Re-factor inclusion of tree.h.
[official-gcc.git] / gcc / ChangeLog
blob643db376af9ba536594f3b42fa4238c455f84d2f
1 2013-10-21  Marek Polacek  <polacek@redhat.com>
3         PR middle-end/58809
4         * fold-const.c (fold_range_test): Return 0 if the type is not
5         an integral type.
7 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
9         * system.h: Move hwint.h include further down.
10         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
11         gcc_checking_asserts.
12         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
14 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
16         Fix volatile issues in optimize_bit_field_compare.
17         * fold-const.c (optimize_bit_field_compare): Bail out if
18         lvolatilep or rvolatilep.
20 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
22         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
23         and get_inner_reference returning different pmode for non-volatile
24         bit-field members dependent on flag_strict_volatile_bitfields.
25         * stor-layout.c (layout_decl): Remove special handling of
26         flag_strict_volatile_bitfields.
27         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
28         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
30 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
32         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
33         calculations.
35 2013-10-21  Jeff Law  <law@redhat.com>
37         * tree-ssa-threadedge.c (thread_through_normal_block): New argument VISITED.
38         Remove VISISTED as a local variable.  When we have a threadable jump, verify
39         the destination of the jump has not been visised.
40         (thread_across_edge): Allocate VISITED bitmap once at function scope and
41         use it throughout.  Make sure to set appropriate bits in VISITED for E (start
42         of jump thread path).
44         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading through
45         a joiner if any edge on the path has a recorded jump thread.
47 2013-10-21  Ian Lance Taylor  <iant@google.com>
49         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
50         don't imply that attributes can solve all problems.
51         (Directory Options): Fix typo.
53 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
55         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
56         extend and extend_add.
58 2013-10-21  Richard Biener  <rguenther@suse.de>
60         PR tree-optimization/58794
61         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
62         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
64 2013-10-21  Richard Biener  <rguenther@suse.de>
66         PR middle-end/58742
67         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
68         to (T) X for sign-changing conversions (or no conversion).
70 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
72         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
74 2013-10-20  Jan Hubicka  <jh@suse.cz>
76         * config/i386/i386-tune.def: Add comment; organize into categories
78 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
80         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
81         argument.  Update function comment.
82         (expand_set_or_movmem_via_rep): New function combining
83         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
84         (expand_movmem_via_rep_mov): Remove.
85         expand_setmem_via_rep_stos): Remove.
86         (expand_movmem_epilogue): Update calls correspondingly.
87         (expand_setmem_epilogue_via_loop): Likewise.
88         (emit_memset): New.
89         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
90         (expand_set_or_movmem_prologue): New function combining
91         expand_movmem_prologue and expand_setmem_prologue.
92         (expand_movmem_prologue): Remove.
93         (expand_setmem_prologue): Remove.
94         (expand_set_or_movmem_constant_prologue): New function combining
95         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
96         (expand_constant_movmem_prologue): Remove.
97         (expand_constant_setmem_prologue): Remove.
98         (promote_duplicated_reg): Allow vector-const0 value.
99         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
100         and ix86_expand_setmem.
101         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
102         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
104 2013-10-21  Diego Novillo  <dnovillo@google.com>
106         * asan.c: Include tree.h
107         * bb-reorder.c: Likewise.
108         * cfgcleanup.c: Likewise.
109         * cfgloopmanip.c: Likewise.
110         * data-streamer-in.c: Likewise.
111         * data-streamer-out.c: Likewise.
112         * data-streamer.c: Likewise.
113         * dwarf2cfi.c: Likewise.
114         * graphite-blocking.c: Likewise.
115         * graphite-clast-to-gimple.c: Likewise.
116         * graphite-dependences.c: Likewise.
117         * graphite-interchange.c: Likewise.
118         * graphite-optimize-isl.c: Likewise.
119         * graphite-poly.c: Likewise.
120         * graphite-scop-detection.c: Likewise.
121         * graphite-sese-to-poly.c: Likewise.
122         * graphite.c: Likewise.
123         * ipa-devirt.c: Likewise.
124         * ipa-profile.c: Likewise.
125         * ipa.c: Likewise.
126         * ira.c: Likewise.
127         * loop-init.c: Likewise.
128         * loop-unroll.c: Likewise.
129         * lower-subreg.c: Likewise.
130         * lto/lto-object.c: Likewise.
131         * recog.c: Likewise.
132         * reginfo.c: Likewise.
133         * tree-loop-distribution.c: Likewise.
134         * tree-parloops.c: Likewise.
135         * tree-ssa-strlen.c: Likewise.
136         * tree-streamer.c: Likewise.
137         * value-prof.c: Likewise.
138         * target-globals.c: Likewise.
139         * expr.h: Include tree-core.h instead of tree.h.
140         * gimple.h: Likewise.
141         * ipa-prop.h: Likewise.
142         * ipa-utils.h: Likewise.
143         * lto-streamer.h: Likewise.
144         * streamer-hooks.h: Likewise.
145         * ipa-reference.h: Include cgraph.h instead of tree.h.
146         * cgraph.h: Include basic-block.h instead of tree.h.
147         * tree-streamer.h: Do not include tree.h.
148         * genattrtab.c (write_header): Generate inclusion of
149         tree.h.
150         * genautomata.c (main): Likewise.
151         * genemit.c: Likewise.
152         * genopinit.c: Likewise.
153         * genoutput.c (output_prologue): Likewise.
154         * genpeep.c: Likewise.
156 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
158         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
159         little endian.
160         (vec_unpacku_hi_v8hi): Likewise.
161         (vec_unpacku_lo_v16qi): Likewise.
162         (vec_unpacku_lo_v8hi): Likewise.
164 2013-10-20  Jan Hubicka  <jh@suse.cz>
166         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
167         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
168         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
170 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
172         * config/mips/mips.h (ISA_HAS_WSBH): Define.
173         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
174         constants.
175         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
177 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
179         PR target/58603
180         * system.h: Undef m_slot.
182 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
184         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
185         all elements for both endian flavors.
187 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
189         PR target/58792
190         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
191         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
192         and SI_REG for 64bit SYSV ABI targets.
194 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
196         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
197         to multi_reg_return.  Clarify that we are skipping USEs of multiple
198         return registers.  Use bool type where appropriate.
200 2013-10-18  Jan Hubicka  <jh@suse.cz>
202         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
203         for cold functions.
204         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
205         (X86_TUNE_PUSH_MEMORY): Likewise.
206         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
207         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
208         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
209         New.
210         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
211         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
212         Remove.
213         (ix86_option_override_internal): Update to use tune features instead
214         of variables.
216 2013-10-18  Cong Hou  <congh@google.com>
218         PR tree-optimization/58508
219         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
220         statement that contains data refs with zero-step.
222 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
224         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
225         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
226         * gimple-low.c (gimple_check_call_arg,
227         gimple_check_call_matching_types): Move to cgraph.c.
228         * gimple-low.h: Remove prototype.
229         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
230         Relocate from gimple-low.c.
231         * cgraph.h: Add prototype.  Don't include basic-block.h.
232         * gimplify.c: Add gimple-low to include list.
233         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
234         * tree-eh.c: Add gimple-low to include list.
235         * tree-nested.c: Likewise.
236         * cfgexpand.c: Add tree-ssa-address.h to include list.
237         * expr.c: Likewise.
238         * gimple-fold.c: Likewise.
239         * gimple-ssa-strength-reduction.c: Likewise.
240         * trans-mem.c: Likewise.
241         * tree-mudflap.c: Likewise.
242         * tree-ssa-loop-ivopts.c: Likewise.
243         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
244         (degenerate_phi_result): Move to tree-phinodes.c.
245         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
246         * tree-ssa-threadedge.c: Likewise.
247         * tree-vrp.c: Likewise.
248         * tree-phinodes.c (degenerate_phi_result): Relocate here.
249         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
250         * tree-phinodes.h (degenerate_phi_result): Add prototype.
251         * tree-ssa-copy.c: Include tree-ssa-dom.h.
252         * tree-ssa-forwprop.c: Likewise.
253         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
254         pass_data_cleanup_cfg_post_optimizing,
255         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
256         * tree-optimize.c: Delete File.
257         * graphite.c: Include tree-cfgcleanup.h.
258         * passes.c: Likewise.
259         * tree-cfg.c: Likewise.
260         * tree-profile.c: Likewise.
261         * tree-ssa-dse.c: Likewise.
262         * tree-ssa-loop-ivcanon.c: Likewise.
263         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
264         * tree-outof-ssa.c: Include sbitmap.h.
265         * tree-ssa-live.c: Likewise.
266         * tree-ssa-propagate.c: Likewise.
267         * tree-ssa-structalias.c: Likewise.
268         * tree-stdarg.c: Likewise.
269         * Makefile.in (OBJS): Delete tree-optimize.o.
270         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
271         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
272         * varasm.c: Include basic-block.h.
273         * cfgloop.h: Include function.h instead of basic-block.h
274         (bb_loop_depth): Move to cfgloop.c.
275         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
277 2013-10-18  Teresa Johnson  <tejohnson@google.com>
279         * predict.c (probably_never_executed): Compare frequency-based
280         count to number of training runs.
281         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
283 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
285         * config/arm/arm.c (cortexa9_extra_costs): New table.
286         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
288 2013-10-18  Jeff Law  <law@redhat.com>
290         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
292         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
293         (register_jump_thread): Add "registered_jump_thread" debug
294         counter support.
295         * dbgcnt.def (registered_jump_thread): New debug counter.
297 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
299         * config/rs6000/rs6000.c: Include cgraph.h.
301 2013-10-18  Teresa Johnson  <tejohnson@google.com>
303         * tree-ssa-tail-merge.c (replace_block_by): Update edge
304         weights during merging.
306 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
308         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
309         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
310         * tree-ssa-operands.h: Remove prototype.
311         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
312         * gimple.c (virtual_operand_p): Relocate from gimple.c.
313         * gimple.h: Add prototype.
314         * gimple-ssa.h: Include tree-ssa-operands.h.
315         * tree-dump.c: Add tree-cfg.h to include list.
316         * tree-ssa-alias.c: Add ipa-reference.h to include list.
317         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
318         * config/i386/i386.c: Don't include tree-flow.h.
319         * config/rs6000/rs6000.c: Likewise.
321 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
323         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
325 2013-10-18  Richard Biener  <rguenther@suse.de>
327         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
328         or TYPE_UNSIGNED of integral types.
329         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
330         NULL_TREE for zero-precision integral types.
332 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
334         * config/aarch64/arm_neon.h
335         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
337 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
339         * opth-gen.awk: Define target_flags_explicit.
341 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
343         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
344         fields to the reg_addr array that describes the valid addressing
345         mode for any register, general purpose registers, floating point
346         registers, and Altivec registers.
347         (FIRST_RELOAD_REG_CLASS): Likewise.
348         (LAST_RELOAD_REG_CLASS): Likewise.
349         (struct reload_reg_map_type): Likewise.
350         (reload_reg_map_type): Likewise.
351         (RELOAD_REG_VALID): Likewise.
352         (RELOAD_REG_MULTIPLE): Likewise.
353         (RELOAD_REG_INDEXED): Likewise.
354         (RELOAD_REG_OFFSET): Likewise.
355         (RELOAD_REG_PRE_INCDEC): Likewise.
356         (RELOAD_REG_PRE_MODIFY): Likewise.
357         (reg_addr): Likewise.
358         (mode_supports_pre_incdec_p): New helper functions to say whether
359         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
360         (mode_supports_pre_modify_p): Likewise.
361         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
362         print the valid address mode bits for each mode.
363         (rs6000_debug_print_mode): Likewise.
364         (rs6000_debug_reg_global): Likewise.
365         (rs6000_setup_reg_addr_masks): New function to set up the address
366         mask bits for each type.
367         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
368         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
369         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
370         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
371         PRE_MODIFY are supported.
372         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
373         registers, instead of {src,dest}_av_p.
374         (rs6000_print_options_internal): Tweak the debug output slightly.
376 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
378         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
379         isa attribute.
381 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
383         * tree-flow.h (struct omp_region): Move to omp-low.c.
384         Remove omp_ prototypes and variables.
385         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
386         (copy_var_decl): Relocate prototype from tree-flow.h.
387         * gimple.c (copy_var_decl): Relocate from omp-low.c.
388         * tree.h: Move prototype to omp-low.h.
389         * omp-low.h: New File.  Relocate prototypes here.
390         * omp-low.c (struct omp_region): Make local here.
391         (root_omp_region): Make static.
392         (copy_var_decl) Move to gimple.c.
393         (new_omp_region): Make static.
394         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
395         * tree-cfg.c: Include omp-low.h.
396         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
397         * gimplify.c: Include omp-low.h.
398         * tree-parloops.c: Likewise.
400 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
402         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
403         operands are in memory, prefer to force non-matched operand 1 to
404         the register.
406 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
408         PR target/58673
409         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
410         restrict TImode addresses to single indirect registers if both
411         -mquad-memory and -mvsx-timode are used.
412         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
413         we should emit load/store quad.  Remove using %y for quad memory
414         addresses.
416         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
417         constraints to allow load/store quad on machines where TImode is
418         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
419         for TImode to load integer constants.
421 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
423         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
425 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
427         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
428         handling of STACK_REG.
430 2013-10-17  Richard Biener  <rguenther@suse.de>
432         PR tree-optimization/58143
433         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
434         New function.
435         (rewrite_to_defined_overflow): Likewise.
436         (move_computations_dom_walker::before_dom): Rewrite stmts
437         with undefined signed overflow that are not always executed
438         into unsigned arithmetic.
440 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
442         PR target/57756
443         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
444         explicit isa flag to be an options variable, instead of using
445         global_options_set.  Remove define from rs6000.h.
446         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
448         * config/rs6000/rs6000.c (rs6000_option_override_internal):
449         Initialize rs6000_isa_flags_explicit.
450         (rs6000_function_specific_save): Add gcc_options* parameter, so
451         that the powerpc builds after the 2013-10-15 changes.
452         (rs6000_function_specific_restore): Likewise.
454 2013-10-16  DJ Delorie  <dj@redhat.com>
456         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
457         op is a REG before checking REGNO.
458         (rl78_alloc_physical_registers): Verify pattern is a SET before
459         checking SET_SRC.
461 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
463         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
464         endianness.
465         (vec_unpacks_lo_v4sf): Likewise.
466         (vec_unpacks_float_hi_v4si): Likewise.
467         (vec_unpacks_float_lo_v4si): Likewise.
468         (vec_unpacku_float_hi_v4si): Likewise.
469         (vec_unpacku_float_lo_v4si): Likewise.
471 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
473         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
474         (vsx_concat_v2sf): Likewise.
476 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
478         * config/aarch64/aarch64.md
479         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
481 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
483         PR tree-optimization/58697
484         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
485         estimated_loop_iterations_int.
486         (max_stmt_executions_int): Call get_max_loop_iterations_int.
487         (get_max_loop_iterations_int): New.  HWINT version of
488         get_max_loop_iterations.
489         * cfgloop.h: Add prototypes.
490         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
491         * loop-unroll.c (decide_peel_once_rolling): Call
492         get_estimated_loop_iterations_int.
493         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
494         * tree-ssa-loop-niter.h: Tweak prototypes.
496 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
498         * gengtype-parse.c (struct_field_seq): Ignore access-control
499         keywords ("public:" etc).
501 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
503         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
504         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
506 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
508         * config/arm/arm.opt (mlra): New option.
509         * config/arm/arm.c (arm_lra_p): New function.
510         (TARGET_LRA_P): Define.
512 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
514         * tree-core.h (tree_code_name): Remove.
515         * tree.h (get_tree_code_name): New prototype.
516         * tree.c (tree_code_name): Make static.
517         (get_tree_code_name): New function.
518         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
519         tree_class_check_failed, tree_range_check_failed,
520         tree_not_class_check_failed, omp_clause_check_failed,
521         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
522         wrapper get_tree_code_name instead of calling tree_code_name directly.
523         * tree-vrp.c (dump_asserts_for): Likewise.
524         * tree-dump.c (dequeue_and_dump): Likewise.
525         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
526         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
527         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
528         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
529         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
530         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
531         dump_gimple_omp_for): Likewise.
532         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
533         * tree-ssa-pre.c (print_pre_expr): Likewise.
534         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
535         * print-tree.c (print_node_brief, print_node): Likewise.
536         * gimple.c (gimple_check_failed): Likewise.
537         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
538         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
539         * config/mep/mep.c (mep_validate_vliw): Likewise.
540         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
541         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
543 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
545         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
546         for AMD bdver3.
548 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
550         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
551         (MULTILIB_MATCHES): Add multilib for -march=v8.
553 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
555         PR target/57756
556         * optc-save-gen.awk: Add extra parameter to the save and restore
557         target calls.
558         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
559         * tree.c (build_optimization_node): New parameter.  Add extra parameter
560         to call to cl_optimization_save.
561         (build_target_option_node): New parameter. Add extra parameter
562         to call to cl_target_option_save.
563         * tree.h (build_optimization_node): New parameter.
564         (build_target_option_node): New parameter.
565         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
566         build_optimization_node and build_target_option_node.
567         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
568         (handle_pragma_push_options): Ditto.
569         * toplev.c (process_options): Ditto.
570         * opts.c (init_options_struct): Check for opts_set non-null.
571         * target.def (target_option.save): New parameter.
572         (target_option.restore): New parameter.
573         * tm.texi: Generate.
574         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
575         (ix86_pragma_target_parse): Ditto.
576         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
577         parameters.
578         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
579         to  build_optimization_node and build_target_option_node.
580         (rs6000_valid_attribute_p): Ditto.
581         (rs6000_pragma_target_parse): Ditto.
582         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
583         data.
584         * config/i386/i386.h:
585         TARGET_64BIT_P: New Macro
586         TARGET_MMX_P: New Macro.
587         TARGET_3DNOW_P: New Macro.
588         TARGET_3DNOW_A_P: New Macro.
589         TARGET_SSE_P: New Macro.
590         TARGET_SSE2_P: New Macro.
591         TARGET_SSE3_P: New Macro.
592         TARGET_SSSE3_P: New Macro.
593         TARGET_SSE4_1_P: New Macro.
594         TARGET_SSE4_2_P: New Macro.
595         TARGET_AVX_P: New Macro.
596         TARGET_AVX2_P: New Macro.
597         TARGET_AVX512F_P: New Macro.
598         TARGET_AVX512PF_P: New Macro.
599         TARGET_AVX512ER_P: New Macro.
600         TARGET_AVX512CD_P: New Macro.
601         TARGET_FMA_P: New Macro.
602         TARGET_SSE4A_P: New Macro.
603         TARGET_FMA4_P: New Macro.
604         TARGET_XOP_P: New Macro.
605         TARGET_LWP_P: New Macro.
606         TARGET_ABM_P: New Macro.
607         TARGET_BMI_P: New Macro.
608         TARGET_BMI2_P: New Macro.
609         TARGET_LZCNT_P: New Macro.
610         TARGET_TBM_P: New Macro.
611         TARGET_POPCNT_P: New Macro.
612         TARGET_SAHF_P: New Macro.
613         TARGET_MOVBE_P: New Macro.
614         TARGET_CRC32_P: New Macro.
615         TARGET_AES_P: New Macro.
616         TARGET_PCLMUL_P: New Macro.
617         TARGET_CMPXCHG16B_P: New Macro.
618         TARGET_FSGSBASE_P: New Macro.
619         TARGET_RDRND_P: New Macro.
620         TARGET_F16C_P: New Macro.
621         TARGET_RTM_P: New Macro.
622         TARGET_HLE_P: New Macro.
623         TARGET_RDSEED_P: New Macro.
624         TARGET_PRFCHW_P: New Macro.
625         TARGET_ADX_P: New Macro.
626         TARGET_FXSR_P: New Macro.
627         TARGET_XSAVE_P: New Macro.
628         TARGET_XSAVEOPT_P: New Macro.
629         TARGET_LP64_P: New Macro.
630         TARGET_X32_P: New Macro.
631         TARGET_FPMATH_DEFAULT_P: New Macro.
632         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
633         * config/i386/i386.c (ix86_option_override_internal): New parameters.
634         opts and opts_set.
635         Change ix86_tune_string to access opts->x_ix86_tune_string.
636         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
637         Change ix86_arch_string to access opts->x_ix86_arch_string.
638         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
639         Change ix86_pmode to access opts->x_ix86_pmode.
640         Change ix86_abi to access opts->x_ix86_abi.
641         Change ix86_cmodel to access opts->x_ix86_cmodel.
642         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
643         Change ix86_isa_flags_explicit to access
644         opts->x_ix86_isa_flags_explicit.
645         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
646         Change ix86_regparm to access opts->x_ix86_regparm.
647         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
648         Change ix86_preferred_stack_boundary_arg to access
649         opts->x_ix86_preferred_stack_boundary_arg.
650         Change ix86_force_align_arg_pointer to access
651         opts->x_ix86_force_align_arg_pointer.
652         Change ix86_incoming_stack_boundar_arg to access
653         opts->x_ix86_incoming_stack_boundar_arg.
654         Change ix86_fpmath to access opts->x_ix86_fpmath.
655         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
656         Change ix86_recip_name to access opts->x_ix86_recip_name.
657         Change ix86_stack_protector_guard to access
658         opts->x_ix86_stack_protector_guard.
659         Change ix86_tune_memcpy_strategy to access
660         opts->x_ix86_tune_memcpy_strategy.
661         Change ix86_tune_memset_strategy to access
662         opts->x_ix86_tune_memset_strategy.
663         Change global_options to access opts.
664         Change global_options_set to access opts_set.
665         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
666         Change TARGET_MMX to TARGET_MMX_P (opts->...).
667         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
668         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
669         Change TARGET_SSE to TARGET_SSE_P (opts->...).
670         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
671         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
672         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
673         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
674         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
675         Change TARGET_AVX to TARGET_AVX_P (opts->...).
676         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
677         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
678         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
679         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
680         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
681         Change TARGET_FMA to TARGET_FMA_P (opts->...).
682         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
683         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
684         Change TARGET_XOP to TARGET_XOP_P (opts->...).
685         Change TARGET_LWP to TARGET_LWP_P (opts->...).
686         Change TARGET_ABM to TARGET_ABM_P (opts->...).
687         Change TARGET_BMI to TARGET_BMI_P (opts->...).
688         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
689         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
690         Change TARGET_TBM to TARGET_TBM_P (opts->...).
691         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
692         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
693         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
694         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
695         Change TARGET_AES to TARGET_AES_P (opts->...).
696         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
697         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
698         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
699         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
700         Change TARGET_F16C to TARGET_F16C_P (opts->...).
701         Change TARGET_RTM to TARGET_RTM_P (opts->...).
702         Change TARGET_HLE to TARGET_HLE_P (opts->...).
703         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
704         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
705         Change TARGET_ADX to TARGET_ADX_P (opts->...).
706         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
707         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
708         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
709         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
710         Change TARGET_X32 to TARGET_X32_P (opts->...).
711         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
712         Change TARGET_FLOAT_RETURNS_IN_80387 to
713         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
714         (ix86_function_specific_save): New parameter. Use opts-> fields
715         to replace global fields.
716         (ix86_function_specific_restore): Ditto.
717         (ix86_valid_target_attribute_inner_p): New parameters.
718         Fix recursive call.
719         Fix call to ix86_handle_option and set_option.
720         (ix86_valid_target_attribute_tree): New parameters.
721         Change global_options to access opts.
722         Change global_options_set to access opts_set.
723         Fix call to ix86_valid_target_attribute_inner_p.
724         Change ix86_tune_string to access opts->x_ix86_tune_string.
725         Change ix86_arch_string to access opts->x_ix86_arch_string.
726         Change ix86_fpmath to access opts->x_ix86_fpmath
727         Fix call to ix86_option_override_internal.
728         Fix call to ix86_add_new_builtins.
729         Fix calls to build_optimization_node and build_target_option_node.
730         (ix86_valid_target_attribute_p): Remove access to global_options.
731         Use new gcc_options structure func_options.
732         Fix call to ix86_valid_target_attribute_tree.
733         Fix call to  build_optimization_node.
734         (get_builtin_code_for_version): Fix call to
735         ix86_valid_target_attribute_tree.
737 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
739         * Makefile.in (PICFLAG): New.
740         (enable_host_shared): New.
741         (INTERNAL_CFLAGS): Use PICFLAG.
742         (LIBIBERTY): Use pic build of libiberty.a if configured with
743         --enable-host-shared.
744         * configure.ac: Add --enable-host-shared, setting up new
745         PICFLAG variable.
746         * configure: Regenerate.
747         * doc/install.texi (--enable-shared): Add note contrasting it with ...
748         (--enable-host-shared): New option.
750 2013-10-15  Richard Biener  <rguenther@suse.de>
752         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
753         for built-in functions.
755 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
757         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
758         (optimize_range_tests_1): New function,
759         extracted from optimize_range_tests.
760         (optimize_range_tests_xor): Similarly.
761         (optimize_range_tests_diff): New function.
762         (optimize_range_tests): Use optimize_range_tests_1.
764 2013-10-15  Cong Hou  <congh@google.com>
766         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
767         requirement of the reduction pattern so that one operand of the
768         reduction operation can come from outside of the loop.
770 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
772         * config/arm/neon-schedgen.ml: Remove.
773         * config/arm/cortex-a9-neon.md: Remove comment regarding
774         neon-schedgen.ml.
776 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
778         * config/arm/types: Remove old neon types.
780 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
782         * config/arm/cortex-a7.md
783         (cortex_a7_neon_type): New.
784         (cortex_a7_neon_mul): Update for new types.
785         (cortex_a7_neon_mla): Likewise.
786         (cortex_a7_neon): Likewise.
788 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
790         * config/arm/cortex-a15-neon.md
791         (cortex_a15_neon_type): New,
793         (cortex_a15_neon_int_1): Remove.
794         (cortex_a15_neon_int_2): Likewise.
795         (cortex_a15_neon_int_3): Likewise.
796         (cortex_a15_neon_int_4): Likewise.
797         (cortex_a15_neon_int_5): Likewise.
798         (cortex_a15_neon_vqneg_vqabs): Likewise.
799         (cortex_a15_neon_vmov): Likewise.
800         (cortex_a15_neon_vaba): Likewise.
801         (cortex_a15_neon_vaba_qqq): Likewise.
802         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
803         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
804         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
805         Likewise.
806         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
807         (cortex_a15_neon_mla_qqq_8_16): Likewise.
808         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
809         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
810         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
811         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
812         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
813         (cortex_a15_neon_shift_1): Likewise.
814         (cortex_a15_neon_shift_2): Likewise.
815         (cortex_a15_neon_shift_3): Likewise.
816         (cortex_a15_neon_vshl_ddd): Likewise.
817         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
818         (cortex_a15_neon_vsra_vrsra): Likewise.
819         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
820         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
821         (cortex_a15_neon_bp_3cycle): Likewise.
822         (cortex_a15_neon_ldm_2): Likewise.
823         (cortex_a15_neon_stm_2): Likewise.
824         (cortex_a15_neon_mcr): Likewise.
825         (cortex_a15_neon_mrc): Likewise.
826         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
827         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
828         (cortex_a15_neon_fp_vmul_ddd): Likewise.
829         (cortex_a15_neon_fp_vmul_qqd): Likewise.
830         (cortex_a15_neon_fp_vmla_ddd): Likewise.
831         (cortex_a15_neon_fp_vmla_qqq): Likewise.
832         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
833         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
834         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
835         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
836         (cortex_a15_neon_bp_simple): Likewise.
837         (cortex_a15_neon_bp_2cycle): Likewise.
838         (cortex_a15_neon_bp_3cycle): Likewise.
839         (cortex_a15_neon_vld1_1_2_regs): Likewise.
840         (cortex_a15_neon_vld1_3_4_regs): Likewise.
841         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
842         (cortex_a15_neon_vld2_4_regs): Likewise.
843         (cortex_a15_neon_vld3_vld4): Likewise.
844         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
845         (cortex_a15_neon_vst1_3_4_regs): Likewise.
846         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
847         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
848         (cortex_a15_neon_vst3_vst4): Rename to...
849         (cortex_a15_neon_vst4): This, update for new attributes.
850         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
851         (cortex_a15_neon_vld3_vld4_lane): Likewise.
852         (cortex_a15_neon_vst1_vst2_lane): Likewise.
853         (cortex_a15_neon_vst3_vst4_lane): Likewise.
854         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
855         (cortex_a15_neon_ldm_2): Likewise.
856         (cortex_a15_neon_stm_2): Likewise.
857         (cortex_a15_neon_mcr): Likewise.
858         (cortex_a15_neon_mcr_2_mcrr): Likewise.
859         (cortex_a15_neon_mrc): Likewise.
860         (cortex_a15_neon_mrrc): Likewise.
862         (cortex_a15_neon_abd): New.
863         (cortex_a15_neon_abd_q): Likewise.
864         (cortex_a15_neon_aba): Likewise.
865         (cortex_a15_neon_aba_q): Likewise.
866         (cortex_a15_neon_acc): Likewise.
867         (cortex_a15_neon_acc_q): Likewise.
868         (cortex_a15_neon_arith_basic): Likewise.
869         (cortex_a15_neon_arith_complex): Likewise.
870         (cortex_a15_neon_multiply): Likewise.
871         (cortex_a15_neon_multiply_q): Likewise.
872         (cortex_a15_neon_mla): Likewise.
873         (cortex_a15_neon_mla_q): Likewise.
874         (cortex_a15_neon_sat_mla_long): Likewise.
875         (cortex_a15_neon_shift_acc): Likewise.
876         (cortex_a15_neon_shift_imm_basic): Likewise.
877         (cortex_a15_neon_shift_imm_complex): Likewise.
878         (cortex_a15_neon_shift_reg_basic): Likewise.
879         (cortex_a15_neon_shift_reg_basic_q): Likewise.
880         (cortex_a15_neon_shift_reg_complex): Likewise.
881         (cortex_a15_neon_shift_reg_complex_q): Likewise.
882         (cortex_a15_neon_fp_negabs): Likewise
883         (cortex_a15_neon_fp_arith): Likewise
884         (cortex_a15_neon_fp_arith_q): Likewise
885         (cortex_a15_neon_fp_cvt_int): Likewise
886         (cortex_a15_neon_fp_cvt_int_q): Likewise
887         (cortex_a15_neon_fp_cvt_16): Likewise
888         (cortex_a15_neon_fp_mul): Likewise
889         (cortex_a15_neon_fp_mul_q): Likewise
890         (cortex_a15_neon_fp_mla): Likewise
891         (cortex_a15_neon_fp_mla_q): Likewise
892         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
893         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
894         (cortex_a15_neon_bitops): Likewise.
895         (cortex_a15_neon_bitops_q): Likewise.
896         (cortex_a15_neon_from_gp): Likewise.
897         (cortex_a15_neon_from_gp_q): Likewise.
898         (cortex_a15_neon_tbl3_tbl4): Likewise.
899         (cortex_a15_neon_zip_q): Likewise.
900         (cortex_a15_neon_to_gp): Likewise.
901         (cortex_a15_neon_load_a): Likewise.
902         (cortex_a15_neon_load_b): Likewise.
903         (cortex_a15_neon_load_c): Likewise.
904         (cortex_a15_neon_load_d): Likewise.
905         (cortex_a15_neon_load_e): Likewise.
906         (cortex_a15_neon_load_f): Likewise.
907         (cortex_a15_neon_store_a): Likewise.
908         (cortex_a15_neon_store_b): Likewise.
909         (cortex_a15_neon_store_c): Likewise.
910         (cortex_a15_neon_store_d): Likewise.
911         (cortex_a15_neon_store_e): Likewise.
