Likewise.
15137         (vsx_nand<mode>3_32bit): Likewise.
15138         (vsx_nand<mode>3_64bit): Likewise.
15139         (vsx_orc<mode>3_32bit): Likewise.
15140         (vsx_orc<mode>3_64bit): Likewise.
15142         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15144         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15145         instruction.
15146         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15147         (p8_vmrgow): Likewise.
15148         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15149         GPRs to be split under VSX.
15150         (p8v_clz<mode>2): Add power8 count leading zero support.
15151         (p8v_popcount<mode>2): Add power8 population count support.
15152         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15153         support.
15155         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15156         instruction.
15158         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15159         builtin functions.
15160         (vec_nand): Likewise.
15161         (vec_vclz): Likewise.
15162         (vec_vclzb): Likewise.
15163         (vec_vclzd): Likewise.
15164         (vec_vclzh): Likewise.
15165         (vec_vclzw): Likewise.
15166         (vec_vgbbd): Likewise.
15167         (vec_vmrgew): Likewise.
15168         (vec_vmrgow): Likewise.
15169         (vec_vpopcnt): Likewise.
15170         (vec_vpopcntb): Likewise.
15171         (vec_vpopcntd): Likewise.
15172         (vec_vpopcnth): Likewise.
15173         (vec_vpopcntw): Likewise.
15175 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15177         PR rtl-optimization/57468
15178         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15179         spilled pseudos.
15181 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15183         PR rtl-optimization/57459
15184         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15185         type when setting live regs.
15187 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15189         * config/s390/s390.opt (mlra): New option.
15190         * config/s390/s390.c (s390_decompose_address): Check displacement
15191         for all registers for LRA.
15192         (s390_secondary_reload): Don't used secondary reloads for LRA.
15193         (s390_lra_p): New function.
15194         (TARGET_LRA_P): Define.
15195         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15196         of attribute cpu_facility to zarch for the last alternative.
15197         (*cmpmem_short): Ditto.
15199 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15201         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15202         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15203         (arm_expand_prologue): Likewise.
15205 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15207         PR c++/53743
15208         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15209         as this is now done by redirect_edge_and_branch_force.
15210         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15211         barriers, and fix interaction with splitting.
15212         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15213         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15214         reflect changes made in the routine.
15215         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15216         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15217         since this is called in cfglayout mode, and replace partition fixup
15218         with assert as that is now done by force_nonfallthru_and_redirect.
15219         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15220         already be marked with region crossing note.
15221         (insert_section_boundary_note): Make non-static, gate on flag
15222         has_bb_partition, rewrite to also check for multiple partitions.
15223         (rest_of_handle_reorder_blocks): Remove call to
15224         insert_section_boundary_note, now done later during free_cfg.
15225         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15226         * bb-reorder.h (insert_section_boundary_note): Declare.
15227         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15228         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15229         invoke insert_section_boundary_note.
15230         (try_redirect_by_replacing_jump): Remove unnecessary
15231         check for region crossing note.
15232         (fixup_partition_crossing): New function.
15233         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15234         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15235         in non-cfglayout mode.
15236         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15237         remove old code that tried to do this. Emit barrier correctly
15238         when we are in cfglayout mode.
15239         (last_bb_in_partition): New function.
15240         (rtl_split_edge): Correctly fixup partition boundaries.
15241         (commit_one_edge_insertion): Remove old code that tried to
15242         fixup region crossing edge since this is now handled in
15243         split_block, and set up insertion point correctly since
15244         block may now end in a jump.
15245         (verify_hot_cold_block_grouping): Guard against checking when not in
15246         linearized RTL mode.
15247         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15248         notes.
15249         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15250         rtl_verify_flow_info, so not called in cfglayout mode.
15251         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15252         (fixup_reorder_chain): Remove old code that attempted to fixup region
15253         crossing note as this is now handled in force_nonfallthru_and_redirect.
15254         (duplicate_insn_chain): Don't duplicate switch section notes.
15255         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15256         note.
15257         * basic-block.h (emit_barrier_after_bb): Declare.
15259 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15261         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15262         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15263         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15265 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15267         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15268         where appropriate.
15269         * config/arm/ldmstm.md: Regenerate.
15271 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15273         * config/arm/sync.md (atomic_loaddi_1):
15274         Disable predication for arm_restrict_it.
15275         (arm_load_exclusive<mode>): Likewise.
15276         (arm_load_exclusivesi): Likewise.
15277         (arm_load_exclusivedi): Likewise.
15278         (arm_load_acquire_exclusive<mode>): Likewise.
15279         (arm_load_acquire_exclusivesi): Likewise.
15280         (arm_load_acquire_exclusivedi): Likewise.
15281         (arm_store_exclusive<mode>): Likewise.
15282         (arm_store_exclusive<mode>): Likewise.
15283         (arm_store_release_exclusivedi): Likewise.
15284         (arm_store_release_exclusive<mode>): Likewise.
15286 2013-06-06  Richard Biener  <rguenther@suse.de>
15288         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15289         after LTO_null.
15290         (lto_tag_is_tree_code_p): Adjust.
15291         (lto_tag_is_gimple_code_p): Likewise.
15292         (lto_gimple_code_to_tag): Likewise.
15293         (lto_tag_to_gimple_code): Likewise.
15294         (lto_tree_code_to_tag): Likewise.
15295         (lto_tag_to_tree_code): Likewise.
15296         * data-streamer.h (streamer_write_hwi_in_range): Use
15297         uhwi streaming to stream the normalized range.
15298         (streamer_read_hwi_in_range): Likewise.
15300 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15302         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15303         (predicable_short_it): Likewise.
15304         (predicated): Likewise.
15305         (enabled): Handle above.
15306         (define_cond_exec): Set predicated attribute to yes.
15308 2013-06-05  Mike Stump  <mikestump@comcast.net>
15310         * gdbinit.in (__FUNCTION__): Add.
15312 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15314         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15315         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15317 2013-06-05  Jan Hubicka  <jh@suse.cz>
15319         * varasm.c (mark_decl_referenced): Revert the removal until targets
15320         are fixed.
15322 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15324         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15325         instead of mark_decl_referenced.
15327 2013-06-05  Jan Hubicka  <jh@suse.cz>
15329         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15330         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15331         and symtab_used_from_object_file_p.
15332         (cgraph_make_node_local_1): Clear forced_by_abi.
15333         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15334         * cgraph.h (symtab_node_base): Add forced_by_abi.
15335         (decide_is_variable_needed): Remove.
15336         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15337         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15338         (decide_is_symbol_needed): ... this one; handle symbols in general;
15339         always analyze virtuals; honnor forced_by_abi.
15340         (cgraph_finalize_function): Update.
15341         (varpool_finalize_decl): Update.
15342         (symbol_defined_and_needed): Remove.
15343         (analyze_functions): Update.
15344         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15345         output_refs, input_overwrite_node): Handle forced_by_abi.
15346         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15347         (address_taken_from_non_vtable_p): ... this one.
15348         (comdat_can_be_unshared_p_1): New function.
15349         (cgraph_comdat_can_be_unshared_p): Rename to ...
15350         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15351         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15352         (function_and_variable_visibility): Clear forced_by_abi as needed.
15353         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15354         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15355         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15356         * varpool.c (decide_is_variable_needed): Remove.
15358 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15360         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15361         (arm_option_override): Override arm_restrict_it where appropriate.
15362         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15363         * config/arm/arm.opt (mrestrict-it): New command-line option.
15364         * doc/invoke.texi: Document -mrestrict-it.
15366 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15368         * tsan.c (tsan_atomic_table): Make const.
15370 2013-06-05  Richard Biener  <rguenther@suse.de>
15372         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15373         index associated with the tree we are supposed to replace.
15374         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15375         TREE_ASM_WRITTEN as zero for everything but SSA names.
15377 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15379         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15381 2013-06-04  Jan Hubicka  <jh@suse.cz>
15383         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15384         (input_node, input_varpool_node): Handle correctly external same
15385         body aliases.
15386         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15387         nodes at ltrans stage.
15389 2013-06-04  Jan Hubicka  <jh@suse.cz>
15391         * ipa-inline.c (update_caller_keys): Fix availability test.
15392         (update_callee_keys): Likewise.
15393         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15394         to follow ELF standard.
15396 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15398         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15399         (mips64r5900el-*-elf*): New configurations.
15400         * config/mips/mips-cpus.def (r5900): New processor.
15401         * config/mips/mips-tables.opt: Regenerate.
15402         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15403         (mips_issue_rate): Handle PROCESSOR_R5900.
15404         (mips_reorg_process_insns): Force reorder mode for the R5900.
15405         * config/mips/mips.h (TARGET_MIPS5900): Define.
15406         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15407         TARGET_MIPS5900.
15408         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15409         TARGET_MIPS5900.
15410         * config/mips/mips.md (processor): Add r5900.
15411         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15413 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15415         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15416         into function to generate MOVI instruction.
15417         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15418         (aarch64_preferred_simd_mode): Turn into wrapper.
15419         (aarch64_output_scalar_simd_mov_immediate): New function.
15420         * config/aarch64/aarch64-protos.h: Add prototype for above.
15422 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15424         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15425         element_char member.
15426         (sizetochar): Return signed char.
15427         (aarch64_simd_valid_immediate): Remove elchar and other
15428         unnecessary variables.
15429         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15430         Calculate element_char as required.
15431         * config/aarch64/aarch64-protos.h: Update and move prototype
15432         for aarch64_output_simd_mov_immediate.
15433         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15434         Update arguments.
15436 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15438         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15439         information completed by aarch64_simd_valid_immediate.
15440         (aarch64_legitimate_constant_p): Update arguments.
15441         (aarch64_simd_valid_immediate): Work with struct rather than many
15442         pointers.
15443         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15444         (aarch64_simd_make_constant): Update arguments.
15445         (aarch64_output_simd_mov_immediate): Work with struct rather than
15446         many pointers.  Output immediate directly rather than as operand.
15447         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15448         Update prototype.
15449         * config/aarch64/constraints.md (Dn): Update arguments.
15451 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15453         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15454         longer static.
15455         (aarch64_simd_immediate_valid_for_move): Remove.
15456         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15457         (aarch64_simd_make_constant): Update call.
15458         (aarch64_output_simd_mov_immediate): Update call.
15459         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15460         Add prototype.
15461         * config/aarch64/constraints.md (Dn): Update call.
15463 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15465         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15466         return type to bool for prototype.
15467         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15468         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15469         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15471 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15473         * config/mips/mips.opt (meva): New.
15474         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15475         (ASM_SPEC): Handle -meva.
15476         * doc/invoke.texi (meva):  Document.
15478 2013-06-04  Alan Modra  <amodra@gmail.com>
15480         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15481         constant output.
15483 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15485         * rtl.def: Add extra fourth optional field to define_cond_exec.
15486         * gensupport.c (process_one_cond_exec): Process attributes from
15487         define_cond_exec.
15488         * doc/md.texi: Document fourth field in define_cond_exec.
15490 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15492         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15493         out the processing order as in store_bit_field_1.
15495 2013-06-04  Jan Hubicka  <jh@suse.cz>
15497         PR middle-end/57500
15498         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15499         non-VAR_DECL node if it does not exist yet.
15501 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15503         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15504         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15505         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15506         target_cpu_default setting.
15508 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15510         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15511         default to -fopt-info=optimized instead of all.
15512         * doc/invoke.texi: Ditto.
15513         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15514         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15515         (execute_vect_slp): Emit BB vectorization success under
15516         MSG_OPTIMIZED_LOCATIONS.
15517         * tree-vect-slp.c (vect_slp_transform_bb): Change
15518         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15519         * tree-vect-loop.c (vect_transform_loop): Ditto.
15521 2013-06-03  Jason Merrill  <jason@redhat.com>
15523         PR c++/57415
15524         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15525         Use TARGET_EXPR for C++.
15527 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15529         PR rtl-optimization/57268
15530         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15531         if DEBUG_INSN_P (insn).
15533         Reapply
15534         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15536         PR rtl-optimization/57268
15537         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15538         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15540 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15542         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15543         (ix86_avoid_lea_for_addr): Likewise.
15544         (exact_dependency_1): Likewise.
15545         (ix86_adjust_cost): Likewise.
15546         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15547         removed.
15548         (do_reorder_for_imul): Fix typo, formatting and
15549         !reload_completed check removed.
15550         (ix86_sched_reorder): Fix typo and formatting.
15551         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15552         list.
15554 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15556         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15558 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15560         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15561         <STRING_CST>: Likewise.
15562         <VECTOR_CST>: Likewise.
15564 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15565             Mikael Morin  <mikael@gcc.gnu.org>
15567         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15568         * config.in: Regenerated.
15569         * configure: Regenerated.
15571 2013-06-01  Jan Hubicka  <jh@suse.cz>
15573         PR middle-end/57366
15574         * cgraphunit.c (compile): When weakref is not supported,
15575         set up transparent aliases before final output pass.
15576         * varasm.c (assemble_alias): Do not try to do it here.
15578 2013-06-01  Jan Hubicka  <jh@suse.cz>
15580         PR middle-end/57467
15581         * passes.c (for_per_function): Skip unanalyzed functions.
15583 2013-06-01  Jan Hubicka  <jh@suse.cz>
15585         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15586         (lto_symtab_merge_symbols_1): ... this one.
15587         (lto_symtab_merge_cgraph_nodes): Rename to ...
15588         (lto_symtab_merge_symbols): ... this one; simplify.
15589         * cgraph.c (same_body_aliases_done): Rename to ...
15590         (cpp_implicit_aliases_done): ... this one.
15591         (cgraph_create_function_alias): Update.
15592         (cgraph_same_body_alias): Update.
15593         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15594         (verify_edge_corresponds_to_fndecl): Simplify.
15595         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15596         (cgraph_node): Remove same_body_alias.
15597         (varpool_node): Remove alias_of and extra_name_alias.
15598         (same_body_aliases_done): Rename to ..
15599         (cpp_implicit_aliases_done): ... this one.
15600         (symtab_alias_ultimate_target): Add default parameter.
15601         (symtab_resolve_alias): New function.
15602         (fixup_same_cpp_alias_visibility): Declare.
15603         (cgraph_function_node): Add default parameter.
15604         (cgraph_node_asm_name): Likewise.
15605         (cgraph_function_or_thunk_node): Add default parameter; do
15606         not ICE when it is NULL.
15607         (varpool_variable_node): Likewise.
15608         * tree-emutls.c (create_emultls_var): Update.
15609         (ipa_lower_emutls): Update.
15610         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15611         (cgraph_reset_node): Reset alias info.
15612         (cgraph_finalize_function): Update.
15613         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15614         (analyze_function): Simplify.
15615         (cgraph_process_same_body_aliases): Simplify.
15616         (analyze_functions): Fixup same body aliases.
15617         (handle_alias_pairs): Simplify.
15618         (assemble_thunk): Update.
15619         (assemble_thunks_and_aliases): Update.
15620         (output_weakrefs): Rewrite.
15621         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15622         (lto_output_varpool_node): Likewise.
15623         (compute_ltrans_boundary): Remve assert.
15624         (get_alias_symbol): New functoin.
15625         (input_node): Rewrite alias handling.
15626         (input_varpool_node): Likewise.
15627         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15628         * ipa.c (process_references): Handle weakrefs correctly.
15629         (symtab_remove_unreachable_nodes): Likewise.
15630         * trans-mem.c (get_cg_data): Update.
15631         (ipa_tm_create_version_alias): Update.
15632         (ipa_tm_execute): Update.
15633         * symtab.c (dump_symtab_base): Dump aliases.
15634         (verify_symtab_base): Verify aliases.
15635         (symtab_node_availability): New function.
15636         (symtab_alias_ultimate_target): Simplify.
15637         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15638         handle all the fixup cases.
15639         (symtab_resolve_alias): New function.
15640         * passes.c (ipa_write_summaries): Handle weakrefs.
15641         * varpool.c (varpool_analyze_node): Simplify.
15642         (assemble_aliases): Update.
15643         (varpool_create_variable_alias): Simplify.
15644         (varpool_extra_name_alias): Simplify.
15645         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15646         (lto_symtab_merge_symbols): ... this one.
15648 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15650         Revert
15651         PR rtl-optimization/57268
15652         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15653         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15655 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15657         Partially reverted:
15658         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15660         PR middle-end/57073
15661         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15662         further up.
15664 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15666         PR rtl-optimization/57268
15667         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15668         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15670 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15672         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15673         unordered comparison operators when -fno-trapping-math is in effect
15674         on the e500.
15675         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15676         and implement unordered comparison operators properly on the e500.
15678 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15680         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15681         for constant scalar integers.
15682         (simplify_relational_operation_1): Likewise.
15684 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15686         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15687         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15688         Fix comment.
15690 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15691             Igor Zamyatin  <igor.zamyatin@intel.com>
15693         Silvermont (SLM) architecture performance tuning.
15694         * config/i386/i386.h (enum ix86_tune_indices): Add
15695         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15696         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15698         * config/i386/i386.c (initial_ix86_tune_features)
15699         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15700         (ix86_lea_outperforms): Handle Silvermont tuning.
15701         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15702         call.
15703         (ix86_use_lea_for_mov): Likewise.
15704         (ix86_avoid_lea_for_addr): Likewise.
15705         (ix86_lea_for_add_ok): Likewise.
15706         (exact_dependency_1): New function.
15707         (exact_store_load_dependency): Likewise.
15708         (ix86_adjust_cost): Handle Silvermont tuning.
15709         (do_reoder_for_imul): Likewise.
15710         (swap_top_of_ready_list): New function.
15711         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15713         * config/i386/i386.md (peepholes that split memory operand in fp
15714         converts): New.
15716 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15718         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15719         Remove un-necessary braces.
15721 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15723         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15724         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15726 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15728         PR middle-end/57073
15729         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15731 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15733         PR target/56315
15734         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15735         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15736         * config/arm/neon.md (iordi3_neon): Remove.
15737         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15738         * config/arm/predicates.md (imm_for_neon_logic_operand):
15739         Move to earlier in the file.
15740         (neon_logic_op2): Likewise.
15741         (arm_iordi_operand_neon): New predicate.
15743 2013-05-31  Richard Biener  <rguenther@suse.de>
15745         PR tree-optimization/57478
15746         PR tree-optimization/57453
15747         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15748         are life as well.
15750 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15752         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15753         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15755 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15756             Thomas Koenig  <tkoenig@gcc.gnu.org>
15758         PR middle-end/57073
15759         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15760         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15762 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15764         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15766 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15768         * target.def (register_usage_leveling_p): New hook.
15769         * targhooks.c (default_register_usage_leveling_p): New.
15770         * targhooks.h (default_register_usage_leveling_p): New prototype.
15771         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15772         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15773         * doc/tm.texi: Update.
15774         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15776 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15778         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15779         (*insv_reg<mode>): New define_insn.
15781 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15783         PR rtl-optimization/57439
15784         * postreload.c (move2add_valid_value_p): Check that we have
15785         a zero subreg_regno_offset when accessing the register in
15786         the requested mode.
15788 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15789             Igor Zamyatin  <igor.zamyatin@intel.com>
15791         Silvermont (SLM) architecture pipeline model, tuning and
15792         insn selection.
15793         * config.gcc: Add slm config options and target.
15795         * config/i386/slm.md: New.
15797         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15799         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15800         PROCESSOR_SLM.
15801         (ix86_target_macros_internal): Likewise.
15803         * config/i386/i386.c (slm_cost): New cost.
15804         (m_SLM): New macro flag.
15805         (initial_ix86_tune_features): Set m_SLM.
15806         (x86_accumulate_outgoing_args): Likewise.
15807         (x86_arch_always_fancy_math_387): Likewise.
15808         (processor_target_table): Add slm cost.
15809         (cpu_names): Add slm cpu name.
15810         (x86_option_override_internal): Set SLM ISA.
15811         (ix86_issue_rate): New case PROCESSOR_SLM.
15812         (ia32_multipass_dfa_lookahead): Likewise.
15813         (fold_builtin_cpu): Add slm.
15815         * config/i386/i386.h (TARGET_SLM): New target macro.
15816         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15817         (processor_type): Add PROCESSOR_SLM.
15819         * config/i386/i386.md (cpu): Add new value "slm".
15820         (slm.md): Include slm.md.
15822 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15823             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15825         * config/arm/arm-protos.h: Add and update function protos.
15826         * config/arm/arm.c (use_simple_return_p): New added.
15827         (thumb2_expand_return): Check simple_return flag.
15828         * config/arm/arm.md: Add simple_return and conditional simple_return.
15829         * config/arm/iterators.md: Add iterator for return and simple_return.
15831 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15833         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15834         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15835         (arm_emit_vfp_multi_reg_pop): Likewise.
15836         (thumb2_emit_ldrd_pop): Likewise.
15837         (arm_expand_epilogue): Add misc REG_CFA notes.
15838         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15840 2013-05-29  Lawrence Crowl  <crowl@google.com>
15842         * config/arm/t-arm: Update for below.
15844         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15845         Change type to hash_table.  Update dependent calls and types.
15847         * config/i386/t-cygming: Update for below.
15849         * config/i386/t-interix: Update for below.
15851         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15852         Change type to hash_table.  Update dependent calls and types.
15853         (i386_find_on_wrapper_list::wrappers): Likewise.
15855         * config/ia64/t-ia64: Update for below.
15857         * config/ia64/ia64.c (bundle_state_table):
15858         Change type to hash_table.  Update dependent calls and types.
15860         * config/mips/mips.c (mips_reorg_process_insns::htab):
15861         Change type to hash_table.  Update dependent calls and types.
15863         * config/sol2.c (solaris_comdat_htab):
15864         Change type to hash_table.  Update dependent calls and types.
15866         * config/t-sol2: Update for above.
15868 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15870         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15871         functions are not yet marked as defined.
15873 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15874             Pat Haugen  <pthaugen@us.ibm.com>
15875             Peter Bergner  <bergner@vnet.ibm.com>
15877         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15878         instructions.
15879         (VEC_A): Likewise.
15880         (VEC_C): Likewise.
15881         (vrotl<mode>3): Likewise.
15882         (vashl<mode>3): Likewise.
15883         (vlshr<mode>3): Likewise.
15884         (vashr<mode>3): Likewise.
15886         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15887         support for power8 V2DI builtins.
15889         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15890         power8 V2DI builtins.
15891         (vupkhsw): Likewise.
15892         (vupklsw): Likewise.
15893         (vaddudm): Likewise.
15894         (vminsd): Likewise.
15895         (vmaxsd): Likewise.
15896         (vminud): Likewise.
15897         (vmaxud): Likewise.
15898         (vpkudum): Likewise.
15899         (vpksdss): Likewise.
15900         (vpkudus): Likewise.
15901         (vpksdus): Likewise.
15902         (vrld): Likewise.
15903         (vsld): Likewise.
15904         (vsrd): Likewise.
15905         (vsrad): Likewise.
15906         (vsubudm): Likewise.
15907         (vcmpequd): Likewise.
15908         (vcmpgtsd): Likewise.
15909         (vcmpgtud): Likewise.
15910         (vcmpequd_p): Likewise.
15911         (vcmpgtsd_p): Likewise.
15912         (vcmpgtud_p): Likewise.
15913         (vupkhsw): Likewise.
15914         (vupklsw): Likewise.
15915         (vaddudm): Likewise.
15916         (vmaxsd): Likewise.
15917         (vmaxud): Likewise.
15918         (vminsd): Likewise.
15919         (vminud): Likewise.
15920         (vpksdss): Likewise.
15921         (vpksdus): Likewise.
15922         (vpkudum): Likewise.
15923         (vpkudus): Likewise.
15924         (vrld): Likewise.
15925         (vsld): Likewise.
15926         (vsrad): Likewise.
15927         (vsrd): Likewise.
15928         (vsubudm): Likewise.
15930         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15931         support for power8 V2DI instructions.
15933         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15934         power8 V2DI instructions.  Combine pack and unpack insns to use an
15935         iterator for each mode.  Check whether a particular mode supports
15936         Altivec instructions instead of just checking TARGET_ALTIVEC.
15937         (UNSPEC_VPKUWUM): Likewise.
15938         (UNSPEC_VPKSHSS): Likewise.
15939         (UNSPEC_VPKSWSS): Likewise.
15940         (UNSPEC_VPKUHUS): Likewise.
15941         (UNSPEC_VPKSHUS): Likewise.
15942         (UNSPEC_VPKUWUS): Likewise.
15943         (UNSPEC_VPKSWUS): Likewise.
15944         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15945         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15946         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15947         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15948         (UNSPEC_VUPKHSB): Likewise.
15949         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15950         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15951         (UNSPEC_VUPKHSH): Likewise.
15952         (UNSPEC_VUPKLSB): Likewise.
15953         (UNSPEC_VUPKLSH): Likewise.
15954         (VI2): Likewise.
15955         (VI_char): Likewise.
15956         (VI_scalar): Likewise.
15957         (VI_unit): Likewise.
15958         (VP): Likewise.
15959         (VP_small): Likewise.
15960         (VP_small_lc): Likewise.
15961         (VU_char): Likewise.
15962         (add<mode>3): Likewise.
15963         (altivec_vaddcuw): Likewise.
15964         (altivec_vaddu<VI_char>s): Likewise.
15965         (altivec_vadds<VI_char>s): Likewise.
15966         (sub<mode>3): Likewise.
15967         (altivec_vsubcuw): Likewise.
15968         (altivec_vsubu<VI_char>s): Likewise.
15969         (altivec_vsubs<VI_char>s): Likewise.
15970         (altivec_vavgs<VI_char>): Likewise.
15971         (altivec_vcmpbfp): Likewise.
15972         (altivec_eq<mode>): Likewise.
15973         (altivec_gt<mode>): Likewise.
15974         (altivec_gtu<mode>): Likewise.
15975         (umax<mode>3): Likewise.
15976         (smax<mode>3): Likewise.
15977         (umin<mode>3): Likewise.
15978         (smin<mode>3): Likewise.
15979         (altivec_vpkuhum): Likewise.
15980         (altivec_vpkuwum): Likewise.
15981         (altivec_vpkshss): Likewise.
15982         (altivec_vpkswss): Likewise.
15983         (altivec_vpkuhus): Likewise.
15984         (altivec_vpkshus): Likewise.
15985         (altivec_vpkuwus): Likewise.
15986         (altivec_vpkswus): Likewise.
15987         (altivec_vpks<VI_char>ss): Likewise.
15988         (altivec_vpks<VI_char>us): Likewise.
15989         (altivec_vpku<VI_char>us): Likewise.
15990         (altivec_vpku<VI_char>um): Likewise.
15991         (altivec_vrl<VI_char>): Likewise.
15992         (altivec_vsl<VI_char>): Likewise.
15993         (altivec_vsr<VI_char>): Likewise.
15994         (altivec_vsra<VI_char>): Likewise.
15995         (altivec_vsldoi_<mode>): Likewise.
15996         (altivec_vupkhsb): Likewise.
15997         (altivec_vupkhs<VU_char>): Likewise.
15998         (altivec_vupkls<VU_char>): Likewise.
15999         (altivec_vupkhsh): Likewise.
16000         (altivec_vupklsb): Likewise.
16001         (altivec_vupklsh): Likewise.
16002         (altivec_vcmpequ<VI_char>_p): Likewise.
16003         (altivec_vcmpgts<VI_char>_p): Likewise.
16004         (altivec_vcmpgtu<VI_char>_p): Likewise.
16005         (abs<mode>2): Likewise.
16006         (vec_unpacks_hi_v16qi): Likewise.
16007         (vec_unpacks_hi_v8hi): Likewise.
16008         (vec_unpacks_lo_v16qi): Likewise.
16009         (vec_unpacks_hi_<VP_small_lc>): Likewise.
16010         (vec_unpacks_lo_v8hi): Likewise.
16011         (vec_unpacks_lo_<VP_small_lc>): Likewise.
16012         (vec_pack_trunc_v8h): Likewise.
16013         (vec_pack_trunc_v4si): Likewise.
16014         (vec_pack_trunc_<mode>): Likewise.
16016         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
16017         V2DI builtins.
16018         (vec_vmaxsd): Likewise.
16019         (vec_vmaxud): Likewise.
16020         (vec_vminsd): Likewise.
16021         (vec_vminud): Likewise.
16022         (vec_vpksdss): Likewise.
16023         (vec_vpksdus): Likewise.
16024         (vec_vpkudum): Likewise.
16025         (vec_vpkudus): Likewise.
16026         (vec_vrld): Likewise.
16027         (vec_vsld): Likewise.
16028         (vec_vsrad): Likewise.
16029         (vec_vsrd): Likewise.
16030         (vec_vsubudm): Likewise.
16031         (vec_vupkhsw): Likewise.
16032         (vec_vupklsw): Likewise.
16034 2013-05-29  Jan Hubicka  <jh@suse.cz>
16036         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16037         flags; reorder rest of fields in more consistent way.
16038         (varpool_node): Remove analyzed, finalized and alias.
16039         (cgraph_ndoe): Likewise.
16040         (symtab_alias_ultimate_target): New function.
16041         (cgraph_function_node): Move offline.
16042         (cgraph_reset_node): Declare.
16043         (cgraph_comdat_can_be_unshared_p): Remove.
16044         (varpool_remove_initializer): Declare.
16045         (varpool_first_defined_variable, varpool_next_defined_variable
16046         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16047         (cgraph_function_with_gimple_body_p): Update.
16048         (varpool_all_refs_explicit_p): Update.
16049         (symtab_alias_target): New function.
16050         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16051         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16052         (cgraph_function_or_thunk_node): Simplify using
16053         symtab_alias_ultimate_target.
16054         (varpool_variable_node): Likewise.
16055         * cgraph.c (cgraph_create_function_alias): Update.
16056         (cgraph_add_thunk): Update.
16057         (cgraph_remove_node): Update.
16058         (dump_cgraph_node): Do not dump removed flags.
16059         (cgraph_function_body_availability): Update.
16060         (cgraph_propagate_frequency): Update.
16061         (verify_cgraph_node): Check sanity of local flag.
16062         (cgraph_function_node): Move here from cgraph.h; revamp for
16063         cgraph_function_or_thunk_node.
16064         * lto-symtab.c (lto_varpool_replace_node): Update.
16065         (lto_symtab_resolve_can_prevail_p): Update.
16066         (lto_symtab_merge_cgraph_nodes): Update.
16067         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16068         propagate_constants_accross_call, devirtualization_time_bonus,
16069         ipcp_propagate_stage): Update.
16070         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16071         * ipa-inline-transform.c (clone_inlined_nodes,
16072         preserve_function_body_p): Update.
16073         * ipa-reference.c (propagate): Update.
16074         (write_node_summary_p): Update.
16075         * toplev.c (wrapup_global_declaration_2): Update.
16076         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16077         (analyze_function) ... this one.
16078         (cgraph_process_new_functions): Update.
16079         (cgraph_reset_node): Export.
16080         (cgraph_finalize_function): Update.
16081         (cgraph_add_new_function): Update.
16082         (process_function_and_variable_attributes): Update.
16083         (varpool_finalize_decl): Update.
16084         (symbol_finalized): Remove.
16085         (symbol_finalized_and_needed): Rename to ...
16086         (symbol_defined_and_needed): ... update.
16087         (cgraph_analyze_functions): Update.
16088         (handle_alias_pairs): Update.
16089         (mark_functions_to_output): Update.
16090         (assemble_thunk): Update.
16091         (output_in_order): Update.
16092         (output_weakrefs): Update.
16093         (finalize_compilation_unit): Update.
16094         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16095         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16096         input_node, input_varpool_node): Update.
16097         * dbxout.c (dbxout_expand_expr): Update.
16098         * cgraphclones.c (cgraph_clone_node): Update.
16099         (cgraph_copy_node_for_versioning): Update.
16100         (cgraph_materialize_clone): Update.
16101         (cgraph_materialize_all_clones): Update.
16102         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16103         propagate_pure_const, propagate_nothrow): Update.
16104         * lto-streamer-out.c (lto_output, write_symbol): Update.
16105         * ipa-utils.c (ipa_reverse_postorder): Update.
16106         * ipa-inline.c (can_inline_edge_p): Update.
16107         (update_caller_keys, ipa_inline): Update.
16108         * dwarf2out.c (reference_to_unused,
16109         premark_types_used_by_global_vars_helper): Update.
16110         * tree-eh.c (tree_could_trap_p): Update.
16111         * ipa-split.c (consider_split, execute_split_functions): Update.
16112         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16113         has_addr_references_p): Update; move ahead in file for better
16114         readability.
16115         (process_references): Simplify.
16116         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16117         bodies are removed.
16118         (cgraph_comdat_can_be_unshared_p): Make static.
16119         (cgraph_externally_visible_p): Update.
16120         (varpool_externally_visible_p): Update.
16121         (function_and_variable_visibility): Update.
16122         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16123         ipa_tm_mark_force_output_node): Update.
16124         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16125         estimate_edge_devirt_benefit, inline_generate_summary,
16126         inline_write_summary): Update.
16127         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16128         * ipa-prop.c (ipa_compute_jump_functions): Update.
16129         (ipa_print_node_params, ipa_prop_read_section,
16130         ipa_update_after_lto_read, read_replacements_section): Update.
16131         * varasm.c (mark_decl_referenced): Update.
16132         (assemble_alias, dump_tm_clone_pairs): Update.
16133         * tree-inline.c (copy_bb): Update.
16134         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16135         Update.
16136         * symtab.c (dump_symtab_base): Print new flags.
16137         (verify_symtab_base): Verify new flags.
16138         (symtab_alias_ultimate_target): New function.
16139         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16140         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16141         Update.
16142         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16143         Update.
16144         * i386.c (ix86_get_function_versions_dispatcher,
16145         ix86_generate_version_dispatcher_body): Update.
16146         (fold_builtin_cpu): Use varpool_add_new_variable.
16147         * varpool.c (varpool_remove_initializer): Break out from ...
16148         (varpool_remove_node): ... this one.
16149         (dump_varpool_node, varpool_node_for_asm,
16150         cgraph_variable_initializer_availability, varpool_analyze_node,
16151         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16152         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16153         Update.
16155 2013-05-29  Jan Hubicka  <jh@suse.cz>
16157         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16159 2013-05-29  Easwaran Raman  <eraman@google.com>
16161         PR tree-optimization/57442
16162         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16163         when control exits the main loop.
16165 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16167         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16168         and RX600.
16169         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16170         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16171         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16172         option.
16174 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16176         PR tree-optimization/57441
16177         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16178         Don't limit size of incr_vec to number of candidates.
16180 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16182         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16183         and mips16 directories.
16184         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16185         (MULTILIB_DIRNAMES): Ditto.
16186         (MULTILIB_EXCEPTIONS): Add new exceptions.
16187         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16188         (MULTILIB_DIRNAMES): Ditto.
16189         (MULTILIB_EXCEPTIONS): Add new exceptions.
16191 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16192             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16194         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16195         SYMBOL_TINY_ABSOLUTE.
16196         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16197         SYMBOL_TINY_ABSOLUTE.
16198         (aarch64_expand_mov_immediate): Likewise.
16199         (aarch64_classify_symbol): Likewise.
16200         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16201         Permit SYMBOL_TINY_ABSOLUTE.
16202         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16204 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16205             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16207         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16208         Refactor if/switch.  Replace gcc_assert with if.
16210 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16212         * config/i386/i386.c (initial_ix86_tune_features): Enable
16213         FP Reassociation for AMD bdver1 and bdver2.
16215 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16217         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16218         and IMAGPART_EXPR do not occur within other handled_components.
16220 2013-05-29  Richard Biener  <rguenther@suse.de>
16222         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16223         access on whether the use is in the BB we currently try to
16224         vectorize.
16225         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16226         vectorize to vect_bb_slp_scalar_cost.
16228 2013-05-29  Richard Biener  <rguenther@suse.de>
16230         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16231         computing scalar cost offsetted by stmts that are kept live
16232         by scalar uses.
16233         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16234         for computation of scalar cost.
16236 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16238         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16240 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16242         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16243         array notation built-in reduction functions.
16244         * doc/passes.texi (Passes): Added documentation about changes done
16245         for Cilk Plus.
16246         * doc/invoke.texi (C Dialect Options): Added documentation about
16247         the -fcilkplus flag.
16248         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16249         (BUILTINS_DEF): Depend on cilkplus.def.
16250         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16251         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16252         * cilkplus.def: New file.
16254 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16256         PR rtl-optimization/57439
16257         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16259 2013-05-28  Easwaran Raman  <eraman@google.com>
16261         PR tree-optimization/57337
16262         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16263         (find_insert_point): Correctly identify the insertion point
16264         when two statements with the same UID is compared.
16266 2013-05-28  Richard Biener  <rguenther@suse.de>
16268         PR tree-optimization/56787
16269         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16270         from the list of data references.
16271         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16272         clobbers.
16273         (vect_analyze_loop_operations): Likewise.
16274         (vect_transform_loop): Remove clobbers.
16276 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16278         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16279         and REALPART_EXPRs have scalar type.
16281 2013-05-28  Richard Biener  <rguenther@suse.de>
16283         PR tree-optimization/57411
16284         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16285         virtual operands.
16286         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16287         virtual operand propagation.
16289 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16291         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16292         destination register for bmasksi_vis.
16293         (vector_init_bshuffle): Likewise.
16294         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16296 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16298         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16299         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16300         mode if the instruction isn't available in the original mode.
16301         * config/sparc/sparc.opt (mfix-ut699): New option.
16302         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16303         (divdf3): Turn into expander.
16304         (divdf3_nofix): New insn.
16305         (divdf3_fix): Likewise.
16306         (divsf3): Disable if -mfix-ut699.
16307         (sqrtdf2): Turn into expander.
16308         (sqrtdf2_nofix): New insn.
16309         (sqrtdf2_fix): Likewise.
16310         (sqrtsf2): Disable if -mfix-ut699.
16312 2013-05-27  Richard Biener  <rguenther@suse.de>
16314         PR middle-end/57412
16315         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16316         block for the new loop.
16318 2013-05-27  Richard Biener  <rguenther@suse.de>
16320         PR tree-optimization/57343
16321         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16322         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16323         (number_of_iterations_cond): Do not build the folded tree.
16325 2013-05-27  Richard Biener  <rguenther@suse.de>
16327         Revert
16328         PR middle-end/57381
16329         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16330         OEP_CONSTANT_ADDRESS_OF retained.
16332         PR tree-optimization/57417
16333         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16334         for unchanged base.
16335         (set_ssa_val_to): Compare addresses using
16336         get_addr_base_and_unit_offset.
16338 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16340         PR rtl-optimization/56833
16341         * postreload.c (move2add_record_mode): New function.
16342         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16343         (move2add_use_add2_insn): Use move2add_record_sym_value.
16344         (move2add_use_add3_insn): Likewise.
16345         (reload_cse_move2add): Use move2add_valid_value_p and
16346         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16347         affected regs by setting reg_mode to VOIDmode.
16348         (move2add_note_store): Don't pretend the inside of a SUBREG is
16349         the actual destination.  Invalidate single/leading registers by
16350         setting reg_mode to VOIDmode.
16351         Use move2add_record_sym_value, move2add_valid_value_p and
16352         move2add_record_mode.
16354 2013-05-27  Richard Biener  <rguenther@suse.de>
16356         PR tree-optimization/57396
16357         * tree-affine.c (double_int_constant_multiple_p): Properly
16358         return false for val == 0 and div != 0.
16360 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16362         * config/mips/mips.h: Use #elif in preprocessor conditions.
16364 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16366         PR target/53916
16367         * config/mips/constraints.md (kl): New constraint.
16368         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16369         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16370         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16371         from LO for MIPS16.
16372         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16374 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16376         PR target/55777
16377         * config/mips/mips.c (mips_can_inline_p): New function.
16378         (TARGET_CAN_INLINE_P): Define.
16380 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16382         * sched-int.h (ds_t, dw_t): Make unsigned int.
16383         Fix documentation that describes how all the ds_t bits are used.
16384         Reserve the last bit for delayed-branch scheduling.
16385         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16386         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16387         (gen_dep_weak_1): Remove prototype.
16388         * sched-deps.c (get_dep_weak_1): Make static.
16389         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16390         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16391         * doc/tm.texi: Regenerate.
16392         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16394 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16396         PR debug/56950
16397         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16399 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16400             Sandra Loosemore  <sandra@codesourcery.com>
16402         * config.gcc (powerpc-*): Allow native for with-cpu.
16404 2013-05-24  Jeff Law  <law@redhat.com>
16406         PR tree-optimization/57124
16407         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16408         conversion feeding a condition if the range has an overflow
16409         if -fstrict-overflow.  Add warnings for when we do make the
16410         transformation.
16412 2013-05-24  Dehao Chen  <dehao@google.com>
16414         * tree-cfg.c (locus_discrim_map): Fix the typo.
16415         (locus_discrim_hasher): Likewise.
16416         (locus_discrim_hasher::hash): Likewise.
16417         (locus_discrim_hasher::equal): Likewise.
16419 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16421         PR tree-optimization/57294
16422         * cgraph.h (ipa_record_stmt_references): Declare.
16423         * cgraphbuild.c (ipa_record_stmt_references): New function.
16424         (build_cgraph_edges): Use ipa_record_stmt_references.
16425         (rebuild_cgraph_edges): Likewise.
16426         (cgraph_rebuild_references): Likewise.
16427         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16428         associated with the old statement and build references from the
16429         newly built statements.
16430         * ipa-ref.c (ipa_remove_stmt_references): New function.
16431         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16433 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16435         * lra-constraints.c (emit_spill_move): Use smaller mode for
16436         mem-mem moves.
16437         (check_and_process_move): Consider mem-reg moves for secondary
16438         too.
16439         (curr_insn_transform): Don't lose insns emitted before for
16440         secondary memory moves.
16441         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16442         reg set up in the current insn.
16444 2013-05-24  Dehao Chen  <dehao@google.com>
16446         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16447         hash function.
16448         (locus_descrim_hasher::equal): Likewise.
16449         (build_gimple_cfg): New discriminator assignment algorithm.
16450         (make_edges): Likewise.
16451         (next_discriminator_for_locus): Likewise.
16452         (same_line_p): Likewise.
16453         (assign_discriminators): Likewise.
16454         (make_cond_expr_edges): Likewise.
16455         (make_gimple_switch_edges): Likewise.
16456         (make_goto_expr_edges): Likewise.
16457         (make_gimple_asm_edges): Likewise.
16459 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16461         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16462         X format specifier to only display bottom 16 bits.
16463         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16464         immediate to match for operand 2, since it will be masked.
16466 2013-05-24  Richard Biener  <rguenther@suse.de>
16468         PR tree-optimization/57287
16469         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16470         all SSA names that occur in abnormal PHIs.
16472 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16474         PR tree-ssa/57385
16475         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16476         that index is not negative.
16478 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16480         PR rtl-optimization/55177
16481         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16482         (simplify_byte_swapping_operation): New.
16483         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16484         (simplify_relational_operation_1): Deal with BSWAP.
16486 2013-05-23  Richard Henderson  <rth@redhat.com>
16488         PR target/56742
16489         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16490         (ix86_reorg): Call it.
16492 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16494         PR target/57379
16495         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16496         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16497         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16499 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16501         PR debug/57351
16502         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16504 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16505             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16507         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16508         * config/aarch64/constraints.md (Usa): Remove.
16509         * doc/md.texi (AArch64 Usa): Remove.
16511 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16512             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16514         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16515         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16516         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16517         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16519 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16521         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16522         instruction (AdvSIMD).
16523         * config/aarch64/aarch64-builtins.c
16524         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16525         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16527 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16529         PR middle-end/57347
16530         * tree.h (contains_bitfld_component_ref_p): Declare.
16531         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16532         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16533         caller.
16534         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16535         not access a bit-field.  Assert all final offsets are byte-aligned.
16537 2013-05-23  Richard Biener  <rguenther@suse.de>
16539         PR tree-optimization/57380
16540         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16541         least one invariant or re-used load.
16542         * passes.c (init_optimization_passes): Move pass_phiprop before
16543         pass_forwprop.
16545 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16547         * config/aarch64/aarch64-simd.md
16548         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16550 2013-05-23  Richard Biener  <rguenther@suse.de>
16552         PR middle-end/57381
16553         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16554         OEP_CONSTANT_ADDRESS_OF retained.
16556 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16558         PR middle-end/57344
16559         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16560         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16562 2013-05-23  Richard Biener  <rguenther@suse.de>
16564         PR rtl-optimization/57341
16565         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16566         instead of true_dependence.
16568 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16570         * bb-reorder.c (branch_threshold): Make const.
16571         (exec_threshold): Ditto.