912         (cortex_a15_neon_store_f): Likewise.
913         (cortex_a15_neon_store_g): Likewise.
914         (cortex_a15_neon_store_h): Likewise.
915         (cortex_a15_vfp_to_from_gp): Likewise.
917 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
919         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
921         (cortex_a9_neon_vshl_ddd): Remove.
922         (cortex_a9_neon_vst3_vst4): Likewise.
923         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
925         (cortex_a9_neon_bit_ops_q): New.
927         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
928         (cortex_a9_neon_int_2): Likewise.
929         (cortex_a9_neon_int_3): Likewise.
930         (cortex_a9_neon_int_4): Likewise.
931         (cortex_a9_neon_int_5): Likewise.
932         (cortex_a9_neon_vqneg_vqabs): Likewise.
933         (cortex_a9_neon_vmov): Likewise.
934         (cortex_a9_neon_vaba): Likewise.
935         (cortex_a9_neon_vaba_qqq): Likewise.
936         (cortex_a9_neon_shift_1): Likewise.
937         (cortex_a9_neon_shift_2): Likewise.
938         (cortex_a9_neon_shift_3): Likewise.
939         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
940         (cortex_a9_neon_vsra_vrsra): Likewise.
941         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
942         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
943         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
944         Likewise.
945         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
946         (cortex_a9_neon_mla_qqq_8_16): Likewise.
947         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
948         Likewise.
949         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
950         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
951         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
952         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
953         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
954         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
955         (cortex_a9_neon_fp_vsum): Likewise.
956         (cortex_a9_neon_fp_vmul_ddd): Likewise.
957         (cortex_a9_neon_fp_vmul_qqd): Likewise.
958         (cortex_a9_neon_fp_vmla_ddd): Likewise.
959         (cortex_a9_neon_fp_vmla_qqq): Likewise.
960         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
961         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
962         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
963         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
964         (cortex_a9_neon_bp_simple): Likewise.
965         (cortex_a9_neon_bp_2cycle): Likewise.
966         (cortex_a9_neon_bp_3cycle): Likewise.
967         (cortex_a9_neon_ldr): Likewise.
968         (cortex_a9_neon_str): Likewise.
969         (cortex_a9_neon_vld1_1_2_regs): Likewise.
970         (cortex_a9_neon_vld1_3_4_regs): Likewise.
971         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
972         (cortex_a9_neon_vld2_4_regs): Likewise.
973         (cortex_a9_neon_vld3_vld4): Likewise.
974         (cortex_a9_neon_vld1_vld2_lane): Likewise.
975         (cortex_a9_neon_vld3_vld4_lane): Likewise.
976         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
977         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
978         (cortex_a9_neon_vst1_3_4_regs): Likewise.
979         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
980         (cortex_a9_neon_vst1_vst2_lane): Likewise.
981         (cortex_a9_neon_vst3_vst4_lane): Likewise.
982         (cortex_a9_neon_mcr): Likewise.
983         (cortex_a9_neon_mcr_2_mcrr): Likewise.
984         (cortex_a9_neon_mrc): Likewise.
985         (cortex_a9_neon_mrrc): Likewise.
987 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
989         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
991         (cortex_a8_neon_vshl_ddd): Remove.
992         (cortex_a8_neon_vst3_vst4): Likewise.
993         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
995         (cortex_a8_neon_bit_ops_q): New.
997         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
998         (cortex_a8_neon_int_2): Likewise..
999         (cortex_a8_neon_int_3): Likewise.
1000         (cortex_a8_neon_int_5): Likewise.
1001         (cortex_a8_neon_vqneg_vqabs): Likewise.
1002         (cortex_a8_neon_int_4): Likewise.
1003         (cortex_a8_neon_vaba): Likewise.
1004         (cortex_a8_neon_vaba_qqq): Likewise.
1005         (cortex_a8_neon_shift_1): Likewise.
1006         (cortex_a8_neon_shift_2): Likewise.
1007         (cortex_a8_neon_shift_3): Likewise.
1008         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1009         (cortex_a8_neon_vsra_vrsra): Likewise.
1010         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1011         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1012         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1013         Likewise.
1014         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1015         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1016         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1017         Likewise.
1018         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1019         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1020         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1021         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1022         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1023         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1024         (cortex_a8_neon_fp_vsum): Likewise.
1025         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1026         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1027         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1028         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1029         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1030         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1031         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1032         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1033         (cortex_a8_neon_bp_simple): Likewise.
1034         (cortex_a8_neon_bp_2cycle): Likewise.
1035         (cortex_a8_neon_bp_3cycle): Likewise.
1036         (cortex_a8_neon_ldr): Likewise.
1037         (cortex_a8_neon_str): Likewise.
1038         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1039         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1040         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1041         (cortex_a8_neon_vld2_4_regs): Likewise.
1042         (cortex_a8_neon_vld3_vld4): Likewise.
1043         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1044         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1045         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1046         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1047         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1048         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1049         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1050         (cortex_a8_neon_mcr): Likewise.
1051         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1052         (cortex_a8_neon_mrc): Likewise.
1053         (cortex_a8_neon_mrrc): Likewise.
1055 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1057         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1058         (fp): New.
1059         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1060         (aarch64_simd_dup<mode>): Add "type" attribute.
1061         (aarch64_dup_lane<mode>): Likewise.
1062         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1063         (*aarch64_simd_mov<mode>): Likewise.
1064         (aarch64_simd_mov_from_<mode>low): Likewise.
1065         (aarch64_simd_mov_from_<mode>high): Likewise.
1066         (orn<mode>3): Likewise.
1067         (bic<mode>3): Likewise.
1068         (add<mode>3): Likewise.
1069         (sub<mode>3): Likewise.
1070         (mul<mode>3): Likewise.
1071         (*aarch64_mul3_elt<mode>): Likewise.
1072         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1073         (*aarch64_mul3_elt_to_128df): Likewise.
1074         (*aarch64_mul3_elt_to_64v2df): Likewise.
1075         (neg<mode>2): Likewise.
1076         (abs<mode>2): Likewise.
1077         (abd<mode>_3): Likewise.
1078         (aba<mode>_3): Likewise.
1079         (fabd<mode>_3): Likewise.
1080         (*fabd_scalar<mode>3): Likewise.
1081         (and<mode>3): Likewise.
1082         (ior<mode>3): Likewise.
1083         (xor<mode>3): Likewise.
1084         (one_cmpl<mode>2): Likewise.
1085         (aarch64_simd_vec_set<mode>): Likewise.
1086         (aarch64_simd_lshr<mode>): Likewise.
1087         (aarch64_simd_ashr<mode>): Likewise.
1088         (aarch64_simd_imm_shl<mode>): Likewise.
1089         (aarch64_simd_reg_sshl<mode): Likewise.
1090         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1091         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1092         (aarch64_simd_vec_setv2di): Likewise.
1093         (aarch64_simd_vec_set<mode>): Likewise.
1094         (aarch64_mla<mode>): Likewise.
1095         (*aarch64_mla_elt<mode>): Likewise.
1096         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1097         (aarch64_mls<mode>): Likewise.
1098         (*aarch64_mls_elt<mode>): Likewise.
1099         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1100         (<su><maxmin><mode>3): Likewise.
1101         (move_lo_quad_<mode>): Likewise.
1102         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1103         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1104         (vec_pack_trunc_<mode>): Likewise.
1105         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1106         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1107         (*aarch64_<su>mlal_lo<mode>): Likewise.
1108         (*aarch64_<su>mlal_hi<mode>): Likewise.
1109         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1110         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1111         (*aarch64_<su>mlal<mode>): Likewise.
1112         (*aarch64_<su>mlsl<mode>): Likewise.
1113         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1114         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1115         (add<mode>3): Likewise.
1116         (sub<mode>3): Likewise.
1117         (mul<mode>3): Likewise.
1118         (div<mode>3): Likewise.
1119         (neg<mode>2): Likewise.
1120         (abs<mode>2): Likewise.
1121         (fma<mode>4): Likewise.
1122         (*aarch64_fma4_elt<mode>): Likewise.
1123         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1124         (*aarch64_fma4_elt_to_128df): Likewise.
1125         (*aarch64_fma4_elt_to_64v2df): Likewise.
1126         (fnma<mode>4): Likewise.
1127         (*aarch64_fnma4_elt<mode>): Likewise.
1128         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1129         (*aarch64_fnma4_elt_to_128df): Likewise.
1130         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1131         (<frint_pattern><mode>2): Likewise.
1132         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1133         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1134         (vec_unpacks_lo_v4sf): Likewise.
1135         (aarch64_float_extend_lo_v2df): Likewise.
1136         (vec_unpacks_hi_v4sf): Likewise.
1137         (aarch64_float_truncate_lo_v2sf): Likewise.
1138         (aarch64_float_truncate_hi_v4sf): Likewise.
1139         (aarch64_vmls<mode>): Likewise.
1140         (<su><maxmin><mode>3): Likewise.
1141         (<maxmin_uns><mode>3): Likewise.
1142         (reduc_<sur>plus_<mode>): Likewise.
1143         (reduc_<sur>plus_v2di): Likewise.
1144         (reduc_<sur>plus_v2si): Likewise.
1145         (reduc_<sur>plus_<mode>): Likewise.
1146         (aarch64_addpv4sf): Likewise.
1147         (clz<mode>2): Likewise.
1148         (reduc_<maxmin_uns>_<mode>): Likewise.
1149         (reduc_<maxmin_uns>_v2di): Likewise.
1150         (reduc_<maxmin_uns>_v2si): Likewise.
1151         (reduc_<maxmin_uns>_<mode>): Likewise.
1152         (reduc_<maxmin_uns>_v4sf): Likewise.
1153         (aarch64_simd_bsl<mode>_internal): Likewise.
1154         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1155         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1156         (aarch64_get_lane<mode>): Likewise.
1157         (*aarch64_combinez<mode>): Likewise.
1158         (aarch64_combine<mode>): Likewise.
1159         (aarch64_simd_combine<mode>): Likewise.
1160         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1161         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1162         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1163         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1164         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1165         (aarch64_<sur>h<addsub><mode>): Likewise.
1166         (aarch64_<sur><addsub>hn<mode>): Likewise.
1167         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1168         (aarch64_pmul<mode>): Likewise.
1169         (aarch64_<su_optab><optab><mode>): Likewise.
1170         (aarch64_<sur>qadd<mode>): Likewise.
1171         (aarch64_sqmovun<mode>): Likewise.
1172         (aarch64_<sur>qmovn<mode>): Likewise.
1173         (aarch64_s<optab><mode>): Likewise.
1174         (aarch64_sq<r>dmulh<mode>): Likewise.
1175         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1176         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1177         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1178         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1179         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1180         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1181         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1182         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1183         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1184         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1185         (aarch64_sqdmull<mode>): Likewise.
1186         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1187         (aarch64_sqdmull_n<mode>): Likewise.
1188         (aarch64_sqdmull2<mode>_internal): Likewise.
1189         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1190         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1191         (aarch64_<sur>shl<mode>): Likewise.
1192         (aarch64_<sur>q<r>shl<mode>
1193         (aarch64_<sur>shll_n<mode>): Likewise.
1194         (aarch64_<sur>shll2_n<mode>): Likewise.
1195         (aarch64_<sur>shr_n<mode>): Likewise.
1196         (aarch64_<sur>sra_n<mode>): Likewise.
1197         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1198         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1199         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1200         (aarch64_cm<optab><mode>): Likewise.
1201         (aarch64_cm<optab>di): Likewise.
1202         (aarch64_cm<optab><mode>): Likewise.
1203         (aarch64_cm<optab>di): Likewise.
1204         (aarch64_cmtst<mode>): Likewise.
1205         (aarch64_cmtstdi): Likewise.
1206         (aarch64_cm<optab><mode>): Likewise.
1207         (*aarch64_fac<optab><mode>): Likewise.
1208         (aarch64_addp<mode>): Likewise.
1209         (aarch64_addpdi): Likewise.
1210         (sqrt<mode>2): Likewise.
1211         (vec_load_lanesoi<mode>): Likewise.
1212         (vec_store_lanesoi<mode>): Likewise.
1213         (vec_load_lanesci<mode>): Likewise.
1214         (vec_store_lanesci<mode>): Likewise.
1215         (vec_load_lanesxi<mode>): Likewise.
1216         (vec_store_lanesxi<mode>): Likewise.
1217         (*aarch64_mov<mode>): Likewise.
1218         (aarch64_ld2<mode>_dreg): Likewise.
1219         (aarch64_ld2<mode>_dreg): Likewise.
1220         (aarch64_ld3<mode>_dreg): Likewise.
1221         (aarch64_ld3<mode>_dreg): Likewise.
1222         (aarch64_ld4<mode>_dreg): Likewise.
1223         (aarch64_ld4<mode>_dreg): Likewise.
1224         (aarch64_tbl1<mode>): Likewise.
1225         (aarch64_tbl2v16qi): Likewise.
1226         (aarch64_combinev16qi): Likewise.
1227         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1228         (aarch64_st2<mode>_dreg): Likewise.
1229         (aarch64_st2<mode>_dreg): Likewise.
1230         (aarch64_st3<mode>_dreg): Likewise.
1231         (aarch64_st3<mode>_dreg): Likewise.
1232         (aarch64_st4<mode>_dreg): Likewise.
1233         (aarch64_st4<mode>_dreg): Likewise.
1234         (*aarch64_simd_ld1r<mode>): Likewise.
1235         (aarch64_frecpe<mode>): Likewise.
1236         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1237         (aarch64_frecps<mode>): Likewise.
1239 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1241         * config/arm/iterators.md (V_elem_ch): New.
1242         (q): Likewise.
1243         (VQH_type): Likewise.
1244         * config/arm/arm.md (is_neon_type): New.
1245         (conds): Use is_neon_type.
1246         (anddi3_insn): Update type attribute.
1247         (xordi3_insn): Likewise.
1248         (one_cmpldi2): Likewise.
1249         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1250         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1251         (*movmisalign<mode>_neon_store): Likewise.
1252         (*movmisalign<mode>_neon_load): Likewise.
1253         (vec_set<mode>_internal): Likewise.
1254         (vec_set<mode>_internal): Likewise.
1255         (vec_setv2di_internal): Likewise.
1256         (vec_extract<mode>): Likewise.
1257         (vec_extract<mode>): Likewise.
1258         (vec_extractv2di): Likewise.
1259         (*add<mode>3_neon): Likewise.
1260         (adddi3_neon): Likewise.
1261         (*sub<mode>3_neon): Likewise.
1262         (subdi3_neon): Likewise.
1263         (fma<VCVTF:mode>4): Likewise.
1264         (fma<VCVTF:mode>4_intrinsic): Likewise.
1265         (*fmsub<VCVTF:mode>4): Likewise.
1266         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1267         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1268         (ior<mode>3): Likewise.
1269         (and<mode>3): Likewise.
1270         (orn<mode>3_neon): Likewise.
1271         (orndi3_neon): Likewise.
1272         (bic<mode>3_neon): Likewise.
1273         (bicdi3_neon): Likewise.
1274         (xor<mode>3): Likewise.
1275         (one_cmpl<mode>2): Likewise.
1276         (abs<mode>2): Likewise.
1277         (neg<mode>2): Likewise.
1278         (negdi2_neon): Likewise.
1279         (*umin<mode>3_neon): Likewise.
1280         (*umax<mode>3_neon): Likewise.
1281         (*smin<mode>3_neon): Likewise.
1282         (*smax<mode>3_neon): Likewise.
1283         (vashl<mode>3): Likewise.
1284         (vashr<mode>3_imm): Likewise.
1285         (vlshr<mode>3_imm): Likewise.
1286         (ashl<mode>3_signed): Likewise.
1287         (ashl<mode>3_unsigned): Likewise.
1288         (neon_load_count): Likewise.
1289         (ashldi3_neon_noclobber): Likewise.
1290         (ashldi3_neon): Likewise.
1291         (signed_shift_di3_neon): Likewise.
1292         (unsigned_shift_di3_neon): Likewise.
1293         (ashrdi3_neon_imm_noclobber): Likewise.
1294         (lshrdi3_neon_imm_noclobber): Likewise.
1295         (<shift>di3_neon): Likewise.
1296         (widen_ssum<mode>3): Likewise.
1297         (widen_usum<mode>3): Likewise.
1298         (quad_halves_<code>v4si): Likewise.
1299         (quad_halves_<code>v4sf): Likewise.
1300         (quad_halves_<code>v8hi): Likewise.
1301         (quad_halves_<code>v16qi): Likewise.
1302         (reduc_splus_v2di): Likewise.
1303         (neon_vpadd_internal<mode>): Likewise.
1304         (neon_vpsmin<mode>): Likewise.
1305         (neon_vpsmax<mode>): Likewise.
1306         (neon_vpumin<mode>): Likewise.
1307         (neon_vpumax<mode>): Likewise.
1308         (*ss_add<mode>_neon): Likewise.
1309         (*us_add<mode>_neon): Likewise.
1310         (*ss_sub<mode>_neon): Likewise.
1311         (*us_sub<mode>_neon): Likewise.
1312         (neon_vadd<mode>_unspec): Likewise.
1313         (neon_vaddl<mode>): Likewise.
1314         (neon_vaddw<mode>): Likewise.
1315         (neon_vhadd<mode>): Likewise.
1316         (neon_vqadd<mode>): Likewise.
1317         (neon_vaddhn<mode>): Likewise.
1318         (neon_vmul<mode>): Likewise.
1319         (neon_vfms<VCVTF:mode>): Likewise.
1320         (neon_vmlal<mode>): Likewise.
1321         (neon_vmls<mode>): Likewise.
1322         (neon_vmlsl<mode>): Likewise.
1323         (neon_vqdmulh<mode>): Likewise.
1324         (neon_vqdmlal<mode>): Likewise.
1325         (neon_vqdmlsl<mode>): Likewise.
1326         (neon_vmull<mode>): Likewise.
1327         (neon_vqdmull<mode>): Likewise.
1328         (neon_vsub<mode>_unspec): Likewise.
1329         (neon_vsubl<mode>): Likewise.
1330         (neon_vsubw<mode>): Likewise.
1331         (neon_vqsub<mode>): Likewise.
1332         (neon_vhsub<mode>): Likewise.
1333         (neon_vsubhn<mode>): Likewise.
1334         (neon_vceq<mode>): Likewise.
1335         (neon_vcge<mode>): Likewise.
1336         (neon_vcgeu<mode>): Likewise.
1337         (neon_vcgt<mode>): Likewise.
1338         (neon_vcgtu<mode>): Likewise.
1339         (neon_vcle<mode>): Likewise.
1340         (neon_vclt<mode>): Likewise.
1341         (neon_vcage<mode>): Likewise.
1342         (neon_vcagt<mode>): Likewise.
1343         (neon_vtst<mode>): Likewise.
1344         (neon_vabd<mode>): Likewise.
1345         (neon_vabdl<mode>): Likewise.
1346         (neon_vaba<mode>): Likewise.
1347         (neon_vabal<mode>): Likewise.
1348         (neon_vmax<mode>): Likewise.
1349         (neon_vmin<mode>): Likewise.
1350         (neon_vpaddl<mode>): Likewise.
1351         (neon_vpadal<mode>): Likewise.
1352         (neon_vpmax<mode>): Likewise.
1353         (neon_vpmin<mode>): Likewise.
1354         (neon_vrecps<mode>): Likewise.
1355         (neon_vrsqrts<mode>): Likewise.
1356         (neon_vqabs<mode>): Likewise.
1357         (neon_vqneg<mode>): Likewise.
1358         (neon_vcls<mode>): Likewise.
1359         (clz<mode>2): Likewise.
1360         (popcount<mode>2): Likewise.
1361         (neon_vrecpe<mode>): Likewise.
1362         (neon_vrsqrte<mode>): Likewise.
1363         (neon_vget_lane<mode>_sext_internal): Likewise.
1364         (neon_vget_lane<mode>_zext_internal): Likewise.
1365         (neon_vdup_n<mode>): Likewise.
1366         (neon_vdup_n<mode>): Likewise.
1367         (neon_vdup_nv2di): Likewise.
1368         (neon_vdup_lane<mode>_interal): Likewise.
1369         (*neon_vswp<mode>): Likewise.
1370         (neon_vcombine<mode>): Likewise.
1371         (float<mode><V_cvtto>2): Likewise.
1372         (floatuns<mode><V_cvtto>2): Likewise.
1373         (fix_trunc<mode><V_cvtto>2): Likewise.
1374         (fixuns_trunc<mode><V_cvtto>2
1375         (neon_vcvt<mode>): Likewise.
1376         (neon_vcvt<mode>): Likewise.
1377         (neon_vcvtv4sfv4hf): Likewise.
1378         (neon_vcvtv4hfv4sf): Likewise.
1379         (neon_vcvt_n<mode>): Likewise.
1380         (neon_vcvt_n<mode>): Likewise.
1381         (neon_vmovn<mode>): Likewise.
1382         (neon_vqmovn<mode>): Likewise.
1383         (neon_vqmovun<mode>): Likewise.
1384         (neon_vmovl<mode>): Likewise.
1385         (neon_vmul_lane<mode>): Likewise.
1386         (neon_vmul_lane<mode>): Likewise.
1387         (neon_vmull_lane<mode>): Likewise.
1388         (neon_vqdmull_lane<mode>): Likewise.
1389         (neon_vqdmulh_lane<mode>): Likewise.
1390         (neon_vqdmulh_lane<mode>): Likewise.
1391         (neon_vmla_lane<mode>): Likewise.
1392         (neon_vmla_lane<mode>): Likewise.
1393         (neon_vmlal_lane<mode>): Likewise.
1394         (neon_vqdmlal_lane<mode>): Likewise.
1395         (neon_vmls_lane<mode>): Likewise.
1396         (neon_vmls_lane<mode>): Likewise.
1397         (neon_vmlsl_lane<mode>): Likewise.
1398         (neon_vqdmlsl_lane<mode>): Likewise.
1399         (neon_vext<mode>): Likewise.
1400         (neon_vrev64<mode>): Likewise.
1401         (neon_vrev32<mode>): Likewise.
1402         (neon_vrev16<mode>): Likewise.
1403         (neon_vbsl<mode>_internal): Likewise.
1404         (neon_vshl<mode>): Likewise.
1405         (neon_vqshl<mode>): Likewise.
1406         (neon_vshr_n<mode>): Likewise.
1407         (neon_vshrn_n<mode>): Likewise.
1408         (neon_vqshrn_n<mode>): Likewise.
1409         (neon_vqshrun_n<mode>): Likewise.
1410         (neon_vshl_n<mode>): Likewise.
1411         (neon_vqshl_n<mode>): Likewise.
1412         (neon_vqshlu_n<mode>): Likewise.
1413         (neon_vshll_n<mode>): Likewise.
1414         (neon_vsra_n<mode>): Likewise.
1415         (neon_vsri_n<mode>): Likewise.
1416         (neon_vsli_n<mode>): Likewise.
1417         (neon_vtbl1v8qi): Likewise.
1418         (neon_vtbl2v8qi): Likewise.
1419         (neon_vtbl3v8qi): Likewise.
1420         (neon_vtbl4v8qi): Likewise.
1421         (neon_vtbl1v16qi): Likewise.
1422         (neon_vtbl2v16qi): Likewise.
1423         (neon_vcombinev16qi): Likewise.
1424         (neon_vtbx1v8qi): Likewise.
1425         (neon_vtbx2v8qi): Likewise.
1426         (neon_vtbx3v8qi): Likewise.
1427         (neon_vtbx4v8qi): Likewise.
1428         (*neon_vtrn<mode>_insn): Likewise.
1429         (*neon_vzip<mode>_insn): Likewise.
1430         (*neon_vuzp<mode>_insn): Likewise.
1431         (neon_vld1<mode>): Likewise.
1432         (neon_vld1_lane<mode>): Likewise.
1433         (neon_vld1_lane<mode>): Likewise.
1434         (neon_vld1_dup<mode>): Likewise.
1435         (neon_vld1_dup<mode>): Likewise.
1436         (neon_vld1_dupv2di): Likewise.
1437         (neon_vst1<mode>): Likewise.
1438         (neon_vst1_lane<mode>): Likewise.
1439         (neon_vst1_lane<mode>): Likewise.
1440         (neon_vld2<mode>): Likewise.
1441         (neon_vld2<mode>): Likewise.
1442         (neon_vld2_lane<mode>): Likewise.
1443         (neon_vld2_lane<mode>): Likewise.
1444         (neon_vld2_dup<mode>): Likewise.
1445         (neon_vst2<mode>): Likewise.
1446         (neon_vst2<mode>): Likewise.
1447         (neon_vst2_lane<mode>): Likewise.
1448         (neon_vst2_lane<mode>): Likewise.
1449         (neon_vld3<mode>): Likewise.
1450         (neon_vld3qa<mode>): Likewise.
1451         (neon_vld3qb<mode>): Likewise.
1452         (neon_vld3_lane<mode>): Likewise.
1453         (neon_vld3_lane<mode>): Likewise.
1454         (neon_vld3_dup<mode>): Likewise.
1455         (neon_vst3<mode>): Likewise.
1456         (neon_vst3qa<mode>): Likewise.
1457         (neon_vst3qb<mode>): Likewise.
1458         (neon_vst3_lane<mode>): Likewise.
1459         (neon_vst3_lane<mode>): Likewise.
1460         (neon_vld4<mode>): Likewise.
1461         (neon_vld4qa<mode>): Likewise.
1462         (neon_vld4qb<mode>): Likewise.
1463         (neon_vld4_lane<mode>): Likewise.
1464         (neon_vld4_lane<mode>): Likewise.
1465         (neon_vld4_dup<mode>): Likewise.
1466         (neon_vst4<mode>): Likewise.
1467         (neon_vst4qa<mode>): Likewise.
1468         (neon_vst4qb<mode>): Likewise.
1469         (neon_vst4_lane<mode>): Likewise.
1470         (neon_vst4_lane<mode>): Likewise.
1471         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1472         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1473         (neon_vec_<US>mult_lo_<mode>): Likewise.
1474         (neon_vec_<US>mult_hi_<mode>): Likewise.
1475         (neon_vec_<US>shiftl_<mode>): Likewise.
1476         (neon_unpack<US>_<mode>): Likewise.
1477         (neon_vec_<US>mult_<mode>): Likewise.
1478         (vec_pack_trunc_<mode>): Likewise.
1479         (neon_vec_pack_trunc_<mode>): Likewise.
1480         (neon_vabd<mode>_2): Likewise.
1481         (neon_vabd<mode>_3): Likewise.
1483 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1485         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1486         (load_pair): Update type attribute.
1487         (store_pair): Update type attribute.
1488         * config/aarch64/iterators.md (q): New.
1490 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1492         * config/arm/types.md: Add new types for Neon insns.
1494 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1495             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1496             Sergey Lega  <sergey.s.lega@intel.com>
1497             Anna Tikhonova  <anna.tikhonova@intel.com>
1498             Ilya Tocar  <ilya.tocar@intel.com>
1499             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1500             Ilya Verbin  <ilya.verbin@intel.com>
1501             Kirill Yukhin  <kirill.yukhin@intel.com>
1502             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1504         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1505         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1506         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1507         (rcp14<mode>): New.
1508         (srcp14<mode>): Ditto.
1509         (rsqrt14<mode>): Ditto.
1510         (rsqrt14<mode>): Ditto.
1511         (avx512f_vmscalef<mode>): Ditto.
1512         (avx512f_scalef<mode>): Ditto.
1513         (avx512f_getexp<mode>): Ditto.
1514         (avx512f_sgetexp<mode>): Ditto.
1515         (avx512f_fixupimm<mode>): Ditto.
1516         (avx512f_sfixupimm<mode>): Ditto.
1517         (avx512f_rndscale<mode>): Ditto.
1518         (*avx512er_exp2<mode>): Ditto.
1519         (*avx512er_rcp28<mode>): Ditto.
1520         (avx512er_rsqrt28<mode>): Ditto.
1521         (avx512f_getmant<mode>): Ditto.
1522         (avx512f_getmant<mode>): Ditto.
1523         (avx512f_rndscale<mode>): Fix formatting.
1525 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1527         * ipa-utils.h (ipa_edge_within_scc): Declare.
1528         * ipa-cp.c (edge_within_scc): Moved...
1529         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1531 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1532             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1533             Sergey Lega  <sergey.s.lega@intel.com>
1534             Anna Tikhonova  <anna.tikhonova@intel.com>
1535             Ilya Tocar  <ilya.tocar@intel.com>
1536             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1537             Ilya Verbin  <ilya.verbin@intel.com>
1538             Kirill Yukhin  <kirill.yukhin@intel.com>
1539             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1541         * config/i386/predicates.md (const_8_to_15_operand): New.
1542         (const_16_to_31_operand): Ditto.
1543         * config/i386/sse.md (V8FI): New.
1544         (V16FI): Ditto.
1545         (reduc_splus_v8df): Ditto.
1546         (reduc_splus_v16sf): Ditto.
1547         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1548         (vec_extract_hi_<mode>): Ditto.
1549         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1550         (vec_set_lo_<mode>): Ditto.
1551         (vec_set_hi_<mode>): Ditto.
1552         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1553         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1554         (avx512f_pshufd_1): Ditto.
1555         (avx512f_broadcast<mode>): Ditto.
1556         (avx512f_broadcast<mode>): Ditto.
1557         (define_split): Split vec_extract_lo into move.
1558         (ssequartermode): Ditto.
1559         (ssedoublemode): Extened with wider modes.
1560         (vec_extract_lo_<mode>): Ditto.
1562 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1563             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1564             Sergey Lega  <sergey.s.lega@intel.com>
1565             Anna Tikhonova  <anna.tikhonova@intel.com>
1566             Ilya Tocar  <ilya.tocar@intel.com>
1567             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1568             Ilya Verbin  <ilya.verbin@intel.com>
1569             Kirill Yukhin  <kirill.yukhin@intel.com>
1570             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1572         * config/i386/predicates.md (register_or_constm1_operand): New.
1573         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1574         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1575         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1576         UNSPEC_SCATTER_PREFETCH
1577         (VI48_512): New.
1578         (avx512f_ucmp<mode>3): Ditto.
1579         (avx512f_vternlog<mode>): Ditto.
1580         (avx512f_align<mode>): Ditto.
1581         (<shift_insn><mode>3): Ditto.
1582         (avx512f_<rotate>v<mode>): Ditto.
1583         (avx512f_<rotate><mode>): Ditto.
1584         (avx512f_eq<mode>3): Ditto.
1585         (avx512f_eq<mode>3_1): Ditto.
1586         (avx512f_gt<mode>3): Ditto.
1587         (avx512f_testm<mode>3): Ditto.
1588         (avx512f_testnm<mode>3): Ditto.
1589         (avx512pf_gatherpf<mode>): Ditto.
1590         (*avx512pf_gatherpf<mode>_mask): Ditto.
1591         (*avx512pf_gatherpf<mode>): Ditto.
1592         (avx512pf_scatterpf<mode>): Ditto.
1593         (*avx512pf_scatterpf<mode>_mask): Ditto.
1594         (*avx512pf_scatterpf<mode>): Ditto.
1595         (avx512f_vec_dup_gpr<mode>): Ditto.
1596         (clz<mode>2): Ditto.
1597         (conflict<mode>): Ditto.
1598         (REDUC_SMINMAX_MODE): Extened with wider modes.
1599         (reduc_<code>_<mode>): Ditto.
1600         (vlshr<mode>3): Ditto.
1601         (vashl<mode>3): Ditto.