16573 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16574             Pat Haugen  <pthaugen@us.ibm.com>
16575             Peter Bergner  <bergner@vnet.ibm.com>
16577         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16578         documentation for the power8 crypto builtins.
16580         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16582         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16583         macros for defining power8 builtin functions.
16584         (BU_P8V_AV_2): Likewise.
16585         (BU_P8V_AV_P): Likewise.
16586         (BU_P8V_VSX_1): Likewise.
16587         (BU_P8V_OVERLOAD_1): Likewise.
16588         (BU_P8V_OVERLOAD_2): Likewise.
16589         (BU_CRYPTO_1): Likewise.
16590         (BU_CRYPTO_2): Likewise.
16591         (BU_CRYPTO_3): Likewise.
16592         (BU_CRYPTO_OVERLOAD_1): Likewise.
16593         (BU_CRYPTO_OVERLOAD_2): Likewise.
16594         (XSCVSPDP): Fix typo, point to the correct instruction.
16595         (VCIPHER): Add power8 crypto builtins.
16596         (VCIPHERLAST): Likewise.
16597         (VNCIPHER): Likewise.
16598         (VNCIPHERLAST): Likewise.
16599         (VPMSUMB): Likewise.
16600         (VPMSUMH): Likewise.
16601         (VPMSUMW): Likewise.
16602         (VPERMXOR_V2DI): Likewise.
16603         (VPERMXOR_V4SI: Likewise.
16604         (VPERMXOR_V8HI: Likewise.
16605         (VPERMXOR_V16QI: Likewise.
16606         (VSHASIGMAW): Likewise.
16607         (VSHASIGMAD): Likewise.
16608         (VPMSUM): Likewise.
16609         (VPERMXOR): Likewise.
16610         (VSHASIGMA): Likewise.
16612         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16613         __CRYPTO__ if the crypto instructions are available.
16614         (altivec_overloaded_builtins): Add support for overloaded power8
16615         builtins.
16617         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16618         support for power8 crypto builtins.
16619         (builtin_function_type): Likewise.
16620         (altivec_init_builtins): Add support for builtins that take vector
16621         long long (V2DI) arguments.
16623         * config/rs6000/crypto.md: New file, define power8 crypto
16624         instructions.
16626 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16627             Pat Haugen  <pthaugen@us.ibm.com>
16628             Peter Bergner  <bergner@vnet.ibm.com>
16630         * doc/invoke.texi (Option Summary): Add power8 options.
16631         (RS/6000 and PowerPC Options): Likewise.
16633         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16634         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16635         wm, wn, wr documentation.
16637         * config/rs6000/constraints.md (wm): New constraint for VSX
16638         registers if direct move instructions are enabled.
16639         (wn): New constraint for no registers.
16640         (wq): New constraint for quad word even GPR registers.
16641         (wr): New constraint if 64-bit instructions are enabled.
16642         (wv): New constraint if power8 vector instructions are enabled.
16643         (wQ): New constraint for quad word memory locations.
16645         * config/rs6000/predicates.md (const_0_to_15_operand): New
16646         constraint for 0..15 for crypto instructions.
16647         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16648         as GPR and floating point registers.
16649         (int_reg_operand): New predicate to match only GPR registers.
16650         (base_reg_operand): New predicate to match base registers.
16651         (quad_int_reg_operand): New predicate to match even GPR registers
16652         for quad memory operations.
16653         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16654         operations in both GPR and VSX registers.
16655         (quad_memory_operand): New predicate for quad memory operations.
16656         (reg_or_indexed_operand): New predicate for direct move support.
16658         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16659         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16660         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16661         (POWERPC_MASKS): Add power8 options.
16662         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16663         various options.
16665         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16666         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16668         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16669         (-mpower8-fusion): New power8 options.
16670         (-mpower8-fusion-sign): Likewise.
16671         (-mpower8-vector): Likewise.
16672         (-mcrypto): Likewise.
16673         (-mdirect-move): Likewise.
16674         (-mquad-memory): Likewise.
16676         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16677         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16678         registers.
16679         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16680         (rs6000_debug_vector_unit): Add p8_vector.
16681         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16682         definitions.  Also print fusion state.
16683         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16684         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16685         (rs6000_option_override_internal): Add support for power8.
16686         (rs6000_common_init_builtins): Add debugging for skipped builtins
16687         if -mdebug=builtin.
16688         (rs6000_adjust_cost): Add power8 support.
16689         (rs6000_issue_rate): Likewise.
16690         (insn_must_be_first_in_group): Likewise.
16691         (insn_must_be_last_in_group): Likewise.
16692         (force_new_group): Likewise.
16693         (rs6000_register_move_cost): Likewise.
16694         (rs6000_opt_masks): Likewise.
16696         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16697         power8 capable assembler, default to power7 options.
16698         (TARGET_DIRECT_MOVE): Likewise.
16699         (TARGET_CRYPTO): Likewise.
16700         (TARGET_P8_VECTOR): Likewise.
16701         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16702         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16703         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16704         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16705         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16706         (TARGET_XSCVDPSPN): Likewise.
16707         (TARGET_XSCVSPDPN): Likewsie.
16708         (TARGET_SYNC_HI_QI): Likewise.
16709         (TARGET_SYNC_TI): Likewise.
16710         (MASK_CRYPTO): Likewise.
16711         (MASK_DIRECT_MOVE): Likewise.
16712         (MASK_P8_FUSION): Likewise.
16713         (MASK_P8_VECTOR): Likewise.
16714         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16715         temporary used by some of the direct move instructions to get two FP
16716         temporary registers does not force creation of a stack frame.
16717         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16718         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16719         that any VSX registers are tieable, even if they are also an
16720         Altivec vector mode.
16721         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16722         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16723         (RS6000_BTM_CRYPTO): Likewise.
16724         (RS6000_BTM_COMMON): Likewise.
16726         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16727         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16728         (enum rs6000_vector): Add power8 vector support.
16730 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16732         PR target/19599
16733         PR target/57340
16734         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16735         (any_sibcall_could_use_r3): this and handle indirect calls.
16736         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16738 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16740         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16742 2013-05-22  Richard Biener  <rguenther@suse.de>
16744         PR middle-end/57349
16745         * profile.c (branch_prob): Do not split blocks that are
16746         abnormally receiving from ECF_RETURNS_TWICE functions.
16748 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16750         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16751         address mode.  Move pointer mode initialization to the same place.
16753 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16755         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16756         while it has any effect.
16758 2013-05-21  Easwaran Raman  <eraman@google.com>
16760         PR tree-optimization/57322
16761         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16762         UID of the statement added to the BB to be 1.
16764 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16766         PR tree-optimization/57331
16767         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16768         of conversion from pointer type to integral type with integer.
16770 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16772         PR lto/57289
16773         * ipa-prop.c (ipa_read_node_info): Process param_used and
16774         controlled_uses in the same order as when writing.
16776 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16778         PR plugins/56754
16779         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16781 2013-05-21  Richard Biener  <rguenther@suse.de>
16783         PR tree-optimization/57318
16784         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16785         estimate stmts with side-effects as likely eliminated.
16787 2013-05-21  Richard Biener  <rguenther@suse.de>
16789         PR tree-optimization/57330
16790         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16791         preserve the call stmts fntype.
16793 2013-05-21  Richard Biener  <rguenther@suse.de>
16795         PR tree-optimization/57303
16796         * tree-ssa-sink.c (statement_sink_location): Improve killing
16797         stmt detection and properly handle self-assignments.
16799 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16801         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16802         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16803         registers. Set register size out of the PARALLEL loop.
16805 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16807         PR target/56547
16808         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16809         (*fmasf4, *fmasf4_media): New insns.
16811 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16813         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16814         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16815         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16816         (mips_idiv_insns): Update the comments to say that the returned
16817         instruction counts are in units of BASE_INSN_LENGTH.
16818         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16819         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16820         using 2 rather than 4 as the length of indirect MIPS16 and
16821         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16822         length of a NOP.  Don't divide MIPS16 lengths by 2.
16823         (mips16_split_long_branches): Assume a branch is long if the
16824         length is greater than 4 rather than 8.
16825         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16826         rather than multiplying them by 2.  Multiply instruction counts
16827         by BASE_INSN_LENGTH rather than 4.
16828         (*jump_mips16, tls_get_tp_mips16_<mode>)
16829         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16831 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16833         * config/mips/mips.md (extended_mips16): Remove branch case.
16834         (length): Remove duplicated extended_mips16 test.
16836 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16838         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16840 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16842         * recog.h (Recog_data): Rename to...
16843         (recog_data_d): ...this.
16844         (recog_data): Update accordingly.
16845         * recog.c (recog_data): Likewise.
16846         * reload.c (save_recog_data): Likewise.
16847         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16848         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16850 2013-05-17  Julian Brown  <julian@codesourcery.com>
16852         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16853         found in a REG_EQUAL note, invalidate it.
16855 2013-05-17  Easwaran Raman  <eraman@google.com>
16857         * tree-ssa-reassoc.c (find_insert_point): New function.
16858         (insert_stmt_after): Likewise.
16859         (get_def_stmt): Likewise.
16860         (ensure_ops_are_available): Likewise.
16861         (not_dominated_by): Likewise.
16862         (rewrite_expr_tree): Do not move statements beyond what is
16863         necessary. Remove call to swap_ops_for_binary_stmt...
16864         (reassociate_bb): ... and move it here.
16865         (build_and_add_sum): Assign UIDs for new statements.
16866         (linearize_expr): Likewise.
16867         (do_reassoc): Renumber gimple statement UIDs.
16869 2013-05-17  Jan Hubicka  <jh@suse.cz>
16871         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16872         weakrefs.
16873         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16874         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16875         weakrefs.
16876         (output_weakrefs): Update.
16878 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16879             Martin Jambor  <mjambor@suse.cz>
16881         PR middle-end/57276
16882         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16883         value that corresponds to the given aggval is found in values vector.
16885 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16887         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16888         sse, sse2, sse3, ssse3 and sse4a flags to options.
16890 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16892         * gengtype-state.c: (s_expr_writer): New class, to handle
16893         prettifying of output layout of s-expressions.
16894         (state_writer): New class, to write out gtype.state.
16895         (state_written_type_count): Move this variable into member data of
16896         state_writer.
16897         (s_expr_writer::s_expr_writer): New code: constructor for new class
16898         (state_writer::state_writer(): ditto
16899         (s_expr_writer::write_new_line): New function
16900         (s_expr_writer::write_any_indent): ditto
16901         (s_expr_writer::begin_s_expr): ditto
16902         (s_expr_writer::end_s_expr): ditto
16903         (write_state_fileloc): convert to method of state_writer...
16904         (state_writer:: write_state_fileloc): ...and use methods of
16905         s_expr_writer to write indentation into the gtype.state output file
16906         to visually represent the hierarchical structure of the list
16907         structures
16908         (write_state_fields): ditto, renaming to...
16909         (state_writer::write_state_fields)
16910         (write_state_a_string): ditto, renaming to...
16911         (state_writer::write_state_a_string)
16912         (write_state_string_option): ditto, renaming to...
16913         (state_writer::write_state_string_option)
16914         (write_state_type_option): ditto, renaming to...
16915         (state_writer::write_state_type_option)
16916         (write_state_nested_option): ditto, renaming to...
16917         (state_writer::write_state_nested_option)
16918         (write_state_option): ditto, renaming to...
16919         (state_writer::write_state_option)
16920         (write_state_options): ditto, renaming to...
16921         (state_writer::write_state_options)
16922         (write_state_lang_bitmap): ditto, renaming to...
16923         (state_writer::write_state_lang_bitmap)
16924         (write_state_version): ditto, renaming to...
16925         (state_writer::write_state_version)
16926         (write_state_scalar_type): ditto, renaming to...
16927         (state_writer::write_state_scalar_type)
16928         (write_state_string_type): ditto, renaming to...
16929         (state_writer::write_state_string_type)
16930         (write_state_undefined_type): ditto, renaming to...
16931         (state_writer::write_state_undefined_type)
16932         (write_state_struct_union_type): ditto, renaming to...
16933         (state_writer::write_state_struct_union_type)
16934         (write_state_struct_type): ditto, renaming to...
16935         (state_writer::write_state_struct_type)
16936         (write_state_user_struct_type): ditto, renaming to...
16937         (state_writer::write_state_user_struct_type)
16938         (write_state_lang_struct_type): ditto, renaming to...
16939         (state_writer::write_state_lang_struct_type)
16940         (write_state_param_struct_type): ditto, renaming to...
16941         (state_writer::write_state_param_struct_type)
16942         (write_state_pointer_type): ditto, renaming to...
16943         (state_writer::write_state_pointer_type)
16944         (write_state_array_type): ditto, renaming to...
16945         (state_writer::write_state_array_type)
16946         (write_state_gc_used): ditto, renaming to...
16947         (state_writer::write_state_gc_used)
16948         (write_state_common_type_content): ditto, renaming to...
16949         (state_writer::write_state_common_type_content)
16950         (write_state_type): ditto, renaming to...
16951         (state_writer::write_state_type)
16952         (write_state_pair_list): ditto, renaming to...
16953         (state_writer::write_state_pair_list)
16954         (write_state_pair): ditto, renaming to...
16955         (state_writer::write_state_pair)
16956         (write_state_typedefs): ditto, renaming to...
16957         (state_writer::write_state_typedefs)
16958         (write_state_structures): ditto, renaming to...
16959         (state_writer::write_state_structures)
16960         (write_state_param_structs): ditto, renaming to...
16961         (state_writer::write_state_param_structs)
16962         (write_state_variables): ditto, renaming to...
16963         (state_writer::write_state_variables)
16964         (write_state_srcdir): ditto, renaming to...
16965         (state_writer::write_state_srcdir)
16966         (write_state_files_list): ditto, renaming to...
16967         (state_writer::write_state_files_list)
16968         (write_state_languages): ditto, renaming to...
16969         (state_writer::write_state_languages)
16970         (write_state): create a state_writer instance and use it when
16971         writing out the state file
16973 2013-05-17  Mike Stump  <mikestump@comcast.net>
16975         PR rtl-optimization/57304
16976         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16977         accessing DF_REF_REAL_LOC.
16979 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16981         PR rtl-optimization/57281
16982         PR rtl-optimization/57300
16983         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16984         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16985         what the other splitter did if the registers are dead.
16987 2013-05-17  Richard Biener  <rguenther@suse.de>
16989         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16990         MEM_REF offsets.
16992 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16994         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16995         linking.
16997 2013-05-17  Marek Polacek  <polacek@redhat.com>
16999         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
17000         length when doing non-zero store of storing '\0' to '\0'.
17002 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17004         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
17005         vect_external_def oprnd1 with loop_vinfo, try to emit
17006         optional cast, negation and and stmts on the loop preheader
17007         edge instead of into the pattern def seq.
17009         PR tree-optimization/57051
17010         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
17011         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
17013 2013-05-16  Nick Clifton  <nickc@redhat.com>
17015         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
17016         (rl78_is_naked_func): New function.
17017         (rl78_expand_prologue): Skip prologue generation for naked functions.
17018         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17019         * doc/extend.texi (naked): Add RL78 to the list of processors
17020         that supports this attribute.
17022 2013-05-16  Jeff Law  <law@redhat.com>
17024         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17026 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17028         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17029         cache parameters using detect_caches_amd also for CYRIX,
17030         NSC and TM2 signatures.
17032 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17033             Dzianis Kahanovich  <mahatma@eu.by>
17035         PR target/45359
17036         PR target/46396
17037         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17038         VIA/Centaur processors and determine their cache parameters
17039         using detect_caches_amd.
17041 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17043         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17044         (rtl_verify_edges): New function.
17045         (rtl_verify_bb_insns): Ditto.
17046         (rtl_verify_bb_pointers): Ditto.
17047         (rtl_verify_bb_insn_chain): Ditto.
17048         (rtl_verify_fallthru): Ditto.
17049         (rtl_verify_bb_layout): Ditto.
17050         (rtl_verify_flow_info_1): Outline checks into new functions.
17051         (rtl_verify_flow_info): Ditto.
17053 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17055         * cfghooks.c (copy_bbs): Add update_dominance argument.
17056         * cfghooks.h (copy_bbs): Update prototype.
17057         * tree-cfg.c (gimple_duplicate_sese_region):
17058         Add update_dominance argument.
17059         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17060         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17061         gimple_duplicate_sese_region call.
17062         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17063         Update copy_bbs call.
17064         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17065         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17067 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17069         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17070         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17071         vect_recog_rotate_pattern.
17072         (vect_recog_rotate_pattern): New function.
17074 2013-05-16  Jason Merrill  <jason@redhat.com>
17076         * Makefile.in (LLINKER): New variable.
17077         (mostlyclean): Remove link mutex.
17078         * configure.ac: Handle --enable-link-mutex.
17079         * lock-and-run.sh: New script.
17081 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17083         PR target/19599
17084         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17085         for NULL decl.
17087 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17089         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17091 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17093         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17094         * config/arm/arm.c (next_consecutive_mem): New function.
17095         (gen_movmem_ldrd_strd): Likewise.
17096         * config/arm/arm.md (movmemqi): Update condition and code.
17097         (unaligned_loaddi, unaligned_storedi): New patterns.
17099 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17101         * config.gcc: Obsolete *-*-solaris2.9*.
17102         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17104 2013-05-16  Richard Biener  <rguenther@suse.de>
17106         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17107         earlier, after GRAPHITE transforms and IV canonicalization.
17109 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17111         * omp-low.c (extract_omp_for_data): For collapsed loops,
17112         if at least one of the loops is known at compile time to
17113         iterate zero times, set count to 0.
17114         (expand_omp_regimplify_p): New function.
17115         (expand_omp_for_generic): For collapsed loops, if at least
17116         one of the loops isn't known to iterate at least once,
17117         add runtime check with setting count to 0.
17118         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17119         For unsigned types if it isn't known at compile time that
17120         the loop will iterate at least once, add runtime check to bypass
17121         the whole loop if initial condition isn't true.
17123 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17125         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17127 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17129         PR middle-end/57286
17130         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17131         transformations to avoid an infinite loop.
17133 2013-05-16  Marek Polacek  <polacek@redhat.com>
17135         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17137 2013-05-15  Leif Ekblad  <leif@rdos.net>
17139         * config/i386/i386.c (ix86_decompose_address): Use
17140         DEFAULT_TLS_SEG_REG to access TLS segment register.
17141         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17142         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17143         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17145 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17147         PR target/57260
17148         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17149         sibling calls to functions that would normally be lazily bound,
17150         unless $gp is call-clobbered.
17152 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17154         * config/i386/i386.c (ix86_option_override_internal): Update
17155         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17156         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17157         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17158         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17159         of TARGET_3DNOW.
17160         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17162 2013-05-15  Andreas Schwab  <schwab@suse.de>
17164         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17165         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17166         third operand.
17168 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17170         * loop-unroll.c (report_unroll_peel): Check decision before
17171         emitting unroll/peel message.
17173 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17175         * function.h (has_bb_partition): New rtl_data flag.
17176         (bb_reorder_complete): Ditto.
17177         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17178         instead of flag_reorder_blocks_and_partition.
17179         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17180         with some enhancements.
17181         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17182         * bb-reorder.c (connect_traces): Check for has_bb_partition
17183         instead of flag_reorder_blocks_and_partition.
17184         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17185         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17186         verify_hot_cold_block_grouping.
17187         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17189 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17191         PR target/19599
17192         * config/arm/predicates.md (call_insn_operand): New predicate.
17193         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17194         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17195         if insn is not a tail call.
17196         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17197         registers.
17198         * config/arm/arm.h (enum reg_class): New caller save register class.
17199         (REG_CLASS_NAMES): Likewise.
17200         (REG_CLASS_CONTENTS): Likewise.
17201         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17202         without decls.
17204 2013-05-15  Richard Biener  <rguenther@suse.de>
17206         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17207         of MSG_OPTIMIZED_LOCATIONS.
17208         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17209         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17210         message.
17211         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17212         of MSG_OPTIMIZED_LOCATIONS.
17213         (execute_vect_slp): Likewise.
17214         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17215         (vect_create_cond_for_alias_checks): Likewise.
17216         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17217         (vect_recog_widen_mult_pattern): Likewise.
17218         (vect_recog_widen_sum_pattern): Likewise.
17219         (vect_recog_over_widening_pattern): Likewise.
17220         (vect_recog_widen_shift_pattern): Likewise.
17221         (vect_recog_vector_vector_shift_pattern): Likewise.
17222         (vect_recog_divmod_pattern): Likewise.
17223         (vect_recog_mixed_size_cond_pattern): Likewise.
17224         (vect_recog_bool_pattern): Likewise.
17225         (vect_pattern_recog_1): Likewise.
17227 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17229         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17230         non-functions to builtin_unreachable.
17231         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17232         correct when new direct edges were discovered.
17234 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17236         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17237         header, print symbol order instead of node uid, print more information
17238         about indirect edge targets.
17239         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17240         uids.
17241         (ipa_make_edge_direct_to_target): Likewise.
17242         (remove_described_reference): Likewise.
17243         (propagate_controlled_uses): Likewise.
17244         (ipa_print_node_params): Also print symbol order.
17245         (ipcp_transform_function): Print symbol order instead of node uids.
17246         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17247         (cgraph_get_create_real_symbol_node): Likewise.
17248         * ipa-cp.c (print_lattice): Likewise.
17249         (print_all_lattices): Likewise.
17250         (determine_versionability): Likewise.
17251         (initialize_node_lattices): Likewise.
17252         (estimate_local_effects): Likewise.
17253         (update_profiling_info): Likewise.
17254         (create_specialized_node): Likewise.
17255         (perhaps_add_new_callers): Likewise.
17256         (decide_about_value): Likewise.
17257         (decide_whether_version_node): Likewise.
17258         (identify_dead_nodes): Likewise.
17259         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17260         (dump_inline_summary): Likewise.
17261         (estimate_node_size_and_time): Likewise.
17262         (inline_analyze_function): Likewise.
17263         * ipa-inline.c (report_inline_failed_reason): Likewise.
17264         (want_early_inline_function_p): Likewise.
17265         (edge_badness): Likewise.
17266         (update_edge_key): Likewise.
17267         (inline_small_functions): Likewise.  Add dumping of order to two other
17268         dumps.
17269         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17270         instead of node uids.
17271         (propagate_pure_const): Likewise.
17272         (propagate_pure_const): Likewise.
17273         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17274         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17275         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17276         of node uids.
17277         * tree-pretty-print.c (dump_function_header): Likewise.
17278         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17279         Print symbol order instead of node uids.
17281 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17283         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17284         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17286 2013-05-15  Richard Biener  <rguenther@suse.de>
17288         PR tree-optimization/57275
17289         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17290         return value for fail to do runtime alias checks for gather loads.
17292 2013-05-15  Jan Hubicka  <jh@suse.cz>
17294         PR lto/57038
17295         PR lto/47375
17296         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17297         weakrefs are not external.
17298         (lto_symtab_merge_decls): Fix thinko when dealing with
17299         non-lto_symtab decls.
17300         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17301         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17302         * varpool.c (dump_varpool_node): Dump more flags.
17304 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17306         * config/i386/i386.c (processor_alias_table): Add instruction
17307         FSGSBASE for AMD bdver3 architecture.
17309 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17311         * tree.c (warn_deprecated_use): Print file:line using locus color.
17312         * diagnostic.c (diagnostic_report_current_module): Print file:line
17313         and file:line:column using locus color.
17315 2013-05-14  Mike Stump  <mikestump@comcast.net>
17317         * gdbinit.in: Add __null.
17319 2013-05-14  Mike Stump  <mikestump@comcast.net>
17321         * recog.h: Rename struct recog_data to Recog_data.
17322         * recog.c: Likewise.
17323         * reload.c (can_reload_into): Likewise.
17324         * config/picochip/picochip.c: Likewise.
17326 2013-05-14  Mike Stump  <mikestump@comcast.net>
17328         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17330 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17332         * resource.h (struct resources): Remove unch_memory member.
17333         (CLEAR_RESOURCE): Don't clear unch_memory.
17334         * resource.c (mark_referenced_resources): Don't set it.
17335         (mark_set_resources): Likewise.
17336         (mark_target_live_regs): Don't clear it.
17337         (init_resource_info): Likewise.
17338         * reorg.c (resource_conflicts_p): Don't compare it.
17339         (redundant_insn): Don't set it.
17341         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17342         Remove prototypes.
17343         * emit-rtl.c (next_label): Remove unused function.
17344         (skip_consecutive_labels, link_cc0_insns): Move to ...
17345         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17346         only place where these functions are used, and make them static.
17348 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17350         * fold-const.c (fold_negate_expr): Handle vectors.
17351         (fold_truth_not_expr): Make it static.
17352         (fold_invert_truthvalue): New static function.
17353         (invert_truthvalue_loc): Handle vectors. Do not call
17354         fold_truth_not_expr directly.
17355         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17356         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17357         (fold_comparison): Handle vector constants.
17358         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17359         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17360         * tree.h (fold_truth_not_expr): Remove declaration.
17362 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17364         * config/aarch64/aarch64-simd.md
17365         (aarch64_vcond_internal<mode>): Rename to...
17366         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17367         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17368         float modes. Clarify all iterator modes.
17369         (vcond<mode><mode>): Use new name for vcond expanders.
17370         (vcond<v_cmp_result><mode>): Likewise.
17371         (vcondu<mode><mode>: Likewise.
17372         * config/aarch64/iterators.md (VDQF_COND): New.
17374 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17376         PR bootstrap/57266
17377         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17378         variable for the shift amount. Check that we shift by non-negative
17379         amounts.
17381 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17383         PR target/42017
17384         * config/arm/arm.h (EPILOGUE_USES): Only return true
17385         for LR_REGNUM after epilogue_completed.
17387 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17389         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17390         is error_mark_node.
17392 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17394         PR target/57261
17395         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17396         and Solaris 11+/x86 with gld.
17397         * configure: Regenerate.
17399 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17401         * expmed.c (expand_shift_1): Canonicalize rotates by
17402         constant bitsize / 2 to bitsize - 1.
17403         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17404         case ROTATERT>: Likewise.
17406         Revert:
17407         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17409         * config/i386/i386.md (rotateinv): New code attr.
17410         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17411         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17412         roll $31, %eax, etc.
17414 2013-05-14  Richard Biener  <rguenther@suse.de>
17416         PR middle-end/57235
17417         * tree-eh.c (sink_clobbers): Give up for successors with
17418         multiple predecessors and no virtual uses.
17420 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17422         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17423         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17425 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17427         PR middle-end/57251
17428         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17429         the case when both op0 and op1 have VOIDmode.
17431 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17433         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17434         in multiply-accumulate mode.
17436 2013-05-13  Guozhi Wei  <carrot@google.com>
17438         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17440 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17442         PR target/56975
17443         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17444         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17445         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17446         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17447         (ix86_expand_prologue): Likewise.
17448         (ix86_expand_split_stack_prologue): Likewise.
17449         (legitimate_pic_address_disp_p): Likewise.
17450         (legitimize_pic_address): Likewise.
17451         (legitimize_tls_address): Likewise.
17452         (legitimize_pe_coff_symbol): Likewise.
17453         (output_pic_addr_const): Likewise.
17454         (construct_plt_address): Likewise.
17455         (ix86_expand_call): Likewise.
17456         (x86_output_mi_thunk): Likewise.
17457         (x86_function_profiler): Likewise.
17459 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17461         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17462         similar switch cases.
17463         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17464         (aarch64_simd_mov_to_<mode>low): Delete.
17465         (aarch64_simd_mov_to_<mode>high): Delete.
17466         (move_lo_quad_<mode>): Add w<-r alternative.
17467         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17468         (aarch64_simd_mov_from_*): Update type attribute.
17469         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17470         statement.
17472 2013-05-13  Jan Hubicka  <jh@suse.cz>
17474         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17475         * config/i386/i386.c (ix86_compute_frame_layout,
17476         ix86_expand_epilogue, emit_i387_cw_initialization,
17477         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17478         ix86_local_alignment): Fix use of size/speed predicates.
17480 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17482         PR tree-optimization/45216
17483         PR tree-optimization/57157
17484         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17485         the (-Y) & (B - 1) variant if OP is |.
17486         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17487         return shifted.  Use (-op1) & (prec - 1) as other_amount
17488         instead of prec - op1.
17490 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17492         PR middle-end/42371
17493         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17494         (ipa_constant_data): New type.
17495         (ipa_jump_func): Use ipa_constant_data to hold information about
17496         constant jump functions.
17497         (ipa_get_jf_constant): Adjust to jump function type changes.
17498         (ipa_get_jf_constant_rdesc): New function.
17499         (ipa_param_descriptor): New field controlled_uses.
17500         (ipa_get_controlled_uses): New function.
17501         (ipa_set_controlled_uses): Likewise.
17502         * ipa-ref.h (ipa_find_reference): Declare.
17503         * ipa-prop.c (ipa_cst_ref_desc): New type.
17504         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17505         changes.
17506         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17507         New parameter cs.  Adjust all callers.
17508         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17509         (remove_described_reference): New function.
17510         (jfunc_rdesc_usable): Likewise.
17511         (try_make_edge_direct_simple_call): Decrement controlled use count,
17512         attempt to remove reference if it hits zero.
17513         (combine_controlled_uses_counters): New function.
17514         (propagate_controlled_uses): Likewise.
17515         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17516         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17517         (ipa_print_node_params): Print described use counter.
17518         (ipa_write_jump_function): Adjust to jump function type changes.
17519         (ipa_read_jump_function): New parameter CS, pass it to
17520         ipa_set_jf_constant.  Adjust caller.
17521         (ipa_write_node_info): Stream controlled use count
17522         (ipa_read_node_info): Likewise.
17523         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17524         asserting.
17525         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17526         count.  Remove cloning-added reference if it reaches zero.
17527         * ipa-ref.c (ipa_find_reference): New function.
17529 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17531         * config/i386/i386.c (processor_target_table): Modified default
17532         alignment values for AMD BD and BT architectures.
17534 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17536         * tree-vect-generic.c (uniform_vector_p): Move ...
17537         * tree.c (uniform_vector_p): ... here.
17538         * tree.h (uniform_vector_p): Declare it.
17539         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17540         into a scalar.
17542 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17544         PR tree-optimization/57230
17545         * tree-ssa-strlen.c (handle_char_store): Record length for
17546         array store from STRING_CST.
17548         PR tree-optimization/57230
17549         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17550         check.
17552 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17554         * config/epiphany/epiphany.c (epiphany_init): Check size of
17555         NUM_MODES_FOR_MODE_SWITCHING.
17556         (epiphany_expand_prologue):
17557         Remove CONFIG_REGNUM initial value handling code.
17558         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17559         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17560         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17561         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17562         Don't return 1 for FP_MODE_NONE.
17563         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17564         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17565         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17566         * config/epiphany/epiphany.md (save_config): New pattern.
17568 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17570         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17572 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17574         * config/i386/i386.md (memory): Handle sseishft1.
17575         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17576         (*vec_extractv2di_1): Ditto.
17578 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17580         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17581         saved registers.
17583 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17585         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17586         Add mthumb/march=armv7-a multilib.
17587         Add mthumb/march=armv7-r multilib.
17588         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17590 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17592         * config/v850/t-rtems: Add more multilibs.
17594 2013-05-10  Richard Biener  <rguenther@suse.de>
17596         PR tree-optimization/57214
17597         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17598         not propagate from SSA names that occur in abnormal PHI nodes.
17600 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17602         * stor-layout.c (element_precision): New function.
17603         * machmode.h (element_precision): Declare it.
17604         * tree.c (build_minus_one_cst): New function.
17605         (element_precision): Likewise.
17606         * tree.h (build_minus_one_cst): Declare new function.
17607         (element_precision): Likewise.
17608         * fold-const.c (operand_equal_p): Use element_precision.
17609         (fold_binary_loc): Handle vector types.
17610         * convert.c (convert_to_integer): Use element_precision.
17611         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17612         separately.
17614 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17616         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17617         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17618         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17619         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17620         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17621         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17622         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17623         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17624         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17625         (Uuw8): New constraints.
17626         (Usb4): Move into alphabetical order.
17627         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17628         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17629         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17630         previously unnamed patterns.
17631         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17632         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17633         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17634         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17635         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17636         of specific lengths.
17638 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17640         * config/i386/i386.md (rotateinv): New code attr.
17641         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17642         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17643         roll $31, %eax, etc.
17645         PR tree-optimization/45216
17646         PR tree-optimization/57157
17647         * tree-ssa-forwprop.c (simplify_rotate): New function.
17648         (ssa_forward_propagate_and_combine): Call it.
17650 2013-05-10  Richard Biener  <rguenther@suse.de>
17652         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17653         disable peeling when we version for aliasing.
17654         (vector_alignment_reachable_p): Honor explicit user alignment.
17655         (vect_supportable_dr_alignment): Likewise.
17656         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17657         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17658         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17659         then peeling to arrange for the cost-model check to come first.
17661 2013-05-10  Alan Modra  <amodra@gmail.com>
17663         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17664         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17665         * configure: Regenerate.
17667 2013-05-10  Alan Modra  <amodra@gmail.com>
17669         PR target/55033
17670         * varasm.c (default_elf_select_section): Move !DECL_P check..
17671         (get_named_section): ..to here before calling get_section_name.
17672         Adjust assertion.
17673         (default_section_type_flags): Add DECL_P check.
17674         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17675         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17677 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17679         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17680         When using gen_stack_adjust_str with a register offset, add a
17681         REG_FRAME_RELATED_EXPR note.
17683 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17685         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17686         (*vec_extractv4si_zext_mem): Ditto.
17687         (*vec_extractv2di): Add 0->x and x->x alternatives.
17688         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17689         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17691 2013-05-09  Jason Merrill  <jason@redhat.com>
17693         N3639 C++1y VLA support
17694         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17695         DECL_VALUE_EXPR.
17697         * tree.c (build_constructor_va): New.
17698         * tree.h: Declare it.
17700 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17702         PR lto/57084
17703         * gimple-fold.c (canonicalize_constructor_val): Call
17704         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17706 2013-05-09  Jan Hubicka  <jh@suse.cz>
17707             Richard Biener  <rguenther@suse.de>
17709         PR lto/54095
17710         * symtab.c (symtab_make_decl_local): Do not add private names.
17712 2013-05-09  Jan Hubicka  <jh@suse.cz>
17714         PR lto/54095
17715         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17716         (unlink_from_assembler_name_hash): Likewise.
17717         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17718         symtab_unregister_node, symtab_initialize_asm_name_hash,
17719         change_decl_assembler_name): Update.
17721 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17723         * config/aarch64/aarch64.md: New movtf split.
17724         (*movtf_aarch64): Update.
17725         (aarch64_movdi_tilow): Handle TF modes and rename to
17726         aarch64_movdi_<mode>low.
17727         (aarch64_movdi_tihigh): Handle TF modes and rename to
17728         aarch64_movdi_<mode>high
17729         (aarch64_movtihigh_di): Handle TF modes and rename to
17730         aarch64_mov<mode>high_di
17731         (aarch64_movtilow_di): Handle TF modes and rename to
17732         aarch64_mov<mode>low_di
17733         (aarch64_movtilow_tilow): Remove spurious whitespace.
17734         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17735         splits.
17736         (aarch64_print_operand): Update.
17738 2013-05-09  Alan Modra  <amodra@gmail.com>
17740         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17741         powerpc64le.
17742         * configure: Regenerate.
17744 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17746         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17747         splitter preparation statements.
17748         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17749         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17750         adjust_address_nv.
17752 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17754         * gimple-ssa-strength-reduction.c (count_candidates): Change
17755         return value to int.
17756         (analyze_candidates_and_replace): Change type of length to int.
17758 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17760         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17761         (*vec_extract<mode>): Use VI12_128 mode iterator.
17762         (*vec_extract<mode>_mem): Ditto.
17763         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17764         attribute.
17766 2013-05-08  Diego Novillo  <dnovillo@google.com>
17768         PR bootstrap/54659
17770         Revert:
17771         2012-08-17  Diego Novillo  <dnovillo@google.com>
17773         PR bootstrap/54281
17774         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17775         * config.in: Regenerate.
17776         * configure: Regenerate.
17777         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17779 2013-05-08  Jan Hubicka  <jh@suse.cz>
17781         PR lto/54095
17782         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17783         * cgraph.h (symtab_node_base): Add unique_name.
17784         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17785         input_overwrite_node, input_varpool_node): Stream unique_name.
17786         * cgraphclones.c (cgraph_create_virtual_clone,
17787         cgraph_function_versioning): Set unique_name.
17788         * ipa.c (function_and_variable_visibility): Set unique_name.
17790 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17792         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17793         (alloc_cand_and_find_basis): Restrict conditional candidate
17794         processing to CAND_MULTs.
17796 2013-05-08  Jan Hubicka  <jh@suse.cz>
17798         PR lto/54095
17799         lto-symtab.c (lto_symtab_symbol_p): New function.
17800         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17801         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17802         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17803         Skip static symbols.
17805 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17807         PR tree-optimization/57200
17808         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17809         Only call inform if the preceding warning_at returns true.
17811 2013-05-07  Han Shen  <shenhan@google.com>
17813         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17814         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17815         * common.opt (fstack-protector-strong): New option.
17816         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17817         * doc/invoke.texi (Optimization Options): Document
17818         "-fstack-protector-strong".
17819         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17821 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17823         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17825 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17827         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17828         it is not lhs of the insn.
17829         (setup_reg_equiv): Remove insn having equiv memory which it is not
17830         lhs of the insn.
17831         * lra-constraints.c (process_address): Try to improve generation
17832         code for address base + disp.
17833         (lra_constraints): Make correct the code for checking insn setting
17834         up backward equivalence.  Remove insn only if it is in the init
17835         insn list.
17836         * lra-eliminations.c (update_reg_eliminate): Change return value.
17837         (lra_eliminate): Use the result.
17839 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17841         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17842         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17843         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17844         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17845         register target operands.
17846         (*vec_extractv8hi_sse2): New pattern.
17847         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17848         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17849         (*vec_extract<mode>_mem): New insn and split pattern.
17851 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17853         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17854         (TARGET_ASAN_SHADOW_OFFSET): Define.
17855         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17856         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17858 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17860         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17861         (incr_vec_index): Return -1 if increment not found.
17862         (create_add_on_incoming_edge): Assert if increment not found.
17863         (record_increment): Limit number of increments recorded.
17864         (all_phi_incrs_profitable): Return false if an increment not found.
17865         (replace_profitable_candidates): Don't process increments that were
17866         not recorded.
17867         (analyze_candidates_and_replace): Limit size of incr_vec.
17869 2013-05-07  Richard Biener  <rguenther@suse.de>
17871         * calls.c (special_function_p): setjmp-like functions are leaf.
17872         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17873         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17875 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17877         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17878         (aarch64_simd_mov<mode>): New expander.
17879         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17880         (aarch64_simd_mov_to_<mode>high): Likewise.
17881         (aarch64_simd_mov_from_<mode>low): Likewise.
17882         (aarch64_simd_mov_from_<mode>high): Likewise.
17883         (aarch64_dup_lane<mode>): Update.
17884         (aarch64_dup_lanedi): New instruction pattern.
17885         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17886         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17888 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17890         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17891         (replace_mult_candidate): Remove unnecessary argument; remove
17892         unnecessary parameter from call to introduce_cast_before_cand.
17893         (replace_unconditional_candidate): Remove unnecessary parameter
17894         from call to replace_mult_candidate.
17895         (replace_conditional_candidate): Likewise.
17896         (insert_initializers): Use make_temp_ssa_name.
17897         (introduce_cast_before_cand): Remove unnecessary argument; use
17898         make_temp_ssa_name.
17899         (replace_one_candidate): Remove unnecessary argument; remove
17900         unnecessary parameter from calls to introduce_cast_before_cand.
17901         (replace_profitable_candidates): Remove unnecessary parameters
17902         from calls to replace_one_candidate.
17904 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17906         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17907         phi def as possibly hiding a basis for a CAND_ADD whose operands
17908         have been commuted in the analysis.
17909         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17911 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17913         * config/aarch64/aarch64.md
17914         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17915         shift value between 0-4.
17917 2013-05-07  Richard Biener  <rguenther@suse.de>
17919         * double-int.h (rshift): New overload.
17920         * double-int.c (rshift): New function.
17921         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17922         (create_reference_ops_from_ref): Remove.
17923         (vn_reference_insert): Use shared ops for constructing the
17924         reference and copy it.
17926 2013-05-07  Richard Biener  <rguenther@suse.de>
17928         PR middle-end/57190
17929         * tree-eh.c (sink_clobbers): Properly propagate
17930         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17932 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17934         PR tree-optimization/57149
17935         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17936         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17937         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17938         uninit_undefined_value_p instead of ssa_undefined_value_p.
17940         PR debug/57184
17941         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17942         for modifier == EXPAND_INITIALIZER.
17944 2013-05-07  Anton Blanchard  <anton@samba.org>
17946         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17947         for powerpc64 little endian.
17948         * configure: Regenerate.
17950 2013-05-06  Graham Stott  <grahams@btinternet.com>
17952         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17953         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17954         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17955         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17957 2013-05-06  Graham Stott  <grahams@btinternet.com>
17959         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17960         codes which allow non-lvalues.
17962 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17964         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17965         components are all 1s.
17966         (integer_minus_onep): New function.
17967         * tree.h (integer_minus_onep): Declare it.
17968         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17969         integer_minus_onep instead of integer_all_onesp.
17971 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17973         PR target/52933
17974         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17975         variations of these patterns.
17977 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17979         * config/i386/i386.md (isa): Add x64_sse4 member.
17980         (enabled): Handle x64_sse4.
17981         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17982         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17983         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17984         instruction for SSE4_1 targets.  Update insn attributes.
17985         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17986         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17987         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17988         const_1 selector.
17989         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17990         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17991         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17993 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17995         PR target/57108
17996         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17998 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
18000         * final.c (do_assembler_dialects): Don't handle curly braces and
18001         vertical bar escaped by % as dialect delimiters.
18002         (output_asm_insn): Print curly braces and vertical bar if escaped
18003         by % and ASSEMBLER_DIALECT defined.
18004         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
18005         * doc/tm.texi: Regenerated.
18007 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18009         * config/mips/mips.c: Include tree-pass.h.
18010         (mips_reorg): Split in pre- and post-dbr_schedule parts.
18011         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
18012         (pass_mips_machine_reorg2): New machine specific pass.
18013         (insert_pass_mips_machine_reorg2): New pass plugin definition.
18014         (mips_option_override): Register the new pass.
18015         * rtl.h (cleanup_barriers): Remove prototype.
18016         (dbr_schedule): Likewise.
18017         * jump.c (cleanup_barriers): Make static.
18018         * reorg.c (dbr_schedule): Likewise.
18020 2013-05-06  Richard Biener  <rguenther@suse.de>
18022         PR tree-optimization/57185
18023         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18024         SSA names properly.
18026 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18028         PR target/57106
18029         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18030         between "first_arg" and "insn", not "last" and "insn".
18032 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18034         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18035         (find_candidates_in_block): Re-enable slsr_process_phi.
18036         (create_phi_basis): Fix double counting of candidate adjustment.
18038 2013-05-06  Richard Biener  <rguenther@suse.de>
18040         PR middle-end/57147
18041         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18042         the edge is also fallthru, preserve it and just clear the
18043         abnormal flag.
18044         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18045         also complex, preserve that and just clear the fallthru flag.
18046         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18047         update virtual operands.
18049 2013-05-06  Alan Modra  <amodra@gmail.com>
18051         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18052         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18053         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18054         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18055         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18056         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18058 2013-05-06  Alan Modra  <amodra@gmail.com>
18060         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18061         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18062         (DEFAULT_ASM_ENDIAN): Define.
18063         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18064         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18065         Update -K PIC clause from sysv4.h.
18066         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18067         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18069 2013-05-06  Alan Modra  <amodra@gmail.com>
18071         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18072         twice for little-endian.
18073         (ashrdi3_no_power, ashrdi3): Support little-endian.
18075 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18077         PR target/55303
18078         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18079         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18080         related expanders.
18081         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18082         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18083         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18084         New predicates.
18086 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18087             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18089         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18090         * config/pa/pa.opt: Make mbig-switch a no-op.
18091         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18092         (CASE_VECTOR_MODE): Always return SImode.
18093         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18094         for the !TARGET_BIG_SWITCH case.
18095         * config/pa/pa-linux.h: Likewise.
18096         * config/pa/pa-openbsd.h: Likewise.
18097         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18098         * config/pa/pa.md (short_jump): Remove define_insn.
18099         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18100         (casesi0): Remove define_insn.
18101         (type): Remove btable_branch.
18102         (pa_combine_type): Likewise.
18103         (in_nullified_branch_delay): Likewise.
18104         (in_call_delay): Likewise.
18105         (define_delay): Likewise.
18106         (define_insn_reservation "Z3"): Likewise.
18107         (define_insn_reservation "Z4"): Likewise.
18108         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18109         (pa_adjust_insn_length): Remove adjustment for btable branches.
18110         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18111         and mno-big-switch
18113 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18115         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18116         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18117         Add m->r,x alternatives.
18118         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18119         splitters using SWI48x mode iterator.
18120         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18121         TARGET_64BIT.  Add m->x alternative.
18122         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18123         Add o->x alternative.  Enable for TARGET_SSE.
18124         (sse_storeq): Remove expander.
18125         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18126         with memory input operand.
18127         (*vec_extractv2di_1 splitter): New.
18128         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18129         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18131 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18133         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18134         (INT_LOWPART): Delete.
18135         (extract_MB): Adjust.
18136         (extract_ME): Adjust.
18137         (print_operand): Adjust.
18139 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18141         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18142         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18143         (reg_or_logical_cint_operand, easy_fp_constant,
18144         logical_const_operand): Delete "CONST_DOUBLE" case.
18145         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18146         "HOST_BITS_PER_WIDE_INT == 64" test.
18147         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18148         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18149         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18150         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18151         test.
18152         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18153         CONST_DOUBLE DImode/VOIDmode case.
18154         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18155         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18156         CONST_DOUBLE VOIDmode case.
18157         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18158         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18159         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18160         Delete CONST_DOUBLE case.
18161         (splitters for mov FMOVE64 const_double): Delete
18162         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18163         "HOST_BITS_PER_WIDE_INT >= 64" test.
18164         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18165         case.
18166         (mov DI const_double): Delete.
18168 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18170         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18171         on op shows all bits zero in mode of a lowpart subreg, return zero.
18173 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18175         PR target/57150
18176         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18177         to save TFmode registers and DImode to save TImode registers for
18178         caller save operations.
18179         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18180         mark being partially clobbered since they only use the first
18181         double word.
18183         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18184         and TDmode only use the upper 64-bits of each VSX register.
18186 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18188         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18189         (find_candidates_in_block): Disable slsr_process_phi.
18191 2013-05-03  Guozhi Wei  <carrot@google.com>
18193         * coverage.c (coverage_obj_init): Move the construction of gcov
18194         constructor to ...
18195         (build_init_ctor): ... here.
18197 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18199         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18200         (slsr_cand_d): Redefine def_phi.
18201         (stride_status, phi_adjust_status, count_phis_status): New enums.
18202         (find_phi_def): New.
18203         (find_basis_for_base_expr): New.
18204         (find_basis_for_candidate): Handle hidden bases.
18205         (alloc_cand_and_find_basis): Handle phi candidates.
18206         (slsr_process_phi): New.
18207         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18208         (create_mul_imm_cand): Likewise.
18209         (create_add_ssa_cand): Exclude phi base candidates.
18210         (create_add_imm_cand): Likewise.
18211         (slsr_process_cast): Likewise.
18212         (slsr_process_copy): Likewise.
18213         (find_candidates_in_block): Handle phi candidates.
18214         (dump_candidate): Likewise.
18215         (unconditional_cands): Delete.
18216         (unconditional_cands_with_known_stride_p): Delete.
18217         (phi_dependent_cand_p): New.
18218         (cand_increment): Handle phi-dependent candidates.
18219         (replace_dependent): Delete.
18220         (replace_mult_candidate): New.
18221         (replace_unconditional_candidate): New.
18222         (incr_vec_index): Move to avoid forward reference.
18223         (create_add_on_incoming_edge): New.
18224         (create_phi_basis): New.
18225         (replace_dependents): Delete.
18226         (replace_conditional_candidate): New.
18227         (phi_add_costs): New.
18228         (replace_uncond_cands_and_profitable_phis): New.
18229         (record_increment): Handle phi adjustments.
18230         (record_phi_increments): New.
18231         (record_increments): Handle phi adjustments.
18232         (phi_incr_cost): New.
18233         (lowest_cost_path): Handle phis.
18234         (total_savings): Likewise.
18235         (analyze_increments): Likewise.
18236         (ncd_with_phi): New.
18237         (ncd_of_cand_and_phis): New.
18238         (nearest_common_dominator_for_cands): Handle phi increments.
18239         (all_phi_incrs_profitable): New.
18240         (replace_profitable_candidates): Handle phi-dependent candidates.
18241         (analyze_candidates_and_replace): Likewise.
18243 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18245         PR bootstrap/57154
18246         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18247         do not exceed REG_BR_PROB_BASE.
18249 2013-05-03  Jeff Law  <law@redhat.com>
18251         PR tree-optimization/57144
18252         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18253         operand of the condition will bit into the new type when eliminating
18254         a cast feeding a condition.
18256 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18258         PR rtl-optimization/57130
18259         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18260         of COMPARE as in_code to the recursive call if needed.
18262 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18264         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18265         (enabled): Handle new members.
18266         * config/i386/sse.md (*vec_concatv2si): Merge from
18267         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18268         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18270 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18272         PR tree-optimization/57027
18273         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18274         for fnms opportunity, check we got the prerequisite kind
18275         of tree / gimple before using accessor functions.
18277 2013-05-03  Richard Biener  <rguenther@suse.de>
18279         * double-int.h (lshift): New overload without precision
18280         and arith argument.
18281         (operator *=, operator +=, operator -=): Move ...
18282         * double-int.c (operator *=, operator +=, operator -=): ... here
18283         and implement more efficiently.
18284         (mul_double_with_sign): Remove.
18285         (lshift_double): Adjust to take unsinged shift argument, push
18286         dispatching code to callers.
18287         (mul_double_wide_with_sign): Add early out for callers that
18288         are not interested in high parts or overflow.
18289         (lshift): New function.
18290         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18291         dispatch code here.
18292         (lrotate, rrotate): Use logical shifts.
18293         * expr.c (get_inner_reference): Use lshift.
18294         * fixed-value.c (do_fixed_divide): Likewise.
18295         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18296         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18297         (indirect_refs_may_alias_p): Likewise.
18298         (stmt_kills_ref_p_1): Likewise.
18300 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18302         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18304 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18306         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18307         scalar form of FABD instruction.
18309 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18311         * lra-constraints.c (process_alt_operands): Add checking alt
18312         number to choose the best alternative.
18314 2013-05-02  Richard Biener  <rguenther@suse.de>
18316         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18317         bitmap and its handling.
18318         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18320 2013-05-02  Richard Biener  <rguenther@suse.de>
18322         PR middle-end/57140
18323         * tree-inline.c (copy_loops): Properly handle removed loops.
18324         (copy_cfg_body): Mark destination loops for fixup if source
18325         loops needed fixup.
18327 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18329         PR target/56732
18330         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18331         generating simple_return for naked functions.
18333 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18335         PR middle-end/56988
18336         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18337         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18338         flags match.
18339         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18340         ipa_agg_replacement_value structures.
18341         (known_aggs_to_agg_replacement_list): Likewise.
18342         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18343         (read_agg_replacement_chain): Likewise.
18344         (ipcp_transform_function): Also check that by_ref flags match.
18346 2013-05-02  Richard Biener  <rguenther@suse.de>
18348         * graphds.h (struct graph): Add obstack member.
18349         * graphds.c (new_graph): Initialize obstack and allocate
18350         vertices from it.
18351         (add_edge): Allocate edge from the obstack.
18352         (free_graph): Free the obstack instead of all edges and vertices.
18354 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18356         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18357         divides.
18358         * cfg.c (update_bb_profile_for_threading): Ditto.
18359         * tree-inline.c (copy_bb): Ditto.
18360         (copy_edges_for_bb): Ditto.
18361         (initialize_cfun): Ditto.
18362         (copy_cfg_body): Ditto.
18363         (expand_call_inline): Ditto.
18364         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18365         (estimate_node_size_and_time): Ditto.
18366         (inline_merge_summary): Ditto.
18367         * cgraphclones.c (cgraph_clone_edge): Ditto.
18368         (cgraph_clone_node): Ditto.
18369         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18370         (compute_trg_info): Ditto.
18372 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18374         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18375         S reg when fp attribute set.
18376         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18378 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18380         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18381         New pattern.
18382         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18383         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18384         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18386 2013-05-02  Richard Biener  <rguenther@suse.de>
18388         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18389         (struct instantiate_cache_entry): New type.
18390         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18391         (struct instantiate_cache_type): New type.
18392         (set_instantiated_value, get_instantiated_value): Remove.
18393         (get_instantiated_value_entry): New function.
18394         (instantiate_scev_name): Use the new cache and adjust.
18395         (instantiate_scev_poly): Adjust.
18396         (instantiate_scev_binary): Likewise.
18397         (instantiate_array_ref): Likewise.
18398         (instantiate_scev_convert): Likewise.
18399         (instantiate_scev_not): Likewise.
18400         (instantiate_scev_3): Likewise.
18401         (instantiate_scev_2): Likewise.
18402         (instantiate_scev_r): Likewise.
18403         (instantiate_scev): Likewise.
18404         (resolve_mixers): Likewise.
18406 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18408         PR target/57091
18409         * lra-constraints.c (best_small_class_operands_num): Remove.
18410         (process_alt_operands): Remove small_class_operands_num.  Take
18411         small classes operands into losers and only if the operand is not
18412         matched.  Modify debugging output.
18413         (curr_insn_transform): Remove best_small_class_operands_num.
18414         Print insn name.
18416 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18418         * config/aarch64/aarch64-builtins.c
18419         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18420         * config/aarch64/aarch64-simd-builtins.def
18421         (reduc_splus_): Add new modes.
18422         (reduc_uplus_): New.
18423         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18424         (reduc_uplus_v4sf): Likewise.
18425         (reduc_splus_v4sf): Likewise.
18426         (aarch64_addv<mode>): Likewise.
18427         (reduc_uplus_<mode>): Likewise.
18428         (reduc_splus_<mode>): Likewise.
18429         (aarch64_addvv2di): Likewise.
18430         (reduc_uplus_v2di): Likewise.
18431         (reduc_splus_v2di): Likewise.
18432         (aarch64_addvv2si): Likewise.
18433         (reduc_uplus_v2si): Likewise.
18434         (reduc_splus_v2si): Likewise.
18435         (reduc_<sur>plus_<mode>): New.
18436         (reduc_<sur>plus_v2di): Likewise.
18437         (reduc_<sur>plus_v2si): Likewise.
18438         (reduc_<sur>plus_v4sf): Likewise.
18439         (aarch64_addpv4sf): Likewise.
18440         * config/aarch64/arm_neon.h
18441         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18442         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18443         add UNSPEC_SADDV, UNSPEC_UADDV.
18444         (SUADDV): New.
18445         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18447 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18449         * config/aarch64/arm_neon.h
18450         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18452 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18454         * config/aarch64/aarch64-builtins
18455         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18457 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18459         * config/aarch64/aarch64-simd-builtins.def
18460         (reduc_smax_): New.
18461         (reduc_smin_): Likewise.
18462         (reduc_umax_): Likewise.
18463         (reduc_umin_): Likewise.
18464         (reduc_smax_nan_): Likewise.
18465         (reduc_smin_nan_): Likewise.
18466         (fmax): Remove.
18467         (fmin): Likewise.
18468         (smax): Update for V2SF, V4SF and V2DF modes.
18469         (smin): Likewise.
18470         (smax_nan): New.
18471         (smin_nan): Likewise.
18472         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18473         (<su><maxmin><mode>3): ...This, refactor.
18474         (s<maxmin><mode>3): New.
18475         (<maxmin_uns><mode>3): Likewise.
18476         (reduc_<maxmin_uns>_<mode>): Refactor.
18477         (reduc_<maxmin_uns>_v4sf): Likewise.
18478         (reduc_<maxmin_uns>_v2si): Likewise.
18479         (aarch64_<fmaxmin><mode>: Remove.
18480         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18481         new builtin names.
18482         (vmin<q>_f<32,64>): Likewise.
18483         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18484         (FMAXMIN): New.
18485         (su): Add mappings for smax, smin, umax, umin.
18486         (maxmin): New.
18487         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18488         (FMAXMIN): Rename as...
18489         (FMAXMIN_UNS): ...This.
18490         (maxminv): Remove.
18491         (fmaxminv): Likewise.
18492         (fmaxmin): Likewise.
18493         (maxmin_uns): New.
18494         (maxmin_uns_op): Likewise.
18496 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18498         * config/aarch64/arm_neon.h
18499         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18500         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18501         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18503 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18505         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18506         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18508 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18510         * config/aarch64/aarch64-simd.md
18511         (vcond<mode>_internal): Handle special cases for constant masks.
18512         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18513         (vcondu<mode><mode>): Likewise.
18514         (vcond<v_cmp_result><mode>): New.
18516 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18518         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18519         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18520         * config/aarch64/aarch64-simd-builtins.def
18521         (cmeq): Update to BUILTIN_VALLDI.
18522         (cmgt): Likewise.
18523         (cmge): Likewise.
18524         (cmle): Likewise.
18525         (cmlt): Likewise.
18526         * config/aarch64/arm_neon.h
18527         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18528         to builtins or C as appropriate.
18530 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18532         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18533         (cmgeu): ...This.
18534         (cmhi): Rename to...
18535         (cmgtu): ...This.
18536         * config/aarch64/aarch64-simd.md
18537         (simd_mode): Add SF.
18538         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18539         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18540         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18541         (cstore<mode>_neg): ...This.
18542         * config/aarch64/iterators.md
18543         (VALLF): new.
18544         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18545         (COMPARISONS): New.
18546         (UCOMPARISONS): Likewise.
18547         (optab): Add missing comparisons.
18548         (n_optab): New.
18549         (cmp_1): Likewise.
18550         (cmp_2): Likewise.
18551         (CMP): Likewise.
18552         (cmp): Remove.
18553         (VCMP_S): Likewise.
18554         (VCMP_U): Likewise.
18555         (V_cmp_result): Add DF, SF modes.
18556         (v_cmp_result): Likewise.
18557         (v): Likewise.
18558         (vmtype): Likewise.
18559         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18561 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18563         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18564         define_insn to define_insn_and_split.
18565         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18566         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18567         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18568         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18569         (thumb2_negscc): Likewise.
18571 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18573         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18575 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18577         * config/arm/thumb2.md: Remove trailing whitespaces.
18579 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18581         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18582         Use gen_int_mode rather than GEN_INT.
18584 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18586         * value-prof.c (stream_in_histogram_value): Remove the strayed
18587         debug_gimple_stmt.
18589 2013-04-30  Richard Biener  <rguenther@suse.de>
18591         PR middle-end/57122
18592         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18594 2013-04-30  Richard Biener  <rguenther@suse.de>
18596         PR middle-end/57107
18597         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18599 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18601         PR rtl-optimization/56957
18602         PR rtl-optimization/57105
18603         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18604         variable.  Use just INSN_UID for determining whether an insn
18605         should be only disconnected from the insn stream.
18606         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18608 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18610         PR tree-optimization/57104
18611         * tsan.c (instrument_expr): Don't instrument accesses to
18612         DECL_HARD_REGISTER VAR_DECLs.
18614 2013-04-30  Richard Biener  <rguenther@suse.de>
18616         * function.h (loops_for_fn): New inline function.
18617         (set_loops_for_fn): Likewise.
18618         * cfgloop.h (place_new_loop): Add struct function parameter.
18619         (get_loop): Likewise.
18620         (get_loops): Likewise.
18621         (number_of_loops): Likewise.
18622         (fel_next): Adjust.
18623         (fel_init): Likewise.
18624         * cfg.c (get_loop_copy): Adjust.
18625         * cfgloop.c (flow_loops_dump): Likewise.
18626         (record_loop_exits): Likewise.
18627         (verify_loop_structure): Likewise.
18628         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18629         (estimate_reg_pressure_cost): Likewise.
18630         (mark_loop_exit_edges): Likewise.
18631         * cfgloopmanip.c (place_new_loop): Likewise.
18632         (add_loop): Likewise.
18633         (duplicate_loop): Likewise.
18634         * graph.c (draw_cfg_nodes): Likewise.
18635         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18636         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18637         (extract_affine_chrec): Likewise.
18638         (build_scop_iteration_domain): Likewise.
18639         * graphite.c (graphite_initialize): Likewise.
18640         * ira-build.c (create_loop_tree_nodes): Likewise.
18641         (more_one_region_p): Likewise.
18642         (rebuild_regno_allocno_maps): Likewise.
18643         (mark_loops_for_removal): Likewise.
18644         (mark_all_loops_for_removal): Likewise.
18645         (remove_unnecessary_regions): Likewise.
18646         (ira_build): Likewise.
18647         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18648         * loop-init.c (fix_loop_structure): Likewise.
18649         (gate_rtl_move_loop_invariants): Likewise.
18650         (gate_rtl_unswitch): Likewise.
18651         (gate_rtl_unroll_and_peel_loops): Likewise.
18652         (rtl_doloop): Likewise.
18653         * lto-streamer-in.c (input_cfg): Likewise.
18654         * lto-streamer-out.c (output_cfg): Likewise.
18655         * modulo-sched.c (sms_schedule): Likewise.
18656         * predict.c (tree_estimate_probability): Likewise.
18657         (tree_estimate_probability_driver): Likewise.
18658         (estimate_loops): Likewise.
18659         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18660         (move_sese_region_to_fn): Likewise.
18661         (debug_loop_num): Likewise.
18662         * tree-chrec.c (chrec_evaluate): Likewise.
18663         (hide_evolution_in_other_loops_than_loop): Likewise.
18664         (chrec_component_in_loop_num): Likewise.
18665         (reset_evolution_in_loop): Likewise.
18666         (evolution_function_is_invariant_rec_p): Likewise.
18667         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18668         * tree-inline.c (copy_loops): Likewise.
18669         (copy_cfg_body): Likewise.
18670         (tree_function_versioning): Likewise.
18671         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18672         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18673         Likewise.
18674         (add_to_evolution_1): Likewise.
18675         (scev_const_prop): Likewise.
18676         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18677         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18678         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18679         (tree_ssa_lim_initialize): Likewise.
18680         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18681         (verify_loop_closed_ssa): Likewise.
18682         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18683         (tree_ssa_loop_im): Likewise.
18684         (tree_ssa_loop_unswitch): Likewise.
18685         (tree_vectorize): Likewise.
18686         (check_data_deps): Likewise.
18687         (tree_ssa_loop_ivcanon): Likewise.
18688         (tree_ssa_loop_bounds): Likewise.
18689         (tree_complete_unroll): Likewise.
18690         (tree_complete_unroll_inner): Likewise.
18691         (tree_parallelize_loops): Likewise.
18692         (tree_ssa_loop_prefetch): Likewise.
18693         (tree_ssa_loop_ivopts): Likewise.
18694         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18695         * tree-vectorizer.c (vectorize_loops): Likewise.
18697 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18699         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18700         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18701         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18702         with EABI_LINK_SPEC.
18704 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18706         PR target/44578
18707         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18708         alternative.
18710 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18712         PR target/57097
18713         * lra-constraints.c (process_alt_operands): Discourage a bit more
18714         using memory for pseudos.  Print cost dump for alternatives.
18715         Modify cost values for conflicts with early clobbers.
18716         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18718 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18720         PR target/57098
18721         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18723 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18725         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18726         from/to S register.
18727         (movdi_aarch64): Support LDR/STR from/to D register.
18729 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18731         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18732         or higher by default.
18734 2013-04-29  Richard Biener  <rguenther@suse.de>
18736         PR middle-end/57075
18737         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18738         even if not adding abnormal edges for calls that can make
18739         abnormal gotos.
18741 2013-04-29  Richard Biener  <rguenther@suse.de>
18743         PR middle-end/57103
18744         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18745         TREE_BLOCK.
18746         (move_stmt_r): Remove redundant checking.
18748 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18750         PR bootstrap/57077
18751         * basic-block.h (apply_scale): New function.
18752         (apply_probability): Use apply_scale.
18753         * gimple-streamer-in.c (input_bb): Ditto.
18754         * lto-streamer-in.c (input_cfg): Ditto.
18755         * lto-cgraph.c (merge_profile_summaries): Ditto.
18756         * tree-optimize.c (execute_fixup_cfg): Ditto.
18757         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18758         (copy_edges_for_bb): Ditto.
18759         (copy_cfg_body): Ditto.
18761 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18763         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18764         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18765         (tail_merge_optimize): Handle current_loops == NULL.
18767 2013-04-26  Jeff Law  <law@redhat.com>
18769         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18770         (simplify_cond_using_ranges): Generalize code to simplify
18771         COND_EXPRs where one argument is a constant and the other
18772         is an SSA_NAME created by an integral type conversion.
18774 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18776         * config/arm/arm.md (store_minmaxsi): Use only when
18777         optimize_insn_for_size_p.
18779 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18781         PR target/57108
18782         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18784 2013-04-29  Richard Biener  <rguenther@suse.de>
18786         PR middle-end/57089
18787         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18788         loop tree make sure to schedule a fixup for the child as well.
18789         (expand_omp_for_generic): Properly add loops.
18790         (expand_omp_for_static_nochunk): Likewise.
18791         (expand_omp_for_static_chunk): Likewise.
18792         (expand_omp_for): For the degenerate case fixup loops.
18793         (expand_omp_sections): Fix default bb placement in loops.
18794         (expand_omp_atomic_pipeline): Properly add loops.
18796 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18798         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18800 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18802         * tree-ssa-tail-merge.c: Update header comment.
18804 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18806         * config/aarch64/arm_neon.h
18807         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18808         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18809         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18810         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18811         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18812         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18813         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18814         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18816 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18818         * config/aarch64/aarch64-simd.md
18819         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18820         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18821         fix_trunc, fixuns_trunc.
18822         (ftrunc<VDQF:mode>2): New.
18823         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18824         (fix_trunc_optab): New.
18826 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18828         * config/aarch64/aarch64-builtins.c
18829         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18830         iceilf, lround, iroundf.
18832 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18834         PR target/54349
18835         * config/i386/i386.h (enum ix86_tune_indices)
18836         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18837         New, split from X86_TUNE_INTER_UNIT_MOVES.
18838         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18839         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18840         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18841         (TARGET_INTER_UNIT_MOVES): Remove.
18842         * config/i386/i386.c (initial_ix86_tune_features): Update.
18843         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18844         (ix86_expand_convert_uns_didf_sse): Use
18845         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18846         (ix86_expand_vector_init_one_nonzero): Ditto.
18847         (ix86_expand_vector_init_interleave): Ditto.
18848         (inline_secondary_memory_needed): Return true for moves from SSE class
18849         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18850         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18851         * config/i386/constraints.md (Yi, Ym): Depend on
18852         TARGET_INTER_UNIT_MOVES_TO_VEC.
18853         (Yj, Yn): New constraints.
18854         * config/i386/i386.md (*movdi_internal): Change constraints of
18855         operand 1 from Yi to Yj and from Ym to Yn.
18856         (*movsi_internal): Ditto.
18857         (*movdf_internal): Ditto.
18858         (*movsf_internal): Ditto.
18859         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18860         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18861         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18862         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18863         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18864         * config/i386/sse.md (movdi_to_sse): Ditto.
18865         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18866         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18867         TARGET_INTER_UNIT_MOVES.
18868         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18869         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18870         instead of TARGET_INTER_UNIT_MOVES.
18871         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18872         operand 1 from Yi to Yj and from Ym to Yn.
18874 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18876         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18877         (float_truncate_hi_): Likewise.
18878         (float_extend_lo_): Likewise.
18879         (float_truncate_lo_): Likewise.
18880         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18881         (aarch64_float_extend_lo_v2df): Likewise.
18882         (vec_unpacks_hi_v4sf): Likewise.
18883         (aarch64_float_truncate_lo_v2sf): Likewise.
18884         (aarch64_float_truncate_hi_v4sf): Likewise.
18885         (vec_pack_trunc_v2df): Likewise.
18886         (vec_pack_trunc_df): Likewise.
18888 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18890         * config/aarch64/aarch64-builtins.c
18891         (aarch64_fold_builtin): Fold float conversions.
18892         * config/aarch64/aarch64-simd-builtins.def
18893         (floatv2si, floatv4si, floatv2di): New.
18894         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18895         * config/aarch64/aarch64-simd.md
18896         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18897         * config/aarch64/iterators.md (FLOATUORS): New.
18898         (optab): Add float, floatuns.
18899         (su_optab): Likewise.
18901 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18903         * config/aarch64/aarch64-builtins.c
18904         (aarch64_builtin_vectorized_function): Use new names for
18905         fcvt builtins.
18906         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18907         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18908         (fcvtzu): Split as...
18909         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18910         (fcvtas): Split as...
18911         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18912         (fcvtau): Split as...
18913         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18914         (fcvtps): Split as...
18915         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18916         (fcvtpu): Split as...
18917         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18918         (fcvtms): Split as...
18919         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18920         (fcvtmu): Split as...
18921         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18922         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18923         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18924         (lfrintnusf, lfrintnudf): Likewise.
18925         * config/aarch64/aarch64-simd.md
18926         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18927         define_insn.
18928         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18929         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18930         (fcvt_pattern): Likewise.
18932 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18934         * config/aarch64/aarch64-simd.md
18935         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18936         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18938 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18940         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18941         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18942         (vrnd<a,m,n,p>_f32): Implement using builtins.
18943         (vrnd<i,x><q>_f<32, 64>): New.
18945 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18947         * config/aarch64/aarch64-builtins.c
18948         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18949         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18950         (frintz): Rename to...
18951         (btrunc): ...this.
18952         (frintp): Rename to...
18953         (ceil): ...this.
18954         (frintm): Rename to...
18955         (floor): ...this.
18956         (frinti): Rename to...
18957         (nearbyint): ...this.
18958         (frintx): Rename to...
18959         (rint): ...this.
18960         (frinta): Rename to...
18961         (round): ...this.
18962         * config/aarch64/aarch64-simd.md
18963         (aarch64_frint<frint_suffix><mode>): Delete.
18964         (<frint_pattern><mode>2): Convert to insn.
18965         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18966         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18967         (frint_pattern): Likewise.
18968         (frint_suffix): Likewise.
18970 2013-04-29  Richard Biener  <rguenther@suse.de>
18972         PR tree-optimization/57081
18973         * loop-init.c: Include tree-flow.h.
18974         (loop_optimizer_finalize): Free number of iteration estimates.
18975         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18977 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18979         PR tree-optimization/57083
18980         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18981         non-singleton shift count range, zero extend low_bound for uns case.
18983         * config/i386/predicates.md (general_vector_operand): New predicate.
18984         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18985         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18986         if they aren't nonimmediate operands.  If their original values
18987         satisfy const_vector_equal_evenodd_p, don't shift them.
18988         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18989         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18990         if not nonimmediate_operand.
18991         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18992         instead of register_operand.
18993         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18995 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18997         * stor-layout.c (finalize_size_functions): Allocate a structure and
18998         reset cfun before dumping the functions.
19000 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
19002         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
19004         PR target/56866
19005         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
19006         use xop_pmacsdqh if uns_p.
19007         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
19008         the immediate rotate count.
19010 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
19012         * rtl.h (struct rtx_def): Add comment for field jump.
19013         (LRA_SUBREG_P): New macro.
19014         * recog.c (register_operand): Check LRA_SUBREG_P.
19015         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19016         stack frame.
19017         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19018         (lra_final_code_change): Skip subreg change for operators.
19019         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19020         if there are no operand changes.
19021         * lra-constraints.c (curr_insn_set): New.
19022         (match_reload): Set LRA_SUBREG_P.
19023         (emit_spill_move): Ditto.
19024         (check_and_process_move): Use curr_insn_set. Process only single
19025         set insns.  Don't initialize sec_mem_p and change_p.
19026         (simplify_operand_subreg): Use LRA_SUBREG_P.
19027         (reg_in_class_p): New function.
19028         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19029         of #ifdef.  Add code to remove cycling.
19030         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19031         non-null disp.  Reload inner instead of disp when base and index
19032         are null.  Try to put lo_sum into register.
19033         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19034         (check_and_process_move): Move code for move cost check to
19035         simple_move_p.  Remove equiv_substitution.
19036         (simple_move_p): New function.
19037         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19038         curr_insn_set.  Call check_and_process_move only for single set
19039         insns.  Use the new function.  Move call of check_and_process_move
19040         after operand equiv substitution and address process.
19042 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19044         PR go/57045
19045         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19046         with nonlocal goto receivers or returns twice calls, ignore
19047         unininitialized values from abnormal edges to nl goto receiver
19048         or returns twice call.
19050 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19052         PR tree-optimization/57051
19053         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19054         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19055         bitsize.
19057 2013-04-26  Richard Biener  <rguenther@suse.de>
19059         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19060         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19061         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19062         (fixup_loop_arrays_after_move): New function.
19063         (move_sese_region_to_fn): Properly outline the loop tree parts
19064         of the SESE region.
19066 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19068         * config/i386/i386.md (type, unit): Fix long lines.
19070 2013-04-26  Richard Biener  <rguenther@suse.de>
19072         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19073         (lto-streamer-out.o): Likewise.
19074         * cfgloop.c (init_loops_structure): Export, add struct function
19075         argument and adjust.
19076         (flow_loops_find): Adjust.
19077         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19078         (init_loops_structure): Declare.
19079         * lto-streamer-in.c: Include cfgloop.h.
19080         (input_cfg): Input the loop tree.
19081         * lto-streamer-out.c: Include cfgloop.h.
19082         (output_cfg): Output the loop tree.
19083         (output_struct_function_base): Do not drop PROP_loops.
19085 2013-03-26  Richard Biener  <rguenther@suse.de>
19087         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19088         (pass_build_cfg): Provide PROP_loops.
19089         (move_sese_region_to_fn): Remove loops that are outlined into fn
19090         for now.
19091         * tree-inline.c: Include cfgloop.h.
19092         (initialize_cfun): Do not drop PROP_loops.
19093         (copy_loops): New function.
19094         (copy_cfg_body): Copy loop structure.
19095         (tree_function_versioning): Initialize destination loop tree.
19096         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19097         (pass_parallelize_loops): Do IL verification.
19098         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19099         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19100         the CFG make sure we fixup loops as well.
19101         * tree-ssa-tail-merge.c: Include cfgloop.h.
19102         (replace_block_by): When merging loop latches mark loops for fixup.
19103         * lto-streamer-out.c (output_struct_function_base): Drop
19104         PROP_loops for now.
19105         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19106         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19107         * ipa-split.c: Include cfgloop.h.
19108         (split_function): Add the new return block to the loop tree root.
19109         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19110         whether we have removed the forwarder block.
19111         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19112         * cfgloop.h (place_new_loop): Declare.
19113         * cfgloopmanip.c (place_new_loop): Export.
19114         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19115         (tree-switch-conversion.o): Likewise.
19116         (tree-complex.o): Likewise.
19117         (tree-inline.o): Likewise.
19118         (tree-ssa-tailmerge.o): Likewise.
19119         (ipa-split.o): Likewise.
19120         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19121         (tree-ssa-copy.o): Likewise.
19122         * tree-switch-conversion.c: Include cfgloop.h
19123         (process_switch): If we emit a bit-test cascade, schedule loops
19124         for fixup.
19125         * tree-complex.c: Include cfgloop.h.
19126         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19127         * asan.c: Include cfgloop.h.
19128         (create_cond_insert_point): Properly add new basic-blocks to
19129         loops, schedule loop fixup.
19130         * cfgloop.c (verify_loop_structure): Check that looks are not
19131         marked for fixup.
19132         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19133         to loops.
19134         (expand_omp_for_generic): Likewise.
19135         (expand_omp_sections): Likewise.
19136         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19137         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19138         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19139         is initialized, not when loops are present.
19140         * tree-parloops.c (parallelize_loops): Remove checking here.
19141         * passes.c (init_optimization_passes): Schedule a copy-propagation
19142         pass before complete unrolling of inner loops.
19144 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19146         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19147         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19148         (colorize_init): Add argument to _WIN32 version.
19149         * toplev.c: Include diagnostic-color.h.
19150         (process_options): Default to -fdiagnostics-color=auto if
19151         GCC_COLORS env var is in the environment.
19152         * common.opt (fdiagnostics-color=): Add Var and Init.
19153         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19154         env var is in the environment, the default is auto rather than never.
19156         * diagnostic.h (file_name_as_prefix): Add context argument.
19157         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19158         the string as locus.
19159         * langhooks.c (lhd_print_error_function): Adjust caller.
19161 2013-04-25  Lawrence Crowl  <crowl@google.com>
19163         * var-tracking.c (shared_hash_def::htab):
19164         Change type to hash_table.  Update dependent calls and types.
19166 2013-04-25  Lawrence Crowl  <crowl@google.com>
19168         * Makefile.in: Update as needed below.
19170         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19171         Move declaration to after the type's method definitons.
19173         * attribs.c (htab_t scoped_attributes::attribute_hash):
19174         Change type to hash_table.  Update dependent calls and types.
19176         * bitmap.c (htab_t bitmap_desc_hash):
19177         Change type to hash_table.  Update dependent calls and types.
19179         * cselib.c (htab_t cselib_hash_table):
19180         Change type to hash_table.  Update dependent calls and types.
19182         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19183         (hash_string_slot_node): Move implementation into lto-streamer.h
19184         struct string_slot_hasher.
19185         (eq_string_slot_node): Likewise.
19187         * data-streamer-out.c: Update output_block::string_hash_table
19188         dependent calls and types.
19190         * dwarf2cfi.c (htab_t trace_index):
19191         Change type to hash_table.  Update dependent calls and types.
19193         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19194         Change type to hash_table.  Update dependent calls and types.
19195         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19196         (htab_t optimize_external_refs::map): Likewise.
19197         (htab_t output_comp_unit::extern_map): Likewise.
19198         (htab_t output_comdat_type_unit::extern_map): Likewise.
19199         (htab_t output_macinfo::macinfo_htab): Likewise.
19200         (htab_t optimize_location_lists::htab): Likewise.
19201         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19203         * except.c (htab_t ehspec_hash_type):
19204         Change type to hash_table.  Update dependent calls and types.
19205         (assign_filter_values::ttypes): Likewise.
19206         (assign_filter_values::ehspec): Likewise.
19207         (sjlj_assign_call_site_values::ar_hash): Likewise.
19208         (convert_to_eh_region_ranges::ar_hash): Likewise.
19210         * gcse.c (htab_t pre_ldst_table):
19211         Change type to hash_table.  Update dependent calls and types.
19213         * ggc-common.c (htab_t saving_htab):
19214         Change type to hash_table.  Update dependent calls and types.
19215         (htab_t loc_hash): Likewise.
19216         (htab_t ptr_hash): Likewise.
19217         (call_count): Rename ggc_call_count.
19218         (call_alloc): Rename ggc_call_alloc.
19219         (loc_descriptor): Rename make_loc_descriptor.
19220         (add_statistics): Rename ggc_add_statistics.
19222         * ggc-common.c (saving_htab):
19223         Change type to hash_table.  Update dependent calls and types.
19225         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19226         (push_gimplify_context): Likewise.
19227         (pop_gimplify_context): Likewise.
19228         (struct gimple_temp_hash_elt): Added.
19229         (struct gimplify_hasher): Likewise.
19230         (struct gimplify_ctx.temp_htab):
19231         Change type to hash_table.  Update dependent calls and types.
19233         * gimple-fold.c: Include gimplify-ctx.h.
19235         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19236         Change type to hash_table.  Update dependent calls and types.
19237         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19238         avoid potential global name collision.
19240         * gimplify.c: Include gimplify-ctx.h.
19241         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19242         (htab_t gimplify_ctx::temp_htab):
19243         Update dependent calls and types for new type hash_table.
19244         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19245         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19247         * gimplify-ctx.h: New.
19248         (struct gimple_temp_hash_elt): Move from gimplify.c.
19249         (class gimplify_hasher): New.
19250         (struct gimplify_ctx): Move from gimple.h.
19251         (htab_t gimplify_ctx::temp_htab):
19252         Change type to hash_table.  Update dependent calls and types.
19254         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19255         (htab_t ivs_params::newivs_index):
19256         Change type to hash_table.  Update dependent calls and types.
19257         (htab_t ivs_params::params_index): Likewise.
19258         (htab_t print_generated_program::params_index): Likewise.
19259         (htab_t gloog::newivs_index): Likewise.
19260         (htab_t gloog::params_index): Likewise.
19262         * graphite.c: Include graphite-htab.h.
19263         4htab_t graphite_transform_loops::bb_pbb_mapping):
19264         Change type to hash_table.  Update dependent calls and types.
19266         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19267         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19268         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19270         * graphite-dependences.c: Include graphite-htab.h.
19271         (loop_is_parallel_p): Change hash table type of parameter.
19273         * graphite-htab.h: New.
19274         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19275         (extern find_pbb_via_hash): Move from graphite-poly.h.
19276         (extern loop_is_parallel_p): Move from graphite-poly.h.
19277         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19279         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19280         (extern loop_is_parallel_p): Move to graphite-htab.h.
19281         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19283         * haifa-sched.c (htab_t delay_htab):
19284         Change type to hash_table.  Update dependent calls and types.
19285         (htab_t delay_htab_i2): Likewise.
19287         * ira-color.c (htab_t allocno_hard_regs_htab):
19288         Change type to hash_table.  Update dependent calls and types.
19290         * ira-costs.c (htab_t cost_classes_htab):
19291         Change type to hash_table.  Update dependent calls and types.
19293         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19294         Change type to hash_table.  Update dependent calls and types.
19296         * loop-iv.c (htab_t bivs):
19297         Change type to hash_table.  Update dependent calls and types.
19299         * loop-unroll.c (htab_t opt_info::insns_to_split):
19300         Change type to hash_table.  Update dependent calls and types.
19301         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19303         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19304         (struct string_slot_hasher): New.
19305         (htab_t output_block::string_hash_table):
19306         Change type to hash_table.  Update dependent calls and types.
19308         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19309         (htab_t file_name_hash_table):
19310         Change type to hash_table.  Update dependent calls and types.
19312         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19313         calls and types.
19315         * lto-streamer.c (htab_t tree_htab):
19316         Change type to hash_table.  Update dependent calls and types.
19318         * omp-low.c: Include gimplify-ctx.h.
19320         * passes.c (htab_t name_to_pass_map):
19321         Change type to hash_table.  Update dependent calls and types.
19322         (pass_traverse): Rename to passes_pass_traverse.
19324         * plugin.c (htab_t event_tab):
19325         Change type to hash_table.  Update dependent calls and types.
19327         * postreload-gcse.c (htab_t expr_table):
19328         Change type to hash_table.  Update dependent calls and types.
19329         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19331         * sese.c (debug_rename_map_1): Make extern.
19332         (htab_t copy_bb_and_scalar_dependences::rename_map):
19333         Change type to hash_table.  Update dependent calls and types.
19335         * sese.h (extern debug_rename_map): Move to .c file.
19337         * store-motion.c (htab_t store_motion_mems_table):
19338         Change type to hash_table.  Update dependent calls and types.
19340         * trans-mem.c (htab_t tm_new_mem_hash):
19341         Change type to hash_table.  Update dependent calls and types.
19343         * tree-browser.c (htab_t TB_up_ht):
19344         Change type to hash_table.  Update dependent calls and types.
19346         * tree-cfg.c (htab_t discriminator_per_locus):
19347         Change type to hash_table.  Update dependent calls and types.
19349         * tree-complex.c: Include tree-hasher.h
19350         (htab_t complex_variable_components):
19351         Change type to hash_table.  Update dependent calls and types.
19353         * tree-eh.c (htab_t finally_tree):
19354         Change type to hash_table.  Update dependent calls and types.
19356         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19357         struct int_tree_hasher.
19358         (extern int_tree_map_eq): Likewise.
19359         (uid_decl_map_hash): Removed.
19360         (extern decl_tree_map_eq): Likewise.
19362         * tree-hasher.h: New.
19363         (struct int_tree_hasher): New.
19364         (typedef int_tree_htab_type): New.
19366         * tree-inline.c: Include gimplify-ctx.h.
19368         * tree-mudflap.c: Include gimplify-ctx.h.
19370         * tree-parloops.c: Include tree-hasher.h.
19371         (htab_t eliminate_local_variables_stmt::decl_address):
19372         Change type to hash_table.  Update dependent calls and types.
19373         (htab_t separate_decls_in_region::decl_copies): Likewise.
19375         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19376         Change type to hash_table.  Update dependent calls and types.
19378         * tree-sra.c (candidates):
19379         Change type to hash_table.  Update dependent calls and types.
19381         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19382         in tree-flow.h.
19383         (int_tree_map_hash): Likewise.
19385         * tree-ssa-dom.c (htab_t avail_exprs):
19386         Change type to hash_table.  Update dependent calls and types.
19388         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19389         Change type to hash_table.  Update dependent calls and types.
19391         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19392         Change type to hash_table.  Update dependent calls and types.
19394         * tree-ssa-phiopt.c (seen_ssa_names):
19395         Change type to hash_table.  Update dependent calls and types.
19397         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19398         Change type to hash_table.  Update dependent calls and types.
19400         * tree-ssa-uncprop.c (equiv):
19401         Change type to hash_table.  Update dependent calls and types.
19403 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19405         PR rtl-optimization/57003
19406         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19407         call note_stores with kill_clobbered_value callback again after
19408         killing regs_invalidated_by_call.
19410 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19412         * config/aarch64/aarch64-simd.md
19413         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19414         (aarch64_simd_bsl<mode>): Likewise.
19415         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19417 2013-04-25  Marek Polacek  <polacek@redhat.com>
19419         PR tree-optimization/57066
19420         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19422 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19424         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19426 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19428         * config/aarch64/aarch64-builtins.c
19429         (aarch64_fold_builtin): New.
19430         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19431         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19432         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19433         * config/aarch64/arm_neon.h
19434         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19436 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19437             Tejas Belagod  <tejas.belagod@arm.com>
19439         * config/aarch64/aarch64-builtins.c
19440         (aarch64_gimple_fold_builtin): New.
19441         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19442         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19443         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19444         (addvv4sf): Update.
19445         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19447 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19449         * config/aarch64/aarch64.md
19450         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19452 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19454         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19455         (*ngcsi_uxtw): New pattern.
19457 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19458             Julian Brown  <julian@codesourcery.com>
19460         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19461         (TB_DREG): Add T_V4HF.
19462         (v4hf_UP): New macro.
19463         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19464         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19465         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19466         built-ins.
19467         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19468         (arm_vector_mode_supported_p): Handle V4HF.
19469         (arm_mangle_map): Handle V4HFmode.
19470         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19471         * config/arm/arm_neon_builtins.def: Add entries for
19472         vcvtv4hfv4sf, vcvtv4sfv4hf.
19473         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19474         (neon_vcvtv4hfv4sf): Likewise.
19475         * config/arm/neon-gen.ml: Handle half-precision floating point
19476         features.
19477         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19478         * config/arm/arm_neon.h: Regenerate.
19479         * config/arm/neon.ml (type elts): Add F16.
19480         (type vectype): Add T_float16x4, T_floatHF.
19481         (type vecmode): Add V4HF.
19482         (type features): Add Requires_FP_bit feature.
19483         (elt_width): Handle F16.
19484         (elt_class): Likewise.
19485         (elt_of_class_width): Likewise.
19486         (mode_of_elt): Refactor.
19487         (type_for_elt): Handle F16, fix error messages.
19488         (vectype_size): Handle T_float16x4.
19489         (vcvt_sh): New function.
19490         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19491         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19492         (string_of_mode): Handle V4HF.
19493         * doc/arm-neon-intrinsics.texi: Regenerate.
19495 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19497         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19498         format specifier in 'X' case.
19500 2013-04-25  Alan Modra  <amodra@gmail.com>
19502         PR target/57052
19503         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19504         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19505         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19506         Repeat for many other rotate/shift and mask patterns using subregs.
19507         Name lshiftrt insns.
19508         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19509         on WORDS_BIG_ENDIAN.
19511 2013-04-25  Alan Modra  <amodra@gmail.com>
19513         * config.gcc: Support little-endian powerpc-linux targets.
19514         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19515         (LINK_OS_LINUX_SPEC): Define.
19516         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19517         Preserve MASK_LITTLE_ENDIAN.
19518         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19519         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19520         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19521         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19522         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19523         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19524         Correct fp word order for little-endian.  Don't shift toc entries
19525         smaller than a word for little-endian.
19526         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19527         (bswapdi2 splits): Correct low-part subreg for little-endian.
19528         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19529         low/high where such is correct only for be.
19530         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19531         little-endian for -mcall-aixdesc.