1603 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1604             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1605             Sergey Lega  <sergey.s.lega@intel.com>
1606             Anna Tikhonova  <anna.tikhonova@intel.com>
1607             Ilya Tocar  <ilya.tocar@intel.com>
1608             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1609             Ilya Verbin  <ilya.verbin@intel.com>
1610             Kirill Yukhin  <kirill.yukhin@intel.com>
1611             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1613         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1614         UNSPEC_SCATTER.
1615         (VI48F_512): New.
1616         (avx512fmaskmode): Ditto.
1617         (bcstscalarsuff): Ditto.
1618         (avx512f_blendm<mode>): Ditto.
1619         (cmp_imm_predicate): Ditto.
1620         (avx512f_cmp<mode>3): Ditto.
1621         (avx512f_vec_dup<mode>): Ditto.
1622         (avx512f_vec_dup_mem<mode>): Ditto.
1623         (avx512f_vpermi2var<mode>3): Ditto.
1624         (avx512f_vpermt2var<mode>3): Ditto.
1625         (vec_init<mode>): Ditto.
1626         (avx512f_gathersi<mode>): Ditto.
1627         (*avx512f_gathersi<mode>): Ditto.
1628         (*avx512f_gathersi<mode>_2): Ditto.
1629         (avx512f_gatherdi<mode>): Ditto.
1630         (*avx512f_gatherdi<mode>): Ditto.
1631         (*avx512f_gatherdi<mode>_2): Ditto.
1632         (avx512f_scattersi<mode>): Ditto.
1633         (*avx512f_scattersi<mode>): Ditto.
1634         (avx512f_scatterdi<mode>): Ditto.
1635         (*avx512f_scatterdi<mode>): Ditto.
1636         (sseintprefix): Extened with wider modes.
1637         (VEC_GATHER_IDXSI): Ditto.
1638         (VEC_GATHER_IDXDI): Ditto.
1639         (VEC_GATHER_SRCDI): Ditto.
1641 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1642             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1644         * config/arm/t-aprofile: New file.
1645         * config.gcc: Handle --with-multilib-list option.
1647 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1649         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1650         the input for r2.
1652 2013-10-15  Richard Biener  <rguenther@suse.de>
1654         * tree-loop-distribution.c (build_empty_rdg): Inline into
1655         single user.
1656         (rdg_flag_vertex): Inline into single user.
1657         (rdg_flag_vertex_and_dependent): Likewise.
1658         (build_rdg_partition_for_vertex): Remove processed bitmap.
1659         (rdg_build_partitions): Simplify.
1661 2013-10-15  Richard Biener  <rguenther@suse.de>
1663         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1664         Restructure forwarding through conversions and copies to
1665         avoid performing copy-propagation the wrong way.  Adjust
1666         recursion invocations.
1667         (forward_propagate_addr_expr): Add argument stating if we
1668         are recursing from a single-use.
1669         (ssa_forward_propagate_and_combine): Adjust.
1671 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1673         * dumpfile.h (gcc::dump_manager): New class, to hold state
1674         relating to dumpfile management.
1675         (get_dump_file_name): Remove in favor of method of dump_manager.
1676         (dump_initialized_p): Likewise.
1677         (dump_start): Likewise.
1678         (dump_finish): Likewise.
1679         (dump_switch_p): Likewise.
1680         (dump_register): Likewise.
1681         (get_dump_file_info): Likewise.
1682         * context.c (gcc::context::context): Construct the dump_manager
1683         instance.
1684         * context.h (gcc::context::get_dumps): New.
1685         (gcc::context::m_dumps): New.
1686         * coverage.c (coverage_init): Port to dump_manager API.
1687         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1688         (extra_dump_files_in_use): Likewise.
1689         (extra_dump_files_alloced): Likewise.
1690         (gcc::dump_manager::dump_manager): New.
1691         (dump_register): Convert to...
1692         (gcc::dump_manager::dump_register): ...method, replacing
1693         function-static next_dump with m_next_dump field.
1694         (get_dump_file_info): Convert to...
1695         (gcc::dump_manager::get_dump_file_info): ...method.
1696         (get_dump_file_name): Convert to...
1697         (gcc::dump_manager::get_dump_file_name): ...method.
1698         (dump_start): Convert to...
1699         (gcc::dump_manager::dump_start): ...method.
1700         (dump_finish): Convert to...
1701         (gcc::dump_manager::dump_finish): ...method.
1702         (dump_begin): Replace body with...
1703         (gcc::dump_manager::dump_begin): ...new method.
1704         (dump_phase_enabled_p): Convert to...
1705         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1706         (dump_phase_enabled_p): Convert to...
1707         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1708         (dump_initialized_p):  Convert to...
1709         (gcc::dump_manager::dump_initialized_p): ...method.
1710         (dump_flag_name): Replace body with...
1711         (gcc::dump_manager::dump_flag_name): ...new method.
1712         (dump_enable_all): Convert to...
1713         (gcc::dump_manager::dump_enable_all): ...new method.
1714         (opt_info_enable_passes): Convert to...
1715         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1716         (dump_switch_p_1): Convert to...
1717         (gcc::dump_manager::dump_switch_p_1): ...new method.
1718         (dump_switch_p):  Convert to...
1719         (gcc::dump_manager::dump_switch_p): ...new method.
1720         (opt_info_switch_p): Port to dump_manager API.
1721         (enable_rtl_dump_file): Likewise.
1722         * opts-global.c (handle_common_deferred_options): Port to new
1723         dump_manager API.
1724         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1725         (pass_manager::register_one_dump_file): Likewise.
1726         (pass_manager::register_pass): Likewise.
1727         (pass_init_dump_file): Likewise.
1728         (pass_fini_dump_file): Likewise.
1729         * statistics.c (statistics_early_init): Likewise.
1731 2013-10-14  Richard Biener  <rguenther@suse.de>
1733         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1734         iterative_hash_canonical_type, gimple_canonical_type_hash,
1735         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1736         gimple_register_canonical_type, print_gimple_types_stats,
1737         free_gimple_type_tables): Move to lto/lto.c
1738         (gt-gimple.h): Do not include.
1739         * gimple.h (gimple_register_canonical_type,
1740         print_gimple_types_stats, free_gimple_type_tables): Remove.
1741         * Makefile.in (GTFILES): Remove gimple.c.
1743 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1745         PR target/58716
1746         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1747         scanning for msp430x targets.
1749 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1751         PR bootstrap/58509
1752         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1753         (registers_ok_for_ldd_peep): Move around.
1754         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1755         * config/sparc/sparc.md (widening peepholes): Use it.
1757 2013-10-14  Richard Biener  <rguenther@suse.de>
1759         PR middle-end/58712
1760         PR middle-end/55358
1761         * gimple.c (iterative_hash_canonical_type): Make sure to
1762         record the hash into the correct hashtable slot.
1764 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1766         PR rtl-optimization/58662
1767         * combine.c (try_combine): Take into account death nodes on I2 when
1768         splitting a PARALLEL of two independent SETs.  Fix dump message.
1770 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1772         PR target/51244
1773         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1774         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1775         * config/sh/t-sh (sh_treg_combine.o): New entry.
1776         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1777         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1778         (register_sh_passes): New function.  Register sh_treg_combine pass.
1779         (sh_option_override): Invoke it.
1780         (sh_canonicalize_comparison): Handle op0_preserve_value.
1781         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1782         opportunities.  Canonicalize branch condition.
1783         (nott): Allow only if pseudos can be created for non-SH2A.
1785 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1787         PR target/58690
1788         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1789         (ix86_expand_movmem): Replace copy_addr_to_reg with
1790         ix86_copy_addr_to_reg.
1791         (ix86_expand_setmem): Likewise.
1793 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1795         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1796         provided by ix86_fp_compare_mode instead of CCFPUmode.
1798 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1800         * config/aarch64/arm_neon.h
1801         (vtbx<1,3>_<psu>8): Fix register constriants.
1803 2013-10-11  Jeff Law  <law@redhat.com>
1805         PR tree-optimization/58640
1806         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1807         threading paths that cross over two loop entry points.
1809 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1811         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1812         handle vector float as well.
1813         (*vsx_le_perm_load_v4si): Likewise.
1814         (*vsx_le_perm_store_v2di): Likewise.
1815         (*vsx_le_perm_store_v4si): Likewise.
1817 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1819         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1820         directly to circumvent subtract from splat{31} workaround.
1821         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1822         prototype.
1823         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1824         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1825         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1826         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1827         separate big and little endian logic.
1828         (*altivec_vperm_<mode>_internal): New define_insn.
1829         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1830         separate big and little endian logic.
1831         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1832         (vec_permv16qi): Add little endian logic.
1834 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1836         * doc/extend.texi (returns_nonnull): Remove arguments.
1838 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1839             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1840             Sergey Lega  <sergey.s.lega@intel.com>
1841             Anna Tikhonova  <anna.tikhonova@intel.com>
1842             Ilya Tocar  <ilya.tocar@intel.com>
1843             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1844             Ilya Verbin  <ilya.verbin@intel.com>
1845             Kirill Yukhin  <kirill.yukhin@intel.com>
1846             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1848         * config/i386/sse.md (VI48F_256_512): New.
1849         (avx2_permvar<mode>): Change to ...
1850         (<avx2_avx512f>_permvar<mode>): This.
1852 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1853             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1854             Sergey Lega  <sergey.s.lega@intel.com>
1855             Anna Tikhonova  <anna.tikhonova@intel.com>
1856             Ilya Tocar  <ilya.tocar@intel.com>
1857             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1858             Ilya Verbin  <ilya.verbin@intel.com>
1859             Kirill Yukhin  <kirill.yukhin@intel.com>
1860             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1862         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1863         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1864         * config/i386/sse.md (VI4_AVX): New.
1865         (sf2simodelower): Ditto.
1866         (sse2_cvtps2dq): Change to ...
1867         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1869 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1870             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1871             Sergey Lega  <sergey.s.lega@intel.com>
1872             Anna Tikhonova  <anna.tikhonova@intel.com>
1873             Ilya Tocar  <ilya.tocar@intel.com>
1874             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1875             Ilya Verbin  <ilya.verbin@intel.com>
1876             Kirill Yukhin  <kirill.yukhin@intel.com>
1877             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1879         * config/i386/sse.md (V_512): New.
1880         (VI_512): Ditto.
1881         (vcond<V_512:mode><VF_512:mode>): Ditto.
1882         (vcond<V_512:mode><VI_512:mode>): Ditto.
1883         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1885 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1886             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1887             Sergey Lega  <sergey.s.lega@intel.com>
1888             Anna Tikhonova  <anna.tikhonova@intel.com>
1889             Ilya Tocar  <ilya.tocar@intel.com>
1890             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1891             Ilya Verbin  <ilya.verbin@intel.com>
1892             Kirill Yukhin  <kirill.yukhin@intel.com>
1893             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1895         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1896         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1897         (FMAMODE): Ditto.
1898         (fma<mode>4): Removed condition.
1899         (fms<mode>4): Ditto.
1900         (fnma<mode>4): Ditto.
1901         (fnms<mode>4): Ditto.
1902         (fma4i_fmadd_<mode>): Ditto.
1903         (*fma_fmadd_<mode>): Ditto.
1904         (*fma_fmsub_<mode>): Ditto.
1905         (*fma_fnmadd_<mode>): Ditto.
1906         (*fma_fnmsub_<mode>): Ditto.
1907         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1908         (*fma_fmaddsub_<mode>): Ditto.
1909         (*fma_fmsubadd_<mode>): Ditto.
1910         (*fmai_fmadd_<mode>): Ditto.
1911         (*fmai_fmsub_<mode>): Ditto.
1912         (*fmai_fnmadd_<mode>): Ditto.
1913         (*fmai_fnmsub_<mode>): Ditto.
1915 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1916             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1917             Sergey Lega  <sergey.s.lega@intel.com>
1918             Anna Tikhonova  <anna.tikhonova@intel.com>
1919             Ilya Tocar  <ilya.tocar@intel.com>
1920             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1921             Ilya Verbin  <ilya.verbin@intel.com>
1922             Kirill Yukhin  <kirill.yukhin@intel.com>
1923             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1925         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1926         (VI124_256): Changed to ...
1927         (VI124_256_48_512): This.
1928         (ssepackmode): Extended with wider modes.
1929         (<code><mode>3): Changed iterator.
1930         (*avx2_<code><mode>3): Ditto.
1931         (vec_pack_trunc_<mode>): Ditto.
1933 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1934             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1935             Sergey Lega  <sergey.s.lega@intel.com>
1936             Anna Tikhonova  <anna.tikhonova@intel.com>
1937             Ilya Tocar  <ilya.tocar@intel.com>
1938             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1939             Ilya Verbin  <ilya.verbin@intel.com>
1940             Kirill Yukhin  <kirill.yukhin@intel.com>
1941             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1943         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1944         (VI8F_256_512): Ditto.
1945         (abs<mode>2): Changed iterator.
1946         (avx2_perm<mode>): Changed to ...
1947         (<avx2_avx512f>_perm<mode>): This.
1948         (avx2_perm<mode>_1): Changed to ...
1949         (<avx2_avx512f>_perm<mode>_1): This.
1951 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1952             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1953             Sergey Lega  <sergey.s.lega@intel.com>
1954             Anna Tikhonova  <anna.tikhonova@intel.com>
1955             Ilya Tocar  <ilya.tocar@intel.com>
1956             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1957             Ilya Verbin  <ilya.verbin@intel.com>
1958             Kirill Yukhin  <kirill.yukhin@intel.com>
1959             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1961         * config/i386/sse.md (VI48_AVX512F): New.
1962         (VI48_AVX2): Changed to ...
1963         (VI48_AVX2_48_AVX512F): This.
1964         (avx2_ashrv<mode>): Changed to ...
1965         (<avx2_avx512f>_ashrv<mode>): This.
1966         (avx2_<shift_insn>v<mode>): Changed to ...
1967         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1969 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1970             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1971             Sergey Lega  <sergey.s.lega@intel.com>
1972             Anna Tikhonova  <anna.tikhonova@intel.com>
1973             Ilya Tocar  <ilya.tocar@intel.com>
1974             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1975             Ilya Verbin  <ilya.verbin@intel.com>
1976             Kirill Yukhin  <kirill.yukhin@intel.com>
1977             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1979         * config/i386/sse.md (VI4_AVX512F): New.
1980         (VI8_AVX2_AVX512F): Ditto.
1981         (mul<mode>3): Extended with wider modes.
1982         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1983         (mul<mode>3): Ditto.
1984         (vec_widen_<s>mult_odd_<mode>): Ditto.
1986 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1987             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1988             Sergey Lega  <sergey.s.lega@intel.com>
1989             Anna Tikhonova  <anna.tikhonova@intel.com>
1990             Ilya Tocar  <ilya.tocar@intel.com>
1991             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1992             Ilya Verbin  <ilya.verbin@intel.com>
1993             Kirill Yukhin  <kirill.yukhin@intel.com>
1994             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1996         * config/i386/sse.md (VI2_AVX512F): New.
1997         (VI124_AVX512F): Ditto.
1998         (sseunpackmode): Extended with wider modes.
1999         (sseunpackfltmode): Ditto.
2000         (vec_unpacks_float_hi_<mode>): Ditto.
2001         (vec_unpacks_float_lo_<mode>): Ditto.
2002         (vec_unpacku_float_hi_<mode>): Ditto.
2003         (vec_unpacku_float_lo_<mode>): Ditto.
2004         (vec_unpacks_lo_<mode>): Ditto.
2005         (vec_unpacks_hi_<mode>): Ditto.
2006         (vec_unpacku_lo_<mode>): Ditto.
2007         (vec_unpacku_hi_<mode>): Ditto.
2009 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2010             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2011             Sergey Lega  <sergey.s.lega@intel.com>
2012             Anna Tikhonova  <anna.tikhonova@intel.com>
2013             Ilya Tocar  <ilya.tocar@intel.com>
2014             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2015             Ilya Verbin  <ilya.verbin@intel.com>
2016             Kirill Yukhin  <kirill.yukhin@intel.com>
2017             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2019         * config/i386/i386.md (multdiv): New.
2020         (multdiv_mnemonic): Ditto.
2021         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2022         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2023         (<sse>_vmdiv<mode>3): Removed.
2025 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2026             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2027             Sergey Lega  <sergey.s.lega@intel.com>
2028             Anna Tikhonova  <anna.tikhonova@intel.com>
2029             Ilya Tocar  <ilya.tocar@intel.com>
2030             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2031             Ilya Verbin  <ilya.verbin@intel.com>
2032             Kirill Yukhin  <kirill.yukhin@intel.com>
2033             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2035         * config/i386/sse.md (V): Extended with wider modes.
2036         (VF2): Ditto.
2037         (ssehalfvecmode): Ditto.
2038         (i128): Ditto.
2039         (ssepackfltmode): Ditto.
2040         (avx_vec_concat<mode>): Ditto.
2041         (V_256_512): New iterator.
2042         (VF2_512_256): Ditto.
2043         (si2dfmode): New attribute.
2044         (si2dfmodelower): Ditto.
2045         (sf2dfmode): Ditto.
2046         (concat_tg_mode): Ditto.
2047         (floatv4siv4df2): Changed to ...
2048         (float<si2dfmodelower><mode>2): This.
2049         (avx_cvtps2pd256): Changed to ...
2050         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2051         (vec_pack_trunc_v4df): Changed to ...
2052         (vec_pack_trunc_<mode>): This.
2053         (avx_vpermil<mode>): Changed to ...
2054         (<sse2_avx_avx512f>_vpermil<mode>): This.
2055         (<fixsuffix>fix_truncv8dfv8si2): New.
2056         (vec_pack_sfix_trunc_v8df): Ditto.
2057         (avx512f_rndscale<mode>): Ditto.
2058         (avx512f_roundpd512): Ditto.
2059         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2061 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2062             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2063             Sergey Lega  <sergey.s.lega@intel.com>
2064             Anna Tikhonova  <anna.tikhonova@intel.com>
2065             Ilya Tocar  <ilya.tocar@intel.com>
2066             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2067             Ilya Verbin  <ilya.verbin@intel.com>
2068             Kirill Yukhin  <kirill.yukhin@intel.com>
2069             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2071         * config/i386/i386.md (any_fix): New iterator.
2072         (fixsuffix): New attribute.
2073         * config/i386/sse.md (VF1): Extened with wider modes.
2074         (VI): Ditto.
2075         (VI_AVX2): Ditto.
2076         (VI8): Ditto.
2077         (sseintvecmodelower): Ditto.
2078         (ssescalarmode): Ditto.
2079         (ssescalarnum): Ditto.
2080         (VF1_128_256): New.
2081         (ssexmmmode): Ditto.
2082         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2083         (<sse>_rcp<mode>2): Change iterator.
2084         (rsqrt<mode>2): Ditto.
2085         (<sse>_rsqrt<mode>2): Ditto.
2086         (avx2_vec_dup<mode>): Ditto.
2087         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2088         (round<mode>2_sfix): Ditto.
2089         (avx2_pbroadcast<mode>): Ditto.
2090         (*andnot<mode>3): Handle XI mode.
2091         (*<code><mode>3): Ditto.
2092         (AVXTOSSEMODE): Removed.
2093         (avx_vpermil<mode>): Changed to ...
2094         (<sse2_avx_avx512f>_vpermil<mode>): This.
2096 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2097             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2098             Sergey Lega  <sergey.s.lega@intel.com>
2099             Anna Tikhonova  <anna.tikhonova@intel.com>
2100             Ilya Tocar  <ilya.tocar@intel.com>
2101             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2102             Ilya Verbin  <ilya.verbin@intel.com>
2103             Kirill Yukhin  <kirill.yukhin@intel.com>
2104             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2106         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2107         (<sse>_comi): Ditto.
2108         (<sse>_ucomi): Ditto.
2109         (sse_cvtss2siq_2): Ditto.
2110         (sse2_cvtsd2si): Ditto.
2111         (sse2_cvtsd2siq): Ditto.
2112         (sse2_cvttsd2si): Ditto.
2113         (sse2_cvttsd2siq): Ditto.
2114         (<shift_insn><mode>3): Ditto.
2115         (sse2_cvtsi2sdq): Update constraint and prefix.
2116         (sse_cvtsi2ss): Update prefix.
2117         (sse_cvtsi2ssq): Ditto.
2119 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2121         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2122         ignore internal calls.
2124 2013-10-11  Richard Biener  <rguenther@suse.de>
2126         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2127         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2128         TREE_OVERFLOW is set.
2130 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2132         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2134         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2135         * gimple.h: Likewise.
2137         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2139         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2140         macros.
2142         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2144         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2145         * configure: Regenerate.
2147 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2149         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2150         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2151         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2152         similarly pp_right_paren.
2153         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2154         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2155         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2156         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2157         clauses.
2158         (omp_declare_simd_clauses_equal,
2159         omp_remove_redundant_declare_simd_attrs): New functions.
2160         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2161         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2162         * tree.h (OMP_LOOP_CHECK): Define.
2163         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2164         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2165         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2166         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2167         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2168         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2169         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2170         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2171         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2172         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2173         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2174         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2175         (omp_remove_redundant_declare_simd_attrs): New prototype.
2176         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2177         GIMPLE_OMP_TEAMS): New codes.
2178         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2179         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2180         fields.
2181         (target_nesting_level): New variable.
2182         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2183         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2184         for collapse > 1 static schedule unless ordered.
2185         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2186         (determine_parallel_type): Adjust get_ws_args_for caller.
2187         (install_var_field): Handle mask & 4 for double indirection.
2188         (scan_sharing_clauses): Ignore shared clause on teams construct.
2189         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2190         (create_omp_child_function): If inside target or declare target
2191         constructs, set "omp declare target" attribute on the child function.
2192         (find_combined_for): New function.
2193         (scan_omp_parallel): Handle combined loops.
2194         (scan_omp_target, scan_omp_teams): New functions.
2195         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2196         restrictions and set ctx->cancellable for cancellable constructs.
2197         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2198         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2199         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2200         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2201         (omp_clause_aligned_alignment): New function.
2202         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2203         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2204         on teams constructs.  Handle user defined reductions and new
2205         OpenMP 4.0 clauses.
2206         (lower_reduction_clauses): Don't set placeholder to address of ref
2207         if it has already the right type.
2208         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2209         (expand_parallel_call): Use the new non-_start suffixed builtins,
2210         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2211         and GOMP_parallel_end after the call.
2212         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2213         (expand_omp_for_init_counts): Handle combined loops.
2214         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2215         loops.
2216         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2217         end of cancellable loops.
2218         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2219         Likewise.  Handle collapse > 1 loops.
2220         (expand_omp_simd): Handle combined loops.
2221         (expand_omp_for): Add inner_stmt argument, adjust callers of
2222         expand_omp_for* functions, use expand_omp_for_static*chunk even
2223         for collapse > 1 unless ordered.
2224         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2225         of cancellable sections.
2226         (expand_omp_single): Remove need_barrier variable, just rely on
2227         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2228         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2229         (expand_omp_atomic_load, expand_omp_atomic_store,
2230         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2231         (expand_omp_target): New function.
2232         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2233         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2234         (build_omp_regions_1): Immediately close region for
2235         GF_OMP_TARGET_KIND_UPDATE.
2236         (maybe_add_implicit_barrier_cancel): New function.
2237         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2238         cancellation.
2239         (lower_omp_single): Likewise.  Add clobber after the barrier.
2240         (lower_omp_taskgroup): New function.
2241         (lower_omp_for): Handle combined loops.  Adjust
2242         lower_rec_input_clauses caller.  Handle cancellation.
2243         (lower_depend_clauses): New function.
2244         (lower_omp_taskreg): Lower depend clauses.  Adjust
2245         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2246         cancellation.
2247         (lower_omp_target, lower_omp_teams): New functions.
2248         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2249         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2250         and GOMP_cancellation_point calls.
2251         (lower_omp): Fold stmts inside of target region.
2252         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2253         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2254         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2255         (BT_FN_VOID_OMPFN_PTR_UINT,
2256         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2257         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2258         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2259         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2260         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2261         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2262         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2263         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2264         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2265         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2266         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2267         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2268         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2269         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2270         BUILT_IN_GOMP_PARALLEL_END.
2271         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2272         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2273         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2274         GF_OMP_FOR_KIND_DISTRIBUTE.
2275         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2276         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2277         (dump_gimple_omp_return): Print lhs if it has any.
2278         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2279         gimple_omp_atomic_seq_cst_p.
2280         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2281         and GIMPLE_OMP_TEAMS.
2282         * langhooks.c (lhd_omp_mappable_type): New function.
2283         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2284         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2285         hook.
2286         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2287         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2288         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2289         (struct gimplify_omp_ctx): Add combined_loop field.
2290         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2291         on stmts inside of target region.
2292         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2293         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2294         ORT_TARGET and ORT_TARGET_DATA.
2295         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2296         Handle GOVD_MAP.
2297         (omp_notice_threadprivate_variable): Complain about threadprivate
2298         variables in target region.
2299         (omp_notice_variable): Complain about vars with non-mappable type
2300         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2301         (omp_check_private): Ignore ORT_TARGET* regions.
2302         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2303         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2304         (find_combined_omp_for): New function.
2305         (gimplify_omp_for): Handle gimplification of combined loops.
2306         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2307         OMP_TEAMS.
2308         (gimplify_omp_target_update): New function.
2309         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2310         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2311         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2312         (gimplify_body): If fndecl has "omp declare target" attribute, add
2313         implicit ORT_TARGET context around it.
2314         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2315         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2316         * tree-nested.c (convert_nonlocal_reference_stmt,
2317         convert_local_reference_stmt, convert_gimple_call): Handle
2318         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2319         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2320         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2321         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2322         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2323         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2324         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2325         BUILT_IN_GOMP_LOOP_END_CANCEL,
2326         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2327         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2328         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2329         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2330         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2331         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2332         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2333         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2334         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2335         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2336         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2337         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2338         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2339         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2340         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2341         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2342         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2343         gimple_build_omp_teams): New functions.
2344         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2345         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2346         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2347         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2348         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2349         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2350         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2351         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2352         GF_OMP_ATOMIC_SEQ_CST): New.
2353         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2354         gimple_build_omp_teams): New prototypes.
2355         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2356         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2357         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2358         GIMPLE_OMP_SINGLE as end of range.
2359         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2360         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2361         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2362         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2363         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2364         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2365         gimple_omp_target_kind, gimple_omp_target_set_kind,
2366         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2367         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2368         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2369         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2370         gimple_omp_teams_set_clauses): New inlines.
2371         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2372         and GIMPLE_OMP_TASKGROUP.
2373         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2374         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2375         enum omp_clause_proc_bind_kind): New.
2376         (union omp_clause_subcode): Add depend_kind, map_kind and
2377         proc_bind_kind fields.
2378         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2379         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2380         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2381         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2382         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2384 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2386         * predict.c (tree_estimate_probability): Add new parameter
2387         for estimate_bb_frequencies.
2388         (estimate_bb_frequencies): Add new parameter to force estimation.
2389         (rebuild_frequencies): When max frequency in function is small,
2390         recompute counts from frequencies.
2391         * predict.h (estimate_bb_frequencies): New parameter.
2393 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2395         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2396         optimizations are disabled.
2398 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2400         * coverage.c (coverage_finish): Fix leak of da_file_name.
2402 2013-10-10  Jan Hubicka  <jh@suse.cz>
2404         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2405         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2406         Bobcat and generic.
2408 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2410         PR middle-end/58670
2411         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2412         if any labels are in FALLTHRU_BB, use a special label emitted
2413         immediately after the asm goto insn rather than label_rtx
2414         of the LABEL_DECL.
2415         (expand_asm_stmt): Adjust caller.
2416         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2417         edge if the last insn in predecessor is a jump with single successor,
2418         but it isn't simplejump_p.
2420 2013-10-10  Richard Biener  <rguenther@suse.de>
2422         PR tree-optimization/58656
2423         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2425 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2427         * gimplify.c: Include expr.h and tm_p.h for targets with special
2428         va-arg padding requirements.
2430 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2432         * tree-flow.h: Move some prototypes to gimple.h.
2433         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2434         * gimple.h: Relocate some prototypes from tree-flow.h
2435         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2436         Move to gimplify.c.
2437         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2438         (build_va_arg_indirect_ref): Relocate and make static.
2439         (std_gimplify_va_arg_expr): Relocate here.
2440         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2441         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2443 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2445         * doc/md.texi: Document the mnemonic attribute.
2447 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2449         PR target/57377
2450         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2451         (set_attr_alternative x ...) when searching for user defined
2452         mnemonic attribute.
2454 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2456         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2458 2013-10-09  Easwaran Raman <eraman@google.com>
2460         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2461         * cfgexpand.c (defer_stack_allocation): ...use here
2462         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2464 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2466         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2467         (operand_equal_for_value_replacement): New function, extracted from
2468         value_replacement and enhanced to catch more cases.
2469         (value_replacement): Use operand_equal_for_value_replacement.
2471 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2473         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2474         get_max_loop_iterations.
2476 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2478         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2479         Place comment above function.
2481 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2483         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2484         are not related to tree-cfg.c.
2485         * tree-ssa-address.h: New file.  Relocate prototypes.
2486         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2487         (addr_for_mem_ref): New.  Combine call to get_address_description and
2488         return addr_for_mem_ref.
2489         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2490         * tree-ssa-live.h: Adjust prototypes.
2491         * passes.c: Include tree-ssa-live.h.
2492         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2493         * graphite.c (graphite_transform_loops): Make static.
2494         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2495         make_pass_graphite, pass_data_graphite_transforms,
2496         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2497         * ipa-pure-const.c (warn_function_noreturn): Make static.
2498         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2499         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2500         Relocate from tree-cfg.c
2501         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2502         static.
2503         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2504         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2505         Move to ipa-pure-const.c.
2506         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2507         Relocate from tree-optimize.c.
2508         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2509         make_pass_fixup_cfg): Move to tree-cfg.c.
2510         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2511         Relocate some prototypes.
2512         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2513         * tree-dump.c (dump_function_to_file): Remove prototype.
2514         Add tree-flow.h to the include file.
2515         * tree-dump.h: Remove prototype.
2516         * tree-parloops.h: New File.  Add prototypes.
2517         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2518         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2519         from tree-ssa-loop.c.
2520         * tree-predcom.c (run_tree_predictive_commoning,
2521         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2522         Relocate here from tree-ssa-loop.c.
2523         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2524         ssa_name_values.release ().
2525         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2526         (ssa_name_values): Relocate from tree-flow.h.
2527         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2528         * tree-ssa-loop.c (run_tree_predictive_commoning,
2529         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2530         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2531         make_pass_graphite, pass_data_graphite_transforms,
2532         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2533         tree_parallelize_loops, pass_data_parallelize_loops,
2534         make_pass_parallelize_loops): Move to other files.
2535         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2536         moved here.
2537         * tree.h: Remove prototypes from tree-address.c.
2539 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2541         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2542         (struct int_tree_map): Move to tree-hasher.h
2543         (SCALE, LABEL, PERCENT): Move to gimple.h
2544         * tree-flow-inline.h: Delete.  Move functions to other files.
2545         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2546         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2547         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2548         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2549         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2550         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2551         * tree-hasher.h: Don't include tree-flow.h.
2552         (struct int_tree_map): Relocate from tree-flow.h.
2553         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2554         tree-flow-inline.h and make static.
2555         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2556         tree-flow-inline.h.
2557         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2558         tree-flow-inline.h and make static.
2559         * tree.h (is_global_var, may_be_aliased): Relocate from
2560         tree-flow-inline.h.
2561         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2562         * value-prof.c: No longer include tree-flow-inline.h.
2563         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2565 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2567         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2568         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2569         (enum move_pos): Move to tree-ssa-loop-im.h
2570         * cfgloop.h: Move some prototypes.