19533 2013-04-25  Alan Modra  <amodra@gmail.com>
19535         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19536         replace_equiv_address_nv.
19538 2013-04-25  Alan Modra  <amodra@gmail.com>
19540         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19542 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19544         Revert:
19545         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19546         * rtl.h (struct rtx_def): ...
19548 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19550         PR rtl-optimizations/57046
19551         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19552         for multi-reg splits.
19554 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19556         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19558 2013-04-24  Sterling Augustine  <saugustine@google.com>
19560         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19561         (comp_dir_string, debug_str_dwo_section): New.
19562         (DEBUG_STR_DWO_SECTION): Rename to ...
19563         (DEBUG_DWO_STR_SECTION): ... this.
19564         (DEBUG_NORM_STR_SECTION): Delete.
19565         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19566         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19567         (find_AT_string): Move most logic to ...
19568         (find_AT_string_in_table): ... here.  New.
19569         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19570         add_skeleton_AT_string.  Delete logic.
19571         (output_skeleton_debug_sections): Remove call to
19572         add_top_level_skeleton_die_attrs.
19573         (add_comp_dir_attribute): Move logic to comp_dir_string.
19574         (dwarf2out_init): Initialize debug_str_dwo_section.
19575         (output_indirect_string): Call find_string_form.
19576         (output_indirect_strings): Rewrite.
19577         (prune_unused_types): Empty skeleton_debug_str_hash.
19578         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19579         (dwarf2out_finish):  Call output_indirect_strings.
19581 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19583         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19585 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19587         * rtl.h (struct rtx_def): Add comment for field jump.
19588         (LRA_SUBREG_P): New macro.
19589         * recog.c (register_operand): Check LRA_SUBREG_P.
19590         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19591         stack frame.
19592         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19593         (lra_final_code_change): Skip subreg change for operators.
19594         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19595         if there are no operand changes.
19596         * lra-constraints.c (curr_insn_set): New.
19597         (match_reload): Set LRA_SUBREG_P.
19598         (emit_spill_move): Ditto.
19599         (check_and_process_move): Use curr_insn_set. Process only single
19600         set insns.  Don't initialize sec_mem_p and change_p.
19601         (simplify_operand_subreg): Use LRA_SUBREG_P.
19602         (reg_in_class_p): New function.
19603         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19604         of #ifdef.  Add code to remove cycling.
19605         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19606         non-null disp.  Reload inner instead of disp when base and index
19607         are null.  Try to put lo_sum into register.
19608         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19609         (check_and_process_move): Move code for move cost check to
19610         simple_move_p.  Remove equiv_substitution.
19611         (simple_move_p): New function.
19612         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19613         curr_insn_set.  Call check_and_process_move only for single set
19614         insns.  Use the new function.  Move call of check_and_process_move
19615         after operand equiv substitution and address process.
19617 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19619         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19620         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19621         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19623 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19625         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19627 2013-04-24  Marek Polacek  <polacek@redhat.com>
19629         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19630         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19631         (select_loops_exit_conditions): Likewise.
19632         (number_of_iterations_for_all_loops): Likewise.
19633         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19634         (scev_analysis): Likewise.
19636 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19637             Chao-ying Fu  <fu@mips.com>
19639         * config/mips/micromips.md (jraddiusp): New pattern.
19640         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19641         instruction if possible.
19643 2013-04-24  Alan Modra  <amodra@gmail.com>
19645         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19647 2013-04-24  Julian Brown  <julian@codesourcery.com>
19648             Chung-Lin Tang  <cltang@codesourcery.com>
19650         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19651         dependency behavior in enumeration type DIE generation. Add TODO note
19652         to comments about future DW_FORM_sdata/udata re-work of related code.
19654 2013-04-23  Lawrence Crowl  <crowl@google.com>
19656         * Makefile.in: Update as needed below.
19658         * hash-table.h (class hash_table):
19659         Correct many methods with parameter types compare_type to the correct
19660         value_type.  (Correct code was unlikely to notice the change.)
19661         (hash_table::elements_with_deleted) New.
19662         (class hashtable::iterator): New.
19663         (hashtable::begin()): New.
19664         (hashtable::end()): New.
19665         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19667         * statistics.c (statistics_hashes):
19668         Change type to hash_table.  Update dependent calls and types.
19670         * tree-into-ssa.c (var_infos):
19671         Change type to hash_table.  Update dependent calls and types.
19673         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19674         Change type to hash_table.  Update dependent calls and types.
19676         * tree-ssa-loop-im.c (struct mem_ref.refs):
19677         Change type to hash_table.  Update dependent calls and types.
19679         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19680         Change type to hash_table.  Update dependent calls and types.
19682         * tree-ssa-sccvn.c (vn_tables_s::nary):
19683         Change type to hash_table.  Update dependent calls and types.
19684         (vn_tables_s::phis): Likewise.
19685         (vn_tables_s::references): Likewise.
19687         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19688         (vn_reference_eq): Update parameter and return types.
19690         * tree-ssa-structalias.c (pointer_equiv_class_table):
19691         Change type to hash_table.  Update dependent calls and types.
19692         (location_equiv_class_table): Likewise.
19694         * tree-vect-data-refs.c: Consequential changes for making
19695         peeling a hash_table.
19697         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19698         (destroy_loop_vec_info): Dependent hash_table update.
19700         * tree-vectorizer.h (peeling_htab):
19701         Change type to hash_table.  Update dependent calls and types.
19703 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19705         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19706         to check the register content is equal or not.
19707         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19708         to assign register content record.
19709         * lra-eliminations.c (update_reg_eliminate): Use
19710         lra_update_reg_val_offset to update register content offset.
19711         * lra-int.h (struct lra_reg): Add offset member.
19712         (lra_reg_val_equal_p): New static inline function.
19713         (lra_update_reg_val_offset): New static inline function.
19714         (lra_assign_reg_val): New static inline function.
19715         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19716         to assign register content record.
19717         (initialize_lra_reg_info_element): Initial offset to zero.
19719 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19721         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19722         operands.  Record compression.
19724 2013-04-23  Xinliang David Li  <davidxl@google.com>
19726         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19728 2013-04-23  Richard Biener  <rguenther@suse.de>
19730         PR middle-end/57036
19731         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19732         parameter, only add abnormal goto edges from the copied body
19733         if the call could perform abnormal gotos.
19734         (copy_cfg_body): Adjust.
19736 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19738         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19740 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19742         * coretypes.h (gimple_stmt_iterator): Add struct to make
19743         compatible with C.
19745 2013-04-23  Richard Biener  <rguenther@suse.de>
19747         PR tree-optimization/57026
19748         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19749         from SSA names occuring in abnormal PHI nodes.
19751 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19753         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19755 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19757         * lto/lto.c (print_lto_report_1): Declare early.
19758         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19760 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19762         * common.opt (-flto-report-wpa): Add.
19763         * doc/invoke.texi (-flto-report-wpa): Add.
19764         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19765         (lto_main): dito.
19767 2013-04-22  Xinliang David Li  <davidxl@google.com>
19769         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19770         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19771         * Makefile.in: New dependency
19773             David Daney <ddaney.cavm@gmail.com>
19775         * configure.ac (gcc_cv_as_micromips_support):  Use the
19776         --fatal-warnings option.
19777         * configure: Regenerate.
19779 2013-04-22  Marek Polacek  <polacek@redhat.com>
19781         PR sanitizer/56990
19782         * tsan.c (instrument_expr): Don't instrument expression
19783         in case its size is zero.
19785 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19787         PR target/57032
19788         Revert:
19789         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19791         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19793 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19795         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19796         (gimple_stmt_iterator): New typedef.
19797         * gimple.h (gimple_stmt_iterator): Rename to...
19798         (gimple_stmt_iterator_d): ... This.
19799         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19800         trees be valid for GIMPLE and GENERIC.
19801         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19802         * gimple-fold.c (gimple_fold_call): Call target hook
19803         gimple_fold_builtin.
19804         * hooks.c (hook_bool_gsiptr_false): New.
19805         * hooks.h (hook_bool_gsiptr_false): New.
19806         * target.def (fold_stmt): New.
19807         * doc/tm.texi: Regenerate.
19809 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19811         PR target/57018
19812         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19813         a set sp if no stack realignment.
19815 2013-04-22  Nick Clifton  <nickc@redhat.com>
19817         * config.gcc (tilegx-linux): Extend extra_objs rather than
19818         overwriting it.
19819         (tilepro-linux): Likewise.
19821 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19823         * config/aarch64/aarch64-builtins.c
19824         (CF): Remove.
19825         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19826         (VAR<1-12>): Add MAP parameter.
19827         (BUILTIN_*): Likewise.
19828         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19829         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19830         (aarch64_ushl_n<mode>): Likewise.
19831         (aarch64_sshr_n<mode>): Likewise.
19832         (aarch64_ushr_n<mode>): Likewise.
19833         (aarch64_<maxmin><mode>): Likewise.
19834         (aarch64_sqrt<mode>): Likewise.
19835         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19836         (vshr<q>_n_*): Likewise.
19838 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19840         * config/aarch64/aarch64-builtins.c
19841         (aarch64_simd_builtin_type_mode): Handle SF types.
19842         (sf_UP): Define.
19843         (BUILTIN_GPF): Define.
19844         (aarch64_init_simd_builtins): Handle SF types.
19845         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19846         (frecps): Likewise.
19847         (frecpx): Likewise.
19848         * config/aarch64/aarch64-simd.md
19849         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19850         (aarch64_frecpe<mode>): New.
19851         (aarch64_frecps<mode>): Likewise.
19852         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19853         (v8type): Add frecp<esx>.
19854         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19855         (aarch64_frecps<mode>): Likewise.
19856         * config/aarch64/iterators.md (FRECP): New.
19857         (frecp_suffix): Likewise.
19858         * config/aarch64/arm_neon.h
19859         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19861 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19863         PR target/56995
19864         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19865         (REG_CLASS_NAMES): Idem.
19866         (REG_CLASS_CONTENTS): Idem.
19867         (REGCLASS_HAS_FP_REG): Idem.
19868         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19869         (sh_conditional_register_usage): Idem.
19871 2013-04-21  Jeff Law  <law@redhat.com>
19873         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19874         (ssa_forward_propagate_and_combine): Use it.
19876 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19878         * lra.c: Update the flow chart diagram.
19880 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19882         PR rtl-optimization/56847
19883         * lra-constraints.c (process_alt_operands): Discourage alternative
19884         with non-matche doffsettable memory constraint fro memory with
19885         known offset.
19887 2013-04-19  Richard Biener  <rguenther@suse.de>
19889         PR tree-optimization/56982
19890         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19891         function.
19892         * gimplify.c (gimplify_call_expr): Notice special calls.
19893         (gimplify_modify_expr): Likewise.
19894         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19895         abnormal control flow receivers.
19896         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19897         in the same way as cfun->has_nonlocal_labels.
19898         (gimple_purge_dead_abnormal_call_edges): Likewise.
19899         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19900         receivers start a basic-block.
19902 2013-04-19  Richard Biener  <rguenther@suse.de>
19904         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19905         member ...
19906         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19907         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19908         (SLP_TREE_LOAD_PERMUTATION): Add.
19909         (vect_transform_slp_perm_load): Adjust prototype.
19910         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19911         (vect_free_slp_instance): Likewise.
19912         (vect_create_new_slp_node): Likewise.
19913         (vect_supported_slp_permutation_p): Remove.
19914         (vect_slp_rearrange_stmts): Adjust.
19915         (vect_supported_load_permutation_p): Likewise.  Inline
19916         vect_supported_slp_permutation_p here.
19917         (vect_analyze_slp_instance): Compute load permutations per
19918         slp node instead of per instance.
19919         (vect_get_slp_defs): Adjust.
19920         (vect_transform_slp_perm_load): Likewise.
19921         (vect_schedule_slp_instance): Remove redundant code.
19922         (vect_schedule_slp): Remove hack for PR56270, add it ...
19923         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19924         CSE loads for SLP.  Adjust.
19926 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19928         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19929         spelling in two comments.
19931 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19933         PR target/56797
19934         * config/arm/arm.c (load_multiple_sequence): Require SP
19935         as base register for loads if SP is in the register list.
19937 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19939         PR tree-optimization/56718
19940         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19941         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19942         and made public.  Adjusted all callers.
19943         (ipa_intraprocedural_devirtualization): New function.
19944         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19945         (ipa_intraprocedural_devirtualization): Likewise.
19946         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19948 2013-04-19  Richard Biener  <rguenther@suse.de>
19950         PR tree-optimization/57000
19951         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19953 2013-04-19  Terry Guo  <terry.guo@arm.com>
19955         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19956         Replace with ...
19957         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19958         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19959         (cortex_m4_fmacs): Use new reservations.
19960         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19962 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19964         PR rtl-optimization/56999
19965         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19966         related code.
19967         (lra_coalesce): Remove split_origin_bitmap and related code.
19968         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19969         ranges if necessary.
19971 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19973         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19974         New array.
19975         (ix86_expand_call): Remove clobbered_registers array and use
19976         x86_64_ms_sysv_extra_clobbered_registers instead.
19977         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19978         Declare here.
19979         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19980         predicate.
19981         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19982         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19983         (*call_value_rex64_ms_sysv): Ditto.
19985 2013-04-18  Cary Coutant  <ccoutant@google.com>
19987         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19988         parent instead of die_mark.
19990 2013-04-18  Diego Novillo  <dnovillo@google.com>
19992         * gimple.c (create_gimple_tmp): New.
19993         (get_expr_type): New.
19994         (build_assign): New.
19995         (build_type_cast): New.
19996         * gimple.h (enum ssa_mode): Define.
19997         (gimple_seq_set_location): New.
19998         * asan.c (build_check_stmt): Change some gimple_build_* calls
19999         to use build_assign and build_type_cast.
20001 2013-04-18  Richard Biener  <rguenther@suse.de>
20003         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
20004         handle negative step.  Remove redundant checks.
20005         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
20006         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
20007         for negative step and grouped loads fail to vectorize.
20009 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20011         * emit-rtl.c (reset_insn_used_flags): New function.
20012         (reset_all_used_flags): Use it.
20013         (verify_insn_sharing): New function.
20014         (verify_rtl_sharing): Fix verification for SEQUENCEs.
20016 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
20018         PR tree-optimization/56984
20019         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20020         and (x >> M) >= N don't register any assertion if N << M is the
20021         minimum value.
20023 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20025         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20026         do not delete_insn insns that have not yet been emitted, only
20027         unlink them with remove_insn.
20028         * df-scan.c (df_insn_delete): Revert r197492.
20030 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20032         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20033         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20035 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20037         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20038         define_insn_and_split.
20039         (and_scc,ior_scc,negscc): Likewise.
20040         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20042 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20044         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20045         can benefit from using a sequence of LDRD instructions in epilogue
20046         instead of a single LDM instruction.
20048 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20050         PR 45688
20051         * doc/extend.texi: Fix typo.
20053 2013-04-17  Richard Biener  <rguenther@suse.de>
20055         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20056         (vect_build_slp_tree): ... here.
20057         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20058         match.  Remove special-casing of mismatched complex loads.
20059         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20060         re-try the match with swapped commutative operands.
20061         (vect_supported_load_permutation_p): Remove special-casing of
20062         mismatched complex loads.
20063         (vect_analyze_slp_instance): Adjust.
20065 2013-04-17  Richard Biener  <rguenther@suse.de>
20067         PR rtl-optimization/56921
20068         * cfgloop.h (struct loop): Add simple_loop_desc member.
20069         (struct niter_desc): Mark with GTY(()).
20070         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20071         * loop-iv.c (get_simple_loop_desc): Likewise.
20072         (free_simple_loop_desc): Likewise.
20074         Revert
20075         2013-04-16  Richard Biener  <rguenther@suse.de>
20077         PR rtl-optimization/56921
20078         * loop-init.c (pass_rtl_move_loop_invariants): Add
20079         TODO_do_not_ggc_collect to todo_flags_finish.
20080         (pass_rtl_unswitch): Same.
20081         (pass_rtl_unroll_and_peel_loops): Same.
20082         (pass_rtl_doloop): Same.
20084 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20086         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20087         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20088         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20089         references.
20090         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20091         * tree-streamer.c (record_common_node): Adjust reference in comment.
20093 2013-04-17  Terry Guo  <terry.guo@arm.com>
20095         * config/arm/cortex-m4.md: Add a new bypass.
20097 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20099         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20100         New pattern.
20101         (*subs_<optab><mode>_multp2): New pattern.
20102         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20103         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20105 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20107         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20108         (*subs_mul_imm_<mode>): New pattern.
20110 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20112         PR target/56948
20113         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20114         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20115         (vsx_movti_32bit): Change j->wa to O->wa.
20117 2013-04-16  Richard Biener  <rguenther@suse.de>
20119         PR rtl-optimization/56921
20120         * loop-init.c (pass_rtl_move_loop_invariants): Add
20121         TODO_do_not_ggc_collect to todo_flags_finish.
20122         (pass_rtl_unswitch): Same.
20123         (pass_rtl_unroll_and_peel_loops): Same.
20124         (pass_rtl_doloop): Same.
20126 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20128         * config/arm/arm.c (emit_multi_reg_push): New declaration
20129         for an existing function.
20130         (arm_emit_strd_push): New function.
20131         (arm_expand_prologue): Used here.
20132         (arm_emit_ldrd_pop): New function.
20133         (arm_expand_epilogue): Used here.
20134         (arm_get_frame_offsets): Update condition.
20135         (arm_emit_multi_reg_pop): Add a special case for load of a single
20136         register with writeback.
20138 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20140         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20141         description.
20143 2013-04-16  Richard Biener  <rguenther@suse.de>
20145         PR tree-optimization/56756
20146         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20147         (first_mem_ref_loc): New.
20148         (execute_sm): Place the load temporarily before a previous
20149         access instead of in the latch edge to ensure its SSA dependencies
20150         are defined at points dominating the load.
20152 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20154         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20155         correct fix by moving header and footer insn to the footer of
20156         the merged basic block.  Clear BB_END of the merged-away block.
20158         PR middle-end/43631
20159         * emit-rtl.c (make_note_raw): New function.
20160         (link_insn_into_chain): New static inline function.
20161         (add_insn): Use it.
20162         (add_insn_before, add_insn_after): Factor insn chain linking code...
20163         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20164         using link_insn_into_chain.
20165         (note_outside_basic_block_p): New helper function for emit_note_after
20166         and emit_note_before.
20167         (emit_note_after): Use nobb variant of add_insn_after if the note
20168         should not be contained in a basic block.
20169         (emit_note_before): Use nobb variant of add_insn_before if the note
20170         should not be contained in a basic block.
20171         (emit_note_copy): Use make_note_raw.
20172         (emit_note): Likewise.
20173         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20174         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20175         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20176         the moved barrier the tail of the basic block it follows.
20177         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20179 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20181         PR tree-optimization/56962
20182         * gimple-ssa-strength-reduction.c (record_increment): Only set
20183         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20184         either rhs1 or rhs2 is equal to c->base_expr.
20186 2013-04-15  Richard Biener  <rguenther@suse.de>
20188         PR tree-optimization/56933
20189         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20190         member.
20191         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20192         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20193         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20194         dependence check ...
20195         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20196         ... here.
20197         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20198         GROUP_READ_WRITE_DEPENDENCE.
20200 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20202         * emit-rtl.c (reset_all_used_flags): New function.
20203         (verify_rtl_sharing): Call reset_all_used_flags before and after
20204         performing the checks.
20206 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20208         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20209         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20210         * config/arm/constraints.md (De): New constraint.
20211         * config/arm/neon.md (anddi3_neon): Delete.
20212         (neon_vand<mode>): Expand to standard anddi3 pattern.
20213         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20214         Move earlier in the file.
20215         (neon_inv_logic_op2): Likewise.
20216         (arm_anddi_operand_neon): New predicate.
20218 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20220         * configure.ac (gcc_cv_ld_as_needed): Set
20221         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20222         Use -z ignore, -z record on *-*-solaris2*.
20223         (HAVE_LD_AS_NEEDED): Update comment.
20224         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20225         * configure: Regenerate.
20226         * config.in: Regenerate.
20227         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20228         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20229         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20230         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20231         equivalents.  Fix markup.
20232         * doc/tm.texi: Regenerate.
20234 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20236         * config/i386/i386.opt: New option mstack-protector-guard=.
20237         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20238         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20239         TARGET_SSP_TLS_GUARD.
20240         * config/i386/i386.c (ix86_option_override_internal): Set
20241         ix86_stack_protector_guard.
20242         * config/i386/i386.md (stack_protect_set): Enable for
20243         TARGET_SSP_TLS_GUARD only.
20244         (stack_protect_set_<mode>): Ditto.
20245         (stack_protect_test): Ditto.
20246         (stack_protect_test_<mode>): Ditto.
20247         * doc/invoke.texi (i386 Option): Document.
20249 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20251         PR target/56890
20252         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20253         (S_MODES): Set H_MODE bit.
20254         (SF_MODES): Set only S_MODE and SF_MODE bits.
20255         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20256         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20257         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20258         <MODE_FLOAT>: Likewise.
20260 2013-04-15  Joey Ye  <joey.ye@arm.com>
20262         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20264 2013-04-15  Joey Ye  <joey.ye@arm.com>
20266         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20267         for real far jump.
20268         (thumb_far_jump_used_p): Count instruction size and set
20269         far_jump_used.
20271 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20273         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20274         * resource.c (mark_target_live_regs): Reformat conditional block.
20276 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20278         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20279         notes, they are emitted only just before final.
20280         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20282 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20284         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20285         * cfgrtl.c (delete_insn): Call it here instead.
20286         * lra-spills.c (lra_final_code_change): Use delete_insn.
20287         * haifa-sched.c (sched_remove_insn): Likewise.
20288         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20289         returning to the nop pool.
20290         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20291         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20293 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20295         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20296         * doc/tm.texi: Regenerated.
20298 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20300         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20301         QImode checks.
20303 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20305         * df-core.c (df_find_def): Compare register numbers.
20306         (df_find_use): Likewise.
20308 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20310         PR target/56903
20311         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20312         lra_in_progress for return.
20314 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20316         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20317         define_insn into define_insn_and_split and emit movsicc patterns.
20319 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20321         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20323 2013-04-12  Richard Biener  <rguenther@suse.de>
20325         * tree-pass.h (TODO_do_not_ggc_collect): New.
20326         * passes.c (execute_one_ipa_transform_pass): Honor
20327         TODO_do_not_ggc_collect.
20328         (execute_one_pass): Likewise.
20330         Revert
20331         2013-04-10  Richard Biener  <rguenther@suse.de>
20333         * passes.c (init_optimization_passes): Remove reload pass.
20334         * ira.c (do_reload): Merge into ...
20335         (ira): ... this.
20336         (rest_of_handle_reload): Remove.
20337         (pass_reload): Likewise.
20338         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20339         of reload for vzeroupper pass placement.
20341 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20343         PR tree-optimization/56918
20344         PR tree-optimization/56920
20345         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20346         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20347         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20348         use wide_mul_with_sign method.
20350 2013-04-12  Richard Biener  <rguenther@suse.de>
20352         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20353         not be considered a gimple constant.
20355 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20357         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20358         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20359         operations.
20361 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20362             Jakub Jelinek  <jakub@redhat.com>
20364         * opts.c: Include diagnostic-color.h.
20365         (common_handle_option): Handle OPT_fdiagnostics_color_.
20366         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20367         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20368         (diagnostic-color.o): New.
20369         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20370         (diagnostic_color_rule): New enum.
20371         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20372         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20373         the location string.
20374         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20375         either NULL, or color kind.
20376         * diagnostic-color.c: New file.
20377         * diagnostic-color.h: New file.
20378         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20379         arguments.
20380         * doc/invoke.texi (-fdiagnostics-color): Document.
20381         * pretty-print.h (pp_show_color): Define.
20382         (struct pretty_print_info): Add show_color field.
20383         * diagnostic.c: Include diagnostic-color.h.
20384         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20385         macros.  Colorize error:, warning: etc. strings and also the location
20386         string.
20387         (diagnostic_show_locus): Colorize the caret line.
20388         * pretty-print.c: Include diagnostic-color.h.
20389         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20390         inside of %< %> quotes or quoted through q format modifier.
20392 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20394         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20396 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20398         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20399         code in CC_NZ mode.
20400         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20401         pattern.
20403 2013-04-11  Marek Polacek  <polacek@redhat.com>
20405         PR tree-optimization/48184
20406         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20408 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20410         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20411         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20412         (skip_simple_arithmetic): Tidy up.
20413         * tree.h (skip_simple_constant_arithmetic): Declare.
20415 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20417         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20419 2013-04-11  Richard Biener  <rguenther@suse.de>
20421         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20422         generate vector constants.
20424 2013-04-11  Richard Biener  <rguenther@suse.de>
20426         PR tree-optimization/56878
20427         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20428         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20429         New function.
20430         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20431         Prefer to align the DR with the most invariant base address.
20433 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20435         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20436         comment.
20438 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20440         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20441         floating-point vector comparisons against 0.
20443 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20445         PR tree-optimization/56899
20446         * fold-const.c (extract_muldiv_1): Apply distributive law
20447         only if TYPE_OVERFLOW_WRAPS (ctype).
20449 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20451         PR target/56124
20452         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20453         loading has side effect.
20455 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20457         * config/sparc/sparc.c: Include tree-pass.h.
20458         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20459         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20460         head of file.  Change return type.  Split off gate function.
20461         (sparc_gate_work_around_errata): New function.
20462         (pass_work_around_errata): New pass definition.
20463         (insert_pass_work_around_errata) New pass insert definition to
20464         insert pass_work_around_errata just after delayed-branch scheduling.
20465         (sparc_option_override): Insert the pass.
20466         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20468 2013-04-10  David S. Miller  <davem@davemloft.net>
20470         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20471         or -mcpu=hypersparc.
20473         * target.def (cstore_mode): New hook.
20474         * target.h: Include insn-codes.h
20475         * targhooks.c: Likewise.
20476         (default_cstore_mode): New function.
20477         * targhooks.h: Declare it.
20478         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20479         * doc/tm.texi: Rebuild.
20480         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20481         target hook, rather than inspecting the insn_data.
20482         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20483         (TARGET_CSTORE_MODE): Redefine.
20484         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20485         result patterns.
20486         * config/sparc/predicates.md (cstore_result_operand): New special
20487         predicate.
20488         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20489         Use it for operand 0.
20490         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20491         (*snesi_special): Likewise.
20492         (*snesi_zero): Likewise.
20493         (*seqsi_zero): Likewise.
20494         (*sltu_insn): Likewise.
20495         (*sgeu_insn): Likewise.
20496         (*seqdi_special): Make operand 0 and comparison operation be of
20497         DImode.
20498         (*snedi_special): Likewise.
20499         (*snedi_special_vis3): Likewise.
20500         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20501         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20502         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20503         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20504         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20505         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20506         (*sltu_extend_sp64): Likewise.
20507         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20508         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20509         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20510         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20511         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20513 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20515         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20516         (aarch64_start_file): Use the new function.
20518 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20519             Jason Merrill  <jason@redhat.com>
20521         * common.opt: Add -gdwarf.
20522         * opts.c (common_handle_option): Handle it.
20523         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20525 2013-04-10  Richard Biener  <rguenther@suse.de>
20527         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20528         (execute_one_ipa_transform_pass): But unconditionally here.
20529         (execute_one_pass): And here.
20530         (init_optimization_passes): Remove reload pass.
20531         * tree-pass.h (TODO_ggc_collect): Remove.
20532         (pass_reload): Likewise.
20533         * ira.c (do_reload): Merge into ...
20534         (ira): ... this.
20535         (rest_of_handle_reload): Remove.
20536         (pass_reload): Likewise.
20537         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20538         of reload for vzeroupper pass placement.
20539         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20540         and todo_flags_finish of all passes.
20542 2013-04-10  Richard Biener  <rguenther@suse.de>
20544         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20545         first_const_oprnd field, rename first_def_type to first_op_type.
20546         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20547         (vect_get_and_check_slp_defs): Always use the type of the
20548         operand.  Allow mixed vect_external_def, vect_constant_def types.
20549         (vect_get_constant_vectors): Handle mixed vect_external_def,
20550         vect_constant_def types.
20552 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20554         PR tree-optimization/55524
20555         * tree-ssa-math-opts.c
20556         (convert_mult_to_fma): Don't use an fms construct
20557         when we don't have an fms operation, but fnma, and it looks
20558         likely that we'll be able to use the latter.
20560 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20562         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20563         function.
20564         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20565         inline fail caused by overwritable functions.
20567 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20569         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20570         unnecessary bits in the constant power of two case.
20572 2013-04-10  Richard Biener  <rguenther@suse.de>
20574         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20575         broken code swapping operands.
20576         (vect_build_slp_tree): Do not compute load permutations here.
20577         (vect_analyze_slp_instance): Compute load permutations here,
20578         after building the SLP tree.
20580 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20582         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20583         of next/prev_real_insn.
20585 2013-04-09  Jan Hubicka  <jh@suse.cz>
20587         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20588         Drop aliased parameter.
20589         (function_and_variable_visibility): Do not handle alias pairs.
20590         * cgraph.c (varpool_externally_visible_p): Update prototype.
20591         * varpool.c (varpool_add_new_variable): Update.
20593 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20595         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20597 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20599         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20601         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20603 2013-04-09  Marek Polacek  <polacek@redhat.com>
20605         PR tree-optimization/48762
20606         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20608 2013-04-09  Richard Biener  <rguenther@suse.de>
20610         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20611         dealing with cost.
20612         (vect_build_slp_tree): Likewise.
20613         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20614         calculating the cost of a SLP instance.
20615         (vect_analyze_slp_instance): Use it from here, after building
20616         the SLP tree.
20618 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20620         PR middle-end/56883
20621         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20622         expand_omp_for_static_chunk): Use simple_p = true in
20623         force_gimple_operand_gsi calls when assigning to addressable decls.
20625 2013-04-09  Jeff Law  <law@redhat.com>
20627         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20628         when the boolean was created by converting a wider object which
20629         had a boolean range.
20631 2013-04-09  Richard Biener  <rguenther@suse.de>
20633         * tree-vectorizer.h (slp_void_p): Remove.
20634         (slp_tree): Typedef before _slp_tree declaration.
20635         (struct _slp_tree): Use a vector of slp_tree as children.
20636         (vect_get_place_in_interleaving_chain): Remove.
20637         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20638         Move ...
20639         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20640         and make static.
20641         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20642         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20643         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20644         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20645         Use slp_node instead of slp_void_p and adjust.
20647 2013-04-09  Richard Biener  <rguenther@suse.de>
20649         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20650         work that is not necessary.
20652 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20654         PR tree-optimization/56854
20655         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20656         forward into clobber stmts if it would change MEM_REF lhs into
20657         non-MEM_REF.
20659 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20661         * tree.c (type_hash_lookup, type_hash_add): Make static.
20662         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20664 2013-04-09  Richard Biener  <rguenther@suse.de>
20666         * tree.h (unsave_expr_now): Remove.
20667         * tree-inline.c (mark_local_for_remap_r): Remove.
20668         (unsave_expr_1): Likewise.
20669         (unsave_r): Likewise.
20670         (unsave_expr_now): Likewise.
20671         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20672         (propagate_tree_value): Likewise.
20674 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20676         * doc/rtl.texi (sequence): Rewrite documentation to match the
20677         current use of SEQUENCE rtl objects.
20678         * rtl.def (SEQUENCE): Likewise.
20680         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20681         Update documentation.
20682         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20683         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20685         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20687 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20689         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20690         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20691         methods.
20692         (estimate_edge_size_and_time): Add comment to suggest using rounding
20693         methods.
20694         (estimate_node_size_and_time): Ditto.
20695         (remap_edge_change_prob): Use helper rounding divide methods.
20696         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20697         (gimple_mod_pow2_value_transform): Ditto.
20698         (gimple_mod_subtract_transform): Ditto.
20699         (gimple_ic_transform): Ditto.
20700         (gimple_stringops_transform): Ditto.
20701         * stmt.c (conditional_probability): Ditto.
20702         (emit_case_dispatch_table): Ditto.
20703         * lto-cgraph.c (merge_profile_summaries): Ditto.
20704         * tree-optimize.c (execute_fixup_cfg): Ditto.
20705         * cfgcleanup.c (try_forward_edges): Ditto.
20706         * cfgloopmanip.c (scale_loop_profile): Ditto.
20707         (loopify): Ditto.
20708         (duplicate_loop_to_header_edge): Ditto.
20709         (lv_adjust_loop_entry_edge): Ditto.
20710         * tree-vect-loop.c (vect_transform_loop): Ditto.
20711         * profile.c (compute_branch_probabilities): Ditto.
20712         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20713         * lto-streamer-in.c (input_cfg): Ditto.
20714         * gimple-streamer-in.c (input_bb): Ditto.
20715         * ipa-cp.c (update_profiling_info): Ditto.
20716         (update_specialized_profile): Ditto.
20717         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20718         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20719         rounding methods.
20720         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20721         (compute_trg_info): Ditto.
20722         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20723         (purge_dead_edges): Ditto.
20724         * loop-unswitch.c (unswitch_loop): Ditto.
20725         * cgraphclones.c (cgraph_clone_edge): Ditto.
20726         (cgraph_clone_node): Ditto.
20727         * tree-inline.c (copy_bb): Ditto.
20728         (copy_edges_for_bb): Ditto.
20729         (initialize_cfun): Ditto.
20730         (copy_cfg_body): Ditto.
20731         (expand_call_inline): Ditto.
20733 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20735         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20736         TARGET_CYGWIN64 by TARGET_64BIT.
20738 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20740         * config/epiphany/epiphany.md (GPR_1): New constant.
20741         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20742         * config/epiphany/epiphany.c (gen_compare_reg):
20743         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20744         is already in place.
20745         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20746         Don't require being called during rtl expansion; If y operlaps r0,
20747         return 0.
20748         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20749         (epiphany_expand_epilogue): Likewise.
20751         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20752         Don't use CC_FPmode for ORDERED / UNORDERED.
20753         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20755         * config/epiphany/constraints.md (CnL): New constraint.
20756         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20757         * config/epiphany/predicates.md (add_operand): Allow 1024.
20759         * config/epiphany/epiphany.md (logical_op): New code iterator.
20760         (op_mnc): New code attribute.
20761         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20762         (mov_f+1, mov_f+2): New peephole2 patterns.
20764         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20765         (cstoresi4): Also allow re-use of zero result when doing a NE
20766         comparison to a non-zero operand.
20767         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20769         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20770         Use gen_addsi3_i / gen_subsi3_i.
20772 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20774         PR c++/34949
20775         PR c++/50243
20776         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20777         contain anything but clobbers, at most one __builtin_stack_restore,
20778         optionally debug stmts and final resx, and if it has at least one
20779         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20780         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20781         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20782         which isn't defaut definition, remove them.
20783         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20784         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20785         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20786         with MEM_REF LHS with SSA_NAME address.
20788 2013-04-08  Jeff Law  <law@redhat.com>
20790         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20792 2013-04-08  Richard Biener  <rguenther@suse.de>
20794         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20795         extra newline.
20796         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20797         determined vector type.
20798         (vect_analyze_data_refs): Likewise.
20799         (vect_get_new_vect_var): Adjust.
20800         (vect_create_destination_var): Preserve SSA name versions.
20801         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20802         not dump anything here.
20804 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20806         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20807         Add member lr_slot_known.
20808         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20809         if necessary.
20810         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20811         Remove code that sets lr_slot_offset according to what a previous
20812         version of epiphany_emit_save_restore used to do.
20813         (epiphany_emit_save_restore): When doing an lr save or restore,
20814         set/verify lr_slot_known and lr_slot_offset.
20816 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20818         PR target/54338
20819         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20820         in ALL_REGS.
20822 2013-04-08  Richard Biener  <rguenther@suse.de>
20824         * alias.c (find_base_term): Fix thinko in previous change.
20826 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20828         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20829         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20830         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20831         if possible to compute val.
20832         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20833         For QImode integers don't require anything about precision.  Use
20834         const_with_all_bytes_same to find out if the constant doesn't have
20835         repeated bytes in it.
20837 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20839         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20840         within mode size.
20842 2013-04-08  Marek Polacek  <polacek@redhat.com>
20844         PR rtl-optimization/48182
20845         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20846         value to 1.
20848 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20850         PR target/55487
20851         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20852         nuses, make sure we have a label.
20854 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20856         PR target/56843
20857         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20858         (rs6000_emit_swdiv_low_precision): Remove.
20859         (rs6000_emit_swdiv): Rewrite to handle between one and four
20860         iterations of Newton-Raphson generally; modify required number of
20861         iterations for some cases.
20862         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20864 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20866         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20867         set-but-unused variable.
20869         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20870         basic blocks of released function bodies garbage-collectable.
20872         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20873         (struct rtl_opt_pass): Add TODO_df_finish.
20875         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20877 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20879         * config/arm/constraints.md (q): New constraint.
20880         * config/arm/ldrdstrd.md: New file.
20881         * config/arm/arm.md (ldrdstrd.md) New include.
20882         (arm_movdi): Use "q" instead of "r" constraint
20883         for double-word memory access.
20884         (movdf_soft_insn): Likewise.
20885         * config/arm/vfp.md (movdi_vfp): Likewise.
20886         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20887         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20888         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20889         (mem_ok_for_ldrd_strd): Likewise.
20890         (output_move_double): Update assertion.
20892 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20894         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20896 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20898         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20899         define_insn_and_split.
20900         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20902 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20904         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20905         define_insn_and_split.
20906         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20907         (shiftsi3_compare): New pattern.
20908         (rrx): New pattern.
20909         * config/arm/unspecs.md (UNSPEC_RRX): New.
20911 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20913         * config/arm/arm.md (negdi_extendsidi): New pattern.
20914         (negdi_zero_extendsidi): Likewise.
20916 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20918         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20919         define_insn_and_split.
20920         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20921         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20923 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20925         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20926         define_insn_and_split.
20927         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20928         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20930 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20932         * config/arm/arm.md (subsi3_carryin): New pattern.
20933         (subsi3_carryin_const): Likewise.
20934         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20935         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20937 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20939         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20941 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20943         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20944         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20946 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20948         * config/arm/arm.c (arm_expand_builtin): Change fcode
20949         type to unsigned int.
20951 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20953         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20955 2013-04-04  Ian Lance Taylor  <iant@google.com>
20957         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20958         language standard.
20960 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20962         PR middle-end/56729
20963         * df-scan.c (df_insn_delete): Disable failing assert.
20965 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20967         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20968         New function prototype.
20969         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20970         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20971         (arm_builtin_vectorized_function): New function.
20973 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20975         * config/arm/arm_neon_builtins.def: New file.
20976         * config/arm/arm.c (neon_builtin_data): Move contents to
20977         arm_neon_builtins.def.
20978         (enum arm_builtins): Include neon builtin definitions.
20979         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20980         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20982 2013-04-04  Marek Polacek  <polacek@redhat.com>
20984         PR tree-optimization/48186
20985         * predict.c (maybe_hot_frequency_p): Return false if
20986         HOT_BB_FREQUENCY_FRACTION is 0.
20987         (cgraph_maybe_hot_edge_p): Likewise.
20989 2013-04-04  Richard Biener  <rguenther@suse.de>
20991         PR tree-optimization/56826
20992         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20993         more accurately.
20995 2013-04-04  Richard Biener  <rguenther@suse.de>
20997         PR tree-optimization/56213
20998         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20999         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
21001 2013-04-04  Richard Biener  <rguenther@suse.de>
21003         PR tree-optimization/56837
21004         * tree-loop-distribution.c (classify_partition): For non-zero
21005         values require that the value has the same precision as its
21006         mode to be useful as memset value.
21008 2013-04-03  Nick Clifton  <nickc@redhat.com>
21010         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
21011         (fmssf4): Use fmsf.s on E3V5 architectures.
21012         (fnmasf4): Use fnmaf.s on E3V5 architectures.
21013         (fnmssf4): Use fnmsf.s on E3V5 architectures.
21015 2013-04-03  Jeff Law  <law@redhat.com>
21017         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
21018         (lra-eliminations.o): Likewise.
21020 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21022         * gcov-io.c (compute_working_sets): Moved most of body of old
21023         compute_working_sets here from profile.c.
21024         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21025         (gcov_working_set_t): Moved typedef here from basic-block.h
21026         (compute_working_set): Declare.
21027         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21028         (get_working_sets): Renamed from compute_working_set,
21029         replace most of body with call to new compute_working_sets.
21030         (get_exec_counts): Replace call to compute_working_sets
21031         to get_working_sets.
21032         * profile.h (get_working_sets): Renamed from compute_working_set.
21033         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21034         to get_working_sets.
21035         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21036         * gcov-dump.c (dump_working_sets): New function.
21038 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21040         * hwint.c (sext_hwi, zext_hwi): New functions.
21041         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21042         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21043         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21044         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21045         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21046         (sext_hwi, zext_hwi): New functions.
21048 2013-04-03  Jeff Law  <law@redhat.com>
21050         PR tree-optimization/56799
21051         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21052         back test for widening conversion erroneously dropped in prior change.
21054 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21056         PR target/56809
21057         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21058         instead of next_real_insn.
21060 2013-04-03  Marek Polacek  <polacek@redhat.com>
21062         PR sanitizer/55702
21063         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21065 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21067         PR target/56809
21068         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21069         next_real_insn.
21070         (thumb1_output_casesi): Likewise.
21071         (thumb2_output_casesi): Likewise.
21073 2013-04-03  Richard Biener  <rguenther@suse.de>
21075         PR tree-optimization/56817
21076         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21077         Split out ...
21078         (tree_unroll_loops_completely_1): ... new function to manually
21079         walk the loop tree, properly defering outer loops of unrolled
21080         loops to later iterations.
21082 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21084         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21085         (vectorizable_load): Likewise.
21086         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21087         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21089 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21091         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21092         BIT_FIELD_REF.
21094 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21096         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21098 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21100         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21102 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21104         PR tree-optimization/56790
21105         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21106         folding.
21108 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21110         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21111         Handle VEC_MERGE.
21112         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21113         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21114         equal arguments.
21116 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21118         PR c/19449
21119         * tree.h (force_folding_builtin_constant_p): New decl.
21120         * builtins.c (force_folding_builtin_constant_p): New variable.
21121         (fold_builtin_constant_p): Fold immediately also if
21122         force_folding_builtin_constant_p.
21124 2013-04-03  Richard Biener  <rguenther@suse.de>
21126         PR tree-optimization/56812
21127         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21128         DRs of the same interleaving chain are independent.
21130 2013-04-02  Jason Merrill  <jason@redhat.com>
21132         * gdbinit.in (pbb): Use debug fn.
21134 2013-04-02  Lawrence Crowl  <crowl@google.com>
21136         * sese.h (struct ivtype_map_elt_s): Remove unused.
21137         (extern debug_ivtype_map): Remove unused.
21138         (extern eq_ivtype_map_elts): Remove unused.
21139         * sese.c (debug_ivtype_map): Removed unused.
21140         (debug_ivtype_map_1): Removed unused.
21141         (debug_ivtype_elt): Remove unused.
21142         (eq_ivtype_map_elts): Remove unused.
21145 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21147         PR target/52790
21148         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21149         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21150         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21151         function.
21152         (legitimize_pe_coff_symbol): Likewise.
21153         (is_imported_p): New helper-function.
21154         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21155         for Windows x64 targets.
21156         (ix86_expand_prologue): Optimize for pe-coff targets.
21157         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21158         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21159         medium/large code-model.
21160         (legitimize_pic_address): Likewise.
21161         (legitimize_tls_address): Likewise.
21162         (ix86_expand_call): Likewise.
21163         (x86_output_mi_thunk): Likewise.
21164         (get_dllimport_decl): Add new beimport argument.
21165         (construct_plt_address): Don't assert for x64 pe-coff targets.
21166         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21167         targets.
21168         (SYMBOL_FLAG_STUBVAR): New macro.
21169         (SYMBOL_REF_STUBVAR_P): Likewise.
21170         * config/i386/winnt.c (stub_list): New structure.
21171         (stub_head): New local variable.
21172         (i386_pe_record_stub): New function.
21173         (i386_pe_file_end): Emit refptr-stubs.
21175 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21177         PR rtl-optimization/56745
21178         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21179         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21181         PR c++/34949
21182         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21183         and both of them are MEM_REFs, just compare first argument for
21184         equality and attempt to deal even with differing offsets.
21186         PR c++/34949
21187         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21188         of gimple_clobber_p to be MEM_REF.
21189         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21190         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21191         after gimplification.