2571         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2572         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2573         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2574         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2575         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2576         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2577         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2578         (enum move_pos): Relocate here.
2579         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2580         tree-ssa-loop.c.
2581         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2582         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2583         make_pass_lim): Relocate here from tree-ssa-loop.c.
2584         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2585         tree-ssa-loop.c.
2586         (loop_edge_to_cancel, unloop_loops): Make static.
2587         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2588         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2589         (tree_complete_unroll, gate_tree_complete_unroll,
2590         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2591         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2592         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2593         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2594         (stmt_invariant_in_loop_p): Remove unused function.
2595         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2596         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2597         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2598         (gcov_type_to_double_int): Move to cfgloop.h.
2599         (double_int_cmp, bound_index,
2600         estimate_numbers_of_iterations_loop): Make static.
2601         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2602         (max_loop_iterations): Factor out get_max_loop_iterations.
2603         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2604         cfgloop.c.
2605         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2606         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2607         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2608         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2609         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2610         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2611         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2612         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2613         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2614         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2615         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2616         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2617         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2618         pass_data_complete_unroll, make_pass_complete_unroll,
2619         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2620         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2621         tree-ssa-loop-ivcanon.c.
2622         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2623         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2624         tree-ssa-loop-prefetch.c.
2625         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2626         tree-ssa-loop-im.c.
2627         (get_lsm_tmp_name): Relocate and add suffix parameter.
2628         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2629         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2630         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2631         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2632         (get_estimated_loop_iterations): Factor out accessor from
2633         estimated_loop_iterations in tree-ssa-loop-niter.c.
2634         (get_max_loop_iterations): Factor out accessor from
2635         _max_loop_iterations in tree-ssa-niter.c.
2636         * loop-unroll.c (decide_unroll_constant_iterations,
2637         decide_unroll_runtime_iterations, decide_peel_simple,
2638         decide_unroll_stupid): Use new get_* accessors.
2640 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2642         PR tree-optimization/20318
2643         * doc/extend.texi (returns_nonnull): New function attribute.
2644         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2645         attribute.
2646         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2647         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2649 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2651         PR middle-end/58570
2652         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2653         false if both components are bitfields.
2655 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2657         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2658         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2659         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2660         * config/aarch64/aarch64.h
2661         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2662         * config/aarch64/aarch64-simd-builtins.def
2663         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2665 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2667         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2669 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2671         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2673 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2675         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2676         (vneg_f64): New intrinsic.
2677         (vneg_s8): Asm replaced with C.
2678         (vneg_s16): Likewise.
2679         (vneg_s32): Likewise.
2680         (vneg_s64): New intrinsic.
2681         (vnegq_f32): Asm replaced with C.
2682         (vnegq_f64): Likewise.
2683         (vnegq_s8): Likewise.
2684         (vnegq_s16): Likewise.
2685         (vnegq_s32): Likewise.
2686         (vnegq_s64): Likewise.
2688 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2690         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2692 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2694         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2695         packed stack special handling.
2696         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2697         back to fixed stack slots for FPRs saved due to stdarg.
2699 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2701         * config/s390/s390.c (s390_frame_info): Restructure function.
2703 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2705         * config/s390/s390.c (struct s390_frame_layout): New field
2706         gpr_save_slots.
2707         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2708         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2709         regs_ever_clobbered to char*.
2710         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2711         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2712         (s390_register_info): Enable FPR save slots.  Move/Copy some
2713         functionality into ...
2714         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2715         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2716         function.
2717         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2718         not imply a stack frame.
2719         (s390_init_frame_layout): Remove variable clobbered_regs.
2720         (s390_update_register_info): Remove function.
2721         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2722         cannot be used for register renaming.
2723         (s390_hard_regno_scratch_ok): New function.
2724         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2725         (s390_initial_elimination_offset): Change offset calculation of
2726         the return address pointer.
2727         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2728         (restore_gprs): Set frame related flag.
2729         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2730         (s390_emit_prologue): Call s390_register_info instead of
2731         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2732         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2733         (s390_optimize_prologue): Call s390_optimize_register_info.
2734         Try to remove also FPR slot save/restore INSNs.  Remove frame
2735         related flags from restore INSNs.
2737 2013-10-08  DJ Delorie  <dj@redhat.com>
2739         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2740         (movhi): Likewise.
2742         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2743         avoid conflict with the MI use of %c.
2744         * config/rl78/rl78-real.md: change %c to %C throughout.
2745         * config/rl78/rl78-virt.md: Likewise.
2747 2013-10-08  Jan Hubicka  <jh@suse.cz>
2749         * config/i386/i386.c (ix86_option_override_internal): Switch
2750         to SSE math for -ffast-math when target ISA supports SSE2.
2752 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2754         * tree-flow.h: Remove some prototypes.
2755         * tree.h: Remove some protypes, add a couple.
2756         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2757         using_eh_for_cleanups_p): Add interface routines for front ends.
2758         * tree-eh.h: New file.  Add protoptyes.
2759         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2760         (add_stmt_to_eh_lp_fn): Make static.
2761         (lower_try_finally): Use new using_eh_for_cleanups_p.
2762         * emit-rtl.c: Include tree-eh.h.
2763         * gimple.h: Include tree-eh.h.
2765 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2767         PR tree-optimization/58480
2768         * tree-vrp.c (infer_nonnull_range): New function.
2769         (infer_value_range): Call infer_nonnull_range.
2771 2013-10-08  Dehao Chen  <dehao@google.com>
2773         PR tree-optimization/58619
2774         * tree-inline.c (copy_phis_for_bb): Combine location data
2775         only if non-null.
2777 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2779         PR target/58423
2780         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2781         RTX_FRAME_RELATED_P on INSN.
2783 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2785         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2786         (altivec_expand_vec_perm_const): Call it.
2788 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2790         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2791         sequences for LE VSX loads and stores at expand time.
2792         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2793         prototype.
2794         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2795         (rs6000_gen_le_vsx_permute): New.
2796         (rs6000_gen_le_vsx_load): New.
2797         (rs6000_gen_le_vsx_store): New.
2798         (rs6000_gen_le_vsx_move): New.
2799         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2800         (*vsx_le_perm_load_v4si): New.
2801         (*vsx_le_perm_load_v8hi): New.
2802         (*vsx_le_perm_load_v16qi): New.
2803         (*vsx_le_perm_store_v2di): New.
2804         (*vsx_le_perm_store_v4si): New.
2805         (*vsx_le_perm_store_v8hi): New.
2806         (*vsx_le_perm_store_v16qi): New.
2807         (*vsx_xxpermdi2_le_<mode>): New.
2808         (*vsx_xxpermdi4_le_<mode>): New.
2809         (*vsx_xxpermdi8_le_V8HI): New.
2810         (*vsx_xxpermdi16_le_V16QI): New.
2811         (*vsx_lxvd2x2_le_<mode>): New.
2812         (*vsx_lxvd2x4_le_<mode>): New.
2813         (*vsx_lxvd2x8_le_V8HI): New.
2814         (*vsx_lxvd2x16_le_V16QI): New.
2815         (*vsx_stxvd2x2_le_<mode>): New.
2816         (*vsx_stxvd2x4_le_<mode>): New.
2817         (*vsx_stxvd2x8_le_V8HI): New.
2818         (*vsx_stxvd2x16_le_V16QI): New.
2820 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2822         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2824 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2826         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2827         loop to work also for 31bit ABI.
2828         Save the stack pointer for frame_size > 0.
2830 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2832         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2833         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2834         constraint letters from expanders.
2835         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2836         retry count to general_operand.
2837         ("tabort"): Give operand 0 a mode.
2838         ("tabort_1"): Add mode and constraint letter for operand 0.
2839         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2841 2013-10-04  Jeff Law  <law@redhat.com>
2843         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2845         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2846         out of ...
2847         (thread_across_edge): Here.  Call it.
2849 2013-10-04  Cary Coutant  <ccoutant@google.com>
2851         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2852         discarding a location list expression (or a piece of one).
2854 2013-10-03  Jan Hubicka  <jh@suse.cz>
2856         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2857         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2858         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2860 2013-10-03  Jan Hubicka  <jh@suse.cz>
2862         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2863         accumulate-outgoing-args when producing unwind info.
2865 2013-10-03  Wei Mi  <wmi@google.com>
2867         * lra-constraints.c (insert_move_for_subreg): New function
2868         extracted from simplify_operand_subreg.
2869         (simplify_operand_subreg): Add reload for paradoxical subreg.
2871 2013-10-03  Rong Xu  <xur@google.com>
2873         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2874         the candidate of builtin_expect such that we should fix the
2875         size/time estimation.
2876         (estimate_function_body_sizes): Do the acutally size/time fix-up
2877         for builtin_expect.
2879 2013-10-03  Rong Xu  <xur@google.com>
2881         * predict.c (tree_predict_by_opcode): Get the probability
2882         for builtin_expect from param builtin_expect_probability.
2883         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2884         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2885         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2887 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2889         PR c++/19476
2890         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2891         * calls.c (alloca_call_p): Use get_callee_fndecl.
2892         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2893         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2894         Likewise.
2895         (vrp_visit_stmt): Remove duplicated code.
2897 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2899         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2900         ceildf2, btruncdf2, instead of vsx_* name.
2902         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2903         iterators to only do V2DF and V4SF here.  Move the DF code to
2904         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2905         just 'v' since only vector operations are handled with these insns
2906         after moving the DF support to rs6000.md.
2907         (vsx_sub<mode>3): Likewise.
2908         (vsx_mul<mode>3): Likewise.
2909         (vsx_div<mode>3): Likewise.
2910         (vsx_fre<mode>2): Likewise.
2911         (vsx_neg<mode>2): Likewise.
2912         (vsx_abs<mode>2): Likewise.
2913         (vsx_nabs<mode>2): Likewise.
2914         (vsx_smax<mode>3): Likewise.
2915         (vsx_smin<mode>3): Likewise.
2916         (vsx_sqrt<mode>2): Likewise.
2917         (vsx_rsqrte<mode>2): Likewise.
2918         (vsx_fms<mode>4): Likewise.
2919         (vsx_nfma<mode>4): Likewise.
2920         (vsx_copysign<mode>3): Likewise.
2921         (vsx_btrunc<mode>2): Likewise.
2922         (vsx_floor<mode>2): Likewise.
2923         (vsx_ceil<mode>2): Likewise.
2924         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2925         (vsx_sminsf3): Likewise.
2926         (vsx_fmadf4): Likewise.
2927         (vsx_fmsdf4): Likewise.
2928         (vsx_nfmadf4): Likewise.
2929         (vsx_nfmsdf4): Likewise.
2930         (vsx_cmpdf_internal1): Likewise.
2932         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2933         simpler to select whether a target has SPE or traditional floating
2934         point support in iterators.
2935         (TARGET_DF_SPE): Likewise.
2936         (TARGET_SF_FPR): Likewise.
2937         (TARGET_DF_FPR): Likewise.
2938         (TARGET_SF_INSN): Macros to say whether floating point support
2939         exists for a given operation for expanders.
2940         (TARGET_DF_INSN): Likewise.
2942         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2943         combining of SF/DF mode operations, using both traditional and VSX
2944         registers.
2945         (Fvsx): Likewise.
2946         (Ff): Likewise.
2947         (Fv): Likewise.
2948         (Fs): Likewise.
2949         (Ffre): Likewise.
2950         (FFRE): Likewise.
2951         (abs<mode>2): Combine SF/DF modes using traditional floating point
2952         instructions.  Add support for using the upper DF registers with
2953         VSX support, and SF registers with power8-vector support.  Update
2954         expanders for operations supported by both the SPE and traditional
2955         floating point units.
2956         (abs<mode>2_fpr): Likewise.
2957         (nabs<mode>2): Likewise.
2958         (nabs<mode>2_fpr): Likewise.
2959         (neg<mode>2): Likewise.
2960         (neg<mode>2_fpr): Likewise.
2961         (add<mode>3): Likewise.
2962         (add<mode>3_fpr): Likewise.
2963         (sub<mode>3): Likewise.
2964         (sub<mode>3_fpr): Likewise.
2965         (mul<mode>3): Likewise.
2966         (mul<mode>3_fpr): Likewise.
2967         (div<mode>3): Likewise.
2968         (div<mode>3_fpr): Likewise.
2969         (sqrt<mode>3): Likewise.
2970         (sqrt<mode>3_fpr): Likewise.
2971         (fre<Fs>): Likewise.
2972         (rsqrt<mode>2): Likewise.
2973         (cmp<mode>_fpr): Likewise.
2974         (smax<mode>3): Likewise.
2975         (smin<mode>3): Likewise.
2976         (smax<mode>3_vsx): Likewise.
2977         (smin<mode>3_vsx): Likewise.
2978         (negsf2): Delete SF operations that are merged with DF.
2979         (abssf2): Likewise.
2980         (addsf3): Likewise.
2981         (subsf3): Likewise.
2982         (mulsf3): Likewise.
2983         (divsf3): Likewise.
2984         (fres): Likewise.
2985         (fmasf4_fpr): Likewise.
2986         (fmssf4_fpr): Likewise.
2987         (nfmasf4_fpr): Likewise.
2988         (nfmssf4_fpr): Likewise.
2989         (sqrtsf2): Likewise.
2990         (rsqrtsf_internal1): Likewise.
2991         (smaxsf3): Likewise.
2992         (sminsf3): Likewise.
2993         (cmpsf_internal1): Likewise.
2994         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2995         (negdf2): Delete DF operations that are merged with SF.
2996         (absdf2): Likewise.
2997         (nabsdf2): Likewise.
2998         (adddf3): Likewise.
2999         (subdf3): Likewise.
3000         (muldf3): Likewise.
3001         (divdf3): Likewise.
3002         (fred): Likewise.
3003         (rsqrtdf_internal1): Likewise.
3004         (fmadf4_fpr): Likewise.
3005         (fmsdf4_fpr): Likewise.
3006         (nfmadf4_fpr): Likewise.
3007         (nfmsdf4_fpr): Likewise.
3008         (sqrtdf2): Likewise.
3009         (smaxdf3): Likewise.
3010         (smindf3): Likewise.
3011         (cmpdf_internal1): Likewise.
3012         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3013         (btrunc<mode>2): Delete separate expander, and combine with the
3014         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3015         (btrunc<mode>2_fpr): Likewise.
3016         (ceil<mode>2): Likewise.
3017         (ceil<mode>2_fpr): Likewise.
3018         (floor<mode>2): Likewise.
3019         (floor<mode>2_fpr): Likewise.
3020         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3021         Add support for using the upper registers with VSX and
3022         power8-vector.  Move insns to be closer to the define_expands. On
3023         VSX systems, prefer the traditional form of FMA over the VSX
3024         version, since the traditional form allows the target not to
3025         overlap with the inputs.
3026         (fms<mode>4_fpr): Likewise.
3027         (nfma<mode>4_fpr): Likewise.
3028         (nfms<mode>4_fpr): Likewise.
3030 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3031             Richard Earnshaw  <richard.earnshaw@arm.com>
3033         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3034         (struct mult_cost_table): Likewise.
3035         (struct mem_cost_table): Likewise.
3036         (struct fp_cost_table): Likewise.
3037         (struct vector_cost_table): Likewise.
3038         (cpu_cost_table): Likewise.
3039         * config/arm/arm.opt (mold-rts-costs): New option.
3040         (mnew-generic-costs): Likewise.
3041         * config/arm/arm.c (generic_extra_costs): New table.
3042         (cortexa15_extra_costs): Likewise.
3043         (arm_slowmul_tune): Use NULL as new costs.
3044         (arm_fastmul_tune): Likewise.
3045         (arm_strongarm_tune): Likewise.
3046         (arm_xscale_tune): Likewise.
3047         (arm_9e_tune): Likewise.
3048         (arm_v6t2_tune): Likewise.
3049         (arm_cortex_a5_tune): Likewise.
3050         (arm_cortex_a9_tune): Likewise.
3051         (arm_v6m_tune): Likewise.
3052         (arm_fa726te_tune): Likewise.
3053         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3054         (arm_cortex_tune): Use generict_extra_costs.
3055         (shifter_op_p): New function.
3056         (arm_unspec_cost): Likewise.
3057         (LIBCALL_COST): Define.
3058         (arm_new_rtx_costs): New function.
3059         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3060         table is available. Use old costs otherwise unless mnew-generic-costs
3061         is specified.
3062         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3063         (cpu_cost_table): Declare.
3065 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3067         PR target/58460
3068         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3069         (*subs_mul_imm_<mode>)
3070         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3071         (*sub_<shift>_<mode>)
3072         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3073         Remove k constraint.
3075 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3077         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3078         code.
3079         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3081 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3083         * predict.c (probably_never_executed): New function.
3084         (probably_never_executed_bb_p): Invoke probably_never_executed.
3085         (probably_never_executed_edge_p): Ditto.
3086         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3087         Treat profile insanities conservatively.
3089 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3091         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3093 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3095         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3096         use it.  Use smaller increase for scratch.  Don't increase reject
3097         for early clobber scratch.
3098         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3099         setting eliminated regs except setting fp from hfp.
3100         (lra_eliminate): Check lra_insn_recog_data on NULL.
3102 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3104         PR target/58587
3105         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3106         setting -mvsx-timode by default until the underlying problem is fixed.
3107         (RS6000_CPU, power7 defaults): Likewise.
3109 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3111         * config/x-linux (host-linux.o): Remove header dependencies.
3112         Use $(COMPILE) and $(POSTCOMPILE).
3113         * config/t-linux-android (linux-android.o): Ditto.
3115 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3117         * Makefile.in (expmed.o-warn): Remove.
3119 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3121         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3122         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3124 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3126         * dojump.c (do_jump_1): Divide probability between
3127         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3129 2013-10-02  Tom Tromey  <tromey@redhat.com>
3131         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3133 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3135         * tree-flow.h: Remove some prototypes.
3136         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3137         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3138         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3139         mark_virtual_phi_result_for_renaming): Relocate here.
3140         * tree-into-ssa.h: Add prototypes.
3141         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3142         single_pred_before_succ_order.
3143         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3144         (nonfreeing_call_p) Move to gimple.c.
3145         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3146         tree-ssa-phiopt.c.
3147         * basic-block.h (single_pred_before_succ_order): Add prototype.
3148         * gimple.c (nonfreeing_call_p): Relocate here.
3149         * gimple.h: Add prototype.
3150         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3151         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3152         * tree-ssa.h: Include tree-ssa-dom.h.
3154 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3156         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3157         Use $(COMPILE) and $(POSTCOMPILE).
3159         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3161 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3163         * tree-flow.h: Remove some prototypes.
3164         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3165         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3166         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3167         propagate_tree_value*): Move from here to...
3168         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3169         propagate_tree_value*): Relocate here.
3170         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3171         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3172         * gimple-fold.c: Remove gimple-fold.h from include list.
3173         * tree-vrp.c: Remove gimple-fold.h from include list.
3174         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3175         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3176         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3177         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3178         * sese.c: Add tree-ssa-propagate.h to include list.
3180 2013-10-02  Richard Biener  <rguenther@suse.de>
3182         * tree-loop-distribution.c: Include tree-vectorizer.h for
3183         find_loop_location.
3184         (enum partition_kind): Remove PKIND_REDUCTION.
3185         (struct partition_s): Remove has_writes member, add reduction_p member.
3186         (partition_alloc): Adjust.
3187         (partition_builtin_p): Likewise.
3188         (partition_has_writes): Remove.
3189         (partition_reduction_p): New function.
3190         (partition_merge_into): Likewise.
3191         (generate_code_for_partition): Commonize builtin partition
3192         handling tail.
3193         (rdg_cannot_recompute_vertex_p): Remove.
3194         (already_processed_vertex_p): Likewise.
3195         (rdg_flag_vertex): Do not set has_writes.
3196         (classify_partition): Adjust.
3197         (rdg_build_partitions): Do not set has_writes, treat all
3198         partitions as useful.
3199         (distribute_loop): Record number of library calls generated.  Adjust.
3200         (tree_loop_distribution): Report number of loops and library
3201         calls generated as opt-info.
3203 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3205         * tree-flow.h: Include new .h files.  Move prototypes.
3206         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3207         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3208         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3209         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3210         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3211         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3212         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3213         * tree.h (get_ref_base_and_extent): Move prototype out.
3214         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3215         tree-dfa.h.
3216         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3217         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3218         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3219         * tree-scalar-evolution.c: Include tree.h.
3220         * sese.c: Include tree.h.
3221         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3222         * dwarf2out.c: Include tree-dfa.h.
3223         * tree-chrec.c: Include tree.h.
3224         * tree-data-ref.c: Include tree.h.
3226 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3228         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3229         Fix whitespace.
3231 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3233         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3234         Use $(COMPILE) and $(POSTCOMPILE).
3235         (sol2-cxx.o): Likewise.
3236         (sol2-stubs.o): Likewise.
3237         (sol2.o): Likewise.
3238         * config/x-solaris (host-solaris.o): Likewise.
3240         * config/sparc/t-sparc (sparc.o): Remove.
3241         (sparc-c.o): Remove header dependencies.
3242         Use $(COMPILE) and $(POSTCOMPILE).
3243         * config/sparc/x-sparc: Likewise.
3245 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3247         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3248         * config/arc/arc700.md: Likewise.
3249         * config/arc/arc-modes.def: Likewise.
3250         * config/arc/arc-simd.h: Likewise.
3251         * config/arc/t-arc-uClibc: Likewise.
3252         * config/arc/t-arc-newlib: Likewise.
3254 2013-10-02  Renlin Li  <renlin.li@arm.com>
3256         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3257         plus_constant.
3258         (aarch64_expand_epilogue): Likewise.
3260 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3261             Yufeng Zhang  <yufeng.zhang@arm.com>
3263         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3264         declaration.
3265         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3266         'base_in' represent a conversion and legal_cast_p_1 holds; set
3267         'base_in' with the returned value from get_unwidened.
3269 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3271         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3272         plus_constant is not used.
3274 2013-10-01  Wei Mi  <wmi@google.com>
3276         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3277         * config/i386/i386.md: Add define_peephole2 to
3278         break partial reg stall for cvtss2sd/cvtsd2ss.
3280 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3282         * config/arc/arc.c (pass_arc_ifcvt::clone):
3283         Update for ctxt_ -> m_ctxt change.
3285 2013-10-01  Jeff Law  <law@redhat.com>
3287         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3288         outgoing_edge and intermediate_edge fields.  Instead store the path.
3289         (redirection_data::hash): Hash on the last edge's destination index.
3290         (redirection_data::equal): Check the entire thread path.
3291         (lookup_redirectio_data): Corresponding changes.
3292         (create_edge_and_update_destination_phis): Likewise.
3293         (thread_single_edge): Likewise.
3295 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3296             Diego Novillo <dnovillo@google.com>
3298         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3299         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3300         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3301         versions of these patterns.
3303         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3304         (__builtin_arc_aligned): Likewise.
3306         * config/arc/arc.md: Expand adc_0 comment stating the intended
3307         purpose and why it isn't ready.
3308         Replace commented out call_value_via_label_mixed with a
3309         plain comment about bl_s.
3311         * config/arc/arc.c (stdio.h): Don't include directly.
3312         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3313         Remove if (1) condition.
3314         (arc_encode_section_info): Fix comment.
3316 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3318         * config/arc/arc.c (arc_conditional_register_usage):
3319         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3320         Also set reg_alloc_order for DMA config regs.
3322 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3323             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3325         * doc/install.texi (--with-cpu): Mention ARC.
3326         (arc-*-elf32): New paragraph.
3327         (arc-linux-uclibc): Likewise.
3328         * doc/md.texi (Machine Constraints): Add ARC part.
3329         * doc/invoke.texi: (menu): Add ARC Options.
3330         (Machine Dependent Options) <ARC Options>: Add synopsis.
3331         (node ARC Options): Add.
3332         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3333         (ARC Built-in Functions): New section defining
3334         generic ARC built-in functions.
3335         (ARC SIMD Built-in Functions): New section defining SIMD specific
3336         built-in functions.
3337         (Declaring Attributes of Functions): Extended
3338         description of short_call and long_call attributes for ARC and
3339         added index entries.
3341 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3342             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3343             Joern Rennecke  <joern.rennecke@embecosm.com>
3344             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3345             Brendan Kehoe  <brendan@zen.org>
3346             Michael Eager  <eager@eagercon.com>
3347             Simon Cook  <simon.cook@embecosm.com>
3348             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3350         * config/arc, common/config/arc: New directories.
3352 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3353             Brendan Kehoe  <brendan@zen.org>
3354             Simon Cook  <simon.cook@embecosm.com>
3356         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3358 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3360         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3361         * tree-ssa-coalesce.h: New. Move prototype to here.
3362         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3363         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3364         (gimple_can_coalesce_p): Move to...
3365         * gimple.c (gimple_can_coalesce_p): Here.
3367 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3369         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3370         (dump_decl_set): Move to gimple.c.
3371         * gimple.h: Don't include tree-ssa-operands.h.
3372         (dump_decl_set): Add prototype.
3373         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3374         Move to gimple-ssa.h.
3375         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3376         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3377         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3378         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3379         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3380         than PHI_ARG_DEF.
3381         (dump_decl_set): Relocate here.
3382         * gimple-ssa.h: New file.
3383         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3384         Relocate from gimple.h.
3385         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3386         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3387         swap_tree_operands and remove non-ssa path.
3388         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3389         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3390         swap_ssa_operands.
3391         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3392         vect_is_simple_reduction_1): Use swap_ssa_operands.
3393         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3394         (enum need_phi_state): Move to tree-into-ssa.c.
3395         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3396         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3397         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3398         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3399         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3400         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3401         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3402         num_imm_uses): Move to ssa-iterators.h.
3403         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3404         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3405         tree-phinodes.h.
3406         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3407         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3408         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3409         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3410         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3411         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3412         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3413         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3414         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3415         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3416         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3417         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3418         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3419         (set_phi_nodes): Move to tree-phinodes.h.
3420         * tree-ssa-operands.h (enum ssa_op_iter_type,
3421         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3422         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3423         (dump_decl_set): Remove prototype.
3424         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3425         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3426         (set_phi_nodes): Relocate from tree-flow-inline.h.
3427         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3428         tree-flow-inline.h
3429         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3430         include list.  Temporarily add gimple.h to include list.
3431         * ssa-iterators.h: New file.
3432         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3433         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3434         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3435         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3436         Relocate from tree-ssa-operands.h.
3437         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3438         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3439         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3440         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3441         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3442         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3443         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3444         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3445         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3446         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3447         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3448         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3449         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3450         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3451         Relocate from tree-flow-inline.h.
3452         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3454 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3456         * aarch64-simd.md
3457         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3458         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3459         Insert '\t' to output template.
3460         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3461         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3462         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3463         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3465 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3467         * doc/install.texi (Host/target specific installation notes for GCC):
3468         Put @anchor before @heading.
3469         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3470         Use @email for email addresses.
3472 2013-10-01  Jeff Law  <law@redhat.com>
3474         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3475         a vec.  Only delete the path if we create one without successfully
3476         registering a jump thread.
3477         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3478         as a pointer.
3479         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3480         (paths): New vector of jump threading paths.
3481         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3482         (THREAD_PATH): New accessor macro for the entire thread path.
3483         (lookup_redirection_data): Get intermediate and final outgoing edge
3484         from the thread path.
3485         (create_edge_and_update_destination_phis): Copy the threading path.
3486         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3487         jump threading path.
3488         (ssa_redirect_edges): Get edges and block types from the jump threading
3489         path.  Free the path vector.
3490         (thread_block): Get edges from the jump threading path.  Look at the
3491         entire path to see if we thread to a loop exit.  If we cancel a jump
3492         thread request, then free the path vector.
3493         (thread_single_edge): Get edges and block types from the jump threading
3494         path.  Free the path vector.
3495         (thread_through_loop_header): Get edges and block types from the jump
3496         threading path.  Free the path vector.
3497         (mark_threaded_blocks): Iterate over the vector of paths and store
3498         the path on the appropriate edge.  Get edges and block types from the
3499         jump threading path.
3500         (mark_threaded_blocks): Get edges and block types from the jump
3501         threading path.  Free the path vector.
3502         (thread_through_all_blocks): Use the vector of paths rather than
3503         a vector of 3-edge sets.
3504         (register_jump_thread): Accept pointer to a path vector rather
3505         than the path vector itself.  Store the path vector for later use.
3506         Simplify.
3508 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3509             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3511         PR target/58574
3512         * config/s390/s390.c (s390_split_branches): Modify check for table
3513         jump insns.
3514         (s390_chunkify_start): Rearrange table jump insn check in order to
3515         deal with compare and branch insns correctly.
3517 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3519         PR target/58578
3520         Revert
3521         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3522         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3523         define_insn_and_split.
3524         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3525         (shiftsi3_compare): New pattern.
3526         (rrx): New pattern.
3527         * config/arm/unspecs.md (UNSPEC_RRX): New.
3529 2013-10-01  Alan Modra  <amodra@gmail.com>
3531         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3532         casing inout operands.
3534 2013-10-01  Richard Biener  <rguenther@suse.de>
3536         PR tree-optimization/58553
3537         * tree-loop-distribution.c (struct partition_s): Add niter member.
3538         (classify_partition): Populate niter member for the partition
3539         and properly identify whether the relevant store happens before
3540         or after the loop exit.
3541         (generate_memset_builtin): Use niter member from the partition.
3542         (generate_memcpy_builtin): Likewise.
3544 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3546         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3547         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3549 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3551         * basic-block.h (edge_list): Prefix member names with "m_".
3552         * context.h (context): Likewise.
3553         * domwalk.h (dom_walker): Likewise.
3554         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3555         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3556         * hash-table.h (hash_table): Likewise.
3557         * machmode.h (bit_field_mode_iterator): Likewise.
3558         * pass_manager.h (pass_list): Likewise.
3559         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3560         * tree-pass.h (pass_data): Likewise.
3561         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3562         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3563         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3564         * asan.c (pass_data_asan): Update accordingly.
3565         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3566         (control_dependences::control_dependences): Likewise.
3567         (control_dependences::~control_dependences): Likewise.
3568         (control_dependences::~control_dependences): Likewise.
3569         (control_dependences::get_edges_dependent_on): Likewise.
3570         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3571         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3572         * context.c (gcc::context::context): Likewise.
3573         * cprop.c (pass_rtl_cprop::clone): Likewise.
3574         * domwalk.c (dom_walker::walk): Likewise.
3575         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3576         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3577         * mode-switching.c (pass_mode_switching::clone): Likewise.
3578         * passes.c (opt_pass::opt_pass): Likewise.
3579         (pass_manager::pass_manager): Likewise.
3580         * predict.c (pass_strip_predict_hints::clone): Likewise.
3581         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3582         (pass_split_all_insns::clone): Likewise.
3583         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3584         Likewise.
3585         (bit_field_mode_iterator::next_mode): Likewise.
3586         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3587         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3588         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3589         * tree-complex.c (pass_lower_complex::clone): Likewise.
3590         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3591         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3592         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3593         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3594         (pass_fold_builtins::clone): Likewise.
3595         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3596         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3597         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3598         (pass_cd_dce::clone): Likewise.
3599         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3600         (pass_phi_only_cprop::clone): Likewise.
3601         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3602         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3603         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3604         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3605         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3606         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3607         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3608         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3609         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3610         * tree-vrp.c (pass_vrp::clone): Likewise.
3611         * tsan.c (pass_tsan::clone): Likewise.
3613 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3615         PR middle-end/58564
3616         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3617         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3619         PR middle-end/58564
3620         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3621         optimization, punt if sign_bit_p looked through any zero extension.