21192         * asan.c (get_mem_ref_of_assignment): Don't instrument
21193         gimple_clobber_p stmts.
21194         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21195         gimple_clobber_p stmt if they have MEM_REF lhs and
21196         are dead because of another gimple_clobber_p stmt.
21197         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21198         gimple_clobber_p stmts like debug stmts.
21199         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21200         that refer to unused VAR_DECLs or uninitialized values.
21201         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21202         gimple_clobber_p stmts if they refer to removed parameters.
21203         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21204         formatting.
21206 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21208         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21209         using SWI48 mode attribute.
21211 2013-04-02  Wei Mi  <wmi@google.com>
21213         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21214         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21215         *<rotate_insn><mode>3_mask in i386.md.
21217 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21219         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21221 2013-04-02  Richard Biener  <rguenther@suse.de>
21223         PR tree-optimization/56778
21224         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21225         Runtime alias tests are not supported for gather loads.
21226         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21227         stmts referenced from SSA operands before updating SSA form.
21229 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21230             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21232         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21233         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21234         * config/arm/cortex-a53.md: New file.
21235         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21236         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21237         * config/arm/arm.c (arm_issue_rate): Likewise.
21238         * config/arm/arm-tune.md: Regenerate
21239         * config/arm/arm-tables.opt: Regenerate.
21240         * config/arm/arm-cores.def: Add cortex-a53.
21242 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21244         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21245         non-static link.
21247 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21249         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21250         scalar load/store operations using B/H registers.
21251         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21253 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21255         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21256         scalar move.
21257         * config/aarch64/aarch64.c
21258         (aarch64_simd_scalar_immediate_valid_for_move): New.
21259         * config/aarch64/aarch64-protos.h
21260         (aarch64_simd_scalar_immediate_valid_for_move): New.
21261         * config/aarch64/constraints.md (Dh, Dq): New.
21262         * config/aarch64/iterators.md (hq): New.
21264 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21266         * reorg.c (get_branch_condition): Deal with conditional returns.
21267         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21269 2013-04-01  Wei Mi  <wmi@google.com>
21271         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21272         Truncate operand 2 using %b asm operand modifier.
21273         (*<shift_insn><mode>3_mask): Ditto.
21274         (*<rotate_insn><mode>3_mask): Ditto.
21276 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21278         PR middle-end/56798
21279         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21281 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21283         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21284         of next_real_insn.
21285         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21287 2013-03-30  Lawrence Crowl  <crowl@google.com>
21289         * dse.c (clear_alias_sets): Remove never set.
21290         (disqualified_clear_alias_sets): Remove never set.
21291         (clear_alias_mode_pool): Remove never set.
21292         (dse_step0): Remove condition that is never true.
21293         (canon_address): Remove condition that is never true.
21294         (dse_step7): Remove condition that is never true.
21295         (rest_of_handle_dse): Remove condition that is never true.
21296         (rest_of_handle_dse::did_global): Remove never read from above.
21297         (dse_step2_spill): Remove never called from above.
21298         (dse_step5_spill): Remove never called from above.
21300 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21302         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21303         JUMP_TABLE_DATA changes.
21304         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21305         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21306         (Insns) <jump_table_data>: New entry.
21307         * doc/tm.texi: Regenerate.
21309         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21311         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21312         for table jump at the end of a basic block using tablejump_p.
21313         * targhooks.c (default_invalid_within_doloop): Likewise.
21314         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21315         target hook implementation that is identical to the default hook.
21316         (rs6000_invalid_within_doloop): Remove.
21318         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21319         unused variable from tablejump_p call.
21321         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21322         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21323         (INSN_DELETED_P): Likewise.
21324         (emit_jump_table_data): New prototype.
21325         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21326         after 4th as unused.
21327         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21328         * sched-vis.c (print_insn): Likewise.
21329         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21330         insn for compatibility with back ends that use next_active_insn to
21331         identify jump table data.
21332         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21333         (remove_insn): Likewise.
21334         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21335         to be emitted.
21336         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21337         (emit_jump_table_data): New function.
21339         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21340         basic block, a JUMP_TABLE_DATA never is.
21341         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21342         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21343         off from code handling real insns.
21344         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21345         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21346         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21347         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21348         is not a NONDEBUG_INSN_P.
21349         * ira-costs.c (scan_one_insn): Likewise.
21350         * jump.c (mark_all_labels): Likewise.
21351         (mark_jump_label_1): Likewise.
21352         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21353         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21354         a basic block.
21355         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21356         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21357         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21358         JUMP_TABLE_DATA_P insns.
21359         (calculate_elim_costs_all_insns): Likewise.
21360         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21361         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21362         (delete_output_reload): Code style fixups.
21363         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21364         insn flags on this non-insn.
21365         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21366         as scheduling barriers, for pre-change compatibility.
21367         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21368         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21370         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21371         redundant JUMP_TABLE_DATA_P test.
21372         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21373         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21374         (frv_for_each_packet): Likewise.
21375         * config/i386/i386.c (min_insn_size): Likewise.
21376         (ix86_avoid_jump_mispredicts): Likewise.
21377         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21378         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21379         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21380         (mips16_insn_length): Robustify.
21381         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21382         (mips16_split_long_branches): Likewise.
21383         * config/pa/pa.c (pa_combine_instructions): Likewise.
21384         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21385         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21386         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21387         as contributing to pool range lengths.
21388         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21389         Remove redundant JUMP_TABLE_DATA_P test.
21390         (sh_loop_align): Likewise.
21391         (split_branches): Likewise.
21392         (sh_insn_length_adjustment): Likewise.
21393         * config/spu/spu.c (get_branch_target): Likewise.
21395 2013-03-29  Jan Hubicka  <jh@suse.cz>
21397         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21398         gcov streaming; stream hot bb threshold to ltrans.
21399         * predict.c (get_hot_bb_threshold): Break out from ....
21400         (maybe_hot_count_p): ... here.
21401         (set_hot_bb_threshold): New function.
21402         * lto-section-in.c (lto_section_name): Add profile.
21403         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21404         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21405         and data-streamer.h
21406         (histogram_entry): New structure.
21407         (histogram, histogram_pool): New global vars.
21408         (histogram_hash): New structure.
21409         (histogram_hash::hash): New method.
21410         (histogram_hash::equal): Likewise.
21411         (account_time_size): New function.
21412         (cmp_counts): New function.
21413         (dump_histogram): New function.
21414         (ipa_profile_generate_summary): New function.
21415         (ipa_profile_write_summary): New function.
21416         (ipa_profile_read_summary): New function.
21417         (ipa_profile): Decide on threshold.
21418         (pass_ipa_profile): Add ipa_profile_write_summary and
21419         ipa_profile_read_summary.
21420         * Makefile.in (ipa.o): Update dependencies.
21421         * lto-streamer.h (LTO_section_ipa_profile): New section.
21423 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21425         * tree.h (VAR_P): New.
21427 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21429         PR lto/56777
21430         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21432 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21434         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21435         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21436         (control_flow_insn_p): Likewise.
21437         * cfgrtl.c (duplicate_insn_chain): Likewise.
21438         * final.c (get_attr_length_1): Likewise.
21439         (shorten_branches): Likewise.
21440         (final_scan_insn): Likewise.
21441         * function.c (instantiate_virtual_regs): Likewise.
21442         * gcse.c (insert_insn_end_basic_block): Likewise.
21443         * ira-costs.c (scan_one_insn): Likewise.
21444         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21445         * lra.c (check_rtl): Likewise.
21446         * reload1.c (elimination_costs_in_insn): Likewise.
21447         * reorg.c (follow_jumps): Likewise.
21449         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21450         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21451         (thumb_far_jump_used_p): Likewise.
21452         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21453         (workaround_speculation): Likewise.
21454         (add_sched_insns_for_speculation): Likewise.
21455         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21456         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21457         (frv_for_each_packet): Likewise.
21458         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21459         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21460         (final_emit_insn_group_barriers): Likewise.
21461         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21462         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21463         (mips16_insn_length): Likewise.
21464         * config/pa/pa.c (pa_reorg): Likewise.
21465         (pa_combine_instructions): Likewise.
21466         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21467         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21468         (sh_reorg): Likewise.
21469         (split_branches): Likewise.
21470         * config/spu/spu.c (get_branch_target): Likewise.
21472         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21473         JUMP_TABLE_DATA_P.
21475 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21477         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21478         Fix declaration name.
21480 2013-03-28  Lawrence Crowl  <crowl@google.com>
21482         * graphds.h (struct graph.indicies): Remove unused.
21483         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21484         (SCOP_ORIGINAL_PDDRS): Remove unused.
21485         * sese.h (extern insert_loop_close_phis): Removed unused.
21486         (extern insert_guard_phis): Removed unused.
21487         (extern ivtype_map_elt_info): Removed unused.
21488         (new_ivtype_map_elt): Removed unused.
21489         * sese.c (ivtype_map_elt_info): Removed unused.
21491 2013-03-28  Lawrence Crowl  <crowl@google.com>
21493         * Makefile.in: Add several missing include dependences.
21494         (DUMPFILE_H): New.
21495         (test-dump.o): New.  This object is not added to any executable,
21496         but is present for ad-hoc testing.
21497         * bitmap.c
21498         (debug (const bitmap_head_def &)): New.
21499         (debug (const bitmap_head_def *)): New.
21500         * bitmap.h
21501         (extern debug (const bitmap_head_def &)): New.
21502         (extern debug (const bitmap_head_def *)): New.
21503         * cfg.c
21504         (debug (edge_def &)): New.
21505         (debug (edge_def *)): New.
21506         * cfghooks.c
21507         (debug (basic_block_def &)): New.
21508         (debug (basic_block_def *)): New.
21509         * dumpfile.h
21510         (dump_node (const_tree, int, FILE *)): Correct source file.
21511         * dwarf2out.c
21512         (debug (die_struct &)): New.
21513         (debug (die_struct *)): New.
21514         * dwarf2out.h
21515         (extern debug (die_struct &)): New.
21516         (extern debug (die_struct *)): New.
21517         * gimple-pretty-print.c
21518         (debug (gimple_statement_d &)): New.
21519         (debug (gimple_statement_d *)): New.
21520         * gimple-pretty-print.h
21521         (extern debug (gimple_statement_d &)): New.
21522         (extern debug (gimple_statement_d *)): New.
21523         * ira-build.c
21524         (debug (ira_allocno_copy &)): New.
21525         (debug (ira_allocno_copy *)): New.
21526         (debug (ira_allocno &)): New.
21527         (debug (ira_allocno *)): New.
21528         * ira-int.h
21529         (extern debug (ira_allocno_copy &)): New.
21530         (extern debug (ira_allocno_copy *)): New.
21531         (extern debug (ira_allocno &)): New.
21532         (extern debug (ira_allocno *)): New.
21533         * ira-lives.c
21534         (debug (live_range &)): New.
21535         (debug (live_range *)): New.
21536         * lra-int.h
21537         (debug (lra_live_range &)): New.
21538         (debug (lra_live_range *)): New.
21539         * lra-lives.c
21540         (debug (lra_live_range &)): New.
21541         (debug (lra_live_range *)): New.
21542         * omega.c
21543         (debug (omega_pb_d &)): New.
21544         (debug (omega_pb_d *)): New.
21545         * omega.h
21546         (extern debug (omega_pb_d &)): New.
21547         (extern debug (omega_pb_d *)): New.
21548         * print-rtl.c
21549         (debug (const rtx_def &)): New.
21550         (debug (const rtx_def *)): New.
21551         * print-tree.c
21552         (debug_tree (tree): Move within file.
21553         (debug_raw (const tree_node &)): New.
21554         (debug_raw (const tree_node *)): New.
21555         (dump_tree_via_hooks (const tree_node *, int)): New.
21556         (debug (const tree_node &)): New.
21557         (debug (const tree_node *)): New.
21558         (debug_verbose (const tree_node &)): New.
21559         (debug_verbose (const tree_node *)): New.
21560         (debug_head (const tree_node &)): New.
21561         (debug_head (const tree_node *)): New.
21562         (debug_body (const tree_node &)): New.
21563         (debug_body (const tree_node *)): New.
21564         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21565         (debug (vec<tree, va_gc> &)): New.
21566         (debug (vec<tree, va_gc> *)): New.
21567         * rtl.h
21568         (extern debug (const rtx_def &)): New.
21569         (extern debug (const rtx_def *)): New.
21570         * sbitmap.c
21571         (debug_raw (simple_bitmap_def &)): New.
21572         (debug_raw (simple_bitmap_def *)): New.
21573         (debug (simple_bitmap_def &)): New.
21574         (debug (simple_bitmap_def *)): New.
21575         * sbitmap.h
21576         (extern debug (simple_bitmap_def &)): New.
21577         (extern debug (simple_bitmap_def *)): New.
21578         (extern debug_raw (simple_bitmap_def &)): New.
21579         (extern debug_raw (simple_bitmap_def *)): New.
21580         * sel-sched-dump.c
21581         (debug (vinsn_def &)): New.
21582         (debug (vinsn_def *)): New.
21583         (debug_verbose (vinsn_def &)): New.
21584         (debug_verbose (vinsn_def *)): New.
21585         (debug (expr_def &)): New.
21586         (debug (expr_def *)): New.
21587         (debug_verbose (expr_def &)): New.
21588         (debug_verbose (expr_def *)): New.
21589         (debug (vec<rtx> &)): New.
21590         (debug (vec<rtx> *)): New.
21591         * sel-sched-dump.h
21592         (extern debug (vinsn_def &)): New.
21593         (extern debug (vinsn_def *)): New.
21594         (extern debug_verbose (vinsn_def &)): New.
21595         (extern debug_verbose (vinsn_def *)): New.
21596         (extern debug (expr_def &)): New.
21597         (extern debug (expr_def *)): New.
21598         (extern debug_verbose (expr_def &)): New.
21599         (extern debug_verbose (expr_def *)): New.
21600         (extern debug (vec<rtx> &)): New.
21601         (extern debug (vec<rtx> *)): New.
21602         * sel-sched-ir.h
21603         (_list_iter_cond_expr): Make inline instead of static.
21604         * sreal.c
21605         (debug (sreal &)): New.
21606         (debug (sreal *)): New.
21607         * sreal.h
21608         (extern debug (sreal &)): New.
21609         (extern debug (sreal *)): New.
21610         * tree.h
21611         (extern debug_raw (const tree_node &)): New.
21612         (extern debug_raw (const tree_node *)): New.
21613         (extern debug (const tree_node &)): New.
21614         (extern debug (const tree_node *)): New.
21615         (extern debug_verbose (const tree_node &)): New.
21616         (extern debug_verbose (const tree_node *)): New.
21617         (extern debug_head (const tree_node &)): New.
21618         (extern debug_head (const tree_node *)): New.
21619         (extern debug_body (const tree_node &)): New.
21620         (extern debug_body (const tree_node *)): New.
21621         (extern debug (vec<tree, va_gc> &)): New.
21622         (extern debug (vec<tree, va_gc> *)): New.
21623         * tree-cfg.c
21624         (debug (struct loop &)): New.
21625         (debug (struct loop *)): New.
21626         (debug_verbose (struct loop &)): New.
21627         (debug_verbose (struct loop *)): New.
21628         * tree-dump.c: Add header dependence.
21629         * tree-flow.h
21630         (extern debug (struct loop &)): New.
21631         (extern debug (struct loop *)): New.
21632         (extern debug_verbose (struct loop &)): New.
21633         (extern debug_verbose (struct loop *)): New.
21634         * tree-data-ref.c
21635         (debug (data_reference &)): New.
21636         (debug (data_reference *)): New.
21637         (debug (vec<data_reference_p> &)): New.
21638         (debug (vec<data_reference_p> *)): New.
21639         (debug (vec<ddr_p> &)): New.
21640         (debug (vec<ddr_p> *)): New.
21641         * tree-data-ref.h
21642         (extern debug (data_reference &)): New.
21643         (extern debug (data_reference *)): New.
21644         (extern debug (vec<data_reference_p> &)): New.
21645         (extern debug (vec<data_reference_p> *)): New.
21646         (extern debug (vec<ddr_p> &)): New.
21647         (extern debug (vec<ddr_p> *)): New.
21648         * tree-ssa-alias.c
21649         (debug (pt_solution &)): New.
21650         (debug (pt_solution *)): New.
21651         * tree-ssa-alias.h
21652         (extern debug (pt_solution &)): New.
21653         (extern debug (pt_solution *)): New.
21654         * tree-ssa-alias.c
21655         (debug (_var_map &)): New.
21656         (debug (_var_map *)): New.
21657         (debug (tree_live_info_d &)): New.
21658         (debug (tree_live_info_d *)): New.
21659         * tree-ssa-alias.h
21660         (extern debug (_var_map &)): New.
21661         (extern debug (_var_map *)): New.
21662         (extern debug (tree_live_info_d &)): New.
21663         (extern debug (tree_live_info_d *)): New.
21665 2013-03-28  Jan Hubicka  <jh@suse.cz>
21667         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21669 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21671         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21672         record only when desired or required.
21674 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21676         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21677         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21679 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21681         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21682         (*andsi3_compare0_uxtw): New pattern.
21683         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21684         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21686 2013-03-28  Jan Hubicka  <jh@suse.cz>
21688         * data-streamer-in.c (streamer_read_gcov_count): New function.
21689         * gimple-streamer-out.c: Include value-prof.h.
21690         (output_gimple_stmt): Output histogram.
21691         (output_bb): Use streamer_write_gcov_count.
21692         * value-prof.c: Include data-streamer.h
21693         (dump_histogram_value): Add HIST_TYPE_MAX.
21694         (stream_out_histogram_value): New function.
21695         (stream_in_histogram_value): New function.
21696         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21697         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21698         * data-streamer-out.c (streamer_write_gcov_count): New function.
21699         (streamer_write_gcov_count_stream): New function.
21700         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21701         (lto_output_node): Likewise.
21702         (input_node, input_edge): Likewise.
21703         * lto-streamer-out.c (output_cfg): Update streaming.
21704         * lto-streamer-in.c (input_cfg): Likewise.
21705         * data-streamer.h (streamer_write_gcov_count,
21706         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21707         * gimple-streamer-in.c: Include value-prof.h
21708         (input_gimple_stmt): Input histograms.
21709         (input_bb): Update profile streaming.
21711 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21713         * genmodes.c (emit_max_int): New function.
21714         (emit_insn_modes_h): Added call to emit_max_function.
21715         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21716         Added doc.
21717         * machmode.def: Fixed comment.
21719 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21721         * combine.c (try_combine): Removed useless assert.
21722         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21724 2013-03-28  Marek Polacek  <polacek@redhat.com>
21725             Richard Biener  <rguenther@suse.de>
21727         PR tree-optimization/56695
21728         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21729         build signed result of a vector comparison.
21730         * tree-cfg.c (verify_gimple_comparison): Check that a result
21731         of a vector comparison has signed type.
21733 2013-03-28  Richard Biener  <rguenther@suse.de>
21735         PR tree-optimization/37021
21736         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21737         do not restrict gaps between groups.
21738         * tree-vect-stmts.c (vectorizable_load): Properly account for
21739         a gap between groups.
21741 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21743         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21744         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21745         is not enabled.
21747 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21749         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21750         * doc/extend.texi (Named Address Spaces): Ditto.
21751         (Variable Attributes): Ditto.
21753 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21755         * config.build: Add support for cygwin x64 target.
21756         * config.gcc: Likewise.
21757         * config.host: Likewise.
21758         * configure.ac: Likewise
21759         * configure: Regenerated.
21761 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21763         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21764         * config/i386/t-cygwin-w64: New file.
21765         * config/i386/cygwin-w64.h: New file.
21766         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21767         and add support for x64-cygwin target.
21768         (CPP_SPEC): Likewise.
21769         (CXX_WRAP_SPEC_LIST): Undefine before define.
21770         (LIBGCJ_SONAME): Use 15 as version.
21772 2013-03-27  Richard Biener  <rguenther@suse.de>
21774         PR tree-optimization/56716
21775         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21776         dumping for ref nodes.
21778 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21780         PR tree-optimization/55334
21781         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21782         restricted pointers to arrays.
21784 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21786         * Makefile.in (.SUFFIXES): Add .cc.
21787         (.c.o): Apply same recipe for implicit rule .cc.o.
21789 2013-03-27  Richard Biener  <rguenther@suse.de>
21791         PR tree-optimization/37021
21792         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21793         REALPART/IMAGPART_EXPRs around the supported refs.
21794         * tree-ssa-structalias.c (find_func_aliases): Assume that
21795         floating-point values are not used to transfer pointers.
21797 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21799         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21800         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21801         * doc/tm.texi: Regenerate.
21802         * targhooks.h (default_has_ifunc_p): New.
21803         * targhooks.c (default_has_ifunc_p): Ditto.
21804         * config/linux-protos.h: New file.
21805         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21806         hook for linux which disables support of indirect functions in android.
21807         * config/linux-android.c: New file.
21808         * config/t-linux-android.c: Ditto.
21809         * config.gcc: Added new object file linux-android.o.
21810         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21811         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21812         * varasm.c (do_assemble_alias): Likewise.
21813         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21814         doesn't support indirect functions.
21815         * configure: Regenerate.
21817 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21819         PR target/56102
21820         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21821         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21822         mult-word mode.
21824 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21826         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21828 2013-03-27  Terry Guo  <terry.guo@arm.com>
21830         * config/arm/arm-cores.def: Added core cortex-r7.
21831         * config/arm/arm-tune.md: Regenerated.
21832         * config/arm/arm-tables.opt: Regenerated.
21833         * doc/invoke.texi: Added entry for core cortex-r7.
21835 2013-03-27  Walter Lee  <walt@tilera.com>
21837         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21838         double-decrement of next_scratch_regno.
21840 2013-03-27  Walter Lee  <walt@tilera.com>
21842         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21843         input operands.
21844         (insn_v1mulus): Ditto.
21845         (insn_v2muls): Ditto.
21847 2013-03-27  Walter Lee  <walt@tilera.com>
21849         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21850         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21852 2013-03-27  Walter Lee  <walt@tilera.com>
21854         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21855         (*sibcall_value): Ditto.
21857 2013-03-27  Walter Lee  <walt@tilera.com>
21859         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21860         (insn_mnz_v8qi): ... this ...
21861         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21862         vector equivalent.
21863         (insn_v<n>mnz): Replaced by ...
21864         (insn_v1mnz): ... this ...
21865         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21866         equivalent.
21867         (insn_mz_<mode>): Replaced by ...
21868         (insn_mz_v8qi): ... this ...
21869         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21870         vector equivalent.
21871         (insn_v<n>mz): Replaced by ...
21872         (insn_v1mz): ... this ...
21873         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21874         equivalent.
21876 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21878         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21880 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21882         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21883         than fprintf with a non-constant, non-format string.
21885 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21887         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21888         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21889         operand 0 predicate.
21890         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21891         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21892         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21893         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21894         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21895         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21896         (mov<mode>_insv_1): Remove expander.  Merge insn with
21897         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21898         Use general_x64nomem_operand as operand 1 predicate.
21899         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21900         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21901         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21902         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21903         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21904         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21905         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21906         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21907         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21908         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21909         (general_x64nomem_operand): Ditto.
21911 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21913         * config/rtems.opt: Add -pthread option.
21915 2013-03-26  Richard Biener  <rguenther@suse.de>
21917         * alias.c (find_base_term): Avoid redundant and not used recursion.
21918         (base_alias_check): Get the initial base term from the caller.
21919         (true_dependence_1): Compute and pass base terms to base_alias_check.
21920         (write_dependence_p): Likewise.
21921         (may_alias_p): Likewise.
21923 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21925         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21926         PC-relative load in SI modes and above only.
21928 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21930         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21931         * config/arm/iwmmxt.md (WCGR0): Update.
21932         (WCGR1, WCGR2, WCGR3): Likewise.
21934 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21936         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21937         Use x64 and nox64 isa attributes.
21939 2013-03-26  Richard Biener  <rguenther@suse.de>
21941         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21942         alignment computations and rely on get_object_alignment_1
21943         for the !TYPE_P case.
21944         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21946 2013-03-26  Walter Lee  <walt@tilera.com>
21948         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21949         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21951 2013-03-25  Jeff Law  <law@redhat.com>
21953         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21954         check for INTEGRAL_TYPE_P that was missing due to checking in
21955         wrong version of prior patch.
21957 2013-03-25  Walter Lee  <walt@tilera.com>
21959         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21960         TILEGX_INSN_SHUFFLEBYTES1.
21961         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21962         shufflebytes1.
21963         (tilegx_builtins): Ditto.
21964         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21966 2013-03-25  Walter Lee  <walt@tilera.com>
21968         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21969         (floatunssisf2): New pattern.
21970         (floatsidf2): New pattern.
21971         (floatunssidf2): New pattern.
21973 2013-03-25  Walter Lee  <walt@tilera.com>
21975         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21976         tests for constraint J, K, N, P.
21978 2013-03-25  Walter Lee  <walt@tilera.com>
21980         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21981         Use indirect/pcrel encoding.
21982         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21983         Ditto.
21985 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21987         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21988         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21989         * config/mips/mips.c (mips_option_override): Set IMADD default.
21990         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21991         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21992         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21993         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21994         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21996 2013-03-25  Jeff Law  <law@redhat.com>
21998         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21999         slightly to avoid creating and folding useless trees.  Simplify
22000         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
22002 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
22004         * config/i386/i386.md (*zero_extendsidi2): Merge with
22005         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
22006         * config/i386/predicates.md (x86_64_zext_operand): Rename from
22007         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
22008         targets.  Clarify comment.
22010 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22012         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
22013         pass-through jump functions differently.
22014         (ipa_read_jump_function): Likewise.  Also use setter functions to set
22015         up jump functions.
22017 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22019         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22020         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22021         process it.
22022         (ipa_get_indirect_edge_target): New function.
22023         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22024         ipa_get_indirect_edge_target.  Update all callers.
22025         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22026         ipa_get_indirect_edge_target_1 instead of calling
22027         ipa_get_indirect_edge_target.
22028         (create_specialized_node): Pass aggvlas to
22029         ipcp_discover_new_direct_edges.
22031 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22033         * config/arm/arm.md (f_sels, f_seld): New types.
22034         (*cmov<mode>): New pattern.
22035         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22036         predicate.
22038 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22040         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22041         POSIX-printf for mingw-hosted  builds.
22043 2013-03-25  Richard Biener  <rguenther@suse.de>
22045         PR middle-end/56694
22046         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22047         must-not-throw stmt location.
22049 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22051         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22052         Emit load-acquire versions when acq is true.
22053         (arm_emit_store_exclusive): Add rel parameter.
22054         Emit store-release versions when rel is true.
22055         (arm_split_compare_and_swap): Use acquire-release instructions
22056         instead.
22057         of barriers when appropriate.
22058         (arm_split_atomic_op): Likewise.
22059         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22060         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22061         (VUNSPEC_SLX): Likewise.
22062         (VUNSPEC_LDA): Likewise.
22063         (VUNSPEC_STL): Likewise.
22064         * config/arm/sync.md (atomic_load<mode>): New pattern.
22065         (atomic_store<mode>): Likewise.
22066         (arm_load_acquire_exclusive<mode>): Likewise.
22067         (arm_load_acquire_exclusivesi): Likewise.
22068         (arm_load_acquire_exclusivedi): Likewise.
22069         (arm_store_release_exclusive<mode>): Likewise.
22071 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22073         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22074         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22075         * config/mip/predicates.md (lwsp_swsp_operand,
22076         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22077         sb16_operand, db4_operand, db7_operand, ib3_operand,
22078         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22079         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22080         andi16_operand): New predicates.
22081         * config/mips/mips.md (compression): New attribute.
22082         (enabled): New attribute.
22083         (length): Consider compression in computing length.
22084         (shift_compression): New code attribute.
22085         (*add<mode>3): New operands. Record compression.
22086         (sub<mode>3): Likewise.
22087         (one_cmpl<mode>2): Likewise.
22088         (*and<mode>3): Likewise.
22089         (*ior<mode>3): Likewise.
22090         (unnamed pattern for xor): Likewise.
22091         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22092         (*<optab><mode>3): Likewise.
22093         (*mov<mode>_internal: Likewise.
22094         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22095         (mips_unsigned_immediate_p): New.
22096         (umips_lwsp_swsp_address_p): New.
22097         (m16_based_address_p): New.
22098         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22099         (mips_unsigned_immediate_p): New prototype.
22100         (lwsp_swsp_address_p): New prototype.
22101         (m16_based_address_p): New prototype.
22102         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22103         (mips_signed_immediate_p): New function.
22104         (m16_based_address_p): New function.
22105         (lwsp_swsp_address_p): New function.
22106         (mips_print_operand_punctuation): Recognize short delay slot insns
22107         for microMIPS.add<mode>3"
22109 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22111         PR target/56720
22112         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22113         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22115 2013-03-25  Richard Biener  <rguenther@suse.de>
22117         PR tree-optimization/56689
22118         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22119         any edge.
22121 2013-03-25  Richard Biener  <rguenther@suse.de>
22123         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22124         of bitmap.
22125         (memory_references): Likewise.
22126         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22127         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22128         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22129         (gather_mem_refs_in_loops): Fold into ...
22130         (analyze_memory_references): ... this.  Move initialization
22131         to tree_ssa_lim_initialize.
22132         (fill_always_executed_in): Rename to ...
22133         (fill_always_executed_in_1): ... this.
22134         (fill_always_executed_in): Move contains_call computation to
22135         this new function from ...
22136         (tree_ssa_lim_initialize): ... here.
22137         (tree_ssa_lim): Call fill_always_executed_in.
22139 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22141         * postreload.c (reload_combine): Fix code detecting returns.
22143 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22145         * function.c (emit_use_return_register_into_block): On cc0 targets,
22146         do not emit the sequence between cc0 setter and user.
22148 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22150         * config/i386/predicates.md (local_symbolic_operand): Interpret
22151         dll-imported symbols as none-local.
22153 2013-03-25  Richard Biener  <rguenther@suse.de>
22155         * tree-ssa-loop-im.c (struct depend): Remove.
22156         (struct lim_aux_data): Make depends a vec of gimples.
22157         (free_lim_aux_data): Adjust.
22158         (add_dependency): Likewise.
22159         (set_level): Likewise.
22161 2013-03-25  Richard Biener  <rguenther@suse.de>
22163         PR middle-end/56434
22164         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22165         the pointer returned by calls with ECF_MALLOC set.
22167 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22169         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22171 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22173         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22174         using MMXMODE mode iterator.
22175         (*move<mode>_internal): Merge with *movv2sf_internal and
22176         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22178 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22180         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22181         (record_last_mem_set_info): Likewise.
22183         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22184         of XNEWVEC followed by memset.
22185         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22187 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22189         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22190         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22191         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22192         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22193         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22194         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22195         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22196         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22197         BARRIER_P instead of GET_CODE.
22199 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22201         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22202         inaccuracy in the probing code.
22204         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22205         (ctrapdi4): Likewise.
22207 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22209         * calls.c (expand_call): Add missing guard to code handling return
22210         of non-BLKmode structures in MSB.
22211         * function.c (expand_function_end): Likewise.
22213 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22215         * combine.c (try_combine): Adjust comment.  Do not add the set of
22216         insn #0 if the destination indirectly is set or dies in insn #2.
22217         Tidy up code to distribute a new note.
22219 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22221         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22222         also for alternatives 16 and 17.
22224 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22226         * config/i386/sse.md (*mov<mode>_internal): Merge with
22227         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22228         Emit insn template depending on type attribute.  Use
22229         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22230         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22231         attribute calculation.  Remove unit attribute calculation.
22232         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22233         Set prefix_data16 attribute for DImode ssemov types.
22234         Use Ym instead of y for SSE-MMX conversion alternatives.
22235         Reorder operand constraints.
22237 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22239         * df.h (df_insn_delete): Adjust prototype.
22240         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22241         and let it decide whether mark the basic block dirty.
22242         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22243         * df-scan.c (df_insn_info_delete): New helper function, split
22244         off from df_insn_delete.
22245         (df_scan_free_bb_info): Use it.
22246         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22247         Likewise.
22248         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22249         that the insn is actually an insn and it has a non-NULL basic block.
22250         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22252 2013-03-22  Richard Biener  <rguenther@suse.de>
22254         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22255         dep_ref members.
22256         (mem_ref_alloc): Do not allocate them.
22257         (refs_independent_p): Do not query or maintain a cache.
22259 2013-03-22  Richard Biener  <rguenther@suse.de>
22261         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22262         (gather_mem_refs_in_loops): Do not compute it.
22263         (analyze_memory_references): Do not allocate it.
22264         (tree_ssa_lim_finalize): Do not free it.
22265         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22267 2013-03-22  Richard Biener  <rguenther@suse.de>
22269         * is-a.h (as_a): Use gcc_checking_assert.
22271 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22273         * config/aarch64/aarch64.c (aarch64_print_operand): New
22274         format specifier for printing a constant in hex.
22275         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22276         format specifier for printing second operand.
22278 2013-03-22  Richard Biener  <rguenther@suse.de>
22280         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22281         bitmaps.
22282         (gather_mem_refs_in_loops): Perform store accumulation here.
22283         (create_vop_ref_mapping_loop): Remove.
22284         (create_vop_ref_mapping): Likewise.
22285         (analyze_memory_references): Initialize refs_stored_in_loop.
22286         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22287         (record_indep_loop): Remove.
22288         (record_dep_loop): New function.
22289         (ref_indep_loop_p_1): Adjust to only walk over references
22290         in the loop, not its subloops.
22291         (ref_indep_loop_p): Rename to ...
22292         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22293         maintaining a more fine-grained cache.
22294         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22295         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22297 2013-03-22  Richard Biener  <rguenther@suse.de>
22299         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22300         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22301         aggregate mem_ref_loc.
22302         (free_mem_ref_locs): Inline into ...
22303         (memref_free): ... this and adjust.
22304         (mem_ref_alloc): Adjust.
22305         (mem_ref_locs_alloc): Remove.
22306         (record_mem_ref_loc): Adjust.
22307         (get_all_locs_in_loop): Rewrite into ...
22308         (for_all_locs_in_loop): ... this iterator.
22309         (rewrite_mem_ref_loc): New functor.
22310         (rewrite_mem_refs): Use for_all_locs_in_loop.
22311         (sm_set_flag_if_changed): New functor.
22312         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22313         (ref_always_accessed): New functor.
22314         (ref_always_accessed_p): Use for_all_locs_in_loop.
22316 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22318         * tree-pass.h (PROP_gimple_lvec): New.
22319         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22320         (init_optimization_passes): Move pass_lower_vector.
22321         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22322         PROP_gimple_lvec.
22323         (pass_lower_vector): Provide PROP_gimple_lvec.
22324         (pass_lower_vector_ssa): Likewise.
22325         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22327 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22329         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22331 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22333         * config/i386/i386.md (*movdi_internal): Disparage slightly
22334         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22335         conversion alternatives.
22337 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22339         PR middle-end/48087
22340         * diagnostic.def (DK_WERROR): New kind.
22341         * diagnostic.h (werrorcount): Define.
22342         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22343         promoted to DK_ERROR, increment DK_WERROR counter instead of
22344         DK_ERROR counter.
22345         * toplev.c (toplev_main): Call print_ignored_options even if
22346         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22347         even if just werrorcount is non-zero.
22349         PR debug/55608
22350         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22351         on failure.
22352         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22353         (string_cst_pool_decl): New function.
22354         (optimize_one_addr_into_implicit_ptr): New function.
22355         (resolve_addr_in_expr): Optimize DWARF location expression
22356         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22357         which doesn't live in memory, but has DW_AT_location or
22358         DW_AT_const_value, or refers to a string literal, into
22359         DW_OP_GNU_implicit_pointer.
22360         (optimize_location_into_implicit_ptr): New function.
22361         (resolve_addr): If removing DW_AT_location of a variable because
22362         it was DW_OP_addr of address of the variable, but the variable doesn't
22363         live in memory, try to emit const value attribute for the initializer.
22365 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22367         * tree.h (VECTOR_TYPE_P): New macro.
22368         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22369         TYPE_MODE): Use it.
22370         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22371         VEC_COND_EXPR cannot be lvalues.
22372         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22374 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22376         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22377         Restrict the transformation to equal modes.
22379 2013-03-21  Richard Biener  <rguenther@suse.de>
22381         PR tree-optimization/39326
22382         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22383         (MEM_ANALYZABLE): Adjust.
22384         (record_mem_ref_loc): Move bitmap ops ...
22385         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22386         unanalyzable refs, do not record locations for it.
22387         (analyze_memory_references): Allocate ref zero as shared
22388         unanalyzable ref.
22389         (refs_independent_p): Do not test for unanalyzed mems here.
22390         (ref_indep_loop_p_1): Special-case disambiguation against
22391         the unanalyzed ref.
22392         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22394 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22396         * config/arm/arm-protos.h (tune_params): Add
22397         prefer_neon_for_64bits field.
22398         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22399         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22400         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22401         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22402         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22403         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22404         (arm_option_override): Handle -mneon-for-64bits new option.
22405         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22406         (prefer_neon_for_64bits): Declare new variable.
22407         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22408         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22409         (arch_enabled): Handle new arch types. Remove support for onlya8
22410         and nota8.
22411         (one_cmpldi2): Use new arch names.
22412         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22413         * config/arm/arm.opt (mneon-for-64bits): Add option.
22414         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22415         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22416         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22417         of onlya8.
22418         * doc/invoke.texi (-mneon-for-64bits): Document.
22420 2013-03-21  Richard Biener  <rguenther@suse.de>
22422         PR tree-optimization/39326
22423         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22424         (sort_bbs_in_loop_postorder_cmp): New function.
22425         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22427 2013-03-21  Richard Biener  <rguenther@suse.de>
22429         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22430         (vect_insert_into_interleaving_chain): Likewise.
22431         (vect_drs_dependent_in_basic_block): Inline ...
22432         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22433         split out from ...
22434         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22435         (vect_check_interleaving): Simplify.
22436         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22437         (vect_slp_analyze_data_ref_dependences): ... this new function.
22438         (dr_group_sort_cmp): New function.
22439         (vect_analyze_data_ref_accesses): Compute data-reference groups
22440         here instead of in vect_analyze_data_ref_dependence.  Use
22441         a more efficient algorithm.
22442         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22443         vect_slp_analyze_data_ref_dependences.  Call
22444         vect_analyze_data_ref_accesses earlier.
22445         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22446         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22447         (vect_slp_analyze_data_ref_dependences): New prototype.
22449 2013-03-21  Richard Biener  <rguenther@suse.de>
22451         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22452         ref is stored in the loop.
22453         (find_refs_for_sm): Walk only over all stores.
22454         (store_motion_loop): Allocate from lim_bitmap_obstack.
22455         (store_motion): Likewise.
22457 2013-03-21  Richard Biener  <rguenther@suse.de>
22459         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22460         Update virtual SSA form.
22462 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22464         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22465         * configure: Regenerate.
22466         * config.in: Regenerate.
22467         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22468         if !HAVE_LD_EH_FRAME_CIEV3.
22470 2013-03-21  Richard Biener  <rguenther@suse.de>
22472         * tree-cfg.c (verify_expr_no_block): New function.
22473         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22474         nor DECL_VALUE_EXPR have locations with associated blocks.
22475         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22476         (clear_unused_block_pointer): Remove code dealing with
22477         blocks in DECL_DEBUG_EXPR locations.
22479 2013-03-21  Richard Biener  <rguenther@suse.de>
22481         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22482         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22483         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22484         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22485         instead of DECL_DEBUG_EXPR_IS_FROM.
22486         * gimplify.c (gimplify_modify_expr): Likewise.
22487         * tree-cfg.c (verify_expr_location_1): Likewise.
22488         * tree-complex.c (create_one_component_var): Likewise.
22489         * tree-sra.c (create_access_replacement): Likewise.
22490         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22491         (clear_unused_block_pointer): Likewise.
22492         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22493         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22494         * var-tracking.c (var_debug_decl): Likewise.
22495         (track_expr_p): Likewise.
22496         * tree-inline.c (add_local_variables): Likewise.  Set
22497         DECL_HAS_DEBUG_EXPR_P after copying it.
22498         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22499         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22501 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22503         PR bootstrap/56656
22504         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22505         * configure: Regenerate.
22506         * config.in: Regenerate.
22507         * config/i386/i386.md (*movdf_internal): Use
22508         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22509         movd instead of movq mnemonic for interunit moves.
22510         (*movdi_internal): Ditto.
22512 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22514         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22515         (abd<mode>_3): New pattern.
22516         (aba<mode>_3): New pattern.
22517         (fabd<mode>_3): New pattern.
22519 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22521         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22522         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22523         occurrence of REGISTER_PREFIX as its empty string.
22525 2013-03-20  Jeff Law  <law@redhat.com>
22527         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22528         addititional equivalences for equality comparisons between an SSA_NAME
22529         and a constant where the SSA_NAME was set from a widening conversion.
22531 2013-03-20  Walter Lee  <walt@tilera.com>
22533         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22535 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22537         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22538         depending on type attribute.
22539         (*movti_internal): Ditto.
22540         (*movtf_internal): Ditto.
22541         (*movxf_internal): Ditto.
22542         (*movdf_internal): Ditto.
22543         (*movsf_internal): Ditto.
22545 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22547         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22548         maybe_vex for sselog1 and ssemov types.
22549         (*movdi_internal): Reorder operand constraints.
22550         (*movsi_internal): Ditto.  Set prefix attribute to
22551         maybe_vex for sselog1 and ssemov types.
22552         (*movtf_internal): Set prefix attribute to maybe_vex
22553         for sselog1 and ssemov types.
22554         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22555         DImode ssemov types.  Reorder operand constraints.
22556         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22557         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22558         attribute for SImode ssemov types.  Reorder operand constraints.
22560 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22562         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22563         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22565 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22567         * config/rs6000/predicates.md (indexed_address, update_address_mem
22568         update_indexed_address_mem): New predicates.
22569         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22570         attribute for load/store instructions.
22571         * config/rs6000/dfp.md (movsd_store): Likewise.
22572         (movsd_load): Likewise.
22573         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22574         (unnamed HI->DI extend define_insn): Likewise.
22575         (unnamed SI->DI extend define_insn): Likewise.
22576         (unnamed QI->SI extend define_insn): Likewise.
22577         (unnamed QI->HI extend define_insn): Likewise.
22578         (unnamed HI->SI extend define_insn): Likewise.
22579         (unnamed HI->SI extend define_insn): Likewise.
22580         (extendsfdf2_fpr): Likewise.
22581         (movsi_internal1): Likewise.
22582         (movsi_internal1_single): Likewise.
22583         (movhi_internal): Likewise.
22584         (movqi_internal): Likewise.
22585         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22586         attribute for load/store instructions.
22587         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22588         instructions.
22589         (mov<mode>_softfloat): Likewise.
22590         (mov<mode>_hardfloat32): Likewise.
22591         (mov<mode>_hardfloat64): Likewise.
22592         (mov<mode>_softfloat64): Likewise.
22593         (movdi_internal32): Likewise.
22594         (movdi_internal64): Likewise.
22595         (probe_stack_<mode>): Likewise.
22597 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22599         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22600         floating point, and decimal floating point to reload iterator.
22602         * config/rs6000/constraints.md (wl constraint): New constraints to
22603         return FLOAT_REGS if certain options are used to reduce the number
22604         of separate patterns that exist in the file.
22605         (wx constraint): Likewise.
22606         (wz constraint): Likewise.
22608         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22609         -mdebug=reg, print wg, wl, wx, and wz constraints.
22610         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22611         Initialize the reload functions for 64-bit binary/decimal floating
22612         point types.
22613         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22614         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22615         create the buffer on the stack to overcome not having a 32-bit
22616         load and store.
22617         (rs6000_emit_move): Likewise.
22618         (rs6000_secondary_memory_needed_rtx): Likewise.
22619         (rs6000_alloc_sdmode_stack_slot): Likewise.
22620         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22621         via xxlxor, just like DFmode 0.0.
22623         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22624         define as 1 if we are running on a power7 or newer.
22625         (enum r6000_reg_class_enum): Add new constraints.
22627         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22628         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22629         with other moves by using conditional constraits (wg).  Use LFIWZX
22630         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22631         (movsd splitter): Likewise.
22632         (movsd_hardfloat): Likewise.
22633         (movsd_softfloat): Likewise.
22635         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22636         binary and decimal floating point moves.
22637         (fmove_ok): New attributes to combine binary and decimal floating
22638         point moves, and to combine power6x (mfpgpr) moves along normal
22639         floating moves.
22640         (real_value_to_target): Likewise.
22641         (f32_lr): Likewise.
22642         (f32_lm): Likewise.