3623 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3625         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3626         Update redirected out edge count in joiner case.
3627         (ssa_redirect_edges): Common the joiner and non-joiner cases
3628         so that joiner case gets profile updates.
3630 2013-09-30  Richard Biener  <rguenther@suse.de>
3632         PR tree-optimization/58554
3633         * tree-loop-distribution.c (classify_partition): Require
3634         unconditionally executed stores for memcpy and memset recognition.
3635         (tree_loop_distribution): Calculate dominance info.
3637 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3639         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3640         (NO_PROFILE_COUNTERS): Likewise.
3641         (PROFILE_HOOK): Likewise.
3642         (FUNCTION_PROFILER): Likewise.
3643         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3645 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3647         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3648         calls and defines in TARGET_MACHO conditional.
3649         (load_macho_picbase_di): Likewise.
3650         (reload_macho_picbase): Likewise.
3651         (reload_macho_picbase_si): Likewise.
3652         (reload_macho_picbase_di): Likewise.
3653         (nonlocal_goto_receiver): Likewise.
3655 2013-09-30  Nick Clifton  <nickc@redhat.com>
3657         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3658         that use the MSP430X ISA.
3659         (msp430_option_override): Scan -mmcu command line option for any
3660         MCU name that supports the MSP430X ISA.
3661         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3662         -mmcu options which enable the MSP430X ISA.
3664 2013-09-30  Richard Biener  <rguenther@suse.de>
3666         PR middle-end/58532
3667         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3668         before looking for setjmp-like calls.
3670 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3672         PR target/10901
3673         * config/darwin-protos.h (machopic_get_function_picbase): New.
3674         * config/darwin.c (machopic_get_function_picbase): New.
3675         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3676         label for a new func.  (load_macho_picbase_di): Likewise.
3677         (reload_macho_picbase): New expand.
3678         (reload_macho_picbase_si): New insn.
3679         (reload_macho_picbase_di): New insn.
3680         (nonlocal_goto_receiver): New define and split.
3681         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3682         (unspecv enum): Add UNSPECV_NLGR.
3684 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3686         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3687         that altivec registers are correctly sized on Darwin.
3689 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3691         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3692         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3693         * config/x-darwin (host-darwin.o): Likewise.
3694         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3695         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3696         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3698 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3700         * doc/invoke.texi: Fix usage of @tie{} command.
3702 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3704         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3706 2013-09-28  Jan Hubicka  <jh@suse.cz>
3708         * config/i386/x86-tune.def: Add documentation for each of the options;
3709         add whitespace.
3711 2013-09-28  Jan Hubicka  <jh@suse.cz>
3713         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3714         generic.
3715         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3716         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3717         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3718         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3719         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3721 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3723         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3724         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3725         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3726         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3727         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3728         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3729         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3730         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3731         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3732         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3733         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3734         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3735         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3736         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3737         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3738         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3739         gimple.h, godump.c, graphite-clast-to-gimple.c,
3740         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3741         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3742         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3743         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3744         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3745         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3746         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3747         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3748         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3749         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3750         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3751         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3752         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3753         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3754         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3755         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3756         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3757         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3758         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3759         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3760         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3761         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3762         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3763         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3764         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3765         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3766         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3767         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3768         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3769         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3770         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3771         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3772         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3773         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3774         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3775         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3776         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3777         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3778         whitespace before "(".
3780 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3782         * expr.h (extract_bit_field): Remove packedp parameter.
3783         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3784         from forward declaration.
3785         (store_split_bit_field): Remove packedp arg from calls to
3786         extract_fixed_bit_field.
3787         (extract_bit_field_1): Remove packedp parameter and packedp
3788         argument from recursive calls and calls to extract_fixed_bit_field.
3789         (extract_bit_field): Remove packedp parameter and corresponding
3790         arg to extract_bit_field_1.
3791         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3792         to issue warnings.
3793         (extract_split_bit_field): Remove packedp arg from call to
3794         extract_fixed_bit_field.
3795         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3796         (copy_blkmode_from_reg): Likewise.
3797         (copy_blkmode_to_reg): Likewise.
3798         (read_complex_part): Likewise.
3799         (store_field): Likewise.
3800         (expand_expr_real_1): Likewise.
3801         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3802         to extract_bit_field.
3803         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3804         call to extract_bit_field.
3805         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3806         call to extract_bit_field.
3807         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3808         and special packedp behavior from -fstrict-volatile-bitfields
3809         documentation.
3811 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3813         * lra-eliminations.c (init_elim_table): Guard value_p.
3815 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3817         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3818         DFmode, DImode, and SFmode in the upper VSX registers based on the
3819         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3820         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3821         of the VSX register handling.
3823         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3824         (f32_sv): Likewise.
3825         (zero_extendsidi2_lfiwzx): Add support for loading into the
3826         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3827         only do VSX memory options on Altivec registers.
3828         (extendsidi2_lfiwax): Likewise.
3829         (extendsfdf2_fpr): Likewise.
3830         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3831         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3832         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3833         (movdi_internal64): Likewise.
3835 2013-09-27  Xinliang David Li  <davidxl@google.com>
3837         * opts.c (finish_options): Adjust parameters
3838         according to vect cost model.
3839         (common_handle_option): Set dynamic vect cost
3840         model for FDO.
3841         targhooks.c (default_add_stmt_cost): Compute stmt cost
3842         unconditionally.
3843         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3844         Use helper function.
3845         * tree-vectorizer.h (unlimited_cost_model): New function.
3846         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3847         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3848         function.
3849         (vect_enhance_data_refs_alignment): Ditto.
3850         * flag-types.h: New enum.
3851         * common/config/i386/i386-common.c (ix86_option_init_struct):
3852         No need to initialize vect_cost_model flag.
3853         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3854         unconditionally.
3856 2013-09-27  Diego Novillo  <dnovillo@google.com>
3858         * gimple.h (enum ssa_mode): Remove.
3860 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3862         * cfgloop.h (number_of_loops): Fix typo in check for null.
3864 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3866         PR middle-end/58551
3867         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3868         are children of outermost saved_cfun's loop, and set it up to
3869         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3870         if loop != loop0 and SESE region contains bbs that belong to loop0.
3872 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3874         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3875         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3876         (set_address_segment, set_address_base, set_address_index)
3877         (set_address_disp): Accept the argument unconditionally.
3878         (baseness): Remove must_be_base_p and must_be_index_p checks.
3879         (decompose_normal_address): Classify as much as possible in the
3880         main loop.
3882 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3884         * cse.c (count_reg_usage): Handle INT_LIST.
3885         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3886         * reginfo.c (reg_scan_mark_refs): Likewise.
3887         * reload1.c (eliminate_regs_1): Likewise.
3889 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3891         PR middle-end/58547
3892         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3893         signed.
3895 2013-09-27  Richard Biener  <rguenther@suse.de>
3897         PR tree-optimization/58459
3898         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3899         restriction not propagating into loops.
3901 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3903         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3904         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3905         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3907 2013-09-26  Richard Biener  <rguenther@suse.de>
3909         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3911 2013-09-26  Richard Biener  <rguenther@suse.de>
3913         * alias.h (component_uses_parent_alias_set): Rename to ...
3914         (component_uses_parent_alias_set_from): ... this.
3915         * alias.c (component_uses_parent_alias_set): Rename to ...
3916         (component_uses_parent_alias_set_from): ... this and return
3917         the desired parent.
3918         (reference_alias_ptr_type_1): Use the result from
3919         component_uses_parent_alias_set_from instead of stripping
3920         components one at a time.
3921         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3923 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3925         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3926         Move prototypes to...
3927         * tree-ssa-ter.h: New File.  Move prototypes here.
3928         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3929         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3930         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3931         from is_replaceable_p.
3932         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3933         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3934         (process_replaceable): Use ter_is_replaceable_p.
3935         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3936         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3937         newly refactored ssa_is_replaceable_p.
3938         * cfgexpand.c: Include tree-outof-ssa.h.
3939         * ssaexpand.h: Delete.
3941 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3943         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3944         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3945         tree-ssa.c
3946         (create_gimple_tmp): Delete.
3947         (get_expr_type, build_assign, build_type_cast): Move to...
3948         * gimple-builder.c: New File.
3949         (get_expr_type): Relocate from gimple.c.
3950         (build_assign, build_type_cast): Change to only create ssanames.
3951         * gimple.h: Move prototypes to...
3952         * gimple-builder.h: New File. Here.
3953         * tree-ssa.h: And here.
3954         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3955         count_uses_and_derefs): Relocate from gimple.c.
3956         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3957         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3958         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3959         gimple_replace_ssa_lhs.
3960         * asan.c: Include gimple-builder.h.
3961         * Makefile.in: Add gimple-builder.o.
3963 2013-09-26  Richard Biener  <rguenther@suse.de>
3965         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3966         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3967         (loe_visit_block): Use gcc_checking_assert.
3968         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3969         gimple_assign_ssa_name_copy_p.
3970         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3971         change.
3973 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3975         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3976         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3978 2013-09-26  Richard Biener  <rguenther@suse.de>
3980         PR tree-optimization/58539
3981         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3982         the fact that debug statements are not taking part in loop-closed
3983         SSA construction.
3985 2013-09-26  Nick Clifton  <nickc@redhat.com>
3987         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3988         time warning message.
3989         (msp430_print_operand_raw): Delete unused letter parameter.
3990         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3991         (msp430_print_operand_address): New function.
3992         (msp430_print_operand): Move address printing code from here to
3993         new function.
3994         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3995         assembler.
3996         (zero_extendpsisi2): Likewise.
3997         (extendpsisi2): New pattern.
3998         (andneghi3): New pattern.
4000 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4002         * config/aarch64/aarch64.opt (mlra): New option.
4003         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4004         (TARGET_LRA_P): Define.
4006 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4008         * expr.c (expand_assignment): Remove obsolete comment.
4010 2013-09-25  Jeff Law  <law@redhat.com>
4012         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4013         tree-ssa-threadupdate.h.
4014         (register_jump_thread): Similarly.
4015         * tree-ssa-threadupdate.h: New header file.
4016         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4017         * tree-vrp.c: Likewise.
4018         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4019         (thread_around_empty_blocks): Change type of path vector argument to
4020         an edge,type pair from just an edge.  Initialize both elements when
4021         appending to a jump threading path.  Tweak references to elements
4022         appropriately.
4023         (thread_across_edge): Similarly.  Release memory for the elements
4024         as needed.
4025         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4026         (dump_jump_thread_path): New function broken out from
4027         register_jump_thread.
4028         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4029         path vector entries.  Search the path for NULL edges and dump
4030         the path if one is found.  Tweak the conversion of path to 3-edge
4031         form to use the block copy type information embedded in the path.
4033 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4035         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4037 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4038             Vladimir Makarov  <vmakarov@redhat.com>
4040         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4041         from the least significant bit.
4042         (strip_address_mutations): Add bitfield operations handling.
4043         (must_be_index_p): Add shifting and rotate operations handling.
4044         (set_address_base): Use must_be_base_p predicate.
4045         (set_address_index): Use must_be_index_p predicate.
4047 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4048             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4049             Sergey Lega  <sergey.s.lega@intel.com>
4050             Anna Tikhonova  <anna.tikhonova@intel.com>
4051             Ilya Tocar  <ilya.tocar@intel.com>
4052             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4053             Ilya Verbin  <ilya.verbin@intel.com>
4054             Kirill Yukhin  <kirill.yukhin@intel.com>
4055             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4057         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4058         Use new names.
4059         (ix86_expand_vector_move_misalign): Support new unaligned load and
4060         stores and use new names.
4061         (CODE_FOR_sse2_storedqu): Rename to ...
4062         (CODE_FOR_sse2_storedquv16qi): ... this.
4063         (CODE_FOR_sse2_loaddqu): Rename to ...
4064         (CODE_FOR_sse2_loaddquv16qi): ... this.
4065         (CODE_FOR_avx_loaddqu256): Rename to ...
4066         (CODE_FOR_avx_loaddquv32qi): ... this.
4067         (CODE_FOR_avx_storedqu256): Rename to ...
4068         (CODE_FOR_avx_storedquv32qi): ... this.
4069         * config/i386/i386.md (fpint_logic): New.
4070         * config/i386/sse.md (VMOVE): Extend for AVX512.
4071         (VF): Ditto.
4072         (VF_128_256): New.
4073         (VF_512): Ditto.
4074         (VI_UNALIGNED_LOADSTORE): Ditto.
4075         (sse2_avx_avx512f): Ditto.
4076         (sse2_avx2): Extend for AVX512.
4077         (sse4_1_avx2): Ditto.
4078         (avx2_avx512f): New.
4079         (sse): Extend for AVX512.
4080         (sse2): Ditto.
4081         (sse4_1): Ditto.
4082         (avxsizesuffix): Ditto.
4083         (sseintvecmode): Ditto.
4084         (ssePSmode): Ditto.
4085         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4086         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4087         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4088         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4089         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4090         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4091         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4092         (STORENT_MODE): Extend for AVX512.
4093         (*absneg<mode>2): Replace constraint "x" with "v".
4094         (*mul<mode>3): Ditto.
4095         (*ieee_smin<mode>3): Ditto.
4096         (*ieee_smax<mode>3): Ditto.
4097         (avx_cmp<mode>3): Replace VF with VF_128_256.
4098         (*<sse>_maskcmp<mode>3_comm): Ditto.
4099         (<sse>_maskcmp<mode>3): Ditto.
4100         (<sse>_andnot<mode>3): Extend for AVX512.
4101         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4102         (<code><mode>3, fpint_logic): New.
4103         (*<code><mode>3): Extend for AVX512.
4104         (avx512flogicsuff): New.
4105         (avx512f_<logic><mode>): Ditto.
4106         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4107         VF_128_256.
4108         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4109         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4110         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4111         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4112         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4113         (xop_vpermil2<mode>3): Ditto.
4114         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4115         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4116         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4117         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4119 2013-09-25  Tom Tromey  <tromey@redhat.com>
4121         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4122         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4123         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4124         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4125         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4126         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4127         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4128         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4129         (GRAPHITE_HTAB_H): Remove.
4131 2013-09-25  Tom Tromey  <tromey@redhat.com>
4133         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4135 2013-09-25  Tom Tromey  <tromey@redhat.com>
4137         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4139 2013-09-25  Tom Tromey  <tromey@redhat.com>
4141         * config/i386/t-i386 (i386.o): Remove.
4142         (i386-c.o): Use COMPILE and POSTCOMPILE.
4144 2013-09-25  Tom Tromey  <tromey@redhat.com>
4146         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4148 2013-09-25  Tom Tromey  <tromey@redhat.com>
4150         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4151         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4152         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4153         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4154         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4155         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4156         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4157         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4158         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4159         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4160         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4161         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4162         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4163         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4164         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4165         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4166         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4167         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4168         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4169         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4170         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4171         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4172         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4173         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4174         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4175         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4176         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4177         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4178         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4179         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4180         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4181         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4182         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4183         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4184         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4185         (graphite-clast-to-gimple.o, graphite-dependences.o)
4186         (graphite-interchange.o, graphite-poly.o)
4187         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4188         (graphite-optimize-isl.o, tree-vect-loop.o)
4189         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4190         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4191         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4192         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4193         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4194         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4195         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4196         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4197         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4198         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4199         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4200         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4201         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4202         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4203         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4204         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4205         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4206         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4207         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4208         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4209         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4210         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4211         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4212         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4213         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4214         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4215         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4216         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4217         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4218         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4219         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4220         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4221         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4222         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4223         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4224         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4225         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4226         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4227         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4228         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4229         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4230         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4231         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4232         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4233         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4234         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4235         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4236         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4237         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4238         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4239         (gcov-dump.o): Remove.
4240         (default-c.o): Use COMPILE and POSTCOMPILE.
4241         (CFLAGS-gcc.o): New variable.
4242         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4244 2013-09-25  Tom Tromey  <tromey@redhat.com>
4246         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4247         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4248         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4249         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4250         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4251         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4252         (c-family/array-notation-common.o, c-family/stub-objc.o)
4253         (c-family/c-ubsan.o): Remove.
4255 2013-09-25  Tom Tromey  <tromey@redhat.com>
4257         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4259 2013-09-25  Tom Tromey  <tromey@redhat.com>
4261         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4262         to add -DENABLE_SHARED_LIBGCC.
4263         (gcc.o): Don't use subshell.
4265 2013-09-25  Tom Tromey  <tromey@redhat.com>
4267         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4268         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4269         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4270         * configure, config.in: Rebuild.
4272 2013-09-25  Tom Tromey  <tromey@redhat.com>
4274         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4275         (COMPILE, POSTCOMPILE): New variables.
4276         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4277         (DEPFILES): New variable.
4278         Include ".Po" files.
4279         * configure.ac: Add checks for dependency checking.
4280         * configure, aclocal.m4: Regenerate.
4282 2013-09-25  Tom Tromey  <tromey@redhat.com>
4284         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4285         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4287 2013-09-25  Tom Tromey  <tromey@redhat.com>
4289         * Makefile.in (generated_files): Add options.h,
4290         target-hooks-def.h, insn-opinit.h,
4291         common/common-target-hooks-def.h, pass-instances.def,
4292         c-family/c-target-hooks-def.h.
4294 2013-09-25  Jeff Law  <law@redhat.com>
4296         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4297         than foo[foo.length () - 1] to access last member in a vec.
4298         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4300 2013-09-25  Richard Biener  <rguenther@suse.de>
4302         PR middle-end/58521
4303         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4305 2013-09-25  Jan Hubicka  <jh@suse.cz>
4307         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4308         test.
4310 2013-09-25  Marek Polacek  <polacek@redhat.com>
4312         PR sanitizer/58420
4313         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4314         when determining the type name.
4316 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4318         * config/sh/sh.md: Fix formatting.
4320 2013-09-24  Xinliang David Li  <davidxl@google.com>
4322         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4323         max peel iterations parameter.
4324         * param.def: New parameter.
4325         * doc/invoke.texi: Document New parameter.
4327 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4329         * gimple-pretty-print.c: Various whitespace tweaks.
4330         * tree-core.h: Likewise.
4331         * tree-pretty-print.c: Likewise.
4332         * tree-ssa-alias.c: Likewise.
4333         * tree-ssa-copy.c: Likewise.
4334         * tree-ssanames.c: Likewise.
4335         * tree-ssanames.h: Likewise.
4336         * tree-vrp.c: Likewise.
4338 2013-09-24  Alan Modra  <amodra@gmail.com>
4340         PR middle-end/57134
4341         PR middle-end/57586
4342         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4343         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4344         inout operands.
4346 2013-09-24  Richard Biener  <rguenther@suse.de>
4348         PR middle-end/58513
4349         * tree.c (reference_alias_ptr_type): Move ...
4350         * alias.c (reference_alias_ptr_type): ... here and implement
4351         in terms of the new reference_alias_ptr_type_1.
4352         (ref_all_alias_ptr_type_p): New helper.
4353         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4354         use ref_all_alias_ptr_type_p.
4355         (get_deref_alias_set): Add flag_strict_aliasing check here.
4356         (reference_alias_ptr_type_1): New function, split out from ...
4357         (get_alias_set): ... here.
4358         (alias_ptr_types_compatible_p): New function.
4359         * alias.h (reference_alias_ptr_type): Declare.
4360         (alias_ptr_types_compatible_p): Likewise.
4361         * tree.h (reference_alias_ptr_type): Remove.
4362         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4363         to compare MEM_REF alias types.
4365 2013-09-24  Richard Biener  <rguenther@suse.de>
4367         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4369 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4371         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4372         reload helper function arrays into a single array reg_addr.
4373         (reload_fpr_gpr): Likewise.
4374         (reload_gpr_vsx): Likewise.
4375         (reload_vsx_gpr): Likewise.
4376         (struct rs6000_reg_addr): Likewise.
4377         (reg_addr): Likewise.
4378         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4379         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4380         (rs6000_init_hard_regno_mode_ok): Likewise.
4381         (rs6000_secondary_reload_direct_move): Likewise.
4382         (rs6000_secondary_reload): Likewise.
4384         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4385         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4386         power8 changes.  Put wg constraint in alphabetical order.
4388         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4389         for future work to add ISA 2.07 VSX single precision support.
4390         (-mvsx-scalar-double): Change default from -1 to 1, update
4391         documentation comment.
4392         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4393         (-mupper-regs-df): New debug switch to control whether DF values
4394         can go in the traditional Altivec registers.
4395         (-mupper-regs-sf): New debug switch to control whether SF values
4396         can go in the traditional Altivec registers.
4398         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4399         and wy constraints.
4400         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4401         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4402         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4403         (rs6000_debug_legitimate_address_p): Print if we are running
4404         before, during, or after reload.
4405         (rs6000_secondary_reload): Add a comment.
4406         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4408         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4409         constraints.  Update documentation string.
4410         (wd constraint): Likewise.
4411         (wf constraint): Likewise.
4412         (wg constraint): Likewise.
4413         (wn constraint): Likewise.
4414         (ws constraint): Likewise.
4415         (wt constraint): Likewise.
4416         (wx constraint): Likewise.
4417         (wz constraint): Likewise.
4418         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4419         instructions.
4420         (ww constraint): Likewise.
4421         (wy constraint): Likewise.
4422         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4423         in the previous submissions.
4424         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4426 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4428         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4429         (REG_BR_PROB): Say that the probability is stored in an int_list.
4430         * reg-notes.def: Update commentary to mention INT_LIST.
4431         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4432         (INT_LIST): New rtx.
4433         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4434         * rtlanal.c (int_reg_note_p): New function.
4435         (alloc_reg_note): Assert that the note does not have an int argument.
4436         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4437         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4438         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4439         rather than an INSN_LIST.  Handle INT_LIST.
4440         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4441         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4442         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4443         Manipulate them as ints rather than rtxes.
4444         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4445         * regmove.c (copy_src_to_dest): Likewise.
4446         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4448         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4449         into the cases that need it.
4450         * config/arm/arm.c (arm_unwind_emit): Likewise.
4452         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4453         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4454         * loop-doloop.c (add_test, doloop_modify): Likewise.
4455         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4456         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4457         * predict.c (combine_predictions_for_insn): Likewise.
4458         * print-rtl.c (print_rtx): Handle INT_LIST.
4459         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4460         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4461         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4462         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4463         (ix86_split_fp_branch, predict_jump): Likewise.
4464         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4465         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4466         * config/spu/spu.c (ea_load_store_inline): Likewise.
4468         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4469         value of a REG_BR_PROB note.
4470         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4471         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4472         * emit-rtl.c (try_split): Likewise.
4473         * predict.c (br_prob_note_reliable_p): Likewise.
4474         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4475         * reorg.c (mostly_true_jump): Likewise.
4476         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4477         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4478         * config/i386/i386.c (ix86_print_operand): Likewise.
4479         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4480         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4481         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4482         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4483         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4484         * config/sparc/sparc.c (output_cbranch): Likewise.
4485         * config/spu/spu.c (get_branch_target): Likewise.
4486         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4487         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4489 2013-09-23  Jan Hubicka  <jh@suse.cz>
4491         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4492         for ipa-devirt.
4493         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4494         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4495         of external calls
4496         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4497         (gimple_fold_call): Dump inconsistent devirtualizations; add
4498         sanity check for type based devirtualizations.
4499         * ipa-prop.c: Include ipa-utils.h
4500         (ipa_intraprocedural_devirtualization): Add sanity check.
4501         (try_make_edge_direct_virtual_call): Likewise.
4503 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4505         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4506         assignment statements.
4508 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4510         * gimple-pretty-print.c (dump_ssaname_info): New function.
4511         (dump_gimple_phi): Call it.
4512         (pp_gimple_stmt_1): Likewise.
4513         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4514         (range_info_def): Declare.
4515         * tree-pretty-print.c (pp_double_int): New function.
4516         (dump_generic_node): Call it.
4517         * tree-pretty-print.h (pp_double_int): Declare.
4518         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4519         * tree-ssanames.h (range_info_def): New structure.
4520         (value_range_type): Move definition here.
4521         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4522         Declare.
4523         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4524         initialization.
4525         (set_range_info): New function.
4526         (get_range_info): Likewise.
4527         (duplicate_ssa_name_range_info): Likewise.
4528         (duplicate_ssa_name_fn): Check pointer type and call
4529         duplicate_ssa_name_range_info.
4530         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4531         * tree-vrp.c (value_range_type): Remove definition, now in
4532         tree-ssanames.h.
4533         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4534         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4535         (SSA_NAME_RANGE_INFO): New macro.
4537 2013-09-23  Richard Biener  <rguenther@suse.de>
4539         PR tree-optimization/58464
4540         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4541         (phi_trans_add): Change to add conditionally on being not present.
4542         (phi_translate_1): Remove recursion detection here.
4543         (phi_translate): Pre-seed the cache with NULL to catch
4544         recursion here in a more generic way.
4545         (bitmap_find_leader): Adjust comment.
4546         (get_representative_for): Dump value-numbers.
4547         (create_expression_by_pieces): Likewise.
4548         (insert_into_preds_of_block): Likewise.
4550 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4552         PR target/58475
4553         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4554         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4556 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4558         Revert r202780:
4559         2013-09-20  Renlin Li  <renlin.li@arm.com>
4561         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4562         plus_constant.
4563         (aarch64_expand_epilogue): Likewise.
4564         (aarch64_legitimize_reload_address): Likewise.
4566 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4568         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4569         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4571 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4573         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4574         comparison operators when -fno-trapping-math is in effect.
4575         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4576         comparison operators in TFmode and assert that unsupported operators
4577         cannot reach here.
4578         (ia64_print_operand): Likewise.
4580 2013-09-21  Jan Hubicka  <jh@suse.cz>
4582         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4583         (sse_typeless_stores): Enable for core
4584         (sse_load0_by_pxor): Likewise.
4585         (four_jump_limit): Disable for core.
4586         (pad_returns): Likewise.
4587         (avoid_vector_decode): Likewise.
4588         (fuse_cmp_and_branch): Enable for cores.
4589         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4591 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4593         PR middle-end/56791
4594         * config/pa/pa.c (pa_option_override): Disable auto increment and
4595         decrement instructions until reload is completed.
4597         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4598         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4599         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4601 2013-09-20  DJ Delorie  <dj@redhat.com>
4602             Nick Clifton  <nickc@redhat.com>
4604         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4605         (need_to_save): Save bank 0 on interrupts.
4606         (characterize_address): Strip far address wrappers.
4607         (rl78_as_legitimate_address): Likewise.
4608         (transcode_memory_rtx): Likewise.
4609         (rl78_peep_movhi_p): Disable this peephole after devirt.
4610         (rl78_propogate_register_origins): Forget all origins when a
4611         CLOBBER is seen.
4612         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4613         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4614         peephole2's.
4615         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4616         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4617         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4618         relocating.
4619         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4621 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4623         * config/pa/pa.md: In "scc" insn patterns, change output template to
4624         handle const0_rtx in reg_or_0_operand operands.
4626 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4628         PR target/56875
4629         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4630         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4632 2013-09-20  Richard Biener  <rguenther@suse.de>
4634         PR middle-end/58484
4635         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4636         remembering SSA name version and block index.
4637         (new_scev_info_str): Adjust.
4638         (hash_scev_info): Likewise.  Also hash the block index.
4639         (eq_scev_info): Adjust.
4640         (find_var_scev_info): Likewise.
4641         (struct instantiate_cache_entry): Remove.
4642         (struct instantiate_cache_type): Use a htab to map name, block
4643         to chrec.
4644         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4645         (get_instantiated_value_entry): Likewise.
4646         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4647         (instantiate_scev_name): Adjust.
4649 2013-09-20  Jeff Law  <law@redhat.com>
4651         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4653 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4655         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4656         Call aarch64_simd_expand_args to update op[argc].
4658 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4660         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4661         plugin argument.
4663 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4665         * gengtype.c (file_rules): Added rule for *.cc files.
4666         (get_output_file_with_visibility): Give fatal message when no
4667         rules found.
4669 2013-09-20  Renlin Li  <renlin.li@arm.com>
4671         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4672         (aarch64_expand_epilogue): Likewise.
4673         (aarch64_legitimize_reload_address): Likewise.
4675 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4677         PR middle-end/57748
4678         * expr.c (expand_assignment): Remove misalignp code path.
4680 2013-09-20  Marek Polacek  <polacek@redhat.com>
4682         PR sanitizer/58413
4683         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4684         TYPE_PRECISION.  Add asserts.
4686 2013-09-20  Richard Biener  <rguenther@suse.de>
4688         PR tree-optimization/58453
4689         * tree-loop-distribution.c (distribute_loop): Apply the cost
4690         model for -ftree-loop-distribute-patterns, too.
4692 2013-09-20  Richard Biener  <rguenther@suse.de>
4694         PR middle-end/58473
4695         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4696         make type comparison less strict.
4698 2013-09-20  Alan Modra  <amodra@gmail.com>
4700         * configure: Regenerate.
4701         * aclocal.m4: Regenerate.
4703 2013-09-20  Marek Polacek  <polacek@redhat.com>
4705         PR other/58467
4706         * doc/extend.texi: Document that attribute used is meant to be used
4707         on variables with static storage duration.
4709 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4711         PR tree-optimization/58472
4712         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4713         simd_lane_access set inv_p = false.
4714         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4715         the simduid magic VAR_DECL.
4717 2013-09-19  Jan Hubicka  <jh@suse.cz>
4719         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4721 2013-09-17  Jeff Law  <law@redhat.com>
4723         * tree-ssa-dom.c (record_temporary_equivalences): New function
4724         split out of dom_opt_dom_walker::after_dom_children.
4725         (dom_opt_dom_walker::thread_across_edge): Move common code
4726         in here from dom_opt_dom_walker::after_dom_children.
4727         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4729 2013-09-19  Jan Hubicka  <jh@suse.cz>
4731         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4732         (TARGET_GENERIC): Use PROCESOR_GENERIC
4733         (enum processor_type): Unify generic32 and 64.
4734         * i386.md (cpu): Likewise.
4735         * x86-tune.def (use_leave): Enable for generic32.
4736         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4737         * athlon.md: Change generic64 to generic in all occurences.
4738         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4739         (ix86_target_macros_internal): Likewise.
4740         * driver-i386.c (host_detect_local_cpu): Likewise.
4741         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4742         to ..
4743         (generic_memcpy, generic_memset, generic_cost): This one.
4744         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4745         (m_GENERIC32, m_GENERIC64): Remove.
4746         (m_GENERIC): Turn into one flag.
4747         (processor_target): Unify generic tunnings.
4748         (ix86_option_override_internal): Replace generic32/64 by generic.
4749         (ix86_issue_rate): Likewise.
4750         (ix86_adjust_cost): Likewise.
4752 2013-09-19  Jan Hubicka  <jh@suse.cz>
4754         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4755         of speculative flag.
4757 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4759         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4760         GOMP_sections_start, even if !exit_reachable.
4762 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4764         * lra-constraints.c (need_for_all_save_p): Use macro
4765         HARD_REGNO_CALL_PART_CLOBBERED.
4766         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4767         set up pseudo conflict hard regs.
4769 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4771         PR target/58452
4772         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4773         operands.
4775 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4777         PR rtl-optimization/58438
4778         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4779         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4780         from previous subpasses.
4782 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4784         * arm.c (arm_get_frame_offsets): Validate architecture supports
4785         LDRD/STRD before accepting the tuning preference.
4786         (arm_expand_prologue): Likewise.
4787         (arm_expand_epilogue): Likewise.
4789 2013-09-18  Richard Biener  <rguenther@suse.de>
4791         PR tree-optimization/58417
4792         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4793         have chrecs with symbols defined in the loop as operands.
4794         (chrec_fold_multiply): Likewise.
4795         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4796         parameters before folding binary operations.
4797         (struct instantiate_cache_entry_hasher): Remove.
4798         (struct instantiate_cache_type): Use a pointer-map.
4799         (instantiate_cache_type::instantiate_cache_type): New function.
4800         (instantiate_cache_type::get): Likewise.
4801         (instantiate_cache_type::set): Likewise.
4802         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4803         (get_instantiated_value_entry): Likewise.
4804         (global_cache): New global.
4805         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4806         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4807         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4808         (instantiate_scev_name): Adjust.
4809         (instantiate_scev): Construct global instead of local cache.
4810         (resolve_mixers): Likewise.
4812 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4813             Paolo Carlini  <paolo.carlini@oracle.com>
4815         PR c++/58458
4816         * doc/implement-cxx.texi: Fix references to the C++ standards.
4818 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4820         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4821         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4822         simd_lane_access drs, update also DR_ALIGNED_TO.
4824 2013-09-18  Marek Polacek  <polacek@redhat.com>
4826         PR sanitizer/58411
4827         * doc/extend.texi: Document no_sanitize_undefined attribute.
4828         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4829         no_sanitize_undefined attribute.
4831 2013-09-18  Nick Clifton  <nickc@redhat.com>
4833         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4834         (ASM_DECLARE_FUNCTION_NAME): Define.
4836 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4838         * compare-elim.c (find_comparison_dom_walker): New class
4839         (find_comparisons_in_bb): Rename to
4840         find_comparison_dom_walker::before_dom_children
4841         (find_comparisons): Adjust
4842         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4843         adjust.
4844         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4845         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4846         (init_walk_dominator_tree): Remove declaration.
4847         (fini_walk_dominator_tree): Remove declaration.
4848         * fwprop.c (single_def_use_dom_walker): New class
4849         (single_def_use_enter_block): Convert to
4850         single_def_use_dom_walker::before_dom_children.
4851         (single_def_use_leave_block): Convert to
4852         single_def_use_dom_walker::after_dom_children.
4853         (build_single_def_use_links): Adjust.
4854         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4855         class.
4856         (find_candidates_in_block): Convert to
4857         find_candidates_dom_walker::before_dom_children.
4858         (execute_strength_reduction): Adjust.
4859         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4860         (sese_dom_walker): New class.
4861         (sese_dom_walker::sese_dom_walker): New constructor.
4862         (sese_dom_walker::~sese_dom_walker): New destructor.
4863         (build_sese_conditions_before): Convert to
4864         sese_dom_walker::before_dom_children.
4865         (build_sese_conditions_after): Convert to
4866         sese_dom_walker::after_dom_children.
4867         (build_poly_scop): Adjust
4868         * tree-into-ssa.c (rewrite_dom_walker): New class
4869         (rewrite_enter_block): Convert to
4870         rewrite_dom_walker::before_dom_children.
4871         (rewrite_leave_block): Convert to
4872         rewrite_dom_walker::after_dom_children.
4873         (rewrite_update_dom_walker): New class.
4874         (rewrite_update_enter_block): Convert to
4875         rewrite_update_dom_walker::before_dom_children.
4876         (rewrite_update_leave_block): Convert to
4877         rewrite_update_dom_walker::after_dom_children.
4878         (rewrite_blocks, rewrite_into_ssa): Adjust.
4879         (mark_def_dom_walker): New class.
4880         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4881         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4882         (mark_def_sites_blocks): Convert to
4883         mark_def_dom_walker::before_dom_children.
4884         (mark_def_site_blocks): Remove.
4885         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4886         (tree_ssa_dominator_optimize): Adjust.
4887         (dom_thread_across_edge): Convert to method
4888         dom_opt_dom_walker::thread_across_edge.
4889         (dom_opt_enter_block): Convert to member function
4890         dom_opt_dom_walker::before_dom_children.
4891         (dom_opt_leave_block): Convert to member function
4892         dom_opt_dom_walker::after_dom_children.
4893         * tree-ssa-dse.c (dse_dom_walker): New class.
4894         (dse_enter_block): Convert to member function
4895         dse_dom_walker::before_dom_children.
4896         (tree_ssa_dse): Adjust.
4897         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4898         (determine_invariantness_stmt): Convert to method
4899         invariantness_dom_walker::before_dom_children.
4900         (determine_invariantness): Remove
4901         (move_computations_dom_walker): New class.
4902         (move_computations_stmt): Convert to method
4903         move_computations_dom_walker::before_dom_children.
4904         (move_computations, tree_ssa_lim): Adjust.
4905         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4906         (nt_init_block): Make method
4907         notrappping_dom_walker::before_dom_children.
4908         (nt_fini_block): Make
4909         method nontrapping_dom_walker::after_dom_children.
4910         (get_non_trapping): Adjust.
4911         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4912         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4913         (eliminate_leave_block): Make method.
4914         eliminate_dom_walker::after_dom_children.
4915         (eliminate): Adjust
4916         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4917         (strlen_enter_block): Make method
4918         strlen_dom_walker::before_dom_children.
4919         (strlen_leave_block): Make
4920         method strlen_dom_walker::after_dom_children.
4921         (tree_ssa_strlen): Adjust.
4922         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4923         (tree_ssa_uncprop): Adjust.
4924         (uncprop_leave_block): Make method
4925         uncprop_dom_walker::after_dom_children.
4926         (uncprop_leave_block): Make method
4927         uncprop_dom_walker::before_dom_children.
4929 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4931         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4932         branch in same basic block.  Check both src and dest of the move insn.
4934 2013-09-17  Nick Clifton  <nickc@redhat.com>
4936         * config/rl78/rl78-real.md (bf): New pattern.
4937         (bt): New pattern.
4938         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4939         (rl78_print_operand): Do not put a # before a %B.
4940         * config/rl78/rl78.opt: Tweak doc strings.
4942 2013-09-17  DJ Delorie  <dj@redhat.com>
4944         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4945         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4946         (rl78_option_override): Likewise, if -mallregs.
4947         (is_virtual_register): Likewise.
4948         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4949         (REGNO_OK_FOR_BASE_P): Likewise.
4950         * config/rl78/rl78.opt (-mallregs): New.
4952 2013-09-17  Nick Clifton  <nickc@redhat.com>
4954         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4955         For interrupt functions: save all call clobbered registers if the
4956         interrupt handler is not a leaf function.
4957         (rl78_expand_prologue): Always recompute the frame information.
4958         For interrupt functions: only select bank 0 if one of the bank 0
4959         registers is going to be psuhed.
4961 2013-09-17  DJ Delorie  <dj@redhat.com>
4963         * config/rl78/constraints.md: For each W* constraint, rename to C*
4964         and create a W* constraint that checks for an optional ES: prefix
4965         pattern also.
4966         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4967         (es_addr): New.  Used to wrap far addresses.
4968         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4969         (rl78_es_base): New.
4970         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4971         wrapped far addresses.
4972         (rl78_print_operand_1): Unwrap far addresses before processing.
4973         (rl78_lo16): Wrap far addresses in unspecs.
4974         (rl78_es_addr): New.
4975         (rl78_es_base): New.
4976         (insn_ok_now): Check for not-yet-wrapped far addresses.
4977         (transcode_memory_rtx): Properly re-wrap far addresses.
4979 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4981         * config/sparc/t-rtems: Add leon3 multilibs.
4983 2013-09-17  Cong Hou  <congh@google.com>
4985         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4986         when checking the dot production pattern. The type of rhs operand
4987         of multiply is now checked correctly.
4989 2013-09-17  Jeff Law  <law@redhat.com>
4991         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4992         edge implied equivalences into successor phis.
4993         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4994         here from tree-ssa-threadedge.c.
4995         (mark_threaded_blocks): When threading through a joiner, if both
4996         successors of the joiner's clone reach the same block, verify the
4997         PHI arguments are equal.  If not, cancel the jump threading request.
4998         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4999         tree-ssa-threadupdate.c
5000         (thread_across_edge): Don't check PHI argument equality when
5001         threading through joiner block here.
5003 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5005         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5006         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5007         (warn_uninit, warn_uninitialized_vars,
5008         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5009         Move to tree-ssa-uninit.c.
5010         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5011         (has_undefined_value_p): New.  Pass dependant parts of
5012         ssa_undefined_value_p.
5013         (uninit_undefined_value_p): Use has_undefined_value_p.
5014         (warn_uninit, warn_uninitialized_vars,
5015         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5016         Move from tree-ssa.c.
5017         * tree-ssa.h: Adjust prototypes.
5019 2013-09-17  Jan Hubicka  <jh@suse.cz>
5021         PR middle-end/58332
5022         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5023         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5024         FUNCTION_NOT_OPTIMIZED.
5025         * ipa-inline-analysis.c (compute_inline_parameters): Function
5026         not optimized is not inlinable unless it is alwaysinline.
5027         (inline_analyze_function): Force calls in not optimized
5028         function not inlinable.
5030 2013-09-17  Jan Hubicka  <jh@suse.cz>
5032         PR middle-end/58329
5033         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5034         to return NULL.
5035         * ipa.c (function_and_variable_visibility): Likewise.
5036         * ipa-profile.c (ipa_profile): Likewise.
5038 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5040         PR ipa/58398
5041         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5042         attribute, and don't inline the resolver in this case.
5044 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5046         * coverage.c (get_coverage_counts): Add missing newline.
5048 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5050         PR tree-optimization/58088
5051         * fold-const.c (mask_with_trailing_zeros): New function.
5052         (fold_binary_loc): Make sure we don't recurse infinitely
5053         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5054         Use mask_with_trailing_zeros where appropriate.
5056 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5058         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5059         of 'prev' var to get better distance estimation.
5061 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5063         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5064         * tree-inline.c (is_parameter_of): New predicate.
5065         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5066         a parameter has been remapped.
5067         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5068         (optimize_inline_calls): Initialize transform_parameter.
5069         (copy_gimple_seq_and_replace_locals): Likewise.
5070         (tree_function_versioning): Likewise.
5071         (maybe_inline_call_in_expr): Likewise.
5073 2013-09-17  Nick Clifton  <nickc@redhat.com>
5075         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5076         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5077         interrupt handlers.
5078         (is_attr_func): New function.
5079         (msp430_is_interrupt_func): New function.
5080         (is_naked_func): New function.
5081         (is_reentrant_func): New function.
5082         (is_critical_func): New function.
5083         (msp430_start_function): Add annotations for function attributes.
5084         (msp430_attr): New function.
5085         (msp430_attribute_table): New.
5086         (msp430_function_section): New function.
5087         (TARGET_ASM_FUNCTION_SECTION): Define.
5088         (msp430_builtin): New enum.
5089         (msp430_init_builtins): New function.
5090         (msp430_builtin_devl): New function.
5091         (msp430_expand_builtin): New function.
5092         (TARGET_INIT_BUILTINS): Define.
5093         (TARGET_EXPAND_BUILTINS): Define.
5094         (TARGET_BUILTIN_DECL): Define.
5095         (msp430_expand_prologue): Add support for naked, interrupt,
5096         critical and reentrant functions.
5097         (msp430_expand_epilogue): Likewise.
5098         (msp430_print_operand): Handle 'O' character.
5099         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5100         NO_TRAMPOLINES.
5101         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5102         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5103         (pushm): Use a 'n' rather than an 'i' constraint.
5104         (msp_return): Add generation of the interrupt return instruction.
5105         (disable_interrupts): New pattern.
5106         (enable_interrupts): New pattern.
5107         (push_intr_state): New pattern.
5108         (pop_intr_state): New pattern.
5109         (bic_SR): New pattern.
5110         (bis_SR): New pattern.
5111         * doc/extend.texi: Document MSP430 function attributes and builtin
5112         functions.
5114 2013-09-17  Richard Biener  <rguenther@suse.de>
5116         PR tree-optimization/58432
5117         * tree-loop-distribution.c (tree_loop_distribution): Also
5118         scan PHIs for outside loop uses and seed a partition from them.
5120 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5122         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5123         (restructure_reference): Call backtrace_base_for_ref.
5125 2013-09-17  Alan Modra  <amodra@gmail.com>
5127         PR target/57589
5128         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5129         patch.
5131 2013-09-16  DJ Delorie  <dj@redhat.com>
5133         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5134         vregs location for RL78/G10.
5135         (rl78_expand_prologue): Avoid SEL on G10.
5136         (rl78_expand_epilogue): Likewise.
5137         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5138         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5139         __RL78_G10__ when appropriate.
5140         (ASM_SPEC): Pass -mg10 along to the assembler.
5141         * config/rl78/rl78.md (sel_rb): Disable for G10.
5142         * config/rl78/rl78.opt: Add -mg10 option.
5143         * config/rl78/t-rl78: Add -mg10 multilib.
5145 2013-09-16  Xinliang David Li  <davidxl@google.com>
5147         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5148         * omp-low.c (omp_max_vf): Ditto.
5149         (expand_omp_simd): Ditto.
5150         * tree-vectorizer.c (vectorize_loops): Ditto.
5151         (gate_vect_slp): Ditto.
5152         (gate_increase_alignment): Ditto.
5153         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5154         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5155         (gate_tree_vectorize): Name change.
5156         (tree_vectorize): Ditto.
5157         (pass_vectorize::gate): Call new function.
5158         (pass_vectorize::execute): Ditto.
5159         * opts.c: O3 default setting change.
5160         (finish_options): Check new flag.
5161         * doc/invoke.texi: Document new flags.
5162         * common.opt: New flags.
5164 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5166         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5167         (Condition Code): Don't reference it.
5168         * doc/tm.texi: Regenerate.
5170 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5172         PR middle-end/58418
5173         * lra-constraints.c (undo_optional_reloads): Consider all optional
5174         reload even if it did not get a hard reg.
5176 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5178         * dumpfile.c (dump_loc): Remove newline emission.
5179         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5180         emission to dump_printf_loc calls where missing.
5181         (vect_mark_for_runtime_alias_test): Ditto.
5182         (vect_analyze_data_ref_dependence): Ditto.
5183         (vect_analyze_data_ref_dependences): Ditto.
5184         (vect_slp_analyze_data_ref_dependence): Ditto.
5185         (vect_slp_analyze_data_ref_dependences): Ditto.
5186         (vect_compute_data_ref_alignment): Ditto.
5187         (vect_update_misalignment_for_peel): Ditto.
5188         (vect_verify_datarefs_alignment): Ditto.
5189         (vector_alignment_reachable_p): Ditto.
5190         (vect_get_data_access_cost): Ditto.
5191         (vect_enhance_data_refs_alignment): Ditto.
5192         (vect_find_same_alignment_drs): Ditto.
5193         (vect_analyze_data_refs_alignment): Ditto.
5194         (vect_analyze_group_access): Ditto.
5195         (vect_analyze_data_ref_access): Ditto.
5196         (vect_analyze_data_ref_accesses): Ditto.
5197         (vect_prune_runtime_alias_test_list): Ditto.
5198         (vect_analyze_data_refs): Ditto.
5199         (vect_create_addr_base_for_vector_ref): Ditto.
5200         (vect_create_data_ref_ptr): Ditto.
5201         (vect_grouped_store_supported): Ditto.
5202         (vect_grouped_load_supported): Ditto.
5203         * value-prof.c (check_counter): Ditto.
5204         (check_ic_target): Ditto.
5205         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5206         (vect_recog_widen_mult_pattern): Ditto.
5207         (vect_recog_widen_sum_pattern): Ditto.
5208         (vect_recog_over_widening_pattern): Ditto.
5209         (vect_recog_widen_shift_pattern): Ditto.
5210         (vect_recog_rotate_pattern): Ditto.
5211         (vect_recog_vector_vector_shift_pattern): Ditto.
5212         (vect_recog_divmod_pattern): Ditto.
5213         (vect_recog_mixed_size_cond_pattern): Ditto.
5214         (vect_recog_bool_pattern): Ditto.
5215         (vect_pattern_recog_1): Ditto.
5216         (vect_pattern_recog): Ditto.
5217         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5218         (vect_is_simple_iv_evolution): Ditto.
5219         (vect_analyze_scalar_cycles_1): Ditto.
5220         (vect_get_loop_niters): Ditto.
5221         (vect_analyze_loop_1): Ditto.
5222         (vect_analyze_loop_form): Ditto.
5223         (vect_analyze_loop_operations): Ditto.
5224         (vect_analyze_loop_2): Ditto.
5225         (vect_analyze_loop): Ditto.
5226         (report_vect_op): Ditto.
5227         (vect_is_slp_reduction): Ditto.
5228         (vect_is_simple_reduction_1): Ditto.
5229         (vect_get_known_peeling_cost): Ditto.
5230         (vect_estimate_min_profitable_iters): Ditto.
5231         (vect_model_reduction_cost): Ditto.
5232         (vect_model_induction_cost): Ditto.
5233         (get_initial_def_for_induction): Ditto.
5234         (vect_create_epilog_for_reduction): Ditto.
5235         (vectorizable_reduction): Ditto.
5236         (vectorizable_induction): Ditto.
5237         (vectorizable_live_operation): Ditto.
5238         (vect_loop_kill_debug_uses): Ditto.
5239         (vect_transform_loop): Ditto.
5240         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5241         (vect_stmt_relevant_p): Ditto.
5242         (process_use): Ditto.
5243         (vect_mark_stmts_to_be_vectorized): Ditto.
5244         (vect_model_simple_cost): Ditto.
5245         (vect_model_promotion_demotion_cost): Ditto.
5246         (vect_model_store_cost): Ditto.
5247         (vect_get_store_cost): Ditto.
5248         (vect_model_load_cost): Ditto.
5249         (vect_get_load_cost): Ditto.
5250         (vect_init_vector_1): Ditto.
5251         (vect_get_vec_def_for_operand): Ditto.
5252         (vect_finish_stmt_generation): Ditto.
5253         (vectorizable_call): Ditto.
5254         (vectorizable_conversion): Ditto.
5255         (vectorizable_assignment): Ditto.
5256         (vectorizable_shift): Ditto.
5257         (vectorizable_operation): Ditto.
5258         (vectorizable_store): Ditto.
5259         (vectorizable_load): Ditto.
5260         (vectorizable_condition): Ditto.
5261         (vect_analyze_stmt): Ditto.
5262         (vect_transform_stmt): Ditto.
5263         (vect_is_simple_use): Ditto.
5264         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5265         (vect_can_advance_ivs_p): Ditto.
5266         (vect_update_ivs_after_vectorizer): Ditto.
5267         (vect_do_peeling_for_loop_bound): Ditto.
5268         (vect_gen_niters_for_prolog_loop): Ditto.
5269         (vect_update_inits_of_drs): Ditto.
5270         (vect_create_cond_for_alias_checks): Ditto.
5271         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5272         (vect_build_slp_tree_1): Ditto.
5273         (vect_supported_load_permutation_p): Ditto.
5274         (vect_analyze_slp_instance): Ditto.
5275         (vect_analyze_slp): Ditto.
5276         (vect_make_slp_decision): Ditto.
5277         (vect_detect_hybrid_slp): Ditto.
5278         (vect_bb_vectorization_profitable_p): Ditto.
5279         (vect_slp_analyze_bb_1): Ditto.
5280         (vect_update_slp_costs_according_to_vf): Ditto.
5281         (vect_get_mask_element): Ditto.
5282         (vect_transform_slp_perm_load): Ditto.
5283         (vect_schedule_slp_instance): Ditto.
5284         (vect_schedule_slp): Ditto.
5285         (vect_slp_transform_bb): Ditto.
5286         * profile.c (read_profile_edge_counts): Ditto.
5287         (compute_branch_probabilities): Ditto.
5288         * coverage.c (get_coverage_counts): Ditto.
5290 2013-09-16  Diego Novillo  <dnovillo@google.com>
5292         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5294 2013-09-16  Jan Hubicka  <jh@suse.cz>
5296         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5297         abstract functions; for static functions check the presence of body.
5299 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5301         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5302         * config/aarch64/aarch64-simd.md
5303         (aarch64_mla_elt<mode>): New.
5304         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5305         (aarch64_mls_elt<mode>): Likewise.
5306         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5307         (aarch64_fma4_elt<mode>): Likewise.
5308         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5309         (aarch64_fma4_elt_to_128v2df): Likewise.
5310         (aarch64_fma4_elt_to_64df): Likewise.
5311         (fnma<mode>4): Likewise.
5312         (aarch64_fnma4_elt<mode>): Likewise.
5313         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5314         (aarch64_fnma4_elt_to_128v2df): Likewise.
5315         (aarch64_fnma4_elt_to_64df): Likewise.
5316         * config/aarch64/iterators.md (VDQSF): New.
5317         * config/aarch64/arm_neon.h
5318         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5319         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5321 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5323         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5324         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5325         (aarch64_mul3_elt_to_128df): Likewise.
5326         (aarch64_mul3_elt_to_64v2df): Likewise.
5327         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5328         (VMUL): New.
5329         (VMUL_CHANGE_NLANES) Likewise.
5330         (h_con): Likewise.
5331         (f): Likewise.
5332         * config/aarch64/arm_neon.h
5333         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5335 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5337         * config/aarch64/arm_neon.h
5338         (vcvtx_high_f32_f64): Fix parameters.
5340 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5341             Uros Bizjak  <ubizjak@gmail.com>
5343         * config/alpha.c: Include tree-ssanames.h.
5345 2013-09-16  Richard Biener  <rguenther@suse.de>
5347         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5348         (dot_rdg_1): Handle control_dd.
5349         (create_edge_for_control_dependence): New function.
5350         (create_rdg_edges): Add control dependences if asked for.
5351         (build_rdg): Likewise.
5352         (generate_loops_for_partition): If there are not necessary
5353         control stmts remove all their dependencies.
5354         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5355         (distribute_loop): Pass on control dependences.
5356         (tree_loop_distribution): Compute control dependences and remove
5357         restriction on number of loop nodes.
5359 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5361         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5362         for internal calls.
5364 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5366         * cse.c (try_const_anchors): Punt on CC modes.
5368 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5370         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5372 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5374         PR target/58382
5375         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5376         calls to word_mode.
5378 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5380         PR target/48094
5381         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5382         seen.
5383         (darwin_objc1_section): Likewise.
5384         (darwin_file_end): Emit Image Info section when required.
5386 2013-09-14  Jan Hubicka  <jh@suse.cz>
5388         * tree-into-ssa.c (gate_into_ssa): New.
5389         (pass_data_build_ssa): Use it.
5390         * cgraph.h (expand_thunk): Update prototype.
5391         * cgraphunit.c (analyze_function): Expand thunks early.
5392         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5393         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5394         set lowered flag; do not add new function.
5395         (assemble_thunks_and_aliases): Update.
5396         * tree-ssa.c (gate_init_datastructures): New gate.
5397         (pass_data_init_datastructures): Use it.
5399 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5401         PR target/58269
5402         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5403         xmm register set described in the psABI.
5405 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5407         * dwarf2out.c (should_emit_struct_debug): Add check
5408         for type_decl variable is not NULL.
5410 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5412         * config.gcc: Use new winnt-c.c target hooks
5413         * config/t-winnt: New file
5414         * config/winnt-c.c: New file
5415         * doc/tm.texi.in: Document new hook
5416         * doc/tm.texi: Regenerated
5418 2013-09-13  Jan Hubicka  <jh@suse.cz>
5420         PR middle-end/58094
5421         * ipa-inline.c (check_callers): New function.
5422         (check_caller_edge): Remove.
5423         (want_inline_function_to_all_callers_p): Also permit alises that are
5424         called dirrectly.
5425         (inline_to_all_callers): Terminate the walk when devirtualization
5426         introduce new calls.
5428 2013-09-13  Jan Hubicka  <jh@suse.cz>
5430         * ipa-inline-analysis.c (struct growth_data): Add node.
5431         (do_estimate_growth_1): Fix detection of recursion.
5433 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5435         PR tree-optimization/58392
5436         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5437         to avoid shadowing of outer loop variable.  If
5438         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5439         replace_by_duplicate_decl simduid of loops that have it set and
5440         set dest_cfun->has_simduid_loops and/or
5441         dest_cfun->has_force_vect_loops.
5442         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5443         instead of maybe_lookup_decl.
5444         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5445         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5446         call.  Copy over force_vect and copy and remap simduid.  Set
5447         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5448         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5449         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5450         caller.  Don't set cfun->has_simduid_loops and/or
5451         cfun->has_force_vect_loops here.
5452         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5453         caller.
5454         (expand_call_inline, tree_function_versioning): Adjust copy_body
5455         callers.
5457 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5459         PR bootstrap/58388
5460         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5461         the assert if the edge was a speculative one.
5463 2013-09-13  Richard Biener  <rguenther@suse.de>
5465         * tree-data-ref.h (known_dependences_p): Move here ...
5466         * tree-loop-distribution.c (known_dependences_p): ... from here.
5467         (dump_rdg_component, debug_rdg_component): Remove.
5468         (dump_rdg): Adjust.
5469         (generate_loops_for_partition): Use gimple_uid instead of
5470         relying on matching stmt visit order.
5471         (rdg_build_partitions): Take starting stmt vector.
5472         (ldist_gen): Merge into ...
5473         (distribute_loop): ... this function.  Do not compute starting
5474         vertices vector.
5475         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5477 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5479         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5480         Set type attribute correctly. Set predicable_short_it attribute.
5481         (cmpsi_shiftsi): Remove %? from output template.
5483 2013-09-13  Richard Biener  <rguenther@suse.de>
5485         * tree-loop-distribution.c (struct rdg_component,
5486         rdg_defs_used_in_other_loops_p, free_rdg_components,
5487         rdg_build_components): Remove.
5488         (stmts_from_loop): Do not record virtual PHIs.
5489         (generate_loops_for_partition): Skip virtual PHIs.
5490         (build_rdg_partition_for_component): Rename to ...
5491         (build_rdg_partition_for_vertex): ... this and adjust.
5492         (rdg_build_partitions): Take a vector of starting vertices
5493         instead of components.  Remove unnecessary leftover handling.
5494         (ldist_gen): Do not build components or record other stores.
5495         (distribute_loop): Do not distribute loops containing stmts
5496         with side-effects.
5498 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5500         PR target/58314
5501         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5503 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5505         * config.gcc: Separate cases for mingw and cygwin targets,
5506         and add 64-bit cygwin target case.
5508         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5509         dll-export inline-functions.
5510         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5513 2013-09-13  Jeff Law  <law@redhat.com>
5515         PR middle-end/58387
5516         Revert:
5517         2013-09-06  Jeff Law  <law@redhat.com>
5519         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5520         edge implied equivalences into successor phis.
5522 2013-09-12  DJ Delorie  <dj@redhat.com>
5524         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5525         separators.
5527 2013-09-12  Brooks Moses  <bmoses@google.com>
5529         PR driver/42955
5530         * Makefile.in: Do not install driver binaries in $(target)/bin.
5532 2013-09-12  DJ Delorie  <dj@redhat.com>
5534         * config/rl78/rl78.opt (mrelax): New.
5535         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5536         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5538         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5539         between SP and FP.
5540         (rl78_expand_epilogue): Likewise.
5542 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5544         PR middle-end/58335
5545         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5546         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5547         pointer elimination with using remove_reg_equal_offset_note.
5549 2013-09-12  DJ Delorie  <dj@redhat.com>
5551         * config/msp430/: New port.
5552         * config.gcc (msp430): Added.
5553         * doc/invoke.texi: Document MSP430 options.
5554         * doc/install.texi: Document msp430-elf
5555         * doc/md.texi: Document msp430-elf
5556         * doc/contrib.texi: Document msp430-elf
5558         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5559         PSImode.
5561 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5563         PR ipa/58389
5564         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5565         reference descriptor is NULL.
5566         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5567         edge to NULL.
5569 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5571         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5572         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5573         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5574         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5575         make_temp_ssa_name): move to tree-ssanames.h
5576         * tree-ssa-alias.h: Move prototype.
5577         * tree-ssa.h: Include tree-ssanames.h.
5578         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5579         * tree-ssanames.h: New.  Move items from tree-flow*.h
5580         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5582 2013-09-12  Richard Biener  <rguenther@suse.de>
5584         PR tree-optimization/58404
5585         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5586         propagate non-invariant addresses into dereferences wrapped
5587         in component references.
5589 2013-09-12  Richard Biener  <rguenther@suse.de>
5591         PR tree-optimization/58402
5592         * passes.def: Move pass_late_warn_uninitialized later.
5594 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5596         * tree-ssa.h: New.  Move content from tree-flow.h and
5597         tree-flow-inline.h.
5598         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5599         Move prototypes belonging to tree-ssa.c.
5600         * tree-flow-inline.h (redirect_edge_var_map_def,
5601         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5602         tree-ssa.h.
5603         * gimple.h: Adjust prototypes.
5604         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5605         to...
5606         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5607         * tree.h: Move prototype to tree-ssa.h.
5608         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5609         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5610         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5611         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5612         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5613         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5614         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5615         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5616         graphite-blocking.c, graphite-clast-to-gimple.c,
5617         graphite-dependences.c, graphite-interchange.c,
5618         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5619         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5620         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5621         ipa-reference.c, ipa-split.c, ipa-utils.c,
5622         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5623         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5624         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5625         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5626         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5627         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5628         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5629         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5630         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5631         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5632         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5633         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5634         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5635         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5636         value-prof.c, var-tracking.c,
5637         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5639 2013-09-12  Richard Biener  <rguenther@suse.de>
5641         PR tree-optimization/58396
5642         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5643         (build_rdg): Take a loop-nest parameter, fix memleaks.
5644         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5646 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5648         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5649         for SLM.
5651 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5653         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5654         The FMA instruction names should have a 'v' prefix.
5656 2013-09-12  Richard Biener  <rguenther@suse.de>
5658         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5659         (dot_rdg): Use popen instead of system in optional code.
5660         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5661         (already_processed_vertex_p): Adjust.
5662         (has_anti_or_output_dependence, predecessor_has_mem_write,
5663         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5664         rdg_flag_uses): Remove.
5665         (rdg_flag_vertex): Simplify.
5666         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5667         remove recursion.
5668         (build_rdg_partition_for_component): Process the first vertex
5669         of a component only.
5670         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5672 2013-09-12  Alan Modra  <amodra@gmail.com>
5674         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5676 2013-09-11  DJ Delorie  <dj@redhat.com>
5677             Nick Clifton  <nickc@redhat.com>
5679         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5680         (rl78_stack_based_mem): New.
5681         * config/rl78/constraints.md (Iv08): New.
5682         (Iv16): New.
5683         (Iv24): New.
5684         (Is09): New.
5685         (Is17): New.
5686         (Is25): New.
5687         (ISsi): New.
5688         (IShi): New.
5689         (ISqi): New.
5690         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5691         (movhi): Likewise.
5692         (movsi): Change from expand to insn-and-split.
5693         (ashrsi3): Clobber AX.
5694         (lshrsi3): New.
5695         (ashlsi3): New.
5696         (cbranchsi4): New.
5697         * config/rl78/rl78.md (CC_REG): Fix.
5698         (addsi3): Allow memory and immediate operands.
5699         (addsi3_internal): Split into...
5700         (addsi3_internal_virt): ...new, and ...
5701         (addsi3_internal_real): ...new.
5702         (subsi): New.
5703         (subsi3_internal_virt): New.
5704         (subsi3_internal_real): New.
5705         (mulsi3): Add memory operand.
5706         (mulsi3_rl78): Likewise.
5707         (mulsi3_g13): Likewise.
5708         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5709         (cbranchqi4_real): Add more constraint options.
5710         (cbranchhi4_real): Expand pattern.
5711         (cbranchhi4_real_signed): New.
5712         (cbranchhi4_real_inverted): New.
5713         (cbranchsi4_real_lt): New.
5714         (cbranchsi4_real_ge): New.