22643         (f32_li): Likewise.
22644         (f32_sr): Likewise.
22645         (f32_sm): Likewise.
22646         (f32_si): Likewise.
22647         (movsf): Combine binary and decimal floating point moves.  Combine
22648         power6x (mfpgpr) moves with other moves by using conditional
22649         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22650         (mov<mode> for SFmode/SDmode); Likewise.
22651         (SFmode/SDmode splitters): Likewise.
22652         (movsf_hardfloat): Likewise.
22653         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22654         (movsf_softfloat): Likewise.
22655         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22657         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22658         wx and wz constraints.
22660         * config/rs6000/constraints.md (wg constraint): New constraint to
22661         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22663         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22664         constraint.
22666         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22667         -mdebug=reg, print wg, wl, wx, and wz constraints.
22668         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22669         Initialize the reload functions for 64-bit binary/decimal floating
22670         point types.
22671         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22672         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22673         create the buffer on the stack to overcome not having a 32-bit
22674         load and store.
22675         (rs6000_emit_move): Likewise.
22676         (rs6000_secondary_memory_needed_rtx): Likewise.
22677         (rs6000_alloc_sdmode_stack_slot): Likewise.
22678         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22679         via xxlxor, just like DFmode 0.0.
22681         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22682         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22683         with other moves by using conditional constraits (wg).  Use LFIWZX
22684         and STFIWX for loading SDmode on power7.
22685         (movdd splitters): Likewise.
22686         (movdd_hardfloat32): Likewise.
22687         (movdd_softfloat32): Likewise.
22688         (movdd_hardfloat64_mfpgpr): Likewise.
22689         (movdd_hardfloat64): Likewise.
22690         (movdd_softfloat64): Likewise.
22692         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22693         64-bit binary and decimal floating point moves.
22694         (FMOVE64X): Likewise.
22695         (movdf): Combine 64-bit binary and decimal floating point moves.
22696         Combine power6x (mfpgpr) moves with other moves by using
22697         conditional constraits (wg).
22698         (mov<mode> for DFmode/DDmode): Likewise.
22699         (DFmode/DDmode splitters): Likewise.
22700         (movdf_hardfloat32): Likewise.
22701         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22702         (movdf_softfloat32): Likewise.
22703         (movdf_hardfloat64_mfpgpr): Likewise.
22704         (movdf_hardfloat64): Likewise.
22705         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22706         (movdf_softfloat64): Likewise.
22707         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22708         (reload_<mode>_load): Move to later in the file so they aren't in
22709         the middle of the floating point move insns.
22710         (reload_<mode>_store): Likewise.
22712         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22713         constraint.
22715         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22716         constraint if -mdebug=reg.
22717         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22718         Enable using dd reload support if needed.
22720         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22721         binary and decimal floating point moves in rs6000.md.
22722         (movtd_internal): Likewise.
22724         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22725         decimal floating point moves.
22726         (movtf): Likewise.
22727         (movtf_internal): Likewise.
22728         (mov<mode>_internal, TDmode/TFmode): Likewise.
22729         (movtf_softfloat): Likewise.
22730         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22732         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22733         movdi_internal64, using wg constraint for move direct operations.
22734         (movdi_internal64): Likewise.
22736         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22737         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22738         the various virtual registers. Use GPR/FPR first/last values,
22739         instead of hard coding the register numbers.  Print which modes
22740         have reload functions registered.
22741         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22742         settings before/after setting cpu, target and subtarget settings.
22743         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22744         and for secondary reload failures in rs6000_secondary_reload_inner.
22745         (rs6000_secondary_reload_fail): Likewise.
22746         (rs6000_secondary_reload_inner): Likewise.
22748         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22749         macros for first/last GPR and FPR registers.
22750         (LAST_GPR_REGNO): Likewise.
22751         (FIRST_FPR_REGNO): Likewise.
22752         (LAST_FPR_REGNO): Likewise.
22754         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22755         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22756         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22757         (vcond<mode><mode>): Likewise.
22758         (vcondu<mode><mode>): Likewise.
22759         (vector_gtu<mode>): Likewise.
22760         (vector_gte<mode>): Likewise.
22761         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22762         to prevent the compiler from converting DImode operations to TImode.
22763         (ior<mode>3): Likewise.
22764         (and<mode>3): Likewise.
22765         (one_cmpl<mode>2): Likewise.
22766         (nor<mode>3): Likewise.
22767         (andc<mode>3): Likewise.
22769         * config/rs6000/constraints.md (wt constraint): New constraint
22770         that returns VSX_REGS if TImode is allowed in VSX registers.
22772         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22773         constant under VSX.
22775         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22776         similar to TImode, but it is restricted to being in the GPRs.
22778         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22779         TImode to occupy a single VSX register.
22781         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22782         -mvsx-timode for power7/power8.
22783         (power7 cpu): Likewise.
22784         (power8 cpu): Likewise.
22786         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22787         sure that TFmode/TDmode take up two registers if they are ever
22788         allowed in the upper VSX registers.
22789         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22790         registers.
22791         (rs6000_init_hard_regno_mode_ok): Likewise.
22792         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22793         constraint.  Print if LRA is turned on.
22794         (rs6000_option_override_internal): Give an error if -mvsx-timode
22795         and VSX is not enabled.
22796         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22797         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22798         to reg+offset addressing.  Use PTImode when checking offset
22799         addresses for validity.
22800         (reg_offset_addressing_ok_p): Likewise.
22801         (rs6000_legitimate_offset_address_p): Likewise.
22802         (rs6000_legitimize_address): Likewise.
22803         (rs6000_legitimize_reload_address): Likewise.
22804         (rs6000_legitimate_address_p): Likewise.
22805         (rs6000_eliminate_indexed_memrefs): Likewise.
22806         (rs6000_emit_move): Likewise.
22807         (rs6000_secondary_reload): Likewise.
22808         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22809         reloads to fpr registers to continue to use reg+offset addressing,
22810         but 64-bit reloads to altivec registers need reg+reg addressing.
22811         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22812         it.  Treat LO_SUM like a PLUS operation.
22813         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22814         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22815         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22816         registers to share a register with a smaller sized type, since VSX
22817         puts scalars in the upper 64-bits.
22818         (print_operand): Add support for PTImode.
22819         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22820         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22821         registers, but don't have arithmetic support.
22822         (rs6000_memory_move_cost): Add test for VSX.
22823         (rs6000_opt_masks): Add -mvsx-timode.
22825         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22826         for TImode.
22827         (VSs): Likewise.
22828         (VSr): Use wt constraint for TImode.
22829         (VSv): Drop TImode support.
22830         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22831         (vsx_movti_64bit): Likewise.
22832         (vsx_movti_32bit): Likewise.
22833         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22834         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22835         one '?' on the appropriate output constraint.  Do not allow TImode
22836         logical operations on 32-bit systems.
22837         (vsx_ior<mode>3): Likewise.
22838         (vsx_xor<mode>3): Likewise.
22839         (vsx_one_cmpl<mode>2): Likewise.
22840         (vsx_nor<mode>3): Likewise.
22841         (vsx_andc<mode>3): Likewise.
22842         (vsx_concat_<mode>): Likewise.
22843         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22845         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22846         OPTION_MASK_VSX_TIMODE.
22847         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22848         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22850         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22851         (TI2 iterator): New iterator for TImode, PTImode.
22852         (wd mode attribute): Add values for vector types.
22853         (movti_string): Replace TI move operations with operations for TImode
22854         and PTImode.  Add support for TImode being allowed in VSX registers.
22855         (mov<mode>_string, TImode/PTImode): Likewise.
22856         (movti_ppc64): Likewise.
22857         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22858         (TI mode splitters): Likewise.
22860         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22861         constraint.
22863 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22865         PR tree-optimization/56355
22866         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22867         Also handle integers with undefined overflow.
22869 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22870             Maciej W. Rozycki  <macro@codesourcery.com>
22871             Tom de Vries  <tom@codesourcery.com>
22872             Nathan Sidwell  <nathan@codesourcery.com>
22873             Iain Sandoe  <iain@codesourcery.com>
22874             Nathan Froyd  <froydnj@codesourcery.com>
22875             Chao-ying Fu  <fu@mips.com>
22877         * doc/extend.texi: (micromips, nomicromips, nocompression):
22878         Document new function attributes.
22879         * doc/invoke.texi (minterlink-compressed, mmicromips,
22880         m14k, m14ke, m14kec): Document new options.
22881         (minterlink-mips16): Update documentation.
22882         * doc/md.texi (ZC, ZD): Document new constraints.
22883         * configure.ac (gcc_cv_as_micromips): Check if linker
22884         supports the .set micromips directive.
22885         * configure: Regenerate.
22886         * config.in: Regenerate.
22887         * config/mips/mips-tables.opt: Regenerate.
22888         * config/mips/micromips.md: New file.
22889         * constraints.md (ZC, ZD): New constraints.
22890         * config/mips/predicates.md (movep_src_register): New predicate.
22891         (movep_src_operand): New predicate.
22892         (non_volatile_mem_operand): New predicate.
22893         * config/mips/mips.md (multimem): New type.
22894         (length): Differentiate between 17-bit and 18-bit branch offsets.
22895         (MOVEP1, MOVEP2): New mode iterator.
22896         (mov_<load>l): Use ZC constraint.
22897         (mov_<load>r): Likewise.
22898         (mov_<store>l): Likewise.
22899         (mov_<store>r): Likewise.
22900         (*branch_equality<mode>_inverted): Add microMIPS support.
22901         (*branch_equality<mode>): Likewise.
22902         (*jump_absolute): Likewise.
22903         (indirect_jump_<mode>): Likewise.
22904         (tablejump_<mode>): Likewise.
22905         (<optab>_internal): Likewise.
22906         (sibcall_internal): Likewise.
22907         (sibcall_value_internal): Likewise.
22908         (prefetch): Use constraint ZD.
22909         * config/mips/mips.opt (minterlink-compressed): New option.
22910         (minterlink-mips16): Now an alias for minterlink-compressed.
22911         (mmicromips): New option.
22912         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22913         (compare_and_swap_12): Likewise.
22914         (sync_add<mode>): Likewise.
22915         (sync_<optab>_12): Likewise.
22916         (sync_old_<optab>_12): Likewise.
22917         (sync_new_<optab>_12): Likewise.
22918         (sync_nand_12): Likewise.
22919         (sync_old_nand_12): Likewise.
22920         (sync_new_nand_12): Likewise.
22921         (sync_sub<mode>): Likewise.
22922         (sync_old_add<mode>): Likewise.
22923         (sync_old_sub<mode>): Likewise.
22924         (sync_new_add<mode>): Likewise.
22925         (sync_new_sub<mode>): Likewise.
22926         (sync_<optab><mode>): Likewise.
22927         (sync_old_<optab><mode>): Likewise.
22928         (sync_new_<optab><mode>): Likewise.
22929         (sync_nand<mode>): Likewise.
22930         (sync_old_nand<mode>): Likewise.
22931         (sync_new_nand<mode>): Likewise.
22932         (sync_lock_test_and_set<mode>): Likewise.
22933         (test_and_set_12): Likewise.
22934         (atomic_compare_and_swap<mode>): Likewise.
22935         (atomic_exchange<mode>_llsc): Likewise.
22936         (atomic_fetch_add<mode>_llsc): Likewise.
22937         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22938         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22939         (umips_save_restore_pattern_p): Likewise.
22940         (umips_load_store_pair_p): Likewise.
22941         (umips_output_load_store_pair): Likewise.
22942         (umips_movep_target_p): Likewise.
22943         (umips_12bit_offset_address_p): Likewise.
22944         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22945         (mips_base_mips16): Rename this...
22946         (mips_base_compression_flags): ...to this. Update all uses.
22947         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22948         (mips_mips16_decl_p): Delete.
22949         (mips_nomips16_decl_p): Delete.
22950         (mips_get_compress_on_flags): New function.
22951         (mips_get_compress_off_flags): New function.
22952         (mips_get_compress_mode): New function.
22953         (mips_get_compress_on_name): New function.
22954         (mips_get_compress_off_name): New function.
22955         (mips_insert_attributes): Support multiple compression types.
22956         (mips_merge_decl_attributes): Likewise.
22957         (umips_12bit_offset_address_p): New function.
22958         (mips_start_function_definition): Emit .set micromips directive.
22959         (mips_call_may_need_jalx_p): New function.
22960         (mips_function_ok_for_sibcall): Add microMIPS support.
22961         (mips_print_operand_punctuation): Support short delay slots and
22962         compact jumps.
22963         (umips_swm_mask, umips_swm_encoding): New.
22964         (umips_build_save_restore): New function.
22965         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22966         (was_mips16_p): Remove.
22967         (old_compression_mode): New.
22968         (mips_set_compression_mode): New function.
22969         (mips_set_current_function): Add microMIPS support.
22970         (mips_option_override): Likewise.
22971         (umips_save_restore_pattern_p): New function.
22972         (umips_output_save_restore): New function.
22973         (umips_load_store_pair_p_1): New function.
22974         (umips_load_store_pair_p): New function.
22975         (umips_output_load_store_pair_1): New function.
22976         (umips_output_load_store_pair): New function.
22977         (umips_movep_target_p) New function.
22978         (mips_prepare_pch_save): Add microMIPS support.
22979         * config/mips/mips.h (TARGET_COMPRESSION): New.
22980         (TARGET_CPU_CPP_BUILTINS): Update macro
22981         to use new compression flags and to support microMIPS.
22982         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22983         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22984         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22985         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22986         (ASM_SPEC): Support mmicromips and mno-micromips.
22987         (M16STORE_REG_P): New macro.
22988         (MIPS_CALL): Support TARGET_MICROMIPS.
22989         (MICROMIPS_J): New macro.
22990         (mips_base_mips16): Rename this...
22991         (mips_base_compression_flags): ...to this.
22992         (UMIPS_12BIT_OFFSET_P): New macro.
22993         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22994         (MULTILIB_DIRNAMES): Likewise.
22995 2013-03-20  Richard Biener  <rguenther@suse.de>
22997         PR tree-optimization/56661
22998         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22999         the result does not have to be distinct.
23001 2013-03-20  Richard Biener  <rguenther@suse.de>
23003         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
23004         remap_gimple_op_r.
23006 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23007             Steven Bosscher  <steven@gcc.gnu.org>
23009         PR rtl-optimization/56605
23010         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
23012 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23014         PR bootstrap/56656
23015         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
23016         that require movd instead of movq.
23018 2013-03-20  Richard Biener  <rguenther@suse.de>
23020         * tree-ssa-structalias.c (struct variable_info): Add pointer
23021         to the first field of an aggregate with sub-vars.  Make
23022         this and the pointer to the next subfield its ID.
23023         (vi_next): New function.
23024         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23025         storedanything_id, integer_id): Increment by one.
23026         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23027         get_call_clobber_vi): Adjust.
23028         (solution_set_expand): Simplify and speedup.
23029         (solution_set_add): Inline into ...
23030         (set_union_with_increment): ... this.  Adjust accordingly.
23031         (do_sd_constraint): Likewise.
23032         (do_ds_constraint): Likewise.
23033         (do_complex_constraint): Simplify.
23034         (build_pred_graph): Adjust.
23035         (solve_graph): Likewise.  Simplify and speedup.
23036         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23037         get_constraint_for_component_ref, get_constraint_for_1,
23038         first_vi_for_offset, first_or_preceding_vi_for_offset,
23039         create_function_info_for, create_variable_info_for_1,
23040         create_variable_info_for, intra_create_variable_infos): Adjust.
23041         (init_base_vars): Push NULL for ID zero.
23042         (compute_points_to_sets): Adjust.
23044 2013-03-20  Richard Biener  <rguenther@suse.de>
23046         * cfgloop.c (verify_loop_structure): Streamline and avoid
23047         ICEing on corrupt loop tree.
23048         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23049         loop tree.
23051 2013-03-20  Richard Biener  <rguenther@suse.de>
23053         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23054         check whether an SSA update is needed.
23056 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23058         * config/mips/constraints.md (T): Rename to...
23059         (Yf): ...this.
23060         (U): Rename to...
23061         (Yd): ...this.
23062         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23063         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23065 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23067         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23068         (*subsi3_carryin_uxtw): Likewise.
23070 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23072         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23073         (*rorsi3_insn_uxtw): Likewise.
23075 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23077         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23078         (*extrsi5_insn_uxtw): Likewise.
23080 2013-03-19  Richard Biener  <rguenther@suse.de>
23082         PR tree-optimization/56273
23083         * passes.c (init_optimization_passes): Move second VRP after DOM.
23085 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23087         * config/i386/i386.md (*movti_internal): Merge from
23088         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23089         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23090         nox64 isa attributes.
23092 2013-03-18  Richard Biener  <rguenther@suse.de>
23094         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23095         (unite): Likewise.
23096         (merge_node_constraints): Likewise.
23097         (build_succ_graph): Likewise.
23098         (valid_graph_edge): Inline into single caller.
23099         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23100         and cache varinfo.
23101         (scc_visit): Fix formatting and variable use.
23102         (do_sd_constraint): Use gcc_checking_assert.
23103         (do_ds_constraint): Likewise.
23104         (do_complex_constraint): Likewise.
23105         (condense_visit): Likewise.  Cleanup.
23106         (dump_pred_graph): New function.
23107         (perform_var_substitution): Dump the pred-graph before
23108         variable substitution.
23109         (find_equivalent_node): Use gcc_checking_assert.
23110         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23112 2013-03-18  Richard Biener  <rguenther@suse.de>
23114         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23115         Remove cond_expr_stmt_list argument and do not gimplify the
23116         built expression.
23117         (vect_loop_versioning): Adjust.
23118         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23119         Cleanup to use less temporaries.
23120         (vect_create_data_ref_ptr): Cleanup.
23122 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23124         PR tree-optimization/56635
23125         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23126         require types_compatible_p types.
23128 2013-03-18  Nick Clifton  <nickc@redhat.com>
23130         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23131         spurious backslash.
23133         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23134         Add missing line to comment describing function.
23136 2013-03-18  Richard Biener  <rguenther@suse.de>
23138         PR tree-optimization/56210
23139         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23140         Handle string / character search functions.
23141         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23143 2013-03-18  Richard Biener  <rguenther@suse.de>
23145         PR middle-end/56483
23146         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23147         and implement properly.
23148         * gimple.h (gimple_cond_single_var_p): Remove.
23150 2013-03-18  Richard Biener  <rguenther@suse.de>
23152         * tree-data-ref.h (find_data_references_in_loop): Declare.
23153         * tree-data-ref.c (get_references_in_stmt): Use a stack
23154         vector pre-allocated in the callers.
23155         (find_data_references_in_stmt): Adjust.
23156         (graphite_find_data_references_in_stmt): Likewise.
23157         (create_rdg_vertices): Likewise.
23158         (find_data_references_in_loop): Export.
23159         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23160         Compute dependences here...
23161         (vect_analyze_data_refs): ...not here.  When we encounter
23162         a non-vectorizable data reference in basic-block vectorization
23163         truncate the data reference vector.  Do not bother to
23164         fixup data-dependence information for gather loads.
23165         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23166         of data references, as reported.
23168 2013-03-18  Richard Biener  <rguenther@suse.de>
23170         PR tree-optimization/3713
23171         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23172         has_constants and expr.
23173         (stmt_has_constants): Properly valueize SSA names when deciding
23174         whether the stmt has constants.
23176 2013-03-18  Richard Biener  <rguenther@suse.de>
23178         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23179         whole function when there is nothing to do.
23180         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23181         * tree-vectorizer.c (vectorize_loops): Update virtual and
23182         loop-closed SSA once.
23183         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23185 2013-03-18  Richard Biener  <rguenther@suse.de>
23187         PR middle-end/56113
23188         * domwalk.c (bb_postorder): New global static.
23189         (cmp_bb_postorder): New function.
23190         (walk_dominator_tree): Replace scheme imposing an order for
23191         visiting dominator sons by one sorting them at the time they
23192         are pushed on the stack.
23194 2013-03-18  Richard Biener  <rguenther@suse.de>
23196         PR tree-optimization/39326
23197         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23198         (struct mem_ref): Replace mem member with ao_ref typed member.
23199         (MEM_ANALYZABLE): Adjust.
23200         (memref_eq): Likewise.
23201         (mem_ref_alloc): Likewise.
23202         (gather_mem_refs_stmt): Likewise.
23203         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23204         (execute_sm_if_changed_flag_set): Adjust.
23205         (execute_sm): Likewise.
23206         (ref_always_accessed_p): Likewise.
23207         (refs_independent_p): Likewise.
23208         (can_sm_ref_p): Likewise.
23210 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23212         PR c/56566
23213         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23214         return 1 even for !unsignedp.
23216 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23218         * config/i386/i386.md (isa): Add x64 and nox64.
23219         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23220         (*pushtf): Enable *roF alternative for x64 isa only.
23221         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23222         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23223         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23224         (*movtf_internal): Merge from *movtf_internal_rex64 and
23225         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23226         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23227         nox64 isa attributes.
23228         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23229         nox64 isa attributes.
23230         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23232 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23234         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23236 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23238         PR target/56640
23239         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23240         class names.  Remove trailing comma after "ALL_REGS".
23242 2013-03-16  Jan Hubicka  <jh@suse.cz>
23244         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23245         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23246         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23247         of cgraph_get_create_node.
23248         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23250 2013-03-16  Jason Merrill  <jason@redhat.com>
23252         PR debug/49090
23253         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23254         with DW_AT_default_value.
23256 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23258         * BASE-VER: Set to 4.9.0.
23260 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23262         PR target/56619
23263         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23264         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23265         Document _x* TSX intrinsics.
23267 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23268             David Holsgrove  <david.holsgrove@xilinx.com>
23270         * configure.ac: Add MicroBlaze TLS support detection.
23271         * configure: Regenerate.
23272         * config/microblaze/microblaze-protos.h
23273         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23274         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23275         * config/microblaze/microblaze.c (microblaze_address_type): Add
23276         ADDRESS_TLS and tls_reloc address types.
23277         (microblaze_address_info): Add tls_reloc.
23278         (TARGET_HAVE_TLS): Define.
23279         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23280          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23281          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23282         load_tls_operand,  microblaze_call_tls_get_addr,
23283         microblaze_legitimize_tls_address): New functions.
23284         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23285         (get_base_reg): Use microblaze_tls_symbol_p.
23286         (microblaze_classify_address): Handle TLS.
23287         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23288         label_mentioned_p and microblaze_tls_referenced_p.
23289         (microblaze_legitimize_address): Handle TLS.
23290         (microblaze_address_insns): Handle ADDRESS_TLS.
23291         (pic_address_needs_scratch): Handle TLS.
23292         (print_operand_address): Handle TLS.
23293         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23294         (microblaze_expand_move): Handle TLS.
23295         (microblaze_legitimate_constant_p): Check
23296         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23297         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23298         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23299         (PIC_OFFSET_TABLE_REGNUM): Set.
23300         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23301         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23302         (addsi3, movsi_internal2, movdf_internal): Update constraints
23303         * config/microblaze/predicates.md (arith_plus_operand): Define
23304         (move_operand): Redefine as move_src_operand,
23305         check microblaze_tls_referenced_p.
23307 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23309         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23310         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23312 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23314         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23315         CC mode for AND.
23317 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23319         PR tree-optimization/53265
23320         * common.opt (Waggressive-loop-optimizations): New option.
23321         * tree-ssa-loop-niter.c: Include tree-pass.h.
23322         (do_warn_aggressive_loop_optimizations): New function.
23323         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23324         if number_of_latch_executions returned constant.
23325         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23326         early.  If number_of_latch_executions returned constant, set
23327         nb_iterations_upper_bound back to it.
23328         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23329         field.
23330         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23331         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23333         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23334         (MULTILIB_OSDIRNAMES): Set.
23335         * genmultilib: If defaultosdirname doesn't start with :: , set
23336         defaultosdirname2 instead, clear it and emit two . multilib_raw
23337         entries instead of just one.
23339 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23341         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23342         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23343         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23344         (SUBTARGET_OVERRIDE_OPTIONS): New.
23346 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23348         PR target/49880
23349         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23350         (musermode): Convert to Var(TARGET_USERMODE).
23351         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23352         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23353         * config/sh/sh.c (sh_option_override): Use
23354         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23355         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23356         condition.
23357         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23358         TARGET_SH4.
23359         (udivsi3_i4_single, divsi3_i4_single): Use
23360         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23362 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23364         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23365         default setting.
23367 2013-03-13  Richard Biener  <rguenther@suse.de>
23369         PR tree-optimization/56608
23370         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23371         calls when vectorizing basic-blocks.
23373 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23375         PR plugins/45078
23376         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23377         tm_file.
23379 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23381         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23383 2013-03-11  Jan Hubicka  <jh@suse.cz>
23385         PR lto/56557
23386         * lto-streamer-out.c (output_symbol_p): Skip references from
23387         constructors of external variables.
23389 2013-03-11  Jan Hubicka  <jh@suse.cz>
23391         PR middle-end/56571
23392         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23393         from pseudos.
23394         * emit-rtl.c (verify_rtx_sharing): Likewise.
23395         (copy_insn_1): Likewise.
23396         * rtl.c (copy_rtx): Likewise.
23398 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23400         PR target/56591
23401         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23402         output_operand_lossage message.
23404 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23406         PR target/56470
23407         * arm.c (shift_op): Validate RTL pattern on the fly.
23408         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23409         the RTL.
23411 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23413         PR target/56347
23414         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23415         new call patterns that clobber %fr12.
23416         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23417         split and postreload patterns.
23418         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23419         registers %fr12 and %fr12R as call used.
23421 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23423         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23424         (canon_address, record_store, replace_read, check_mem_read_rtx,
23425         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23426         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23427         rest_of_handle_dse): Likewise.
23429 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23431         PR middle-end/56524
23432         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23433         Add base_optabs.
23434         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23435         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23436         (save_optabs_if_changed): Replace with...
23437         (init_tree_optimization_optabs): ...this.
23438         * optabs.c (save_optabs_if_changed): Rename to...
23439         (init_tree_optimization_optabs): ...this.  Take the optimization node
23440         as argument.  Do nothing if the base optabs are already correct.
23441         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23442         to recompute optabs.
23443         * function.h (function): Remove optabs field.
23444         * function.c (invoke_set_current_function_hook): Call
23445         init_tree_optimization_optabs.  Use the result to initialize
23446         this_fn_optabs.
23448 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23450         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23451         if GTMA_HAS_NO_INSTRUMENTATION.
23452         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23453         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23454         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23455         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23456         GTMA_HAS_NO_INSTRUMENTATION.
23458 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23460         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23461         libasan_preinit.o.
23463 2013-03-08  Marek Polacek  <polacek@redhat.com>
23464             Jakub Jelinek  <jakub@redhat.com>
23466         PR tree-optimization/56478
23467         * predict.c (is_comparison_with_loop_invariant_p): Change the
23468         type of loop_step to tree.
23469         (predict_loops): Adjust.
23470         (predict_iv_comparison): Perform the computations on double_ints.
23472 2013-03-08  Richard Biener  <rguenther@suse.de>
23474         PR tree-optimization/56570
23475         * tree-cfg.c (verify_expr_location_1): Verify locations for
23476         DECL_DEBUG_EXPR.
23477         * tree-sra.c (create_access_replacement): Strip locations
23478         from DECL_DEBUG_EXPRs.
23480 2013-03-08  Richard Biener  <rguenther@suse.de>
23482         * tree-inline.c (expand_call_inline): Do not associate
23483         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23484         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23486 2013-03-08  Richard Biener  <rguenther@suse.de>
23488         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23489         or block changes with -Og.  Fix for location / block encoding
23490         changes and PHI arguments with locations.
23492 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23494         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23495         for all counters.
23496         (struct output_info): Likewise.
23497         (register_overhead): Remove bad gcc_assert.
23498         (bitmap_find_bit): If there is only a single bitmap element, do not
23499         count a miss as a search.
23500         (print_statistics): Update for counter type changes.
23501         (dump_bitmap_statistics): Likewise.  Print headers such that they
23502         are properly lined up with the printed counters.
23504 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23506         PR tree-optimization/56559
23507         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23508         check that it has only a single use.
23510 2013-03-07  Richard Biener  <rguenther@suse.de>
23512         * doc/invoke.texi (fwhole-program): Discourage use in combination
23513         with -flto.
23515 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23517         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23519         PR tree-optimization/56539
23520         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23521         instead of GSI_CONTINUE_LINKING as last argument to
23522         force_gimple_operand_gsi.  Adjust function comment.
23524         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23525         aarch64-cores.def.
23527         PR middle-end/56548
23528         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23529         promoted mode, convert the result back to the original mode.
23531 2013-03-06  Richard Biener  <rguenther@suse.de>
23533         PR middle-end/56294
23534         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23535         (insert_updated_phi_nodes_compare_uids): New function.
23536         (update_ssa): Sort symbols_to_rename after UID before
23537         traversing it to insert PHI nodes.
23539 2013-03-06  Richard Biener  <rguenther@suse.de>
23541         PR middle-end/50494
23542         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23543         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23545         Revert
23546         2013-02-13  Richard Biener  <rguenther@suse.de>
23548         PR lto/50494
23549         * varasm.c (output_constant_def_1): Get the decl representing
23550         the constant as argument.
23551         (output_constant_def): Wrap output_constant_def_1.
23552         (make_decl_rtl): Use output_constant_def_1 with the decl
23553         representing the constant.
23554         (build_constant_desc): Optionally re-use a decl already
23555         representing the constant.
23556         (tree_output_constant_def): Adjust.
23558 2013-03-06  Joey Ye  <joey.ye@arm.com>
23560         PR lto/50293
23561         * gcc.c (convert_white_space): New function.
23562         (main): Handles white space in function name.
23564 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23566         PR target/56529
23567         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23568         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23569         to SH_DIV_CALL_TABLE for TARGET_SH2.
23570         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23571         list.
23572         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23573         call-table options.
23575 2013-03-05  Sterling Augustine  <saugustine@google.com>
23576             Cary Coutant  <ccoutant@google.com>
23578         PR debug/55364
23579         * dwarf2out.c (resolve_addr): Don't call
23580         remove_loc_list_addr_table_entries a second time for the same
23581         expression.
23583 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23585         PR debug/56510
23586         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23587         (avoid_complex_debug_insns): New function.
23588         (expand_debug_locations): Call it.
23590         PR rtl-optimization/56484
23591         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23592         lifetimes of hard registers on small register class machines.
23594 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23596         * config/microblaze/microblaze-protos.h: Rename
23597         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23598         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23599         fast_interrupt.
23600         (microblaze_fast_interrupt_function_p): New function.
23601         (microblaze_is_interrupt_handler): Rename to
23602         microblaze_is_interrupt_variant and add fast_interrupt check.
23603         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23604         (save_restore_insns): Likewise.
23605         (compute_frame_size): Likewise.
23606         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23607         (microblaze_globalize_label): Likewise.
23608         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23609         * config/microblaze/microblaze.md: Use wrapper
23610         microblaze_is_interrupt_variant.
23612 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23614         * sdbout.c (sdbout_one_type): Switch to current function's section
23615         supporting cold/hot.
23617 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23619         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23620         -mxl-reorder.
23622 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23624         PR middle-end/56461
23625         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23626         if VALGRIND_GET_VBITS is defined, temporarily make object
23627         memory all defined, and restore previous valgrind addressability
23628         and definability afterwards.  Free this_object at the end.
23630         PR middle-end/56461
23631         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23632         right before calling lra_create_live_ranges, also call it
23633         when clearing live_p.  Only call lra_clear_live_ranges
23634         at the end if live_p.
23636         PR middle-end/56461
23637         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23639 2013-03-05  Richard Biener  <rguenther@suse.de>
23641         PR tree-optimization/56521
23642         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23643         value-id.
23645 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23647         PR c++/55135
23648         * except.h (remove_unreachable_eh_regions): New prototype.
23649         * except.c (remove_eh_handler_splicer): New function, split out
23650         of remove_eh_handler.
23651         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23652         warning about running it on many EH regions one at a time.
23653         (remove_unreachable_eh_regions_worker): New function, walk the
23654         EH tree in depth-first order and remove non-marked regions.
23655         (remove_unreachable_eh_regions): New function.
23656         * tree-eh.c (mark_reachable_handlers): New function, split out
23657         from remove_unreachable_handlers.
23658         (remove_unreachable_handlers): Use mark_reachable_handlers and
23659         remove_unreachable_eh_regions.
23660         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23661         and remove_unreachable_eh_regions.
23663 2013-03-05  Richard Biener  <rguenther@suse.de>
23665         PR middle-end/56525
23666         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23667         not freeing them until the end.
23669 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23671         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23673 2013-03-05  Richard Biener  <rguenther@suse.de>
23675         PR tree-optimization/56270
23676         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23677         of loads after scheduling an SLP instance.
23679 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23681         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23682         tic6x.exp.
23683         (check_gcc_parallelize): Run guality.exp as a separate job from
23684         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23685         struct-layout-1.exp with stackalign.exp.
23687         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23689         PR middle-end/56461
23690         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23691         load_index sbitmap even if some bit in it isn't set.
23693         PR middle-end/56461
23694         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23695         (discover_iteration_bound_by_body_walk): Change queues to
23696         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23697         spelling in comment.  Call safe_push on queues[bound_index] directly.
23698         Release queues[queue_index] in every iteration unconditionally.
23699         Release bounds vector.
23701         PR middle-end/56461
23702         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23703         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23704         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23705         inner_phis vector.
23707 2013-03-05  Richard Biener  <rguenther@suse.de>
23709         PR lto/56515
23710         * tree-inline.c (remap_blocks_to_null): New function.
23711         (expand_call_inline): When expanding a call stmt without
23712         an associated block inline remap all callee blocks to NULL.
23714 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23716         PR rtl-optimization/56494
23717         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23718         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23719         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23721         PR middle-end/56461
23722         * sel-sched-ir.c (free_sched_pools): Release
23723         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23724         if succs_info_pool.max_top isn't -1.
23726         PR bootstrap/56509
23727         * opts.c (opts_obstack, opts_concat): Moved to...
23728         * opts-common.c (opts_obstack, opts_concat): ... here.
23730 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23732         PR middle-end/56461
23733         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23735 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23737         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23738         all appropriate places.
23740 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23742         PR tree-optimization/56424
23743         * ipa-split.c (split_function): Do not set the RSO flag if result is
23744         not by reference and its type is a register type.
23746 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23748         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23749         (microblaze_legitimate_pic_operand): Likewise
23750         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23751         new function microblaze_legitimate_pic_operand
23752         * config/microblaze/microblaze-protos.h
23753         (microblaze_legitimate_pic_operand): Declare.
23755 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23757         * config/microblaze/predicates.md (call_insn_simple_operand):
23758         New predicate for supported rtx code types.
23759         * config/microblaze/microblaze.md (call_internal1): Use
23760         call_insn_simple_operand predicate.
23762 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23764         PR middle-end/56461
23765         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23766         partitions.ordered_remove.
23768         PR middle-end/56461
23769         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23770         vec_oprnds0.create (1) for modifier == NONE.
23772         PR middle-end/56461
23773         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23774         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23775         vec_oprnds1 right before pushing anything to it for
23776         scalar_shift_arg.
23778         PR middle-end/56461
23779         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23780         set nbbs to 0 instead of having separate code path.
23781         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23782         instead of false as last argument if returning NULL.
23784 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23786         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23787         the attribute is now called "target" instead of "option".
23788         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23789         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23790         attribute/pragma name for TARGET_OPTION_VALID_P and
23791         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23792         * doc/tm.texi: Regenerated.
23794 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23796         * config/microblaze/microblaze.c:
23797         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23798         * config/microblaze/microblaze.h: Add -mxl-reorder to
23799         DRIVER_SELF_SPECS.
23800         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23801         instructions emitted if TARGET_REORDER.
23802         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23803         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23804         separately.
23806 2013-03-01  Xinliang David Li  <davidxl@google.com>
23808         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23809         walk length.
23811 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23813         PR middle-end/56461
23814         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23815         vector even when returning true.  Fix up function comment formatting.
23817         PR middle-end/56461
23818         * ira-build.c (ira_loop_nodes_count): New variable.
23819         (create_loop_tree_nodes): Initialize it.
23820         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23822         PR middle-end/56461
23823         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23824         method on dr_chain and result_chain.
23825         * tree-vect-stmts.c (vectorizable_store): Only call
23826         result_chain.create if j == 0.
23828         PR middle-end/56461
23829         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23830         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23831         before overwriting it.
23833 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23835         * doc/extended.texi (C Extensions): Change order in @menu
23836         to match @node.
23837         (Other MIPS Built-in Functions): Move last MIPS entry before
23838         "picoChip Built-in Functions".
23839         (SH Built-in Functions): Move after RX Built-in Functions.
23840         * doc/gcc.texi (Introduction): Change order in @menu
23841         to match @node.
23842         * doc/md.texi (Constraints): Ditto.
23843         * gty.texi (Type Information): Ditto.
23844         (User-provided marking routines for template types): Make
23845         subsection.
23846         * doc/invoke.texi (AArch64 Options): Move before
23847         "Adapteva Epiphany Options".
23849 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23850             Jakub Jelinek  <jakub@redhat.com>
23852         PR sanitizer/56454
23853         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23854         no_address_safety_analysis attribute.
23855         * doc/extend.texi (no_address_safety_attribute): Rename to
23856         no_sanitize_address attribute, mention no_address_safety_analysis
23857         attribute as deprecated alias.
23859 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23861         PR middle-end/56461
23862         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23863         type to vec<vec<tree> > *.
23864         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23865         to be vec<tree> instead of vec<tree> *, set vec_defs
23866         to vNULL and call vec_defs.create (number_of_vects), adjust other
23867         uses of vec_defs.
23868         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23869         vectorizable_condition): Adjust vect_get_slp_defs callers.
23871 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23873         * config/aarch64/aarch64.c
23874         (aarch64_float_const_representable): Remove unused variable.
23876 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23878         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23880 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23882         * config/aarch64/aarch64-builtins.c
23883         (aarch64_init_simd_builtins): Make static.
23885 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23887         * config/aarch64/aarch64.c
23888         (aarch64_simd_make_constant): Make static.
23890 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23892         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23893         with no initialization to the RHS of debug statements.
23895 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23897         PR tree-optimization/56294
23898         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23899         Adjust dumping.
23900         (get_access_replacement): Do not call create_access_replacement.
23901         Assert a replacement exists.
23902         (get_repl_default_def_ssa_name): Create the replacement declaration
23903         itself.
23905 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23907         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23908         final_end_function.
23910 2013-02-28  Marek Polacek  <polacek@redhat.com>
23912         PR rtl-optimization/56466
23913         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23914         if we're changing a loop.
23915         (peel_loops_completely): Likewise.
23917 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23919         PR c++/55813
23920         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23922 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23924         PR target/56445
23925         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23926         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23927         INTX_FTYPE_FX, FX_FTYPE_INTX.
23928         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23930 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23932         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23933         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23934         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23935         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23936         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23937         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23938         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23939         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23940         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23941         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23942         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23943         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23944         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23945         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23946         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23947         (avrxmega6): Increase max flash segments from 5 to 6.
23948         * config/avr/t-multilib: Regenerate.
23949         * config/avr/avr-tables.opt: Regenerate.
23950         * doc/avr-mmcu.texi: Regenerate.
23952 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23954         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23955         (avr_device_to_arch): Rename to avr_device_to_ld.
23956         (avr_device_to_as): New prototype.
23957         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23958         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23959         * config/avr/driver-avr.c (avr_device_to_as): New.
23960         (avr_device_to_arch): Rename to avr_device_to_ld.
23962 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23964         PR middle-end/56461
23965         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23966         method on dr_chain and result_chain.
23968         PR middle-end/56461
23969         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23970         pointer_set_destroy on not_executed_last_iteration.
23972         PR middle-end/56461
23973         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23975         PR middle-end/56461
23976         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23977         FOR_EACH_DEFINED_FUNCTION when freeing state.
23979         PR middle-end/56461
23980         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23981         pool_free.
23982         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23983         overwriting it.
23985         PR middle-end/56461
23986         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23987         known_aggs[i].items and release known_aggs vector.
23989         PR middle-end/56461
23990         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23992 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23994         * config/microblaze/microblaze.c (microblaze_emit_compare):
23995         Use xor for EQ/NE comparisions.
23996         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23997         (cbranchsf4): Adjust operator to comparison_operator.
23999 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24001         PR middle-end/56461
24002         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
24003         vector.
24004         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
24005         vec_safe_push, always update *slot.
24006         (redirect_edge_var_map_clear): Use vec_free.
24007         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
24008         (free_var_map_entry): Use vec_free.
24009         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
24010         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
24012 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
24014         PR middle-end/45472
24015         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
24016         when the may_trap_p bit of the exprs being merged differs.
24017         Reorder tests for speculativeness in the logical and operator.
24019 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24021         * incpath.c (add_standard_paths): Use reconcat instead of concat
24022         where appropriate and avoid leaking memory.
24024         * opts.h: Include obstack.h.
24025         (opts_concat): New prototype.
24026         (opts_obstack): New declaration.
24027         * opts.c (opts_concat): New function.
24028         (opts_obstack): New variable.
24029         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24030         (finish_options): Use opts_concat instead of concat
24031         and XOBNEWVEC instead of XNEWVEC.
24032         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24033         generate_option): Likewise.
24034         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24035         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24037         PR target/56455
24038         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24039         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24041 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24043         PR middle-end/56461
24044         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24046 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24048         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24049         (arm_block_move_unaligned_straight): Likewise.
24050         (arm_adjust_block_mem): Likewise.
24052 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24054         PR target/48901
24055         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24056         temp, cond and label.
24057         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24059         PR target/52500
24060         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24061         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24063         PR target/52501
24064         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24065         of prologue/epilogue functions.
24067         PR target/52550
24068         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24069         Remove unused variable cfa_offset.
24070         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24072         PR target/54639
24073         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24074         type promotion to unsigned.
24076         PR target/54640
24077         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24078         for HOST_WIDE_INT of 32 bit / same size as int.
24079         (arm_block_move_unaligned_straight): Likewise.
24080         (arm_adjust_block_mem): Likewise.
24082         PR target/54662
24083         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24084         ALL_CFLAGS.
24086 2013-02-26  Marek Polacek  <polacek@redhat.com>
24088         PR tree-optimization/56426
24089         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24091 2013-02-26  Richard Biener  <rguenther@suse.de>
24093         PR target/56444
24094         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24095         unused variable loops.
24097 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24099         PR tree-optimization/56448
24100         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24101         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24102         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24103         later operands of the references, or even first operand for
24104         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24106         PR tree-optimization/56443
24107         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24108         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24109         to type_for_mode langhook.
24111 2013-02-25  Matt Turner  <mattst88@gmail.com>
24113         * doc/invoke.texi: Document r4700.
24115 2013-02-25  Richard Biener  <rguenther@suse.de>
24117         PR tree-optimization/56175
24118         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24119         split out from ...
24120         (simplify_bitwise_binary): ... here.  Also guard the conversion
24121         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24123 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24125         Revert:
24126         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24127             Maciej W. Rozycki  <macro@codesourcery.com>
24128             Tom de Vries  <tom@codesourcery.com>
24129             Nathan Sidwell  <nathan@codesourcery.com>
24130             Iain Sandoe  <iain@codesourcery.com>
24131             Nathan Froyd  <froydnj@codesourcery.com>
24132             Chao-ying Fu  <fu@mips.com>
24134         * doc/extend.texi: (micromips, nomicromips, nocompression):
24135         Document new function attributes.
24136         * doc/invoke.texi (minterlink-compressed, mmicromips,
24137         m14k, m14ke, m14kec): Document new options.
24138         (minterlink-mips16): Update documentation.
24139         * doc/md.texi (ZC, ZD): Document new constraints.
24140         * configure.ac (gcc_cv_as_micromips): Check if linker
24141         supports the .set micromips directive.
24142         * configure: Regenerate.
24143         * config.in: Regenerate.
24144         * config/mips/mips-tables.opt: Regenerate.
24145         * config/mips/micromips.md: New file.
24146         * constraints.md (ZC, AD): New constraints.
24147         * config/mips/predicates.md (movep_src_register): New predicate.
24148         (movep_src_operand): New predicate.
24149         (non_volatile_mem_operand): New predicate.
24150         * config/mips/mips.md (multimem): New type.
24151         (length): Differentiate between 17-bit and 18-bit branch offsets.
24152         (MOVEP1, MOVEP2): New mode iterator.
24153         (mov_<load>l): Use ZC constraint.
24154         (mov_<load>r): Likewise.
24155         (mov_<store>l): Likewise.