5715         (cbranchsi4_real_signed): New.
5716         (cbranchsi4_real): New.
5717         (peephole2): New.
5718         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5719         constant shifts.
5720         (lshrsi3_virt): Likewise.
5721         (ashlsi3_virt): Likewise.
5722         (cbranchqi4_virt_signed): New.
5723         (cbranchhi4_virt_signed): New.
5724         (cbranchsi4_virt): New.
5725         * config/rl78/rl78.c: Whitespace fixes throughout.
5726         (move_elim_pass): New.
5727         (pass_data_rl78_move_elim): New.
5728         (pass_rl78_move_elim): New.
5729         (make_pass_rl78_move_elim): New.
5730         (rl78_devirt_info): Run devirt earlier.
5731         (rl78_move_elim_info): New.
5732         (rl78_asm_file_start): Register it.
5733         (rl78_split_movsi): New.
5734         (rl78_as_legitimate_address): Allow virtual base registers when
5735         appropriate.
5736         (rl78_addr_space_convert): Remove spurious debug stuff.
5737         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5738         (rl78_print_operand): More cases for not printing '#'.
5739         (rl78_expand_compare): Remove most of the logic.
5740         (content_memory): New.
5741         (clear_content_memory): New.
5742         (get_content_index): New.
5743         (get_content_name): New.
5744         (display_content_memory): New.
5745         (update_content): New.
5746         (record_content): New.
5747         (already_contains): New.
5748         (insn_ok_now): Re-recog insns with virtual registers.
5749         (add_postponed_content_update): New.
5750         (process_postponed_content_update): New.
5751         (gen_and_emit_move): New.
5752         (transcode_memory_rtx): Record new location content.
5753         Use gen_and_emit_move.
5754         (force_into_acc): New.
5755         (move_to_acc): Use gen_and_emit_move.
5756         (move_from_acc): Likewise.
5757         (move_acc_to_reg): Likewise.
5758         (move_to_x): Likewise.
5759         (move_to_hl): Likewise.
5760         (move_to_de): Likewise.
5761         (rl78_alloc_physical_registers_op1): Record location content.
5762         (has_constraint): New.
5763         (rl78_alloc_physical_registers_op2): Record location content.
5764         Optimize use of HL.
5765         (rl78_alloc_physical_registers_ro1): Likewise.
5766         (rl78_alloc_physical_registers_cmp): Likewise.
5767         (rl78_alloc_physical_registers_umul): Likewise.
5768         (rl78_alloc_address_registers_macax): New.
5769         (rl78_alloc_physical_registers): Initialize and set location
5770         content memory as needed.
5771         (rl78_reorg): Make sure split2 is called.
5772         (rl78_rtx_costs): New.
5774 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5776         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5777         for (not (neg ...)) and (neg (not ...)) cases.
5779 2013-09-11  Richard Biener  <rguenther@suse.de>
5781         PR middle-end/58377
5782         * passes.def: Split critical edges before late uninit warning passes.
5783         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5785 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5787         PR tree-optimization/58385
5788         * fold-const.c (build_range_check): If both low and high are NULL,
5789         use omit_one_operand_loc to preserve exp side-effects.
5791 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5793         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5795 2013-09-11  Richard Biener  <rguenther@suse.de>
5797         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5798         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5799         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5800         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5801         stmts_from_loop, known_dependences_p, build_empty_rdg,
5802         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5803         * tree-loop-distribution.c: ... here.
5804         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5805         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5806         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5807         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5808         * tree-loop-distribution.c: ... here.
5809         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5810         (struct partition_s): Add loops member.
5811         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5812         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5813         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5815 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5816             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5817             Sergey Lega  <sergey.s.lega@intel.com>
5818             Anna Tikhonova  <anna.tikhonova@intel.com>
5819             Ilya Tocar  <ilya.tocar@intel.com>
5820             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5821             Ilya Verbin  <ilya.verbin@intel.com>
5822             Kirill Yukhin  <kirill.yukhin@intel.com>
5823             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5825         * config/i386/constraints.md (k): New.
5826         (Yk): Ditto.
5827         * config/i386/i386.c (const regclass_map): Add new mask registers.
5828         (dbx_register_map): Ditto.
5829         (dbx64_register_map): Ditto.
5830         (svr4_dbx_register_map): Ditto.
5831         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5832         disabled.
5833         (ix86_preferred_reload_class): Disable constants for mask registers.
5834         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5835         (ix86_hard_regno_mode_ok): Support new mask registers.
5836         (x86_order_regs_for_local_alloc): Ditto.
5837         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5838         (FIXED_REGISTERS): Add new mask registers.
5839         (CALL_USED_REGISTERS): Ditto.
5840         (REG_ALLOC_ORDER): Ditto.
5841         (VALID_MASK_REG_MODE): New.
5842         (FIRST_MASK_REG): Ditto.
5843         (LAST_MASK_REG): Ditto.
5844         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5845         (MAYBE_MASK_CLASS_P): New.
5846         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5847         (REG_CLASS_CONTENTS): Ditto.
5848         (MASK_REGNO_P): New.
5849         (ANY_MASK_REG_P): Ditto.
5850         (HI_REGISTER_NAMES): Add new mask registers.
5851         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5852         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5853         mask registers.
5854         (attribute "type"): Add mskmov, msklog.
5855         (attribute "length_immediate"): Support them.
5856         (attribute "memory"): Ditto.
5857         (attribute "prefix_0f"): Ditto.
5858         (*movhi_internal): Support new mask registers.
5859         (*movqi_internal): Ditto.
5860         (define_split): Split out clobber pattern is a logic
5861         insn on mask registers.
5862         (*k<logic><mode>): New.
5863         (*andhi_1): Extend to support mask regs.
5864         (*andqi_1): Extend to support mask regs.
5865         (kandn<mode>): New.
5866         (define_split): Split and-not to and and not if operands
5867         are not mask regs.
5868         (*<code><mode>_1): Separate HI mode to new pattern...
5869         (*<code>hi_1): This.
5870         (*<code>qi_1): Extend to support mask regs.
5871         (kxnor<mode>): New.
5872         (kortestzhi): Ditto.
5873         (kortestchi): Ditto.
5874         (kunpckhi): Ditto.
5875         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5876         (*one_cmplhi2_1): ...Here, now with mask registers support.
5877         (*one_cmplqi2_1): Support new mask registers.
5878         (HI/QImode arithmetics splitter): Don't split if mask registers
5879         are used.
5880         (HI/QImode not splitter): Ditto.
5881         * config/i386/predicated.md (mask_reg_operand): New.
5882         (general_reg_operand): Ditto.
5884 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5886         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5887         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5889 2013-09-10  Jeff Law  <law@redhat.com>
5891         PR tree-optimization/58380
5892         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5893         of threading through a buried loop header.
5895         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5896         return value for single successor case.
5898 2013-09-10  Jan Hubicka  <jh@suse.cz>
5900         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5902 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5904         PR target/58361
5905         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5906         support conditional execution.
5907         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5909 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5911         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5912         constraint pass.
5913         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5914         Check destination too to check move insn.
5915         (undo_optional_reloads): Add check that the original peudo did not
5916         changed its allocation and the optional reload was inherited on last
5917         inheritance pass.  Break loop after deciding to keep optional reload.
5918         (lra_undo_inheritance): Add check that inherited pseudo still in
5919         memory.
5921 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5923         * config/aarch64/aarch64.md (generic_sched): New.
5924         * config/aarch64/aarch64-generic.md (load): Make conditional
5925         on generic_sched attribute.
5926         (nonload): Likewise.
5928 2013-09-10  Jan Hubicka  <jh@suse.cz>
5930         * lto-cgraph.c: Include ipa-utils.h.
5931         (compute_ltrans_boundary): Also add possible targets into the boundary.
5933 2013-09-10  Jan Hubicka  <jh@suse.cz>
5935         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5936         VAR_DECL of vtable rather than full expression.
5938 2013-09-10  Jan Hubicka  <jh@suse.cz>
5939             Paolo Carlini  <paolo.carlini@oracle.com>
5941         * cgraphunit.c (analyze_functions): Save input_location, set it
5942         to UNKNOWN_LOCATION and restore it at the end.
5944 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5946         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5947         represented by a thunk.
5949 2013-09-10  Jeff Law  <law@redhat.com>
5951         PR tree-optimization/58343
5952         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5953         jump threading requests through loop headers buried in the middle
5954         of a jump threading path.
5956         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5957         in return value/type.
5959 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5961         PR rtl-optimization/58365
5962         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5963         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5964         it differs.
5966 2013-09-10  Richard Biener  <rguenther@suse.de>
5968         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5969         * tree-data-ref.c (create_rdg_vertices): Collect all data
5970         references, signal failure to the caller, use data-ref API.
5971         (build_rdg): Compute data references only once.  Maintain lifetime
5972         of data references and data dependences from within RDG.
5973         (free_rdg): Free dependence relations.
5974         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5975         inventing extra dependences.
5976         (distribute_loop): Update for RDG API changes.
5978 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5980         * doc/invoke.texi (fms-extensions): Document changed
5981         behavior for ms-abi targets.
5982         * config/i386/i386.c (ix86_option_override_internal):
5983         Set default value of option -fms-extension for ms-abi targets.
5985 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5987         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5989 2013-09-10  Alan Modra  <amodra@gmail.com>
5991         PR target/58330
5992         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5994 2013-09-10  Alan Modra  <amodra@gmail.com>
5996         * config/rs6000/predicates.md (add_cint_operand): New.
5997         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5998         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5999         using add_cint_operand.
6000         (largetoc_high_plus_aix): Likewise.
6002 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6004         PR tree-optimization/58364
6005         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6006         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6007         the current range can't be an unconditional true or false.
6009 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6011         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6013 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6015         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6017 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6019         PR c++/43452
6020         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6022 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6024         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6025         NO_REGS for immediate that can't be moved directly into FP_REGS.
6027 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6029         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6030         comparison with negated operand.
6031         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6032         RTL form.
6034 2013-09-09  Richard Biener  <rguenther@suse.de>
6036         PR middle-end/58326
6037         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6038         of a subloop record all its block as affecting loop-closed SSA form.
6040 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6042         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6043         of an rtx/bitpos pair.
6044         (store_fixed_bit_field): Update accordingly.
6046 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6048         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6049         GEN_INT.
6050         * builtins.c (expand_errno_check): Likewise.
6051         * dwarf2cfi.c (init_return_column_size): Likewise.
6052         * except.c (sjlj_mark_call_sites): Likewise.
6053         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6054         * lra-constraints.c (emit_inc): Likewise.
6055         * ree.c (combine_set_extension): Likewise.
6056         * regmove.c (fixup_match_2): Likewise.
6057         * reload1.c (inc_for_reload): Likewise.
6059 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6061         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6062         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6063         Use gen_int_mode with the mode of the associated simplify_* call.
6064         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6065         * expmed.c (expand_shift_1): Likewise.
6066         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6067         * loop-iv.c (iv_number_of_iterations): Likewise.
6068         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6069         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6071 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6073         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6074         of the associated expand_* call.
6075         (asan_emit_stack_protection): Likewise.
6076         * builtins.c (round_trampoline_addr): Likewise.
6077         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6078         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6079         (emit_store_flag): Likewise.
6080         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6081         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6082         Likewise.
6083         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6084         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6085         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6086         * modulo-sched.c (generate_prolog_epilog): Likewise.
6087         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6088         (expand_ctz, expand_ffs, expand_unop): Likewise.
6090 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6092         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6093         of the associated gen_rtx_* call.
6094         * caller-save.c (init_caller_save): Likewise.
6095         * combine.c (find_split_point, make_extraction): Likewise.
6096         (make_compound_operation): Likewise.
6097         * dwarf2out.c (mem_loc_descriptor): Likewise.
6098         * explow.c (plus_constant, probe_stack_range): Likewise.
6099         * expmed.c (expand_mult_const): Likewise.
6100         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6101         * reload1.c (init_reload): Likewise.
6102         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6103         * var-tracking.c (adjust_mems): Likewise.
6104         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6105         rather than gen_rtx_fmt_ee.
6107 2013-09-09  Jan Hubicka  <jh@suse.cz>
6109         PR middle-end/58294
6110         * value-prof.c (gimple_ic): Copy also abnormal edges.
6112 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6114         * asan.c (asan_shadow_cst): Use gen_int_mode.
6116 2013-09-08  Jan Hubicka  <jh@suse.cz>
6118         * ipa-profile.c: Add toplevel comment.
6119         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6120         (contains_hot_call_p): New function.
6121         (ipa_propagate_frequency): Set frequencies based on counts when
6122         profile is read.
6123         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6124         profile; do not tamper with profile after inlining if it is read.
6126 2013-09-08  Jan Hubicka  <jh@suse.cz>
6128         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6129         speculative edges.
6131 2013-09-08  Jan Hubicka  <jh@suse.cz>
6133         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6134         summary generation.
6136 2013-09-08  Jeff Law  <law@redhat.com>
6138         PR bootstrap/58340
6139         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6140         of 'found'.
6142 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6144         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6146 2013-09-08  Jan Hubicka  <jh@suse.cz>
6148         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6150 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6152         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6153         for non-debug insns.
6154         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6155         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6156         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6157         accordingly.
6159 2013-09-08  Jan Hubicka  <jh@suse.cz>
6161         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6162         targets and devirtualize to BUILT_IN_UNREACHABLE.
6163         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6164         * ipa.c (walk_polymorphic_call_targets): New function.
6165         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6166         functions; use the new timevar.
6167         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6168         was removed from the program.
6169         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6170         not consider it in the walk when its vtable is dead.
6171         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6172         record_binfo.
6173         (devirt_variable_node_removal_hook): New function.
6174         (possible_polymorphic_call_targets): Also register
6175         devirt_variable_node_removal_hook.
6176         (ipa_devirt): Do not do non-speculative devirtualization.
6177         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6179 2013-09-08  Jan Hubicka  <jh@suse.cz>
6181         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6182         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6183         varpool_remove_variable_insertion_hook): Declare.
6184         * varpool.c (varpool_node_hook_list): New structure.
6185         (first_varpool_node_removal_hook,
6186         first_varpool_variable_insertion_hook): New variables.
6187         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6188         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6189         varpool_remove_variable_insertion_hook,
6190         varpool_call_variable_insertion_hooks): New functions.
6191         (varpool_remove_node): Use it.
6193 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6195         PR c++/54941
6196         * diagnostic.c (diagnostic_build_prefix): When s.file is
6197         "<built-in>" don't output line and column numbers.
6199 2013-09-06  Jan Hubicka  <jh@suse.cz>
6201         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6202         * lto-streamer-out.c: Stream thunks, too.
6203         * lto-streamer-in.c (input_function): Pop cfun here
6204         (lto_read_body): Instead of here.
6206 2013-09-06  Caroline Tice  <cmtice@google.com>
6208         * doc/install.texi: Add documentation for the --enable-vtable-verify
6209         and the --disable-libvtv configure options.
6211 2013-09-06  Jeff Law  <law@redhat.com>
6213         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6214         edge implied equivalences into successor phis.
6216 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6218         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6220 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6222         * resource.c (mark_target_live_regs): Compute resources taking
6223         into account if a call is predicated or not.
6225 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6227         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6228         the compiler in the function names.
6230 2013-09-06  Jan Hubicka  <jh@suse.cz>
6232         PR middle-end/58094
6233         * ipa-inline.c (has_caller_p): New function.
6234         (want_inline_function_to_all_callers_p): Use it.
6235         (sum_callers, inline_to_all_callers): Break out from ...
6236         (ipa_inline): ... here.
6238 2013-09-06  Jan Hubicka  <jh@suse.cz>
6240         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6241         only when AVX is enabled.
6243 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6245         * config/aarch64/aarch64.md
6246         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6247         is fpsimd_<load/store>2.
6248         (load_pair<mode>): Likewise.
6249         (store_pair<mode>): Likewise.
6251 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6253         * config/arm/types.md (type): Add "mrs" type.
6254         * config/aarch64/aarch64.md
6255         (aarch64_load_tp_hard): Make type "mrs".
6256         * config/arm/arm.md
6257         (load_tp_hard): Make type "mrs".
6258         * config/arm/cortex-a15.md: Update with new attributes.
6259         * config/arm/cortex-a5.md: Update with new attributes.
6260         * config/arm/cortex-a53.md: Update with new attributes.
6261         * config/arm/cortex-a7.md: Update with new attributes.
6262         * config/arm/cortex-a8.md: Update with new attributes.
6263         * config/arm/cortex-a9.md: Update with new attributes.
6264         * config/arm/cortex-m4.md: Update with new attributes.
6265         * config/arm/cortex-r4.md: Update with new attributes.
6266         * config/arm/fa526.md: Update with new attributes.
6267         * config/arm/fa606te.md: Update with new attributes.
6268         * config/arm/fa626te.md: Update with new attributes.
6269         * config/arm/fa726te.md: Update with new attributes.
6271 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6273         * config/aarch64/aarch64.md
6274         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6275         (*movtf_aarch64): Likewise.
6276         * config/arm/arm.md
6277         (thumb1_movdi_insn): Use "multiple" for type where more than one
6278         instruction is used for a move.
6279         (*arm32_movhf): Likewise.
6280         (*thumb_movdf_insn): Likewise.
6282 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6284         * config/arm/types.md (type): Rename fcpys to fmov.
6285         * config/arm/vfp.md
6286         (*arm_movsi_vfp): Rename type fcpys as fmov.
6287         (*thumb2_movsi_vfp): Likewise
6288         (*movhf_vfp_neon): Likewise
6289         (*movhf_vfp): Likewise
6290         (*movsf_vfp): Likewise
6291         (*thumb2_movsf_vfp): Likewise
6292         (*movsfcc_vfp): Likewise
6293         (*thumb2_movsfcc_vfp): Likewise
6294         * config/aarch64/aarch64-simd.md
6295         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6296         * config/aarch64/aarch64.md
6297         (*movsi_aarch64): Replace type mov_reg with fmovs.
6298         (*movdi_aarch64): Likewise
6299         (*movsf_aarch64): Likewise
6300         (*movdf_aarch64): Likewise
6301         * config/arm/arm.c
6302         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6303         * config/arm/iwmmxt.md
6304         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6305         * config/arm/arm1020e.md: Update with new attributes.
6306         * config/arm/cortex-a15-neon.md: Update with new attributes.
6307         * config/arm/cortex-a5.md: Update with new attributes.
6308         * config/arm/cortex-a53.md: Update with new attributes.
6309         * config/arm/cortex-a7.md: Update with new attributes.
6310         * config/arm/cortex-a8-neon.md: Update with new attributes.
6311         * config/arm/cortex-a9.md: Update with new attributes.
6312         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6313         * config/arm/cortex-r4f.md: Update with new attributes.
6314         * config/arm/marvell-pj4.md: Update with new attributes.
6315         * config/arm/vfp11.md: Update with new attributes.
6317 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6319         * config/aarch64/aarch64.md
6320         (*madd<mode>): Fix type attribute.
6321         (*maddsi_uxtw): Likewise.
6322         (*msub<mode>): Likewise.
6323         (*msubsi_uxtw): Likewise.
6324         (<su_optab>maddsidi4): Likewise.
6325         (<su_optab>msubsidi4): Likewise.
6327 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6329         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6330         * config/arm/arm.md (core_cycles): Remove fdiv.
6331         * config/arm/vfp.md:
6332         (*sqrtsf2_vfp): Update for attribute changes.
6333         (*sqrtdf2_vfp): Likewise.
6334         * config/aarch64/aarch64.md:
6335         (sqrt<mode>2): Update for attribute changes.
6336         * config/arm/arm1020e.md: Update with new attributes.
6337         * config/arm/cortex-a15-neon.md: Update with new attributes.
6338         * config/arm/cortex-a5.md: Update with new attributes.
6339         * config/arm/cortex-a53.md: Update with new attributes.
6340         * config/arm/cortex-a7.md: Update with new attributes.
6341         * config/arm/cortex-a8-neon.md: Update with new attributes.
6342         * config/arm/cortex-a9.md: Update with new attributes.
6343         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6344         * config/arm/cortex-r4f.md: Update with new attributes.
6345         * config/arm/marvell-pj4.md: Update with new attributes.
6346         * config/arm/vfp11.md: Update with new attributes.
6348 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6350         * config/arm/types.md
6351         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6352         * config/aarch64/aarch64.md
6353         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6354         new attributes.
6355         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6356         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6357         (float<GPI:mode><GPF:mode>2): Likewise.
6358         * config/arm/vfp.md
6359         (*truncsisf2_vfp): Update with new attributes.
6360         (*truncsidf2_vfp): Likewise.
6361         (fixuns_truncsfsi2): Likewise.
6362         (fixuns_truncdfsi2): Likewise.
6363         (*floatsisf2_vfp): Likewise.
6364         (*floatsidf2_vfp): Likewise.
6365         (floatunssisf2): Likewise.
6366         (floatunssidf2): Likewise.
6367         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6368         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6369         * config/arm/arm1020e.md: Update with new attributes.
6370         * config/arm/cortex-a15-neon.md: Update with new attributes.
6371         * config/arm/cortex-a5.md: Update with new attributes.
6372         * config/arm/cortex-a53.md: Update with new attributes.
6373         * config/arm/cortex-a7.md: Update with new attributes.
6374         * config/arm/cortex-a8-neon.md: Update with new attributes.
6375         * config/arm/cortex-a9.md: Update with new attributes.
6376         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6377         * config/arm/cortex-r4f.md: Update with new attributes.
6378         * config/arm/marvell-pj4.md: Update with new attributes.
6379         * config/arm/vfp11.md: Update with new attributes.
6381 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6383         * config/aarch64/arm_neon.h
6384         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6385         (vqtbx<1,2,3,4><q>_s8): Likewise.
6387 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6389         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6390         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6391         (add<mode>3): Add type attribute.
6392         (add<mode>3): Likewise.
6393         (usadd<mode>3): Likewise.
6394         (ssadd<mode>3): Likewise.
6395         (sub<mode>3): Likewise.
6396         (sub<mode>3): Likewise.
6397         (ussub<mode>3): Likewise.
6398         (sssub<mode>3): Likewise.
6399         (ssmulsa3): Likewise.
6400         (usmulusa3): Likewise.
6401         (arm_usatsihi): Likewise.
6402         * config/arm/vfp.md
6403         (*movdi_vfp): Add types for all instructions.
6404         (*movdi_vfp_cortexa8): Likewise.
6405         (*movhf_vfp_neon): Likewise.
6406         (*movhf_vfp): Likewise.
6407         (*movdf_vfp): Likewise.
6408         (*thumb2_movdf_vfp): Likewise.
6409         (*thumb2_movdfcc_vfp): Likewise.
6410         * config/arm/arm.md: Add type attribute to all insn patterns.
6411         (*thumb1_adddi3): Add type attribute.
6412         (*arm_adddi3): Likewise.
6413         (*adddi_sesidi_di): Likewise.
6414         (*adddi_zesidi_di): Likewise.
6415         (*thumb1_addsi3): Likewise.
6416         (addsi3_compare0): Likewise.
6417         (*addsi3_compare0_scratch): Likewise.
6418         (*compare_negsi_si): Likewise.
6419         (cmpsi2_addneg): Likewise.
6420         (*addsi3_carryin_<optab>): Likewise.
6421         (*addsi3_carryin_alt2_<optab>): Likewise.
6422         (*addsi3_carryin_clobercc_<optab>): Likewise.
6423         (*subsi3_carryin): Likewise.
6424         (*subsi3_carryin_const): Likewise.
6425         (*subsi3_carryin_compare): Likewise.
6426         (*subsi3_carryin_compare_const): Likewise.
6427         (*arm_subdi3): Likewise.
6428         (*thumb_subdi3): Likewise.
6429         (*subdi_di_zesidi): Likewise.
6430         (*subdi_di_sesidi): Likewise.
6431         (*subdi_zesidi_di): Likewise.
6432         (*subdi_sesidi_di): Likewise.
6433         (*subdi_zesidi_ze): Likewise.
6434         (thumb1_subsi3_insn): Likewise.
6435         (*arm_subsi3_insn): Likewise.
6436         (*anddi3_insn): Likewise.
6437         (*anddi_zesidi_di): Likewise.
6438         (*anddi_sesdi_di): Likewise.
6439         (*ne_zeroextracts): Likewise.
6440         (*ne_zeroextracts): Likewise.
6441         (*ite_ne_zeroextr): Likewise.
6442         (*ite_ne_zeroextr): Likewise.
6443         (*anddi_notdi_di): Likewise.
6444         (*anddi_notzesidi): Likewise.
6445         (*anddi_notsesidi): Likewise.
6446         (andsi_notsi_si): Likewise.
6447         (thumb1_bicsi3): Likewise.
6448         (*iordi3_insn): Likewise.
6449         (*iordi_zesidi_di): Likewise.
6450         (*iordi_sesidi_di): Likewise.
6451         (*thumb1_iorsi3_insn): Likewise.
6452         (*xordi3_insn): Likewise.
6453         (*xordi_zesidi_di): Likewise.
6454         (*xordi_sesidi_di): Likewise.
6455         (*arm_xorsi3): Likewise.
6456         (*andsi_iorsi3_no): Likewise.
6457         (*smax_0): Likewise.
6458         (*smax_m1): Likewise.
6459         (*arm_smax_insn): Likewise.
6460         (*smin_0): Likewise.
6461         (*arm_smin_insn): Likewise.
6462         (*arm_umaxsi3): Likewise.
6463         (*arm_uminsi3): Likewise.
6464         (*minmax_arithsi): Likewise.
6465         (*minmax_arithsi_): Likewise.
6466         (*satsi_<SAT:code>): Likewise.
6467         (arm_ashldi3_1bit): Likewise.
6468         (arm_ashrdi3_1bit): Likewise.
6469         (arm_lshrdi3_1bit): Likewise.
6470         (*arm_negdi2): Likewise.
6471         (*thumb1_negdi2): Likewise.
6472         (*arm_negsi2): Likewise.
6473         (*thumb1_negsi2): Likewise.
6474         (*negdi_extendsid): Likewise.
6475         (*negdi_zero_extend): Likewise.
6476         (*arm_abssi2): Likewise.
6477         (*thumb1_abssi2): Likewise.
6478         (*arm_neg_abssi2): Likewise.
6479         (*thumb1_neg_abss): Likewise.
6480         (one_cmpldi2): Likewise.
6481         (extend<mode>di2): Likewise.
6482         (*compareqi_eq0): Likewise.
6483         (*arm_extendhisi2addsi): Likewise.
6484         (*arm_movdi): Likewise.
6485         (*thumb1_movdi_insn): Likewise.
6486         (*arm_movt): Likewise.
6487         (*thumb1_movsi_insn): Likewise.
6488         (pic_add_dot_plus_four): Likewise.
6489         (pic_add_dot_plus_eight): Likewise.
6490         (tls_load_dot_plus_eight): Likewise.
6491         (*thumb1_movhi_insn): Likewise.
6492         (*thumb1_movsf_insn): Likewise.
6493         (*movdf_soft_insn): Likewise.
6494         (*thumb_movdf_insn): Likewise.
6495         (cbranchsi4_insn): Likewise.
6496         (cbranchsi4_scratch): Likewise.
6497         (*negated_cbranchsi4): Likewise.
6498         (*tbit_cbranch): Likewise.
6499         (*tlobits_cbranch): Likewise.
6500         (*tstsi3_cbranch): Likewise.
6501         (*cbranchne_decr1): Likewise.
6502         (*addsi3_cbranch): Likewise.
6503         (*addsi3_cbranch_scratch): Likewise.
6504         (*arm_cmpdi_insn): Likewise.
6505         (*arm_cmpdi_unsig): Likewise.
6506         (*arm_cmpdi_zero): Likewise.
6507         (*thumb_cmpdi_zero): Likewise.
6508         (*deleted_compare): Likewise.
6509         (*mov_scc): Likewise.
6510         (*mov_negscc): Likewise.
6511         (*mov_notscc): Likewise.
6512         (*cstoresi_eq0_thumb1_insn): Likewise.
6513         (cstoresi_nltu_thumb1): Likewise.
6514         (cstoresi_ltu_thu): Likewise.
6515         (thumb1_addsi3_addgeu): Likewise.
6516         (*arm_jump): Likewise.
6517         (*thumb_jump): Likewise.
6518         (*check_arch2): Likewise.
6519         (arm_casesi_internal): Likewise.
6520         (thumb1_casesi_dispatch): Likewise.
6521         (*arm_indirect_jump): Likewise.
6522         (*thumb1_indirect_jump): Likewise.
6523         (nop): Likewise.
6524         (*and_scc): Likewise.
6525         (*ior_scc): Likewise.
6526         (*compare_scc): Likewise.
6527         (*cond_move): Likewise.
6528         (*cond_arith): Likewise.
6529         (*cond_sub): Likewise.
6530         (*cmp_ite0): Likewise.
6531         (*cmp_ite1): Likewise.
6532         (*cmp_and): Likewise.
6533         (*cmp_ior): Likewise.
6534         (*ior_scc_scc): Likewise.
6535         (*ior_scc_scc_cmp): Likewise.
6536         (*and_scc_scc): Likewise.
6537         (*and_scc_scc_cmp): Likewise.
6538         (*and_scc_scc_nod): Likewise.
6539         (*negscc): Likewise.
6540         (movcond_addsi): Likewise.
6541         (movcond): Likewise.
6542         (*ifcompare_plus_move): Likewise.
6543         (*if_plus_move): Likewise.
6544         (*ifcompare_move_plus): Likewise.
6545         (*if_move_plus): Likewise.
6546         (*ifcompare_arith_arith): Likewise.
6547         (*if_arith_arith): Likewise.
6548         (*ifcompare_arith_move): Likewise.
6549         (*if_arith_move): Likewise.
6550         (*ifcompare_move_arith): Likewise.
6551         (*if_move_arith): Likewise.
6552         (*ifcompare_move_not): Likewise.
6553         (*if_move_not): Likewise.
6554         (*ifcompare_not_move): Likewise.
6555         (*if_not_move): Likewise.
6556         (*ifcompare_shift_move): Likewise.
6557         (*if_shift_move): Likewise.
6558         (*ifcompare_move_shift): Likewise.
6559         (*if_move_shift): Likewise.
6560         (*ifcompare_shift_shift): Likewise.
6561         (*ifcompare_not_arith): Likewise.
6562         (*ifcompare_arith_not): Likewise.
6563         (*if_arith_not): Likewise.
6564         (*ifcompare_neg_move): Likewise.
6565         (*if_neg_move): Likewise.
6566         (*ifcompare_move_neg): Likewise.
6567         (*if_move_neg): Likewise.
6568         (prologue_thumb1_interwork): Likewise.
6569         (*cond_move_not): Likewise.
6570         (*sign_extract_onebit): Likewise.
6571         (*not_signextract_onebit): Likewise.
6572         (stack_tie): Likewise.
6573         (align_4): Likewise.
6574         (align_8): Likewise.
6575         (consttable_end): Likewise.
6576         (consttable_1): Likewise.
6577         (consttable_2): Likewise.
6578         (consttable_4): Likewise.
6579         (consttable_8): Likewise.
6580         (consttable_16): Likewise.
6581         (*thumb1_tablejump): Likewise.
6582         (prefetch): Likewise.
6583         (force_register_use): Likewise.
6584         (thumb_eh_return): Likewise.
6585         (load_tp_hard): Likewise.
6586         (load_tp_soft): Likewise.
6587         (tlscall): Likewise.
6588         (*arm_movtas_ze): Likewise.
6589         (*arm_rev): Likewise.
6590         (*arm_revsh): Likewise.
6591         (*arm_rev16): Likewise.