24156         (mov_<store>r): Likewise.
24157         (*branch_equality<mode>_inverted): Add microMIPS support.
24158         (*branch_equality<mode>): Likewise.
24159         (*jump_absolute): Likewise.
24160         (indirect_jump_<mode>): Likewise.
24161         (tablejump_<mode>): Likewise.
24162         (<optab>_internal): Likewise.
24163         (sibcall_internal): Likewise.
24164         (sibcall_value_internal): Likewise.
24165         (prefetch): Use constraint ZD.
24166         * config/mips/mips.opt (minterlink-compressed): New option.
24167         (minterlink-mips16): Now an alias for minterlink-compressed.
24168         (mmicromips): New option.
24169         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24170         (compare_and_swap_12): Likewise.
24171         (sync_add<mode>): Likewise.
24172         (sync_<optab>_12): Likewise.
24173         (sync_old_<optab>_12): Likewise.
24174         (sync_new_<optab>_12): Likewise.
24175         (sync_nand_12): Likewise.
24176         (sync_old_nand_12): Likewise.
24177         (sync_new_nand_12): Likewise.
24178         (sync_sub<mode>): Likewise.
24179         (sync_old_add<mode>): Likewise.
24180         (sync_old_sub<mode>): Likewise.
24181         (sync_new_add<mode>): Likewise.
24182         (sync_new_sub<mode>): Likewise.
24183         (sync_<optab><mode>): Likewise.
24184         (sync_old_<optab><mode>): Likewise.
24185         (sync_new_<optab><mode>): Likewise.
24186         (sync_nand<mode>): Likewise.
24187         (sync_old_nand<mode>): Likewise.
24188         (sync_new_nand<mode>): Likewise.
24189         (sync_lock_test_and_set<mode>): Likewise.
24190         (test_and_set_12): Likewise.
24191         (atomic_compare_and_swap<mode>): Likewise.
24192         (atomic_exchange<mode>_llsc): Likewise.
24193         (atomic_fetch_add<mode>_llsc): Likewise.
24194         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24195         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24196         (umips_save_restore_pattern_p): Likewise.
24197         (umips_load_store_pair_p): Likewise.
24198         (umips_output_load_store_pair): Likewise.
24199         (umips_movep_target_p): Likewise.
24200         (umips_12bit_offset_address_p): Likewise.
24201         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24202         (mips_base_mips16): Rename this...
24203         (mips_base_compression_flags): ...to this. Update all uses.
24204         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24205         (mips_mips16_decl_p): Delete.
24206         (mips_nomips16_decl_p): Delete.
24207         (mips_get_compress_on_flags): New function.
24208         (mips_get_compress_off_flags): New function.
24209         (mips_get_compress_mode): New function.
24210         (mips_get_compress_on_name): New function.
24211         (mips_get_compress_off_name): New function.
24212         (mips_insert_attributes): Support multiple compression types.
24213         (mips_merge_decl_attributes): Likewise.
24214         (umips_12bit_offset_address_p): New function.
24215         (mips_start_function_definition): Emit .set micromips directive.
24216         (mips_call_may_need_jalx_p): New function.
24217         (mips_function_ok_for_sibcall): Add microMIPS support.
24218         (mips_print_operand_punctuation): Support short delay slots and
24219         compact jumps.
24220         (umips_swm_mask, umips_swm_encoding): New.
24221         (umips_build_save_restore): New function.
24222         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24223         (was_mips16_p): Remove.
24224         (old_compression_mode): New.
24225         (mips_set_compression_mode): New function.
24226         (mips_set_current_function): Add microMIPS support.
24227         (mips_option_override): Likewise.
24228         (umips_save_restore_pattern_p): New function.
24229         (umips_output_save_restore): New function.
24230         (umips_load_store_pair_p_1): New function.
24231         (umips_load_store_pair_p): New function.
24232         (umips_output_load_store_pair_1): New function.
24233         (umips_output_load_store_pair): New function.
24234         (umips_movep_target_p) New function.
24235         (mips_prepare_pch_save): Add microMIPS support.
24236         * config/mips/mips.h (TARGET_COMPRESSION): New.
24237         (TARGET_CPU_CPP_BUILTINS): Update macro
24238         to use new compression flags and to support microMIPS.
24239         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24240         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24241         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24242         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24243         (ASM_SPEC): Support mmicromips and mno-micromips.
24244         (M16STORE_REG_P): New macro.
24245         (MIPS_CALL): Support TARGET_MICROMIPS.
24246         (MICROMIPS_J): New macro.
24247         (mips_base_mips16): Rename this...
24248         (mips_base_compression_flags): ...to this.
24249         (UMIPS_12BIT_OFFSET_P): New macro.
24250         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24251         (MULTILIB_DIRNAMES): Likewise.
24253 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24255         PR rtl-optimization/56131
24256         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24257         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24258         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24260 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24262         * doc/invoke.texi (-fsanitize=): Move from optimization
24263         to debugging options.
24265 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24267         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24269 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24270             Alexander Monakov  <amonakov@ispras.ru>
24272         PR middle-end/56077
24273         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24274         flush pending lists also on non-jumps.  Adjust comment.
24276 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24277             Maciej W. Rozycki  <macro@codesourcery.com>
24278             Tom de Vries  <tom@codesourcery.com>
24279             Nathan Sidwell  <nathan@codesourcery.com>
24280             Iain Sandoe  <iain@codesourcery.com>
24281             Nathan Froyd  <froydnj@codesourcery.com>
24282             Chao-ying Fu  <fu@mips.com>
24284         * doc/extend.texi: (micromips, nomicromips, nocompression):
24285         Document new function attributes.
24286         * doc/invoke.texi (minterlink-compressed, mmicromips,
24287         m14k, m14ke, m14kec): Document new options.
24288         (minterlink-mips16): Update documentation.
24289         * doc/md.texi (ZC, ZD): Document new constraints.
24290         * configure.ac (gcc_cv_as_micromips): Check if linker
24291         supports the .set micromips directive.
24292         * configure: Regenerate.
24293         * config.in: Regenerate.
24294         * config/mips/mips-tables.opt: Regenerate.
24295         * config/mips/micromips.md: New file.
24296         * constraints.md (ZC, AD): New constraints.
24297         * config/mips/predicates.md (movep_src_register): New predicate.
24298         (movep_src_operand): New predicate.
24299         (non_volatile_mem_operand): New predicate.
24300         * config/mips/mips.md (multimem): New type.
24301         (length): Differentiate between 17-bit and 18-bit branch offsets.
24302         (MOVEP1, MOVEP2): New mode iterator.
24303         (mov_<load>l): Use ZC constraint.
24304         (mov_<load>r): Likewise.
24305         (mov_<store>l): Likewise.
24306         (mov_<store>r): Likewise.
24307         (*branch_equality<mode>_inverted): Add microMIPS support.
24308         (*branch_equality<mode>): Likewise.
24309         (*jump_absolute): Likewise.
24310         (indirect_jump_<mode>): Likewise.
24311         (tablejump_<mode>): Likewise.
24312         (<optab>_internal): Likewise.
24313         (sibcall_internal): Likewise.
24314         (sibcall_value_internal): Likewise.
24315         (prefetch): Use constraint ZD.
24316         * config/mips/mips.opt (minterlink-compressed): New option.
24317         (minterlink-mips16): Now an alias for minterlink-compressed.
24318         (mmicromips): New option.
24319         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24320         (compare_and_swap_12): Likewise.
24321         (sync_add<mode>): Likewise.
24322         (sync_<optab>_12): Likewise.
24323         (sync_old_<optab>_12): Likewise.
24324         (sync_new_<optab>_12): Likewise.
24325         (sync_nand_12): Likewise.
24326         (sync_old_nand_12): Likewise.
24327         (sync_new_nand_12): Likewise.
24328         (sync_sub<mode>): Likewise.
24329         (sync_old_add<mode>): Likewise.
24330         (sync_old_sub<mode>): Likewise.
24331         (sync_new_add<mode>): Likewise.
24332         (sync_new_sub<mode>): Likewise.
24333         (sync_<optab><mode>): Likewise.
24334         (sync_old_<optab><mode>): Likewise.
24335         (sync_new_<optab><mode>): Likewise.
24336         (sync_nand<mode>): Likewise.
24337         (sync_old_nand<mode>): Likewise.
24338         (sync_new_nand<mode>): Likewise.
24339         (sync_lock_test_and_set<mode>): Likewise.
24340         (test_and_set_12): Likewise.
24341         (atomic_compare_and_swap<mode>): Likewise.
24342         (atomic_exchange<mode>_llsc): Likewise.
24343         (atomic_fetch_add<mode>_llsc): Likewise.
24344         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24345         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24346         (umips_save_restore_pattern_p): Likewise.
24347         (umips_load_store_pair_p): Likewise.
24348         (umips_output_load_store_pair): Likewise.
24349         (umips_movep_target_p): Likewise.
24350         (umips_12bit_offset_address_p): Likewise.
24351         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24352         (mips_base_mips16): Rename this...
24353         (mips_base_compression_flags): ...to this. Update all uses.
24354         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24355         (mips_mips16_decl_p): Delete.
24356         (mips_nomips16_decl_p): Delete.
24357         (mips_get_compress_on_flags): New function.
24358         (mips_get_compress_off_flags): New function.
24359         (mips_get_compress_mode): New function.
24360         (mips_get_compress_on_name): New function.
24361         (mips_get_compress_off_name): New function.
24362         (mips_insert_attributes): Support multiple compression types.
24363         (mips_merge_decl_attributes): Likewise.
24364         (umips_12bit_offset_address_p): New function.
24365         (mips_start_function_definition): Emit .set micromips directive.
24366         (mips_call_may_need_jalx_p): New function.
24367         (mips_function_ok_for_sibcall): Add microMIPS support.
24368         (mips_print_operand_punctuation): Support short delay slots and
24369         compact jumps.
24370         (umips_swm_mask, umips_swm_encoding): New.
24371         (umips_build_save_restore): New function.
24372         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24373         (was_mips16_p): Remove.
24374         (old_compression_mode): New.
24375         (mips_set_compression_mode): New function.
24376         (mips_set_current_function): Add microMIPS support.
24377         (mips_option_override): Likewise.
24378         (umips_save_restore_pattern_p): New function.
24379         (umips_output_save_restore): New function.
24380         (umips_load_store_pair_p_1): New function.
24381         (umips_load_store_pair_p): New function.
24382         (umips_output_load_store_pair_1): New function.
24383         (umips_output_load_store_pair): New function.
24384         (umips_movep_target_p) New function.
24385         (mips_prepare_pch_save): Add microMIPS support.
24386         * config/mips/mips.h (TARGET_COMPRESSION): New.
24387         (TARGET_CPU_CPP_BUILTINS): Update macro
24388         to use new compression flags and to support microMIPS.
24389         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24390         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24391         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24392         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24393         (ASM_SPEC): Support mmicromips and mno-micromips.
24394         (M16STORE_REG_P): New macro.
24395         (MIPS_CALL): Support TARGET_MICROMIPS.
24396         (MICROMIPS_J): New macro.
24397         (mips_base_mips16): Rename this...
24398         (mips_base_compression_flags): ...to this.
24399         (UMIPS_12BIT_OFFSET_P): New macro.
24400         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24401         (MULTILIB_DIRNAMES): Likewise.
24403 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24405         PR target/52555
24406         * target-globals.c (save_target_globals): For init_reg_sets and
24407         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24409 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24411         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24412         * config/aarch64/t-aarch64
24413         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24415 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24417         PR inline-asm/56148
24418         * lra-constraints.c (process_alt_operands): Reload operand
24419         conflicting with earlier clobber only if no more other conflicting
24420         operands.
24422 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24424         PR sanitizer/56393
24425         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24426         if not linking a shared library.
24428 2013-02-22  Seth LaForge  <sethml@google.com>
24430         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24432 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24434         * config/arm/arm.md (split for extendsidi): Update condition.
24435         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24436         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24437         (qhs_zextenddi_cstr): Likewise.
24439 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24441         PR middle-end/56420
24442         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24443         avoid signed wrapping.
24444         (expand_mult): Handle properly multiplication by
24445         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24446         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24447         in the compiler if coeff is HOST_WIDE_INT_MIN.
24448         (expand_divmod): Don't make ext_op1 static, change it's type to
24449         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24451         PR rtl-optimization/50339
24452         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24453         field.
24454         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24455         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24456         into splitting_ashiftrt field.
24457         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24458         ASHIFTRT.
24459         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24460         choices.
24462 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24464         PR middle-end/56108
24465         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24466         are sure to go irrevocable.
24468 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24470         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24471         scalars are valid operands.
24473 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24475         PR tree-optimization/56310
24476         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24477         only matching indices and non-negative final offsets.
24478         (intersect_aggregates_with_edge): Pass src_idx to
24479         agg_replacements_to_vector.  Pass src_idx insstead of index to
24480         intersect_with_agg_replacements.
24482 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24484         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24485         instead of hard-wired defaults.
24487 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24489         * doc/invoke.texi (MIPS Options): Update documentation of the
24490         floating-point multiply-accumulate instruction restrictions.
24492 2013-02-21  Kostya Serebryany  <kcc@google.com>
24494         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24495         asan_shadow_offset on x86_64 linux.
24497 2013-02-21  Richard Biener  <rguenther@suse.de>
24499         PR tree-optimization/56415
24500         Revert
24501         2013-02-11  Richard Biener  <rguenther@suse.de>
24503         PR tree-optimization/56273
24504         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24505         first VRP run.
24507 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24509         PR bootstrap/56258
24510         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24511         instead of @itemx.
24513         PR inline-asm/56405
24514         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24515         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24517 2013-02-20  Jan Hubicka  <jh@suse.cz>
24519         PR tree-optimization/56265
24520         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24521         when target is referenced for first time.
24523 2013-02-20  Richard Biener  <rguenther@suse.de>
24525         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24526         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24527         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24528         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24529         not return anything.
24530         (rename_ssa_copies): Do not remove unused locals.
24531         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24532         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24533         * passes.c (execute_function_todo): Do not schedule unused locals
24534         removal if cleanup_tree_cfg did something.
24535         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24536         about the number of removed locals.
24538 2013-02-20  Richard Biener  <rguenther@suse.de>
24540         PR tree-optimization/56398
24541         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24543 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24545         PR tree-optimization/55334
24546         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24547         restricted pointers to arrays.
24549 2013-02-20  Richard Biener  <rguenther@suse.de>
24550             Jakub Jelinek  <jakub@redhat.com>
24552         PR tree-optimization/56396
24553         * tree-ssa-ccp.c (n_const_val): New static variable.
24554         (get_value): Return NULL for SSA names we don't have a lattice
24555         entry for.
24556         (ccp_initialize): Initialize n_const_val.
24557         * tree-ssa-copy.c (n_copy_of): New static variable.
24558         (init_copy_prop): Initialize n_copy_of.
24559         (get_value): Return NULL_TREE for SSA names we don't have a
24560         lattice entry for.
24562 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24564         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24566 2013-02-20  Richard Biener  <rguenther@suse.de>
24568         * genpreds.c (write_lookup_constraint): Do not compare first
24569         letter of the constraint again.
24571 2013-02-20  Richard Biener  <rguenther@suse.de>
24573         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24574         and ceil_log2.
24575         (get_use_iv_cost): Terminate hashtable walk when coming across
24576         an empty entry.
24578 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24580         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24581         reassociation for avx2 targets.
24583 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24585         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24586         Add version check for v8.10.a to enable microblaze_has_clz
24587         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24588         version and TARGET_PATTERN_COMPARE check
24589         * config/microblaze/microblaze.md: New clzsi2 instruction
24591 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24593         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24594         function before branching.
24596 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24598         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24599         DUMP_INSN_RTX_UID.
24600         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24602 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24604         PR middle-end/55889
24605         * sel-sched.c: Include ira.h.
24606         (implicit_clobber_conflict_p): New function.
24607         (moveup_expr): Use it.
24608         * Makefile.in (sel-sched.o): Depend on ira.h.
24610 2013-02-19  Richard Biener  <rguenther@suse.de>
24612         PR tree-optimization/56384
24613         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24614         (vn_hash_type): Split out from ...
24615         (vn_hash_constant_with_type): ... here.
24616         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24617         (vn_phi_eq): Compare types from vn_phi_s structure.
24618         (vn_phi_lookup): Populate vn_phi_s type.
24619         (vn_phi_insert): Likewise.
24621 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24623         PR tree-optimization/56350
24624         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24625         if haven't found reduction or nested cycle operand, rather than
24626         asserting we must find it.
24628         PR tree-optimization/56381
24629         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24630         to fold_build3.
24632 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24633             Jakub Jelinek  <jakub@redhat.com>
24635         PR target/52555
24636         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24637         (swap_optab_enable): Same.
24638         (init_all_optabs): Use argument instead of global.
24639         * tree.h (struct tree_optimization_option): New field target_optabs.
24640         * expr.h (init_all_optabs): Add argument to prototype.
24641         (TREE_OPTIMIZATION_OPTABS): New.
24642         (save_optabs_if_changed): Protoize.
24643         * optabs.h: Declare this_fn_optabs.
24644         * optabs.c (save_optabs_if_changed): New.
24645         Declare this_fn_optabs.
24646         (init_optabs): Add argument to init_all_optabs() call.
24647         * function.c (invoke_set_current_function_hook): Handle per
24648         function optabs.
24649         * function.h (struct function): New field optabs.
24650         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24651         optimization_current_node has changed.
24652         * target-globals.h (save_target_globals_default_opts): Protoize.
24653         * target-globals.c (save_target_globals_default_opts): New.
24655 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24657         PR target/56347
24658         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24659         registers %fr12 and %fr12R as call used.
24661         PR target/56214
24662         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24663         and HImode, require all displacements to be an integer multiple of
24664         their mode size.
24665         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24666         only allow QImode and HImode when reload is in progress and strict is
24667         true.  Likewise for symbolic addresses.  Use base14_operand to check
24668         displacements in REG+BASE addresses.
24670 2013-02-18  Richard Biener  <rguenther@suse.de>
24672         PR tree-optimization/56366
24673         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24674         handle sign-conversion of outer-loop initial induction value.
24676 2013-02-18  Richard Biener  <rguenther@suse.de>
24678         PR middle-end/56349
24679         * cfghooks.c (merge_blocks): If we merge a latch into another
24680         block adjust references to it.
24681         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24682         (verify_loop_structure): Verify that a recorded latch is in fact
24683         a latch.
24685 2013-02-18  Richard Biener  <rguenther@suse.de>
24687         PR tree-optimization/56321
24688         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24689         order SSA name release and virtual operand unlinking.
24691 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24693         * config/microblaze/microblaze.md (save_stack_block): Define.
24694         (restore_stack_block): Likewise.
24696 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24698         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24699         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24700         * config/microblaze/microblaze.c (microblaze_option_override):
24701         Bail out early for PIC modes when target does not support PIC.
24703 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24705         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24706         Replace with a microblaze version.
24707         (microblaze_trampoline_init): Adapt for microblaze.
24708         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24709         microblaze.
24711 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24712             Dodji Seketeli  <dodji@redhat.com>
24714         PR asan/56330
24715         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24716         (instrument_mem_region_access): Do not forget to always put
24717         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24718         0) statement, even for cases where either 'base' or 'base + len'
24719         are not instrumented -- because they have been previously
24720         instrumented.  Simplify the logic by putting all the statements
24721         instrument 'base + len' inside a sequence, and then insert that
24722         sequence right before the current insertion point.  Then, to
24723         instrument 'base + len', just get an iterator on that statement.
24724         And do not forget to update the pointer to iterator the function
24725         received as argument.
24727 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24729         PR rtl-optimization/56348
24730         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24732 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24734         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24735         (clean_graph_dump_file): Pass base to start_graph_dump.
24737 2013-02-14  Richard Henderson  <rth@redhat.com>
24739         PR target/55941
24740         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24742 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24744         * collect2-aix.h: Define F_LOADONLY.
24746 2013-02-14  Richard Biener  <rguenther@suse.de>
24748         PR lto/50494
24749         * varasm.c (output_constant_def_1): Get the decl representing
24750         the constant as argument.
24751         (output_constant_def): Wrap output_constant_def_1.
24752         (make_decl_rtl): Use output_constant_def_1 with the decl
24753         representing the constant.
24754         (build_constant_desc): Optionally re-use a decl already
24755         representing the constant.
24756         (tree_output_constant_def): Adjust.
24758 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24760         Fix an asan crash
24761         * asan.c (instrument_builtin_call):  Really put the length of the
24762         second source argument into src1_len.
24764 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24766         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24767         argument.  If it is false, don't create edge from then_bb to
24768         fallthru_bb.
24769         (insert_if_then_before_iter): Pass true to it.
24770         (build_check_stmt): Pass false to it.
24771         (transform_statements): Flush hash table only on extended basic
24772         block boundaries, rather than at the beginning of every bb.
24773         Don't flush hash table on nonfreeing_call_p calls.
24774         * tree-flow.h (nonfreeing_call_p): New prototype.
24775         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24777 2013-02-13  David S. Miller  <davem@davemloft.net>
24779         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24781 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24783         PR target/56184
24784         * ira.c (max_regno_before_ira): Move from ...
24785         (ira): ... here.
24786         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24787         vec_safe_length.
24789 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24791         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24793 2013-02-13  Richard Biener  <rguenther@suse.de>
24795         PR lto/56295
24796         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24797         globals in MEM_REFs.
24799 2013-02-13  Richard Biener  <rguenther@suse.de>
24801         * loop-init.c (loop_optimizer_init): Clear loop state when
24802         re-initializing preserved loops.
24803         * loop-unswitch.c (unswitch_single_loop): Return whether
24804         we unswitched the loop.  Do not verify loop state here.
24805         (unswitch_loops): When we unswitched a loop discover new loops.
24807 2013-02-13  Kostya Serebryany  <kcc@google.com>
24809         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24810         on x86_64 linux.
24811         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24813 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24815         Avoid instrumenting duplicated memory access in the same basic block
24816         * Makefile.in (asan.o): Add new dependency on hash-table.h
24817         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24818         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24819         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24820         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24821         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24822         (get_mem_ref_of_assignment): New functions.
24823         (get_mem_refs_of_builtin_call): Extract from
24824         instrument_builtin_call and tweak a little bit to make it fit with
24825         the new signature.
24826         (instrument_builtin_call): Use the new
24827         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24828         of is_gimple_builtin_call.
24829         (instrument_derefs, instrument_mem_region_access): Insert the
24830         instrumented memory reference into the hash table.
24831         (maybe_instrument_assignment): Renamed instrument_assignment into
24832         this, and change it to advance the iterator when instrumentation
24833         actually happened and return true in that case.  This makes it
24834         homogeneous with maybe_instrument_assignment, and thus give a
24835         chance to callers to be more 'regular'.
24836         (transform_statements): Clear the memory reference hash table
24837         whenever we enter a new BB, when we cross a function call, or when
24838         we are done transforming statements.  Use
24839         maybe_instrument_assignment instead of instrumentation.  No more
24840         need to special case maybe_instrument_assignment and advance the
24841         iterator after calling it; it's now handled just like
24842         maybe_instrument_call.  Update comment.
24844 2013-02-13  Richard Biener  <rguenther@suse.de>
24846         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24847         Fix loop discovery code.
24849 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24851         PR inline-asm/56148
24852         * lra-constraints.c (process_alt_operands): Match early clobber
24853         operand with itself.  Check conflicts with earlyclobber only if
24854         the operand is not reloaded.  Prefer to reload conflicting operand
24855         if earlyclobber and matching operands are the same.
24857 2013-02-12  Richard Biener  <rguenther@suse.de>
24859         PR lto/56297
24860         * lto-streamer-out.c (write_symbol): Do not output symbols
24861         for hard register variables.
24863 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24865         PR target/54222
24866         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24867         (umulsidi3_insn, mulsidi3_insn): New insns.
24869 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24871         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24872         (struct tune_params): Add vec_costs field.
24873         * config/arm/arm.c (arm_builtin_vectorization_cost)
24874         (arm_add_stmt_cost): New functions.
24875         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24876         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24877         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24878         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24879         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24880         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24881         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24883 2013-02-12  Richard Biener  <rguenther@suse.de>
24885         PR lto/56295
24886         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24887         decls again if possible.
24889 2013-02-12  Richard Biener  <rguenther@suse.de>
24891         PR middle-end/56288
24892         * tree-ssa.c (verify_ssa_name): Fix check, move
24893         SSA_NAME_IN_FREE_LIST check up.
24895 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24896             Steven Bosscher   <steven@gcc.gnu.org>
24898         PR rtl-optimization/56151
24899         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24900         equal to op0 or op1, and last_insn pattern is CODE operation
24901         with MEM dest and one of the operands matches that MEM.
24903 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24905         * doc/extend.texi: Document Function Multiversioning and "default"
24906         parameter string to target attribute.
24907         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24908         target attribute parameter is "default".
24909         (ix86_compare_version_priority): Remove checks for target attribute.
24910         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24911         Remove check for target attribute equal to NULL. Add assert.
24912         (ix86_generate_version_dispatcher_body): Change error to sorry.
24914 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24915             Jack Howarth  <howarth@bromo.med.uc.edu>
24916             Patrick Marlier  <patrick.marlier@gmail.com>
24918         PR libitm/55693
24919         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24920         define ENDFILE_SPEC as TM_DESTRUCTOR.
24921         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24923 2013-02-11  Alexander Potapenko  <glider@google.com>
24924             Jack Howarth  <howarth@bromo.med.uc.edu>
24925             Jakub Jelinek  <jakub@redhat.com>
24927         PR sanitizer/55617
24928         * config/darwin.c (cdtor_record): Rename ctor_record.
24929         (sort_cdtor_records): Rename sort_ctor_records.
24930         (finalize_dtors): New routine to sort destructors by
24931         priority before use in assemble_integer.
24932         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24934 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24936         PR rtl-optimization/56275
24937         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24938         offset is non-negative and less than cmode size before
24939         calling simplify_subreg.
24941 2013-02-11  Richard Biener  <rguenther@suse.de>
24943         PR tree-optimization/56264
24944         * cfgloop.h (fix_loop_structure): Adjust prototype.
24945         * loop-init.c (fix_loop_structure): Return the number of
24946         newly discovered loops.
24947         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24948         are discovered, do a full loop-closed SSA rewrite.
24950 2013-02-11  Richard Biener  <rguenther@suse.de>
24952         PR tree-optimization/56273
24953         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24954         first VRP run.
24955         (check_array_ref): Fix missing newline in dumps.
24956         (search_for_addr_array): Likewise.
24958 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24960         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24962 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24964         PR target/56256
24965         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24967 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24969         PR rtl-optimization/56246
24970         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24971         reload pseudo.
24972         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24973         constraints are satisfied.
24975 2013-02-08  Jeff Law  <law@redhat.com>
24977         PR debug/53948
24978         * emit-rtl.c (reg_is_parm_p): New function.
24979         * regs.h (reg_is_parm_p): New prototype.
24980         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24981         callee-clobbered registers.
24983 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24985         PR target/56043
24986         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24987         If there is no implicit builtin declaration, just return NULL.
24989 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24991         * config/i386/sse.md (FMAMODEM): New mode iterator.
24992         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24993         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24995 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24997         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24998         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24999         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
25001 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25003         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
25004         (microblaze*-*-elf): Likewise.
25005         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
25006         LINK_SPEC.
25007         * config/microblaze/microblaze-c.c: Add builtin defines for
25008         _LITTLE_ENDIAN and _BIG_ENDIAN.
25009         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
25010         add to TARGET_DEFAULT flags.
25011         Expand ASM_SPEC and LINK_SPEC.
25012         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
25013         * config/microblaze/microblaze.md: Update extendsidi2 and
25014         movdi_internal instructions to use low-order / high-order reg
25015         print_operands.
25016         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
25017         options and inversemask / mask of LITTLE_ENDIAN.
25018         * config/microblaze/t-microblaze: Expand multilib options to
25019         include mlittle-endian (le) and update exceptions patterns.
25021 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25023         PR rtl-optimization/56195
25024         * lra-constraints.c (get_reload_reg): Don't reuse regs
25025         if they have smaller mode than requested, if they have
25026         wider mode than requested, try to return a SUBREG.
25028         PR tree-optimization/56250
25029         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25030         if type is unsigned and code isn't MULT_EXPR.
25032 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25034         PR tree-optimization/56064
25035         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25036         bits according to mode.
25037         * fixed-value.h (fixed_from_double_int)
25038         (const_fixed_from_double_int): Adjust comments.
25040 2013-02-08  Richard Biener  <rguenther@suse.de>
25042         PR lto/56231
25043         * lto-streamer.h (struct data_in): Remove current_file, current_line
25044         and current_col members.
25045         * lto-streamer-out.c (lto_output_location): Stream changed bits
25046         en-block for efficiency.
25047         * lto-streamer-in.c (clear_line_info): Remove.
25048         (lto_input_location): Cache current file, line and column
25049         globally via local statics.  Read changed bits en-block.
25050         (input_function): Do not call clear_line_info.
25051         (lto_read_body): Likewise.
25052         (lto_input_toplevel_asms): Likewise.
25054 2013-02-08  Michael Matz  <matz@suse.de>
25056         PR tree-optimization/52448
25057         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25058         (nt_call_phase): New static.
25059         (add_or_mark_expr): Only mark accesses with newer phase than any
25060         call seen.
25061         (nonfreeing_call_p): New.
25062         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25063         (nt_fini_block): Keep blocks marked as visited.
25064         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25066 2013-02-08  Richard Biener  <rguenther@suse.de>
25068         * ira.c (ira): Free broken dominator information.
25070 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25072         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25074 2013-02-08  Marek Polacek  <polacek@redhat.com>
25076         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25078 2013-02-08  Richard Biener  <rguenther@suse.de>
25080         PR middle-end/56181
25081         * cfgloop.h (flow_loops_find): Adjust.
25082         (bb_loop_header_p): Declare.
25083         * cfgloop.c (bb_loop_header_p): New function split out from ...
25084         (flow_loops_find): ... here.  Adjust function signature,
25085         support incremental loop structure update.
25086         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25087         * cfgloopmanip.c (fix_loop_structure): Move ...
25088         * loop-init.c (fix_loop_structure): ... here.
25089         (apply_loop_flags): Split out from ...
25090         (loop_optimizer_init): ... here.
25091         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25092         in incremental mode, only remove dead loops here.
25094 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25096         PR target/54222
25097         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25098         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25099         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25100         (*round<mode>3.libgcc): New insns for fixed-modes.
25101         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25102         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25103         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25104         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25105         implementations.  Define to __builtin_avr_absFX,
25106         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25107         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25108         __builtin_avr_countlsFX, respectively.
25109         * config/avr/avr-c.c (target.h): Include it.
25110         (enum avr_builtin_id): New enum.
25111         (avr_resolve_overloaded_builtin): New static function.
25112         (avr_register_target_pragmas): Use it to set
25113         targetm.resolve_overloaded_builtin.
25114         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25115         tree nodes used by DEF_BUILTIN.
25116         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25117         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25118         <AVR_BUILTIN_xxBITS>: Same.
25120 2013-02-08  Richard Biener  <rguenther@suse.de>
25122         * cfgloop.c (verify_loop_structure): Properly handle
25123         a loop exiting to another loop header.
25124         * ira-int.h (ira_loops): Remove.
25125         * ira.c (ira_loops): Remove.
25126         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25127         (do_reload): Use loop_optimizer_finalize.
25128         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25129         number_of_loops to access the loop tree.
25130         (more_one_region_p): Likewise.
25131         (finish_loop_tree_nodes): Likewise.
25132         (rebuild_regno_allocno_maps): Likewise.
25133         (mark_loops_for_removal): Likewise.
25134         (mark_all_loops_for_removal): Likewise.
25135         (remove_unnecessary_regions): Likewise.
25136         (ira_build): Likewise.
25137         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25139 2013-02-08  Richard Biener  <rguenther@suse.de>
25141         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25142         * ipa-pure-const.c (analyze_function): Avoid calling
25143         mark_irreducible_loops twice.
25144         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25146 2013-02-07  David S. Miller  <davem@davemloft.net>
25148         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25149         on 'reg'.
25150         * var-tracking.c (vt_add_function_parameter): Test the presence of
25151         HAVE_window_save properly and do not remap argument registers when
25152         we have a leaf function.
25154 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25156         PR bootstrap/56227
25157         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25158         instead of "ll".
25159         * config/i386/i386.c (ix86_print_operand): Ditto.
25161 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25163         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25165 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25167         PR rtl-optimization/56225
25168         * lra-constraints.c (process_alt_operands): Check that reload hard
25169         reg can hold value for strict_low_part.
25171 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25173         PR debug/56154
25174         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25175         dwarf2out_end_function.
25176         (in_first_function_p, maybe_at_text_label_p,
25177         first_loclabel_num_not_at_text_label): New variables.
25178         (dwarf2out_var_location): In the first function find out
25179         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25180         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25181         functions.
25183 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25185         PR rtl-optimization/56178
25186         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25187         SUBREG of a register.  Tidy up related block of code.
25188         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25189         note if the source is a register or a SUBREG of a register.
25191 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25193         PR target/56228
25194         * config/rs6000/rs6000.md (ptrm): New mode attr.
25195         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25196         call_value_indirect_aix<pttrsize>,
25197         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25198         m in constraints.
25200 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25202         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25203         if -bnortl. Convert to strcmp and strncmp.
25205 2013-02-07  Alan Modra  <amodra@gmail.com>
25207         PR target/54009
25208         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25209         addresses won't wrap when offsetting.
25210         (rs6000_secondary_reload): Provide secondary reloads needed for
25211         wrapping LO_SUM addresses.
25213 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25215         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25216         MACH, just __MACH__.
25218 2013-02-06  Richard Biener  <rguenther@suse.de>
25220         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25221         instead of calling fix_loop_structure.
25223 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25225         PR middle-end/56217
25226         * omp-low.c (use_pointer_for_field): Return false if
25227         lower_send_shared_vars doesn't generate any copy-out code.
25229 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25231         PR rtl-optimization/56131
25232         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25233         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25234         of the label is NULL.  Add comment.
25236 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25238         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25240         PR sanitizer/55374
25241         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25242         (STATIC_LIBTSAN_LIBS): Likewise.
25243         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25244         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25245         is defined, don't add anything else beyond that.
25246         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25247         (LINK_COMMAND_SPEC): Use them.
25249         PR tree-optimization/56205
25250         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25251         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25252         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25254 2013-02-05  Richard Biener  <rguenther@suse.de>
25256         PR tree-optimization/53342
25257         PR tree-optimization/53185
25258         * tree-vectorizer.h (vect_check_strided_load): Remove.
25259         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25260         not disallow peeling for vectorized strided loads.
25261         (vect_check_strided_load): Make static and simplify.
25262         (vect_analyze_data_refs): Adjust.
25263         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25264         correctly when vectorizing strided loads.
25266 2013-02-05  Richard Biener  <rguenther@suse.de>
25268         * doc/install.texi: Refer to ISL, not PPL.
25270 2013-02-05  Jan Hubicka  <jh@suse.cz>
25272         PR tree-optimization/55789
25273         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25275 2013-02-05  Jan Hubicka  <jh@suse.cz>
25277         PR tree-optimization/55789
25278         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25279         the dead call anyway.
25281 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25283         PR sanitizer/55374
25284         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25286 2013-02-04  Alexander Potapenko  <glider@google.com>
25287             Jack Howarth  <howarth@bromo.med.uc.edu>
25288             Jakub Jelinek  <jakub@redhat.com>
25290         PR sanitizer/55617
25291         * config/darwin.c (sort_ctor_records): Stabilized qsort
25292         on constructor priority by using original position.
25293         (finalize_ctors): New routine to sort constructors by
25294         priority before use in assemble_integer.
25295         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25297 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25299         PR libstdc++/54314
25300         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25301         about visibility on artificial decls.
25302         * config/sol2.c (solaris_assemble_visibility): Likewise.
25304 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25306         PR target/56186
25307         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25308         argument and improve checking of return-argument types for 16-byte
25309         modes.
25310         (ix86_function_value_1): Add additional valtype argument on call
25311         of function_value_64.
25312         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25313         handling infunction_value_64 function.
25315 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25317         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25319 2013-02-04  Richard Biener  <rguenther@suse.de>
25321         PR tree-optimization/56188
25322         * tree-ssa-structalias.c (label_visit): Consider case with
25323         initially non-empty points-to set.
25324         (perform_var_substitution): Dump node mapping and clean up.
25326 2013-02-04  Richard Guenther  <rguenther@suse.de>
25328         PR lto/56168
25329         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25330         node prevail as last resort.
25331         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25332         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25334 2013-02-04  Richard Biener  <rguenther@suse.de>
25336         PR tree-optimization/56113
25337         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25338         Merge into ...
25339         (equiv_class_lookup_or_add): ... this.
25340         (label_visit): Adjust and fix error in previous patch.
25341         (perform_var_substitution): Adjust.
25343 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25345         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25346         * config/sh/sh4-300.md: Likewise.
25347         * config/sh/sh4a.md: Likewise.
25348         * config/sh/constraints.md: Likewise.
25349         * config/sh/sh.md: Likewise.
25350         * config/sh/netbsd-elf.h: Likewise.
25351         * config/sh/predicates.md: Likewise.
25352         * config/sh/sh-protos.h: Likewise.
25353         * config/sh/ushmedia.h: Likewise.
25354         * config/sh/linux.h: Likewise.
25355         * config/sh/sh.c: Likewise.
25356         * config/sh/superh.h: Likewise.
25357         * config/sh/elf.h: Likewise.
25358         * config/sh/sh4.md: Likewise.
25359         * config/sh/sh.h: Likewise.
25361 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25363         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25364         constraint to match_test floating_point_store_memory_operand().
25365         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25366         (base14_operand): New.
25367         (floating_point_store_memory_operand): New.
25368         (integer_store_memory_operand): Revise to use base14_operand and
25369         reg_plus_base_memory_operand.
25370         (move_dest_operand): Allow symbolic_memory_operands.
25371         (symbolic_memory_operand): Check for LO_SOM.
25372         (symbolic_operand): Change default case to break.
25373         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25374         CONST_DOUBLE values to be reloaded by putting them into memory when
25375         the destination is a floating point register.
25376         (movdf): Remove code to handle CONST_DOUBLE.
25377         (movsf): Likewise.
25378         (reload_indf_r1): New.
25379         (reload_insf_r1): New.
25380         Consistently use "Q" and "T" constraints with integer and floating
25381         point move instructions, respectively.
25382         (movdi): Remove FAIL.
25383         Change predicate for source operand unamed DImode move from
25384         general_operand to move_src_operand.
25385         (umulsidi3): Change predicate for destination operand to
25386         register_operand.
25387         Likewise for similar unamed patterns.
25388         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25389         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25390         (hppa_legitimize_address): Simplify mask calculation.
25391         (pa_emit_move_sequence): Revised handling of secondary reloads from
25392         REG+D addresses for floating point loads and stores.  Directly handle
25393         loading CONST0_RTX (mode) to a floating point register.
25394         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25395         to floating point registers.  Don't restrict secondary reloads to
25396         floating point registers to integer modes.  Revise some comments and
25397         cleanup some code.
25398         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25399         (pa_legitimate_address_p): New.
25400         (pa_legitimize_reload_address): New.
25401         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25402         (STRICT_REG_OK_FOR_BASE_P): New.
25403         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25404         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25406 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25407             Andrew Dixie  <andrewd@gentrack.com>
25409         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25410         flag set.
25412 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25414         * expmed.c (extract_bit_field_1): Pass the full width of the
25415         structure to get_best_reg_extraction_insn.
25417 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25419         PR target/54601
25420         * configure.ac (use_cxa_atexit): Add AIX.
25421         * configure: Regenerate.
25423         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25425 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25427         PR debug/54793
25428         * final.c (need_profile_function): New variable.
25429         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25430         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25431         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25432         notes, targetm.asm_out.function_prologue doesn't emit anything,
25433         HAVE_prologue and profiler should be emitted before prologue,
25434         set need_profile_function instead of emitting it.
25435         (final_scan_insn): If need_profile_function, emit
25436         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25437         NOTE_INSN_FUNCTION_BEG note.
25439 2013-02-01  Richard Henderson  <rth@redhat.com>
25441         * config/rs6000/rs6000.md (smulditi3): New.
25442         (umulditi3): New.
25444         * config/alpha/alpha.md (umulditi3): New.
25446 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25448         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25449         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25451 2013-02-01  Richard Biener  <rguenther@suse.de>
25453         PR tree-optimization/56113
25454         * tree-ssa-structalias.c (label_visit): Reduce work for
25455         single-predecessor nodes.
25457 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25459         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25460         range isn't testing for zero.
25462 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25464         PR middle-end/56113
25465         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25467 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25468             Nick Clifton  <nickc@redhat.com>
25470         * config/v850/constraints.md (Q): Define as a memory constraint.
25471         * config/v850/predicates.md (label_ref_operand): New predicate.
25472         (e3v5_shift_operand): New predicate.
25473         (ior_operator): New predicate.
25474         * config/v850/t-v850: Add e3v5 multilib.
25475         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25476         (v850_gen_movdi): Prototype.
25477         * config/v850/v850.c: Add support for e3v5 architecture.
25478         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25479         TARGET_V850E_UP.
25480         (construct_save_jarl): Add e3v5 long JARL support.
25481         (v850_adjust_insn_length): New function.  Adjust length of call
25482         insns when using e3v5 instructions.
25483         (v850_gen_movdi): New function: Generate instructions to move a
25484         DImode value.
25485         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25486         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25487         (TARGET_USE_FPU): Enable for e3v5.
25488         (CONST_OK_FOR_W): New macro.
25489         (ADJUST_INSN_LENGTH): Define.
25490         * config/v850/v850.md (UNSPEC_LOOP): Define.
25491         (attr cpu): Add v850e3v5.
25492         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25493         (movdi): New pattern.
25494         (movdi_internal): New pattern.
25495         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25496         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25497         (cstoresf4): Likewise.
25498         (cstoredf4): Likewise.
25499         (insv): New pattern.
25500         (rotlso3_a): New pattern.
25501         (rotlsi3_b): New pattern
25502         (rotlsi3_v850e3v5): New pattern.
25503         (doloop_begin): New pattern.
25504         (fix_loop_counter): New pattern.
25505         (doloop_end): New pattern.
25506         (branch_normal): Add e3v5 long branch support.
25507         (branch_invert): Likewise.
25508         (branch_z_normal): Likewise.
25509         (branch_z_invert): Likewise.
25510         (branch_nz_normal): Likewise.
25511         (branch_nz_invert): Likewise.
25512         (call_internal_short): Add e3v5 register-indirect JARL support.
25513         (call_internal_long): Likewise.
25514         (call_value_internal_short): Likewise.
25515         (call_value_internal_long): Likewise.
25516         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25517         (mloop): New option.
25518         * config.gcc: Add support for configuring v840e3v5 target.
25519         * doc/invoke.texi: Document new v850 specific command line options.
25521 2013-01-31  Paul Koning  <ni1d@arrl.net>
25523         PR debug/55059
25524         PR debug/54508
25525         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25526         children if parent is a class.
25527         (prune_unused_types_prune): Don't add DW_AT_declaration.
25529 2013-01-31  Richard Biener  <rguenther@suse.de>
25531         PR tree-optimization/56157
25532         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25533         match up operand with SLP child.
25535 2013-01-31  Jason Merrill  <jason@redhat.com>
25537         PR debug/54410
25538         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25539         parameters the first time.
25540         (gen_scheduled_generic_parms_dies): Check completeness here.
25542 2013-01-31  Richard Biener  <rguenther@suse.de>
25544         PR middle-end/53073
25545         * common.opt (faggressive-loop-optimizations): New flag,
25546         enabled by default.
25547         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25548         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25549         infer_loop_bounds_from_undefined by it.
25551 2013-01-31  Richard Biener  <rguenther@suse.de>
25553         PR tree-optimization/56150
25554         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25555         visit virtual operands.
25556         (find_uses_to_rename_bb): Likewise.
25558 2013-01-31  Richard Biener  <rguenther@suse.de>
25560         PR tree-optimization/56150
25561         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25562         mixed store non-store stmts.
25564 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25566         PR sanitizer/55374
25567         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25568         LIBASAN_EARLY_SPEC is defined.
25569         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25570         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25571         before %o.
25572         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25574         PR c++/55742
25575         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25576         invalid args instead of ICEing on it.
25577         (ix86_valid_target_attribute_tree): Return error_mark_node if
25578         ix86_valid_target_attribute_inner_p failed.
25579         (ix86_valid_target_attribute_p): Return false only if
25580         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25581         target("default") attribute.
25582         (sorted_attr_string): Change argument from const char * to tree,
25583         merge in all target attribute arguments rather than just one.
25584         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25585         instead of free.  Avoid using strcat.
25586         (ix86_mangle_function_version_assembler_name): Mangle
25587         target("default") as if no target attribute is present.  Adjust
25588         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25589         instead of xmalloc and XDELETEVEC instead of free.
25590         (ix86_function_versions): Don't return true if one of the decls
25591         doesn't have target attribute.  If they don't and one of the decls
25592         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25593         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25594         (ix86_supports_function_versions): Remove.
25595         (make_name): Fix up formatting.
25596         (make_dispatcher_decl): Remove resolver_name and its initialization.
25597         Avoid leaking memory.
25598         (is_function_default_version): Return true if there is
25599         target("default") attribute rather than no target attribute at all.
25600         (make_resolver_func): Avoid leaking memory.
25601         (ix86_generate_version_dispatcher_body): Likewise.
25602         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25603         * target.def (supports_function_versions): Remove.
25604         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25605         * doc/tm.texi: Regenerated.
25607 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25609         PR rtl-optimization/56144
25610         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25611         for values with side effects.
25613 2013-01-30  Richard Biener  <rguenther@suse.de>
25615         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25616         (sparseset_pop): Likewise.
25617         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25618         to be able to use quick_push in the worker loop.
25620 2013-01-30  Marek Polacek  <polacek@redhat.com>
25622         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25624 2013-01-30  Richard Biener  <rguenther@suse.de>
25626         PR lto/56147
25627         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25629 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25631         PR tree-optimization/56064
25632         * fixed-value.c (fixed_from_double_int): New function.
25633         * fixed-value.h (fixed_from_double_int): New prototype.
25634         (const_fixed_from_double_int): New static inline function.
25635         * fold-const.c (native_interpret_fixed): New static function.
25636         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25637         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25638         (native_encode_fixed): New static function.
25639         (native_encode_expr) <FIXED_CST>: Use it.
25640         (native_interpret_int): Move double_int worker code to...
25641         * double-int.c (double_int::from_buffer): ...this new static method.
25642         * double-int.h (double_int::from_buffer): Prototype it.
25644 2013-01-30  Richard Biener  <rguenther@suse.de>
25646         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25647         New pointer-map and obstack.
25648         (init_alias_vars): Allocate pointer-map and obstack.
25649         (delete_points_to_sets): Free them.
25650         (find_what_var_points_to): Cache result.
25651         (find_what_p_points_to): Adjust for changed interface of
25652         find_what_var_points_to.
25653         (compute_points_to_sets): Likewise.
25654         (ipa_pta_execute): Likewise.
25656 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25658         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25659         * configure: Regenerate.
25660         * config.in: Regenerate.
25661         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25662         #nobits/#progbits if supported.
25664 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25666         PR target/56121
25667         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25668         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25669         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25671 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25673         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25674         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25676 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25678         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25679         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25681 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25683         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25684         declaration.
25685         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25686         * config/arm/cortex-a7.md: New bypasses using
25687         arm_mac_accumulator_is_result.
25689 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25691         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25692         (cortex_a7_neon_mla): Likewise.
25693         (cortex_a7_fpfmad): New reservation.
25694         (cortex_a7_fpmacs): Use ffmas and update required units.
25695         (cortex_a7_fpmuld): Update required units and latency.
25696         (cortex_a7_fpmacd): Likewise.
25697         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25698         (cortex_a7_neon). Likewise.
25699         (bypass) Update participating units.
25701 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25703         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25704         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25705         from fmac to ffma.
25706         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25707         (vfp_fmul): Use ffmad.
25708         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25709         (cortex_r4_fmacd): Use ffmad.
25710         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25711         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25712         (cortex_a9_fmacd): Use ffmad.
25713         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25714         (cortex_a8_vfp_macd): Use ffmad.
25715         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25716         (cortex_a5_fpmacd): Use ffmad.
25717         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25718         (cortex_a15_vfp_macd): Use ffmad.
25719         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25721 2013-01-29  Jason Merrill  <jason@redhat.com>
25723         PR libstdc++/54314
25724         * varasm.c (default_assemble_visibility): Don't warn about
25725         visibility on artificial decls.
25727 2013-01-29  Richard Biener  <rguenther@suse.de>
25729         PR tree-optimization/56113
25730         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25731         the bitmap leader.
25732         (label_visit): Free duplicate bitmaps and record the leader instead.
25733         (perform_var_substitution): Adjust.
25735 2013-01-29  Richard Biener  <rguenther@suse.de>
25737         PR tree-optimization/55270
25738         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25739         the CFG, schedule loops for fixup.
25741 2013-01-29  Nick Clifton  <nickc@redhat.com>
25743         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25744         SP_REG.
25746 2013-01-28  Leif Ekblad  <leif@rdos.net>
25748         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25749         * config/i386/i386.h (TARGET_RDOS): New macro.
25750         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25751         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25752         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25753         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25754         DEFAULT_LARGE_SECTION_THRESHOLD.
25755         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25756         * config/i386/rdos.h: New file.
25757         * config/i386/rdos64.h: New file.
25759 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25761         PR other/54814
25762         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25763         TEST_HARD_REG_BIT.
25765 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25767         PR rtl-optimization/56117
25768         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25769         call cselib_lookup_from_insn on the MEM before calling
25770         add_insn_mem_dependence.
25772 2013-01-28  Richard Biener  <rguenther@suse.de>
25774         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25775         to a stmt that didn't have one.
25776         (copy_phis_for_bb): Likewise for PHI arguments.
25777         (copy_debug_stmt): Likewise for debug stmts.
25779 2013-01-28  Richard Biener  <rguenther@suse.de>
25781         PR tree-optimization/56034
25782         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25783         (partition_builtin_p): Adjust.
25784         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25785         it is the last partition.
25786         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25787         up the vertex for the definition.
25788         (classify_partition): Classify whether a partition is a
25789         PKIND_REDUCTION, thus has uses outside of the loop.
25790         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25791         Merge all PKIND_REDUCTION partitions into the last partition.
25792         (tree_loop_distribution): Seed partitions from reductions as well.
25794 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25796         PR tree-optimization/56125
25797         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25798         pow(x,c) into sqrt(x) * powi(x, n/2) or
25799         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25800         optimizing for size.
25801         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25802         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25803         integer.
25805         PR tree-optimization/56094
25806         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25807         to UNKNOWN_LOCATION while gimplifying expr.
25809 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25811         PR target/56114
25812         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25813         operand 0 in movabs insn template for -masm=intel asm alternative.
25814         (*movabs<mode>_2): Ditto for operand 1.
25816 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25818         PR target/54663
25819         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25820         of microblaze-c.o
25822 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25824         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25825         tm_file.
25827 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25829         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25830         Undef to avoid warning.
25832 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25834         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25835         * configure: Regenerate.
25837 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25839         PR tree-optimization/56098
25840         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25841         for stmts with volatile ops.
25842         (cond_store_replacement): Don't optimize if assign has volatile ops.
25843         (cond_if_else_store_replacement_1): Don't optimize if either
25844         then_assign or else_assign have volatile ops.
25845         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25846         volatile ops.
25848 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25850         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25852 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25854         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25855         missing ':' in asm example.
25857 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25859         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25860         entries into lane and laneq entries.
25861         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25862         Remove AdvSIMD scalar modes.
25863         (aarch64_sq<r>dmulh_laneq<mode>): New.
25864         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25865         modes.
25866         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25867         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25868         * config/aarch64/iterators.md (VCOND): New.
25869         (VCONQ): New.
25871 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25873         PR target/54222
25874         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25875         Add NULL LIBNAME argument to existing definitions.
25876         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25877         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25878         * config/avr/avr.c (DEF_BUILTIN): Same.
25879         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25880         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25881         implementation is available (DECL_ASSEMBLER_NAME is set).
25882         (avr_fold_absfx): New static function.
25883         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25884         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25885         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25886         AVR_BUILTIN_ABSLLK.
25887         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25888         (abshk, absk, abslk, absllk): Provide as static inline functions.
25890 2013-01-25  Marek Polacek  <polacek@redhat.com>
25892         PR tree-optimization/56035
25893         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25895 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25897         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25898         (*movtf_internal_rex64): Add (!o,C) alternative
25899         (*movxf_internal_rex64): Ditto.
25900         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25902 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25904         * doc/invoke.texi: fix typo.
25905         * doc/objc.texi: fix typo.
25907 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25909         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25910         for the first two alternatives.
25912 2013-01-24  Diego Novillo  <dnovillo@google.com>
25914         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25915         (ggc-zone.o): Remove.
25916         * configure.ac: Remove option --with-gc.
25917         * configure: Re-generate.
25918         * doc/install.texi: Remove documentation for --with-gc.
25919         * gengtype.c (write_enum_defn): Remove.  Update all users.
25920         (write_Types_process_field): Remove generation of gt_e_* argument.
25921         (output_type_enum): Remove.  Update all users.
25922         (write_enum_defn): Remove.  Update all users.
25923         (enum alloc_zone): Remove.  Update all users.
25924         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25925         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25926         Update all callers.
25927         (struct ptr_data): Remove field TYPE.  Update all users.
25928         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25929         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25930         Update all users.
25931         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25932         (struct alloc_zone): Remove.
25933         (ggc_internal_alloc_zone_stat): Remove.
25934         (ggc_internal_cleared_alloc_zone_stat): Remove.
25935         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25936         (ggc_pch_count_object): Remove last argument.  Update all users.
25937         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25938         (struct alloc_zone): Remove.
25939         * ggc-zone.c: Remove.
25940         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25941         (struct alloc_zone): Remove.
25942         (ggc_alloc_typed_stat): Remove.
25943         (ggc_alloc_typed): Remove.
25944         (ggc_splay_alloc): Remove first argument.
25945         (rtl_zone): Remove.  Update all users.
25946         (tree_zone): Remove.  Update all users.
25947         (tree_id_zone): Remove.  Update all users.
25948         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25949         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25950         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25951         * tree-ssanames.c: Remove references to zone allocator in comments.
25953 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25955         * config/avr/avr.c (avr_out_fract): Make register numbers that
25956         might be outside of source operand signed.
25958 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25960         * config/i386/constraints.md (Yf): New constraint.
25961         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25962         of f constraint to conditionaly disable x87 register preferences.
25963         (*movdf_internal): Ditto.
25964         (*movsf_internal): Ditto.
25966 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25968         PR inline-asm/55934
25969         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25970         that have operands with impossible constraints.
25971         Add a FIXME for a speed-up opportunity.
25972         * lra-constraints.c (process_alt_operands): Verify that a class
25973         selected from constraints on asms is valid for the operand mode.
25974         (curr_insn_transform): Remove incorrect comment.
25976 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25978         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25979         TOC operand is a valid symbol ref in the constant pool.
25981 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25983         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25985 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25987         PR target/54222
25988         * config/avr/stdfix.h: New file.
25989         * t-avr (stdfix-gcc.h): New rule to build it.
25990         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25992 2013-01-23  Kostya Serebryany  <kcc@google.com>
25994         * config/darwin.h: remove dependency on
25995         CoreFoundation (asan on Mac OS).
25997 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25999         PR target/49069
26000         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
26001         instead of cmpdi_operand for first comparison operand.
26002         Don't assert that comparison operands aren't both constants.
26004 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26006         * doc/install.texi (Downloading the Source): Update references to
26007         downloading separate components.
26009 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26011         * doc/extend.texi (__int128): Improve grammar.
26013 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
26015         PR target/56028
26016         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
26017         alternative to (o,r).
26018         (*movdi_internal_rex64): Remove (!o,n) alternative.
26019         (DImode immediate->memory splitter): Remove.
26020         (DImode immediate->memory peephole2): Remove.
26021         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26022         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26023         alternative to (!o,*r).
26024         (*movtf_internal_sse): New pattern.
26025         (*movxf_internal_rex64): New pattern.
26026         (*movxf_internal): Disable for TARGET_64BIT.
26027         (*movdf_internal_rex64): Remove (!o,F) alternative.
26029 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26031         PR middle-end/56074
26032         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26033         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26034         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26035         stmt locations where LOCATION_LOCUS of the stmt location is
26036         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26038         PR target/55686
26039         * config/i386/i386.md (UNSPEC_STOS): New.
26040         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26041         *strsetqi_1): Add UNSPEC_STOS.
26043 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26045         PR c++/56067
26046         * doc/invoke.texi: Remove left over -Wsynth example.
26048 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26050         PR tree-optimization/56051
26051         * fold-const.c (fold_binary_loc): Don't fold
26052         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26053         a narrowing conversion, or widening conversion from signed
26054         to unsigned.
26056 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26058         PR rtl-optimization/56023
26059         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26060         dependent on debug instruction.
26062 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26064         PR middle-end/56022
26065         * function.c (allocate_struct_function): Call
26066         invoke_set_current_function_hook earlier.
26068 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26070         * reload1.c (init_reload): Only initialize reload_obstack
26071         during the first call.
26073 2013-01-21  Marek Polacek  <polacek@redhat.com>
26075         * cfgloop.c (verify_loop_structure): Fix up grammar.
26077 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26079         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26080         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26082 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26084         PR target/56058
26085         * config/arm/marvell-pj4.md: Update copyright year.
26086         Fix up use of alu to alu_reg and simple_alu_imm.
26088 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26090         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26092 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26094         PR target/55433
26095         * lra-constraints.c (curr_insn_transform): Don't reuse original
26096         insn for secondary memory move when memory mode should be different.
26098 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26100         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26101         atomic_storedi_1): New patterns.
26103 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26105         btver2 pipeline descriptions.
26106         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26107         descriptions.
26108         * config/i386/i386.md (btver2_decode): New type attributes.
26109         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26110         type attributes.
26111         * config/i386/btver2.md: New file describing btver2 pipelines.
26113 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26115         PR tree-optimization/52631
26116         * tree-ssa-sccvn (visit_use): Before looking up the original
26117         statement, try looking up the simplified expression.
26119 2013-01-19  Anthony Green  <green@moxielogic.com>
26121         * config/moxie/moxie.c (moxie_expand_prologue): Set
26122         current_function_static_stack_size.
26124 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26126         PR tree-optimization/56029
26127         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26128         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26130 2013-01-18  Sharad Singhai  <singhai@google.com>
26132         PR tree-optimization/55995
26133         * dumpfile.c (dump_loc): Print location only if available.
26134         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26136 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26138         PR target/55433
26139         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26140         secondary memory move.
26141         (inherit_reload_reg): Use rclass instead of cl for
26142         check_secondary_memory_needed_p.
26144 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26146         PR middle-end/56015
26147         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26148         the case where writing real complex part of target modifies op1.
26150 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26152         * config/aarch64/aarch64-simd.md
26153         (aarch64_vcond_internal<mode>): Handle unordered cases.
26154         * config/aarch64/iterators.md (v_cmp_result): New.
26156 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26157             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26159         * config/arm/marvell-pj4.md: New file.
26160         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26161         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26162         (generic_vfp): Likewise.
26163         * config/arm/arm-cores.def: Add marvell-pj4.
26164         * config/arm/arm-tune.md: Regenerate.
26165         * config/arm/arm-tables.opt: Regenerate.
26166         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26167         * doc/invoke.texi: Document marvell-pj4.
26169 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26171         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26173 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26175         PR debug/54114
26176         PR debug/54402
26177         PR debug/49888
26178         * var-tracking.c (negative_power_of_two_p): New.
26179         (global_get_addr_cache, local_get_addr_cache): New.
26180         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26181         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26182         heading comment.
26183         (vt_stack_offset_p): Remove.
26184         (vt_canon_true_dep): Always canonicalize loc's address.
26185         (clobber_overlapping_mems): Make sure we have a MEM.
26186         (local_get_addr_clear_given_value): New.
26187         (val_reset): Clear local cached entries.
26188         (compute_bb_dataflow): Create and release the local cache.
26189         Disable duplicate MEMs clobbering.
26190         (emit_notes_in_bb): Clobber MEMs likewise.
26191         (vt_emit_notes): Create and release the local cache.
26192         (vt_initialize, vt_finalize): Create and release the global
26193         cache, respectively.
26194         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26196 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26198         PR libmudflap/53359
26199         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26200         not found in the symtab.
26202 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26204         PR debug/56006
26205         PR rtl-optimization/55547
26206         PR rtl-optimization/53827
26207         PR debug/53671
26208         PR debug/49888
26209         * alias.c (offset_overlap_p): New, factored out of...
26210         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26211         the conservative special case for symbolic constants.  Don't
26212         adjust zero sizes on alignment.
26214 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26216         PR rtl-optimization/52573
26217         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26218         REG_UNUSED for the same register.
26220 2013-01-17  Richard Biener  <rguenther@suse.de>
26221             Marek Polacek  <polacek@redhat.com>
26223         PR rtl-optimization/55833
26224         * loop-unswitch.c (unswitch_loops): Move loop verification...
26225         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26226         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26227         Set it to true when we're removing a loop from hierarchy tree in
26228         an irreducible region.
26229         (fix_bb_placements): Adjust caller.
26230         (fix_loop_placements): Likewise.
26232 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26234         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26235         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26236         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26237         Remove ID.  Adjust comments.
26238         * config/avr/avr-c.c (avr_builtin_name): Remove.
26239         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26240         * config/avr/avr.c (avr_tolower): New static function.
26241         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26242         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26243         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26244         default expansion.
26246 2013-01-17  Jan Hubicka  <jh@suse.cz>
26248         PR tree-optimization/55273
26249         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26251 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26253         PR target/55981
26254         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26255         store through atomic_store<mode>_1.
26256         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26258 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26260         PR tree-optimizations/55264
26261         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26262         for virtual methods.
26263         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26264         virtual methods before inlining is over.
26265         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26266         virtual functions.
26267         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26268         non-virtual.
26270 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26272         PR rtl-optimization/56005
26273         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26274         pending reads for prefetch.
26276 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26278         * config/aarch64/aarch64.md
26279         (*cstoresi_neg_uxtw): New pattern.
26280         (*cmovsi_insn_uxtw): New pattern.
26281         (*<optab>si3_uxtw): New pattern.
26282         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26283         (*<optab>si3_insn_uxtw): New pattern.
26284         (*bswapsi2_uxtw): New pattern.
26286 2013-01-16  Richard Biener  <rguenther@suse.de>
26288         * tree-inline.c (tree_function_versioning): Remove set but
26289         never used variable.
26291 2013-01-16  Richard Biener  <rguenther@suse.de>
26293         PR tree-optimization/55964
26294         * tree-flow.h (rename_variables_in_loop): Remove.
26295         (rename_variables_in_bb): Likewise.
26296         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26297         (copy_loop_before): Adjust and delete update-ssa status.
26298         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26299         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26300         (rename_variables_in_loop): Remove.
26301         (slpeel_update_phis_for_duplicate_loop): Likewise.
26302         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26303         use available cfg machinery instead of duplicating it.
26304         Update PHI nodes and perform poor-mans SSA update here.
26305         (slpeel_tree_peel_loop_to_edge): Adjust.
26307 2013-01-16  Richard Biener  <rguenther@suse.de>
26309         PR tree-optimization/54767
26310         PR tree-optimization/53465
26311         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26312         (vrp_visit_phi_node): For PHI arguments coming via backedges
26313         drop all symbolical range information.
26314         (execute_vrp): Compute backedges.
26316 2013-01-16  Richard Biener  <rguenther@suse.de>
26318         * doc/install.texi: Update CLooG and ISL requirements to
26319         0.18.0 and 0.11.1.
26321 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26323         PR target/55301
26324         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26325         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26326         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26328 2013-01-16  DJ Delorie  <dj@redhat.com>
26330         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26331         (UNSPECV_SP_SWITCH_E): New.
26332         (sp_switch_1): Change to an unspec.
26333         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26334         replace $r15.
26336 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26338         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26339         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26340         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26341         (expand_mem_thread_fence): Ditto.
26342         (expand_mem_signal_fence): Ditto.
26343         (expand_atomic_load): Ditto.
26344         (expand_atomic_store): Ditto.
26346 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26348         PR rtl-optimization/55547
26349         PR rtl-optimization/53827
26350         PR debug/53671
26351         PR debug/49888
26352         * alias.c (memrefs_conflict_p): Set sizes to negative after
26353         AND adjustments.
26355 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26357         PR target/55940
26358         * function.c (thread_prologue_and_epilogue_insns): Always
26359         add crtl->drap_reg to set_up_by_prologue.set, even if
26360         stack_realign_drap is false.
26362 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26364         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26365         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26366         *call): Fix indention.
26368 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26370         PR target/55876
26371         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26372         Update comment.
26374 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26376         PR rtl-optimization/55153
26377         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26379 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26381         PR tree-optimization/55920
26382         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26383         accesses as grp_to_be_debug_replaced.
26385 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26387         PR tree-optimization/55920
26388         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26389         there is non-useless type conversion needed from debug rhs to lhs,
26390         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26392 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26393             Mikael Pettersson  <mikpe@it.uu.se>
26395         PR target/43961
26396         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26397         Thumb.
26398         (ASM_OUTPUT_CASE_LABEL): Remove.
26399         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26400         * final.c (shorten_branches): Update alignment of labels before
26401         jump tables if CASE_VECTOR_SHORTEN_MODE.
26403 2013-01-15  Richard Biener  <rguenther@suse.de>
26405         PR bootstrap/55961
26406         * system.h: Do not include gmp.h for building host tools.
26408 2013-01-15  Richard Biener  <rguenther@suse.de>
26410         PR middle-end/55882
26411         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26412         account for bitpos when computing alignment.
26414 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26416         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26417         (ix86_target_macros_internal): Likewise.
26419         * config/i386/i386.c (m_CORE2I7): Removed.
26420         (m_CORE_HASWELL): New macro.
26421         (m_CORE_ALL): Likewise.
26422         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26423         (initial_ix86_arch_features): Likewise.
26424         (processor_target_table): Initializations for Core avx2.
26425         (cpu_names): New names "core-avx2".
26426         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26427         PROCESSOR_CORE_HASWELL.
26428         (ix86_issue_rate): New case.
26429         (ia32_multipass_dfa_lookahead): Likewise.
26430         (ix86_sched_init_global): Likewise.
26432         * config/i386/i386.h (TARGET_HASWELL): New macro.
26433         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26434         (processor_type): New PROCESSOR_HASWELL.
26436 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26438         PR tree-optimization/55955
26439         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26440         *SHIFT_EXPR and *ROTATE_EXPR codes.
26442         PR tree-optimization/48766
26443         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26444         -ftrapv disable -fwrapv.
26446 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26448         PR target/55974
26449         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26450         etc. to 1 and not to __flash.
26451         Use LL suffix for __INT24_MAX__ with -mint8.
26452         Use ULL suffix for __UINT24_MAX__ with -mint8.
26454 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26456         * config/avr/avr-arch.h
26457         (struct base_arch_s): Use typedef avr_arch_t instead.
26458         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26459         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26460         * config/avr/avr.c: Same.
26461         * config/avr/avr-devices.c: Same.
26462         * config/avr/driver-avr.c: Same.
26463         * config/avr/gen-avr-mmcu-texi.c: Same.
26464         * config/avr/avr-mcus.def: Adjust comment.
26466 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26468         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26469         * config/aarch64/iterators.md (VALLDI): New.
26471 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26472             Andi Kleen  <ak@linux.intel.com>
26474         PR target/55948
26475         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26476         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26477         memmodel flag.
26479 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26481         * config/avr/avr-stdint.h: Remove trailing blanks.
26482         * config/avr/avr-log.h: Same.
26483         * config/avr/avr-arch.h: Same.
26484         * config/avr/avr-devices.c: Same.
26485         * config/avr/avr-dimode.md: Same.
26486         * config/avr/predicates.md: Same.
26487         * config/avr/avr-c.c: Same.  And fix typo.
26489         * config/avr/avr-protos.h: Same.  And:
26490         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26491         (init_cumulative_args): Rename to avr_init_cumulative_args.
26492         (expand_prologue): Rename to avr_expand_prologue.
26493         (expand_epilogue): Rename to avr_expand_epilogue.
26494         (adjust_insn_length): Rename to avr_adjust_insn_length.
26495         (notice_update_cc): Rename to avr_notice_update_cc.
26496         (final_prescan_insn): Rename to avr_final_prescan_insn.
26497         * config/avr/avr.c: Same.
26498         * config/avr/avr.h: Same.
26499         * config/avr/avr.md: Remove trailing blanks.
26500         (prologue): Use avr_expand_prologue.
26501         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26503 2013-01-14  Richard Biener  <rguenther@suse.de>
26505         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26506         verify_location, collect_subblocks): New functions.
26507         (verify_gimple_in_cfg): Verify that locations only reference
26508         BLOCKs in the functions BLOCK tree.
26510 2013-01-14  Richard Biener  <rguenther@suse.de>
26512         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26513         PHI argument.
26514         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26515         unshare reference.
26516         (insert_out_of_ssa_copy_on_edge): Likewise.
26517         (rewrite_close_phi_out_of_ssa): Likewise.
26518         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26519         debug expressions.
26520         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26521         propagated constants.
26522         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26523         can not be shared.
26525 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26527         * config/avr/avr-modes.def: Add GPL copyright notice.
26529 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26531         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26532         MEMMODEL_MASK to determine memory model.
26533         (atomic_store<mode>): Ditto from operands[2].
26534         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26536 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26538         PR fortran/55935
26539         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26540         (fold_gimple_assign): Don't call unshare_expr here.
26541         (fold_ctor_reference): Call unshare_expr.
26543 2013-01-13  Terry Guo  <terry.guo@arm.com>
26545         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26546         * doc/fragments.texi: Document MULTILIB_REUSE.
26547         * gcc.c (multilib_reuse): New internal spec.
26548         (set_multilib_dir): Also search multilib from multilib_reuse.
26549         * genmultilib (tmpmultilib3): Refactor code.
26550         (tmpmultilib4): Ditto.
26551         (multilib_reuse): New multilib argument.
26553 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26555         * Makefile.in: Update copyright.
26557 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26559         PR middle-end/55890
26560         * calls.c (expand_call): Check if arg_nr is valid.
26562 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26564         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26565         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26566         documentation.  Add missing '__' in front of
26567         __builtin_ia32_packssdw256.
26569 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26571         PR target/55719
26572         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26573         NO_REGS for larl operands.
26574         (s390_reload_larl_operand): Use s390_load_address instead of
26575         emit_move_insn.
26577 2013-01-11  Richard Biener  <rguenther@suse.de>
26579         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26580         (verify_node_sharing): ... here.
26581         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26583 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26585         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26586         Substitute TREECHECKING.
26587         * configure: Regenerate.
26588         * Makefile.in (TREECHECKING): New.
26590 2013-01-11  Richard Guenther  <rguenther@suse.de>
26592         PR tree-optimization/44061
26593         * tree-vrp.c (extract_range_basic): Compute zero as
26594         value-range for __builtin_constant_p of function parameters.
26596 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26598         Update copyright years.
26600 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26602         PR rtl-optimization/55672
26603         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26604         const to be eliminable.
26606 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26608         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26609         * configure: Regenerate.
26611 2013-01-10  Richard Biener  <rguenther@suse.de>
26613         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26615 2013-01-10  Richard Biener  <rguenther@suse.de>
26617         PR bootstrap/55792
26618         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26619         locations for virtual PHI arguments.
26620         (rewrite_update_phi_arguments): Likewise.
26622 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26624         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26625         on to assembler.
26627 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26629         PR tree-optimization/55921
26630         * tree-complex.c (expand_complex_asm): New function.
26631         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26633 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26635         PR target/55718
26636         * config/s390/s390.c (s390_symref_operand_p)
26637         (s390_loadrelative_operand_p): Merge the two functions.
26638         (s390_check_qrst_address, print_operand_address): Add parameters
26639         to s390_loadrelative_operand_p invokation.
26640         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26641         (s390_reload_larl_operand, s390_secondary_reload): Use
26642         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26643         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26645 2013-01-09  Mike Stump  <mikestump@comcast.net>
26647         * dse.c (record_store): Remove unnecessary assert.
26649 2013-01-09  Jan Hubicka  <jh@suse.cz>
26651         PR tree-optimization/55569
26652         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26653         * cfgloop.h (scale_loop_profile): Likewise.
26655 2013-01-09  Jan Hubicka  <jh@suse.cz>
26657         PR lto/45375
26658         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26659         functions.
26660         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26662 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26664         PR middle-end/55114
26665         * expr.h (maybe_emit_group_store): Declare.
26666         * expr.c (maybe_emit_group_store): New function.
26667         * builtins.c (expand_builtin_int_roundingfn): Call it.
26668         (expand_builtin_int_roundingfn_2): Likewise.
26670 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26672         PR rtl-optimization/55829
26673         * lra-constraints.c (match_reload): Add code for absent output.
26674         (curr_insn_transform): Add code for reloads of matched inputs
26675         without output.
26677 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26679         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26680         attribute of movddup insn to DF.
26681         (*vec_interleave_lowv2df): Ditto.
26682         (vec_dupv2df): Ditto.
26684 2013-01-09  Jan Hubicka  <jh@suse.cz>
26686         PR tree-optimiation/55875
26687         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26688         EVERY_ITERATION parameter.
26689         (number_of_iterations_exit): Check if exit is executed every iteration.
26690         (idx_infer_loop_bounds): Similarly here.
26691         (n_of_executions_at_most): Simplify
26692         to only test for cases where statement is dominated by the
26693         particular bound; handle correctly the "postdominance" test.
26694         (scev_probably_wraps_p): Use max loop iterations info
26695         as a global bound first.
26697 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26698             Nick Clifton  <nickc@redhat.com>
26700         * config/v850/v850.md (cbranchsf4): New pattern.
26701         (cstoresf4): New pattern.
26702         (cbranchdf4): New pattern.
26703         (cstoredf4): New pattern.
26704         (movsicc): Disallow floating point comparisons.
26705         (cmpsf_le_insn): Fix order of operators.
26706         (cmpsf_lt_insn): Likewise.
26707         (cmpsf_eq_insn): Likewise.
26708         (cmpdf_le_insn): Likewise.
26709         (cmpdf_lt_insn): Likewise.
26710         (cmpdf_eq_insn): Likewise.
26711         (cmpsf_ge_insn): Use LE comparison.
26712         (cmpdf_ge_insn): Likewise.
26713         (cmpsf_gt_insn): Use LT comparison.
26714         (cmpdf_gt_insn): Likewise.
26715         (cmpsf_ne_insn): Delete pattern.
26716         (cmpdf_ne_insn): Delete pattern.
26717         * config/v850/v850.c (v850_gen_float_compare): Use
26718         gen_cmpdf_eq_insn for NE comparison.
26719         (v850_float_z_comparison_operator)
26720         (v850_float_nz_comparison_operator): Move from here ...
26721         * config/v850/predicates.md: ... to here.  Move GT and GE
26722         comparisons into v850_float_z_comparison_operator.
26723         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26724         Delete prototype.
26725         (v850_float_nz_comparison_operator): Likewise.
26727 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26729         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26730         with calls to gen_insvsi/gen_insvdi.
26732 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26734         * config/i386/i386.c (initial_ix86_tune_features): Set up
26735         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26737 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26738             Jakub Jelinek  <jakub@redhat.com>
26740         PR tree-optimization/48189
26741         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26742         If nitercst is 0, don't predict the exit edge.
26744 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26746         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26747         in asm_fprintf with reg_names.
26748         (aarch64_print_operand_address): Likewise.
26749         (aarch64_return_addr): Likewise.
26750         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26752 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26754         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26755         (INT_U6_BITS): Likewise.
26756         * config/pa/predicates.md (uint6_operand): New predicate.
26757         (shift5_operand, shift6_operand): Likewise.
26758         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26759         arith32_operand.
26760         (lshrdi3): Use shift6_operand.
26761         (shrpsi4, shrpdi4): New insn patterns.
26762         (extzv): Delete expander.
26763         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26764         predicates in unamed zero extract patterns.  Tighten common constraint.
26765         (extv): Delete expander.
26766         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26767         predicates in unamed sign extract patterns.  Tighten common constraint.
26768         (insv): Delete expander.
26769         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26770         predicates in unamed insert patterns.  Tighten common constraint.
26771         Change uint32_operand predicate to uint6_operand predicate in unamed
26772         DImode pattern to insert constant values of type 1...1xxxx.
26774 2013-01-04  Jan Hubicka  <jh@suse.cz>
26776         PR tree-optimization/55823
26777         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26778         issue.
26780 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26781             Uros Bizjak  <ubizjak@gmail.com>
26783         PR rtl-optimization/55845
26784         * df-problems.c (can_move_insns_across): Stop scanning at
26785         volatile_insn_p source instruction or give up if
26786         across_from .. across_to range contains any volatile_insn_p
26787         instructions.
26789 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26791         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26792         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26793         Declare.
26794         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26795         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26797 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26799         PR fortran/55341
26800         * asan.c (asan_clear_shadow): New function.
26801         (asan_emit_stack_protection): Use it.
26803 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26805         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26806         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26807         with tab instead of space.
26809 2013-01-08  Nick Clifton  <nickc@redhat.com>
26811         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26812         register bank 0 at the start of an interrupt handler.
26813         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26814         MDBH registers.
26816 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26818         * config/aarch64/aarch64-simd.md
26819         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26820         (aarch64_simd_bsl): Likewise.
26821         (aarch64_vcond_internal<mode>): Likewise.
26822         (vcond<mode><mode>): Likewise.
26823         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26824         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26826 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26828         * config/aarch64/aarch64-builtins.c
26829         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26831 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26833         PR debug/55579
26834         * tree-sra.c (analyze_access_subtree): Return true also after
26835         potentially creating a debug-only replacement.
26837 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26839         PR middle-end/55890
26840         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26842         PR tree-optimization/54120
26843         * tree-vrp.c (range_fits_type_p): Don't allow
26844         src_precision < precision from signed vr to unsigned_p
26845         if vr->min or vr->max is negative.
26846         (simplify_float_conversion_using_ranges): Test can_float_p
26847         against CODE_FOR_nothing.
26849 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26850             Richard Biener  <rguenther@suse.de>
26852         PR middle-end/55851
26853         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26854         types instead of just INTEGER_TYPE types.
26856 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26858         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26859         TF_SIZE): Define.
26861 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26863         PR target/42661
26864         * config/mips/mips.opt: Change mad to mmad to match documentation.
26866 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26868         PR target/55897
26869         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26870         .progmemx.data now.
26872 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26874         PR target/55897
26875         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26876         (avr_addrspace_t): Add .section_name field.
26877         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26878         array size.
26879         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26880         NULL entry.  Put __memx into .progmemx.data.
26881         (progmem_section_prefix): Remove.
26882         (avr_asm_init_sections): No need to initialize progmem_section.
26883         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26884         section name prefix.
26885         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26886         retrieve the progmem section.
26887         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26888         boundary to run over avr_addrspace[].
26889         (avr_register_target_pragmas): Ditto.
26891 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26893         * varasm.c (output_constant_def_contents): For asan_protect_global
26894         protected strings, adjust DECL_ALIGN if needed, before testing for
26895         anchored symbols.
26896         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26897         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26898         normal decls.
26899         (output_object_block): For asan protected decls, emit asan padding
26900         after their contents.
26901         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26902         (asan_finish_file): Test it here instead.
26904 2013-01-07  Nick Clifton  <nickc@redhat.com>
26905             Matthias Klose  <doko@debian.org>
26906             Doug Kwan  <dougkwan@google.com>
26907             H.J. Lu  <hongjiu.lu@intel.com>
26909         PR driver/55470
26910         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26912         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26914         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26916         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26918         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26920 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26922         PR target/54461
26923         * doc/install.texi (Cross-Compiler-Specific Options): Document
26924         --with-avrlibc.
26926 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26928         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26929         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26930         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26931         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26932         vqmovun_high_s64): Fix source operand number and update copyright.
26934 2013-01-07  Richard Biener  <rguenther@suse.de>
26936         PR middle-end/55890
26937         * gimple.h (gimple_call_builtin_p): New overload.
26938         * gimple.c (validate_call): New function.
26939         (gimple_call_builtin_p): Likewise.
26940         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26941         Use gimple_call_builtin_p.
26942         (find_func_clobbers): Likewise.
26943         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26944         (strlen_optimize_stmt): Likewise.
26946 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26948         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26949         (vld1q_dup_*): Likewise.
26950         (vld1_*): Likewise.
26951         (vld1q_*): Likewise.
26952         (vld1_lane_*): Likewise.
26953         (vld1q_lane_*): Likewise.
26955 2013-01-07  Richard Biener  <rguenther@suse.de>
26957         * lto-streamer.h (LTO_minor_version): Bump to 2.
26959 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26961         * config/aarch64/aarch64-protos.h
26962         (aarch64_const_double_zero_rtx_p): Rename to...
26963         (aarch64_float_const_zero_rtx_p): ...this.
26964         (aarch64_float_const_representable_p): New.
26965         (aarch64_output_simd_mov_immediate): Likewise.
26966         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26967         move immediate case.
26968         * config/aarch64/aarch64.c
26969         (aarch64_const_double_zero_rtx_p): Rename to...
26970         (aarch64_float_const_zero_rtx_p): ...this.
26971         (aarch64_print_operand): Allow printing of new constants.
26972         (aarch64_valid_floating_const): New.
26973         (aarch64_legitimate_constant_p): Check for valid floating-point
26974         constants.
26975         (aarch64_simd_valid_immediate): Likewise.
26976         (aarch64_vect_float_const_representable_p): New.
26977         (aarch64_float_const_representable_p): Likewise.
26978         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26979         (aarch64_output_simd_mov_immediate): New.
26980         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26981         (*movdf_aarch64): Likewise.
26982         * config/aarch64/constraints.md (Ufc): New.
26983         (Y): call aarch64_float_const_zero_rtx.
26984         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26986 2013-01-07  Richard Biener  <rguenther@suse.de>
26988         PR tree-optimization/55888
26989         PR tree-optimization/55862
26990         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26991         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26992         not if it is contained therein.
26994 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26996         * config/avr/t-avr: Typo.
26998 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27000         PR55243
27001         * config/avr/t-avr: Don't automatically rebuild
27002         $(srcdir)/config/avr/t-multilib
27003         $(srcdir)/config/avr/avr-tables.opt
27004         $(srcdir)/doc/avr-mmcu.texi
27005         (avr-mcus): New phony target to build them on request.
27006         (s-avr-mlib, s-avr-mmcu-texi): Remove.
27007         * avr/avr-mcus.def: Adjust comments.
27009 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
27011         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
27013 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27015         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
27017 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27019         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27021 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27023         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27024         to generate profiling.
27025         * config/rs6000/aix64.h (LIB_SPEC): Same.
27027 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27029         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27030         New function.
27031         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27033 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27035         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27036         unconditionally.
27037         (ix86_expand_move): Ditto.
27038         (ix86_zero_extend_to_Pmode): Ditto.
27039         (ix86_expand_call): Ditto.
27040         (ix86_expand_special_args_builtin): Ditto.
27041         (ix86_expand_builtin): Ditto.
27043 2013-01-04  Richard Biener  <rguenther@suse.de>
27045         PR tree-optimization/55862
27046         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27047         translating them through PHI nodes.
27049 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27051         PR tree-optimization/55755
27052         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27053         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27055 2013-01-04  Marek Polacek  <polacek@redhat.com>
27057         PR middle-end/55859
27058         * opts.c (default_options_optimization): Clarify error message.
27060 2013-01-04  Richard Biener  <rguenther@suse.de>
27062         PR middle-end/55863
27063         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27064         reassociation.
27066 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27068         PR target/53789
27069         * config/pa/pa.md (movsi): Revert previous change.
27070         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27071         references.
27073 2013-01-03  Richard Henderson  <rth@redhat.com>
27075         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27076         after eliminating TLS symbols.
27078 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27080         PR bootstrap/50167
27081         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27082         * graphite-poly.c (debug_gmp_value): Likewise.
27084 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27086         PR target/55712
27087         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27088         selected code model, define __code_mode_small__, __code_model_medium__,
27089         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27090         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27091         xchg temporary register with %k.  Declare temporary register as
27092         early clobbered.
27093         [__x86_64__]: For medium and large code models, preserve %rbx register.
27095 2013-01-03  Richard Biener  <rguenther@suse.de>
27097         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27098         (dump_subscript): Adjust.
27099         (finalize_ddr_dependent): Do not dump redundant info.
27100         (analyze_siv_subscript): Adjust.
27101         (subscript_dependence_tester): Likewise.
27102         (compute_affine_dependence): Likewise.
27104 2013-01-03  Richard Biener  <rguenther@suse.de>
27106         Revert
27107         2013-01-03  Richard Biener  <rguenther@suse.de>
27109         PR tree-optimization/55857
27110         * tree-vect-stmts.c (vectorizable_load): Do not setup
27111         re-alignment for invariant loads.
27113         2013-01-02  Richard Biener  <rguenther@suse.de>
27115         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27116         invariant load do not generate a vector load from the scalar location.
27118 2013-01-03  Richard Biener  <rguenther@suse.de>
27120         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27121         for not vectorizing.
27122         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27123         not build INDIRECT_REFs, call get_name once only.
27124         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27125         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27127 2013-01-03  Richard Biener  <rguenther@suse.de>
27129         PR tree-optimization/55857
27130         * tree-vect-stmts.c (vectorizable_load): Do not setup
27131         re-alignment for invariant loads.
27133 2013-01-03  Richard Biener  <rguenther@suse.de>
27135         PR lto/55848
27136         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27137         prefer a built-in decl.
27139 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27141         * gcc.c (process_command): Update copyright notice dates.
27142         * gcov.c (print_version): Likewise.
27143         * gcov-dump.c (print_version): Likewise.
27145         PR rtl-optimization/55838
27146         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27147         iv0.step, iv1.step and step.
27149 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27150             Marc Glisse  <marc.glisse@inria.fr>
27152         PR tree-optimization/55832
27153         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27154         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27155         integer_{one,zero}_node.
27157 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27159         PR debug/54402
27160         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27161         * var-tracking.c (reverse_op): Don't add reverse ops to
27162         VALUEs that have already
27163         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27165 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27167         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27169 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27171         * dumpfile.c (dump_loc): Print filename with location.
27172         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27173         new location_t parameter to emit complete unroll message with
27174         new dump framework.
27175         (canonicalize_loop_induction_variables): Compute loops location
27176         and pass to try_unroll_loop_completely.
27177         * loop-unroll.c (report_unroll_peel): New function.
27178         (peel_loops_completely): Use new dump format with location
27179         for main dumpfile message, and invoke report_unroll_peel on success.
27180         (decide_unrolling_and_peeling): Ditto.
27181         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27182         by report_unroll_peel.
27183         (decide_peel_completely): Ditto.
27184         (decide_unroll_constant_iterations): Ditto.
27185         (decide_unroll_runtime_iterations): Ditto.
27186         (decide_peel_simple): Ditto.
27187         (decide_unroll_stupid): Ditto.
27188         * cfgloop.c (get_loop_location): New function.
27189         * cfgloop.h (get_loop_location): Declare.
27191 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27193         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27194         NULL.
27196 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27198         PR middle-end/55198
27199         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27200         BLKmode objects when EXPAND_MEMORY is specified.
27202 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27204         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27205         in loop predicate.
27206         (fold_builtin_cpu): Do not share cpu model decls across statements.
27208 2013-01-02  Jason Merrill  <jason@redhat.com>
27210         PR c++/55804
27211         * tree.c (build_array_type_1): Revert earlier change.
27213 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27215         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27216         "cortex-a57".
27217         * config/aarch64/aarch64-tune.md: Re-generate.
27219 2013-01-02  Richard Biener  <rguenther@suse.de>
27221         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27222         invariant load do not generate a vector load from the scalar location.
27224 2013-01-02  Richard Biener  <rguenther@suse.de>
27226         PR bootstrap/55784
27227         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27228         * configure: Regenerate.
27230 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27232         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27233         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27234         (expand_builtin_int_roundingfn_2): Keep the original target around
27235         for the fallback case.
27237 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27239         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27240         to be clear for sign changes.
27242 2013-01-01  Jan Hubicka  <jh@suse.cz>
27244         * ipa-inline-analysis.c: Fix formatting.
27246 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27248         PR tree-optimization/55831
27249         * tree-vect-loop.c (get_initial_def_for_induction): Use
27250         gsi_after_labels instead of gsi_start_bb.
27252 Copyright (C) 2013 Free Software Foundation, Inc.
27254 Copying and distribution of this file, with or without modification,
27255 are permitted in any medium without royalty provided the copyright
27256 notice and this notice are preserved.