6592         * config/arm/thumb2.md
6593         (*thumb2_smaxsi3): Likewise.
6594         (*thumb2_sminsi3): Likewise.
6595         (*thumb32_umaxsi3): Likewise.
6596         (*thumb2_uminsi3): Likewise.
6597         (*thumb2_negdi2): Likewise.
6598         (*thumb2_abssi2): Likewise.
6599         (*thumb2_neg_abss): Likewise.
6600         (*thumb2_movsi_insn): Likewise.
6601         (tls_load_dot_plus_four): Likewise.
6602         (*thumb2_movhi_insn): Likewise.
6603         (*thumb2_mov_scc): Likewise.
6604         (*thumb2_mov_negs): Likewise.
6605         (*thumb2_mov_negs): Likewise.
6606         (*thumb2_mov_nots): Likewise.
6607         (*thumb2_mov_nots): Likewise.
6608         (*thumb2_movsicc_): Likewise.
6609         (*thumb2_movsfcc_soft_insn): Likewise.
6610         (*thumb2_indirect_jump): Likewise.
6611         (*thumb2_and_scc): Likewise.
6612         (*thumb2_ior_scc): Likewise.
6613         (*thumb2_ior_scc_strict_it): Likewise.
6614         (*thumb2_cond_move): Likewise.
6615         (*thumb2_cond_arith): Likewise.
6616         (*thumb2_cond_ari): Likewise.
6617         (*thumb2_cond_sub): Likewise.
6618         (*thumb2_negscc): Likewise.
6619         (*thumb2_movcond): Likewise.
6620         (thumb2_casesi_internal): Likewise.
6621         (thumb2_casesi_internal_pic): Likewise.
6622         (*thumb2_alusi3_short): Likewise.
6623         (*thumb2_mov<mode>_shortim): Likewise.
6624         (*thumb2_addsi_short): Likewise.
6625         (*thumb2_subsi_short): Likewise.
6626         (thumb2_addsi3_compare0): Likewise.
6627         (*thumb2_cbz): Likewise.
6628         (*thumb2_cbnz): Likewise.
6629         (*thumb2_one_cmplsi2_short): Likewise.
6630         (*thumb2_negsi2_short): Likewise.
6631         (*orsi_notsi_si): Likewise.
6632         * config/arm/arm1020e.md: Update with new attributes.
6633         * config/arm/arm1026ejs.md: Update with new attributes.
6634         * config/arm/arm1136jfs.md: Update with new attributes.
6635         * config/arm/arm926ejs.md: Update with new attributes.
6636         * config/arm/cortex-a15.md: Update with new attributes.
6637         * config/arm/cortex-a5.md: Update with new attributes.
6638         * config/arm/cortex-a53.md: Update with new attributes.
6639         * config/arm/cortex-a7.md: Update with new attributes.
6640         * config/arm/cortex-a8.md: Update with new attributes.
6641         * config/arm/cortex-a9.md: Update with new attributes.
6642         * config/arm/cortex-m4.md: Update with new attributes.
6643         * config/arm/cortex-r4.md: Update with new attributes.
6644         * config/arm/fa526.md: Update with new attributes.
6645         * config/arm/fa606te.md: Update with new attributes.
6646         * config/arm/fa626te.md: Update with new attributes.
6647         * config/arm/fa726te.md: Update with new attributes.
6649 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6651         * config/aarch64/aarch64-simd.md
6652         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6653         <vwx> iterator to ensure correct register choice.
6654         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6655         (aarch64_sqdmull_n<mode>): Likewise.
6656         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6657         * config/aarch64/arm_neon.h
6658         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6659         (vml<as><q>_n_<su>16): Likewise.
6660         (vml<as>l_high_lane<q>_<su>16): Likewise.
6661         (vml<as>l_high_n_<su>16): Likewise.
6662         (vml<as>l_lane<q>_<su>16): Likewise.
6663         (vml<as>l_n_<su>16): Likewise.
6664         (vmul<q>_lane<q>_<su>16): Likewise.
6665         (vmul<q>_n_<su>16): Likewise.
6666         (vmull_lane<q>_<su>16): Likewise.
6667         (vmull_n_<su>16): Likewise.
6668         (vmull_high_lane<q>_<su>16): Likewise.
6669         (vmull_high_n_<su>16): Likewise.
6670         (vqrdmulh<q>_n_s16): Likewise.
6672 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6674         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6675         have the correct lane parameter.
6677 2013-09-06  Richard Biener <rguenther@suse.de>
6679         * cfganal.c (control_dependences::~control_dependences):
6680         Properly free all of the vector.
6682 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6684         PR target/58269
6685         * config/i386/i386.c (ix86_conditional_register_usage):
6686         Proper initialize extended SSE registers.
6688 2013-09-06  Jan Hubicka  <jh@suse.cz>
6690         PR tree-optimization/58311
6691         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6693 2013-09-06  Jan Hubicka  <jh@suse.cz>
6695         * Makefile.in (tree-sra.o): Update dependencies.
6696         * tree-sra.c: Include ipa-utils.h
6697         (scan_function): Use recursive_call_p.
6698         (has_caller_p): New function.
6699         (cgraph_for_node_and_aliases): Count also callers of aliases.
6701 2013-09-06  Jan Hubicka  <jh@suse.cz>
6703         PR middle-end/58094
6704         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6705         * tree-tailcall.c: Include ipa-utils.h.
6706         (find_tail_calls): Use it.
6707         * ipa-pure-const.c (check_call): Likewise.
6708         * ipa-utils.c (recursive_call_p): New function.
6709         * ipa-utils.h (recursive_call_p): Dclare.
6710         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6711         (symtab_semantically_equivalent_p): New function.
6712         * Makefile.in (tree-tailcall.o): Update dependencies.
6714 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6716         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6717         non-inlinable part.
6719 2013-09-06  Richard Biener  <rguenther@suse.de>
6721         * lto-streamer.h (lto_global_var_decls): Remove.
6722         * Makefile.in (OBJS): Remove lto-symtab.o.
6723         (lto-symtab.o): Remove.
6724         (GTFILES): Remove lto-symtab.c
6725         * lto-symtab.c: Move to lto/
6727 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6729         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6730         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6731         (UNSPEC_FPINT_RINT): New constant definitions.
6732         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6733         definition with 2 attributes.
6734         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6735         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6736         definitions.
6738 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6740         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6741         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6742         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6743         attribute to "z196_alone".
6744         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6745         "zEC12_simple".
6747 2013-09-06  Richard Biener  <rguenther@suse.de>
6749         * basic-block.h (class control_dependences): New.
6750         * tree-ssa-dce.c (control_dependence_map): Remove.
6751         (cd): New global.
6752         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6753         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6754         find_pdom, find_control_dependence, find_all_control_dependences):
6755         Move to cfganal.c.
6756         (mark_control_dependent_edges_necessary,
6757         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6758         tree_dce_done, perform_tree_ssa_dce): Adjust.
6759         * cfganal.c (set_control_dependence_map_bit,
6760         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6761         find_all_control_dependences): Move from tree-ssa-dce.c and
6762         implement as methods of control_dependences class.
6763         (control_dependences::control_dependences): New.
6764         (control_dependences::~control_dependences): Likewise.
6765         (control_dependences::get_edges_dependent_on): Likewise.
6766         (control_dependences::get_edge): Likewise.
6768 2013-09-04  Jan Hubicka  <jh@suse.cz>
6770         * tree.c (types_same_for_odr): Drop overactive check.
6771         * ipa-devirt.c (hash_type_name): Likewise.
6773 2013-09-04  Jan Hubicka  <jh@suse.cz>
6775         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6776         (analyze_functions): ... here.
6778 2013-09-04  Jan Hubicka  <jh@suse.cz>
6780         PR middle-end/58201
6781         * cgraphunit.c (analyze_functions): Clear AUX fields
6782         after processing; initialize assembler name has.
6784 2013-09-05  Jeff Law  <law@redhat.com>
6786         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6787         from thread_around_empty_block.  Record threading path into PATH.
6788         Recurse if threading through the initial block is successful.
6789         (thread_across_edge): Corresponding changes to slightly simplify.
6791 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6793         * config/aarch64/aarch64.md
6794         (type): Remove frecpe, frecps, frecpx.
6795         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6796         fix to be a TARGET_SIMD instruction.
6797         (aarch64_frecps): Remove.
6798         * config/aarch64/aarch64-simd.md
6799         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6800         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6802 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6803             Sofiane Naci  <sofiane.naci@arm.com>
6805         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6806         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6807         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6808         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6809         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6810         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6811         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6812         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6813         into "clz, "rbit".  Rename "shift" to "shift_imm".
6814         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6815         changes.  Update for attribute changes all occurrences of arlo_* and
6816         shift* types.
6817         * config/arm/arm-fixed.md: Update for attribute changes
6818         all occurrences of arlo_* types.
6819         * config/arm/thumb2.md: Update for attribute changes all occurrences
6820         of arlo_* types.
6821         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6822         (cortexa7_older_only): Likewise.
6823         (cortexa7_younger):  Likewise.
6824         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6825         (1020alu_shift_op): Likewise.
6826         (1020alu_shift_reg_op): Likewise.
6827         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6828         (alu_shift_op): Likewise.
6829         (alu_shift_reg_op): Likewise.
6830         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6831         (11_alu_shift_op): Likewise.
6832         (11_alu_shift_reg_op): Likewise.
6833         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6834         (9_alu_shift_reg_op): Likewise.
6835         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6836         attribute changes.
6837         (cortex_a15_alu_shift): Likewise.
6838         (cortex_a15_alu_shift_reg): Likewise.
6839         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6840         attribute changes.
6841         (cortex_a5_alu_shift): Likewise.
6842         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6843         attribute changes.
6844         (cortex_a53_alu_shift): Likewise.
6845         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6846         attribute changes.
6847         (cortex_a7_alu_reg): Likewise.
6848         (cortex_a7_alu_shift): Likewise.
6849         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6850         attribute changes.
6851         (cortex_a8_alu_shift): Likewise.
6852         (cortex_a8_alu_shift_reg): Likewise.
6853         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6854         (cortex_a9_dp_shift): Likewise.
6855         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6856         attribute changes.
6857         * config/arm/cortex-r4.md
6858         (cortex_r4_alu): Update for attribute changes.
6859         (cortex_r4_mov): Likewise.
6860         (cortex_r4_alu_shift_reg): Likewise.
6861         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6862         (526_alu_shift_op): Likewise.
6863         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6864         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6865         (626te_alu_shift_op): Likewise.
6866         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6867         (726te_alu_shift_op): Likewise.
6868         (726te_alu_shift_reg_op): Likewise.
6869         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6870         (mp626_alu_shift_op): Likewise.
6871         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6872         (pj4_alu_conds): Likewise.
6873         (pj4_shift): Likewise.
6874         (pj4_shift_conds): Likewise.
6875         (pj4_alu_shift): Likewise.
6876         (pj4_alu_shift_conds): Likewise.
6877         * config/aarch64/aarch64.md: Update for attribute change
6878         all occurrences of arlo_* and shift* types.
6880 2013-09-05  Mike Stump  <mikestump@comcast.net>
6882         * tree.h: Move documentation for tree_function_decl to tree-core.h
6883         with the declaration.
6885 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6887         PR target/58139
6888         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6889         looking for widest mode.
6891 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6893         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6895 2013-09-05  Richard Biener  <rguenther@suse.de>
6897         PR tree-optimization/58137
6898         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6899         Do not create vectors of pointers.
6900         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6901         types for the components of the vector initializer.
6902         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6903         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6905 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6907         * ipa-prop.c (remove_described_reference): Accept missing references,
6908         return false if that hppens, otherwise return true.
6909         (cgraph_node_for_jfunc): New function.
6910         (try_decrement_rdesc_refcount): Likewise.
6911         (try_make_edge_direct_simple_call): Use them.
6912         (ipa_edge_removal_hook): Remove references from rdescs.
6913         (ipa_edge_duplication_hook): Clone rdescs and their references
6914         when the new edge has the same caller as the old one.
6915         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6916         reference before removing any edges.
6918 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6920         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6921         initial store.
6922         * thumb2.md (thumb2_storewb_parisi): New pattern.
6924 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6926         * config/aarch64/aarch64-option-extensions.def: Add
6927         AARCH64_OPT_EXTENSION of 'crc'.
6928         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6929         (AARCH64_ISA_CRC): Ditto.
6930         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6931         description of the CRC extension.
6933 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6935         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6936         * config/rs6000/linux.h: Ditto.
6937         * alpha/linux.h: Ditto.
6938         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6939         no_c99_libc_has_function.
6940         * config/c6x/uclinux-elf.h: Ditto.
6941         * config/lm32/uclinux-elf.h: Ditto.
6942         * config/m68k/uclinux.h: Ditto.
6943         * config/moxie/uclinux.h: Ditto.
6944         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6945         (crisv32-*-linux*, cris-*-linux*): Ditto.
6946         * config/bfin/bfin.c: Include "tm_p.h".
6948 2013-09-05  Richard Biener  <rguenther@suse.de>
6950         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6951         check for a definition without a basic-block.
6953 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6954             Sofiane Naci <sofiane.naci@arm.com>
6956         * config/aarch64/aarch64.md
6957         (*movti_aarch64): Rename r_2_f and f_2_r.
6958         (*movsf_aarch64): Likewise.
6959         (*movdf_aarch64): Likewise.
6960         (*movtf_aarch64): Likewise.
6961         (aarch64_movdi_<mode>low): Likewise.
6962         (aarch64_movdi_<mode>high): Likewise.
6963         (aarch64_mov<mode>high_di): Likewise.
6964         (aarch64_mov<mode>low_di): Likewise.
6965         (aarch64_movtilow_tilow): Likewise.
6966         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6967         values to config/arm/types.md
6968         (attribute "conds"): Update for attribute change.
6969         (anddi3_insn): Likewise.
6970         (iordi3_insn): Likewise.
6971         (xordi3_insn): Likewise.
6972         (one_cmpldi2): Likewise.
6973         * config/arm/types.md (type): Add Neon types.
6974         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6975         use "type" attribute.
6976         (movmisalign<mode>_neon_store): Likewise.
6977         (movmisalign<mode>_neon_load): Likewise.
6978         (vec_set<mode>_internal): Likewise.
6979         (vec_setv2di_internal): Likewise.
6980         (vec_extract<mode>): Likewise.
6981         (vec_extractv2di): Likewise.
6982         (add<mode>3_neon): Likewise.
6983         (adddi3_neon): Likewise.
6984         (sub<mode>3_neon): Likewise.
6985         (subdi3_neon): Likewise.
6986         (mul<mode>3_neon): Likewise.
6987         (mul<mode>3add<mode>_neon): Likewise.
6988         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6989         (fma<VCVTF:mode>4)): Likewise.
6990         (fma<VCVTF:mode>4_intrinsic): Likewise.
6991         (fmsub<VCVTF:mode>4)): Likewise.
6992         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6993         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6994         (ior<mode>3): Likewise.
6995         (and<mode>3): Likewise.
6996         (anddi3_neon): Likewise.
6997         (orn<mode>3_neon): Likewise.
6998         (orndi3_neon): Likewise.
6999         (bic<mode>3_neon): Likewise.
7000         (bicdi3_neon): Likewise.
7001         (xor<mode>3): Likewise.
7002         (one_cmpl<mode>2): Likewise.
7003         (abs<mode>2): Likewise.
7004         (neg<mode>2): Likewise.
7005         (umin<mode>3_neon): Likewise.
7006         (umax<mode>3_neon): Likewise.
7007         (smin<mode>3_neon): Likewise.
7008         (smax<mode>3_neon): Likewise.
7009         (vashl<mode>3): Likewise.
7010         (vashr<mode>3_imm): Likewise.
7011         (vlshr<mode>3_imm): Likewise.
7012         (ashl<mode>3_signed): Likewise.
7013         (ashl<mode>3_unsigned): Likewise.
7014         (neon_load_count): Likewise.
7015         (ashldi3_neon_noclobber): Likewise.
7016         (signed_shift_di3_neon): Likewise.
7017         (unsigned_shift_di3_neon): Likewise.
7018         (ashrdi3_neon_imm_noclobber): Likewise.
7019         (lshrdi3_neon_imm_noclobber): Likewise.
7020         (widen_ssum<mode>3): Likewise.
7021         (widen_usum<mode>3): Likewise.
7022         (quad_halves_<code>v4si): Likewise.
7023         (quad_halves_<code>v4sf): Likewise.
7024         (quad_halves_<code>v8hi): Likewise.
7025         (quad_halves_<code>v16qi): Likewise.
7026         (reduc_splus_v2di): Likewise.
7027         (neon_vpadd_internal<mode>): Likewise.
7028         (neon_vpsmin<mode>): Likewise.
7029         (neon_vpsmax<mode>): Likewise.
7030         (neon_vpumin<mode>): Likewise.
7031         (neon_vpumax<mode>): Likewise.
7032         (ss_add<mode>_neon): Likewise.
7033         (us_add<mode>_neon): Likewise.
7034         (ss_sub<mode>_neon): Likewise.
7035         (us_sub<mode>_neon): Likewise.
7036         (neon_vadd<mode>_unspec): Likewise.
7037         (neon_vaddl<mode>): Likewise.
7038         (neon_vaddw<mode>): Likewise.
7039         (neon_vhadd<mode>): Likewise.
7040         (neon_vqadd<mode>): Likewise.
7041         (neon_vaddhn<mode>): Likewise.
7042         (neon_vmul<mode>): Likewise.
7043         (neon_vmla<mode>): Likewise.
7044         (neon_vmlal<mode>): Likewise.
7045         (neon_vmls<mode>): Likewise.
7046         (neon_vmlsl<mode>): Likewise.
7047         (neon_vqdmulh<mode>): Likewise.
7048         (neon_vqdmlal<mode>): Likewise.
7049         (neon_vqdmlsl<mode>): Likewise.
7050         (neon_vmull<mode>): Likewise.
7051         (neon_vqdmull<mode>): Likewise.
7052         (neon_vsub<mode>_unspec): Likewise.
7053         (neon_vsubl<mode>): Likewise.
7054         (neon_vsubw<mode>): Likewise.
7055         (neon_vqsub<mode>): Likewise.
7056         (neon_vhsub<mode>): Likewise.
7057         (neon_vsubhn<mode>): Likewise.
7058         (neon_vceq<mode>): Likewise.
7059         (neon_vcge<mode>): Likewise.
7060         (neon_vcgeu<mode>): Likewise.
7061         (neon_vcgt<mode>): Likewise.
7062         (neon_vcgtu<mode>): Likewise.
7063         (neon_vcle<mode>): Likewise.
7064         (neon_vclt<mode>): Likewise.
7065         (neon_vcage<mode>): Likewise.
7066         (neon_vcagt<mode>): Likewise.
7067         (neon_vtst<mode>): Likewise.
7068         (neon_vabd<mode>): Likewise.
7069         (neon_vabdl<mode>): Likewise.
7070         (neon_vaba<mode>): Likewise.
7071         (neon_vabal<mode>): Likewise.
7072         (neon_vmax<mode>): Likewise.
7073         (neon_vmin<mode>): Likewise.
7074         (neon_vpaddl<mode>): Likewise.
7075         (neon_vpadal<mode>): Likewise.
7076         (neon_vpmax<mode>): Likewise.
7077         (neon_vpmin<mode>): Likewise.
7078         (neon_vrecps<mode>): Likewise.
7079         (neon_vrsqrts<mode>): Likewise.
7080         (neon_vqabs<mode>): Likewise.
7081         (neon_vqneg<mode>): Likewise.
7082         (neon_vcls<mode>): Likewise.
7083         (clz<mode>2): Likewise.
7084         (popcount<mode>2): Likewise.
7085         (neon_vrecpe): Likewise.
7086         (neon_vrsqrte): Likewise.
7087         (neon_vget_lane<mode>_sext_internal): Likewise.
7088         (neon_vget_lane<mode>_zext_internal): Likewise.
7089         (neon_vdup_n<mode>): Likewise.
7090         (neon_vdup_nv2di): Likewise.
7091         (neon_vdpu_lane<mode>_internal): Likewise.
7092         (neon_vswp<mode>): Likewise.
7093         (float<mode><V_cvtto>2): Likewise.
7094         (floatuns<mode><V_cvtto>2): Likewise.
7095         (fix_trunc<mode><V_cvtto>)2): Likewise
7096         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7097         (neon_vcvt<mode>): Likewise.
7098         (neon_vcvtv4sfv4hf): Likewise.
7099         (neon_vcvtv4hfv4sf): Likewise.
7100         (neon_vcvt_n<mode>): Likewise.
7101         (neon_vmovn<mode>): Likewise.
7102         (neon_vqmovn<mode>): Likewise.
7103         (neon_vqmovun<mode>): Likewise.
7104         (neon_vmovl<mode>): Likewise.
7105         (neon_vmul_lane<mode>): Likewise.
7106         (neon_vmull_lane<mode>): Likewise.
7107         (neon_vqdmull_lane<mode>): Likewise.
7108         (neon_vqdmulh_lane<mode>): Likewise.
7109         (neon_vmla_lane<mode>): Likewise.
7110         (neon_vmlal_lane<mode>): Likewise.
7111         (neon_vqdmlal_lane<mode>): Likewise.
7112         (neon_vmls_lane<mode>): Likewise.
7113         (neon_vmlsl_lane<mode>): Likewise.
7114         (neon_vqdmlsl_lane<mode>): Likewise.
7115         (neon_vext<mode>): Likewise.
7116         (neon_vrev64<mode>): Likewise.
7117         (neon_vrev32<mode>): Likewise.
7118         (neon_vrev16<mode>): Likewise.
7119         (neon_vbsl<mode>_internal): Likewise.
7120         (neon_vshl<mode>): Likewise.
7121         (neon_vqshl<mode>): Likewise.
7122         (neon_vshr_n<mode>): Likewise.
7123         (neon_vshrn_n<mode>): Likewise.
7124         (neon_vqshrn_n<mode>): Likewise.
7125         (neon_vqshrun_n<mode>): Likewise.
7126         (neon_vshl_n<mode>): Likewise.
7127         (neon_vqshl_n<mode>): Likewise.
7128         (neon_vqshlu_n<mode>): Likewise.
7129         (neon_vshll_n<mode>): Likewise.
7130         (neon_vsra_n<mode>): Likewise.
7131         (neon_vsri_n<mode>): Likewise.
7132         (neon_vsli_n<mode>): Likewise.
7133         (neon_vtbl1v8qi): Likewise.
7134         (neon_vtbl2v8qi): Likewise.
7135         (neon_vtbl3v8qi): Likewise.
7136         (neon_vtbl4v8qi): Likewise.
7137         (neon_vtbx1v8qi): Likewise.
7138         (neon_vtbx2v8qi): Likewise.
7139         (neon_vtbx3v8qi): Likewise.
7140         (neon_vtbx4v8qi): Likewise.
7141         (neon_vtrn<mode>_internal): Likewise.
7142         (neon_vzip<mode>_internal): Likewise.
7143         (neon_vuzp<mode>_internal): Likewise.
7144         (neon_vld1<mode>): Likewise.
7145         (neon_vld1_lane<mode>): Likewise.
7146         (neon_vld1_dup<mode>): Likewise.
7147         (neon_vld1_dupv2di): Likewise.
7148         (neon_vst1<mode>): Likewise.
7149         (neon_vst1_lane<mode>): Likewise.
7150         (neon_vld2<mode>): Likewise.
7151         (neon_vld2_lane<mode>): Likewise.
7152         (neon_vld2_dup<mode>): Likewise.
7153         (neon_vst2<mode>): Likewise.
7154         (neon_vst2_lane<mode>): Likewise.
7155         (neon_vld3<mode>): Likewise.
7156         (neon_vld3qa<mode>): Likewise.
7157         (neon_vld3qb<mode>): Likewise.
7158         (neon_vld3_lane<mode>): Likewise.
7159         (neon_vld3_dup<mode>): Likewise.
7160         (neon_vst3<mode>): Likewise.
7161         (neon_vst3qa<mode>): Likewise.
7162         (neon_vst3qb<mode>): Likewise.
7163         (neon_vst3_lane<mode>): Likewise.
7164         (neon_vld4<mode>): Likewise.
7165         (neon_vld4qa<mode>): Likewise.
7166         (neon_vld4qb<mode>): Likewise.
7167         (neon_vld4_lane<mode>): Likewise.
7168         (neon_vld4_dup<mode>): Likewise.
7169         (neon_vst4<mode>): Likewise.
7170         (neon_vst4qa<mode>): Likewise.
7171         (neon_vst4qb<mode>): Likewise.
7172         (neon_vst4_lane<mode>): Likewise.
7173         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7174         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7175         (neon_vec_<US>mult_lo_<mode>): Likewise.
7176         (neon_vec_<US>mult_hi_<mode>): Likewise.
7177         (neon_vec_<US>shiftl_<mode>): Likewise.
7178         (neon_unpack<US>_<mode>): Likewise.
7179         (neon_vec_<US>mult_<mode>): Likewise.
7180         (vec_pack_trunc_<mode>): Likewise.
7181         (neon_vec_pack_trunk_<mode>): Likewise.
7182         (neon_vabd<mode>_2): Likewise.
7183         (neon_vabd<mode>_3): Likewise.
7184         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7185         (thumb2_movsi_vfp): Likewise.
7186         (movdi_vfp): Likewise.
7187         (movdi_vfp_cortexa8): Likewise.
7188         (movhf_vfp_neon): Likewise.
7189         (movhf_vfp): Likewiwse.
7190         (movsf_vfp): Likewiwse.
7191         (thumb2_movsf_vfp): Likewiwse.
7192         (movdf_vfp): Likewise.
7193         (thumb2_movdf_vfp): Likewise.
7194         (movsfcc_vfp): Likewise.
7195         (thumb2_movsfcc_vfp): Likewise.
7196         (movdfcc_vfp): Likewise.
7197         (thumb2_movdfcc_vfp): Likewise.
7198         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7199         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7200         (v10_v2c): Likewise.
7201         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7202         attribute change.
7203         (cortex_a15_neon_int_2): Likewise.
7204         (cortex_a15_neon_int_3): Likewise.
7205         (cortex_a15_neon_int_4): Likewise.
7206         (cortex_a15_neon_int_5): Likewise.
7207         (cortex_a15_neon_vqneg_vqabs): Likewise.
7208         (cortex_a15_neon_vmov): Likewise.
7209         (cortex_a15_neon_vaba): Likewise.
7210         (cortex_a15_neon_vaba_qqq): Likewise.
7211         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7212         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7213         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7214         scalar_64_32_long_scalar): Likewise.
7215         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7216         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7217         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7218         lotype_qdd_64_32_long): Likewise.
7219         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7220         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7221         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7222         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7223         (cortex_a15_neon_shift_1): Likewise.
7224         (cortex_a15_neon_shift_2): Likewise.
7225         (cortex_a15_neon_shift_3): Likewise.
7226         (cortex_a15_neon_vshl_ddd): Likewise.
7227         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7228         (cortex_a15_neon_vsra_vrsra): Likewise.
7229         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7230         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7231         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7232         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7233         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7234         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7235         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7236         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7237         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7238         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7239         (cortex_a15_neon_bp_simple): Likewise.
7240         (cortex_a15_neon_bp_2cycle): Likewise.
7241         (cortex_a15_neon_bp_3cycle): Likewise.
7242         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7243         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7244         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7245         (cortex_a15_neon_vld2_4_regs): Likewise.
7246         (cortex_a15_neon_vld3_vld4): Likewise.
7247         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7248         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7249         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7250         (cortex_a15_neon_vst3_vst4): Likewise.
7251         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7252         (cortex_a15_neon_vld3_vld4_lane" 10
7253         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7254         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7255         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7256         (cortex_a15_neon_ldm_2): Likewise.0
7257         (cortex_a15_neon_stm_2): Likewise.
7258         (cortex_a15_neon_mcr): Likewise.
7259         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7260         (cortex_a15_neon_mrc): Likewise.
7261         (cortex_a15_neon_mrrc): Likewise.
7262         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7263         change.
7264         (cortex_a15_alu_shift): Likewise.
7265         (cortex_a15_alu_shift_reg): Likewise.
7266         (cortex_a15_mult32): Likewise.
7267         (cortex_a15_mult64): Likewise.
7268         (cortex_a15_block): Likewise.
7269         (cortex_a15_branch): Likewise.
7270         (cortex_a15_load1): Likewise.
7271         (cortex_a15_load3): Likewise.
7272         (cortex_a15_store1): Likewise.
7273         (cortex_a15_store3): Likewise.
7274         (cortex_a15_call): Likewise.
7275         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7276         (cortex_a5_f2r): Likewise.
7277         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7278         change.
7279         (cortex_a53_f2r): Likewise.
7280         * config/arm/cortex-a7.md
7281         (cortex_a7_branch): Update for attribute change.
7282         (cortex_a7_call): Likewise.
7283         (cortex_a7_alu_imm): Likewise.
7284         (cortex_a7_alu_reg): Likewise.
7285         (cortex_a7_alu_shift): Likewise.
7286         (cortex_a7_mul): Likewise.
7287         (cortex_a7_load1): Likewise.
7288         (cortex_a7_store1): Likewise.
7289         (cortex_a7_load2): Likewise.
7290         (cortex_a7_store2): Likewise.
7291         (cortex_a7_load3): Likewise.
7292         (cortex_a7_store3): Likewise.
7293         (cortex_a7_load4): Likewise.
7294         (cortex_a7_store4): Likewise.
7295         (cortex_a7_fpalu): Likewise.
7296         (cortex_a7_fconst): Likewise.
7297         (cortex_a7_fpmuls): Likewise.
7298         (cortex_a7_neon_mul): Likewise.
7299         (cortex_a7_fpmacs): Likewise.
7300         (cortex_a7_neon_mla: Likewise.
7301         (cortex_a7_fpmuld: Likewise.
7302         (cortex_a7_fpmacd: Likewise.
7303         (cortex_a7_fpfmad: Likewise.
7304         (cortex_a7_fdivs: Likewise.
7305         (cortex_a7_fdivd: Likewise.
7306         (cortex_a7_r2f: Likewise.
7307         (cortex_a7_f2r: Likewise.
7308         (cortex_a7_f_flags: Likewise.
7309         (cortex_a7_f_loads: Likewise.
7310         (cortex_a7_f_loadd: Likewise.
7311         (cortex_a7_f_stores: Likewise.
7312         (cortex_a7_f_stored: Likewise.
7313         (cortex_a7_neon): Likewise.
7314         * config/arm/cortex-a8-neon.md
7315         (cortex_a8_neon_mrc): Update for attribute change.
7316         (cortex_a8_neon_mrrc): Likewise.
7317         (cortex_a8_neon_int_1): Likewise.
7318         (cortex_a8_neon_int_2): Likewise.
7319         (cortex_a8_neon_int_3): Likewise.
7320         (cortex_a8_neon_int_4): Likewise.
7321         (cortex_a8_neon_int_5): Likewise.
7322         (cortex_a8_neon_vqneg_vqabs): Likewise.
7323         (cortex_a8_neon_vmov): Likewise.
7324         (cortex_a8_neon_vaba): Likewise.
7325         (cortex_a8_neon_vaba_qqq): Likewise.
7326         (cortex_a8_neon_vsma): Likewise.
7327         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7328         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7329         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7330         long_scalar): Likewise.
7331         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7332         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7333         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7334         long_scalar_qdd_64_32_long): Likewise.
7335         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7336         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7337         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7338         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7339         (cortex_a8_neon_shift_1): Likewise.
7340         (cortex_a8_neon_shift_2): Likewise.
7341         (cortex_a8_neon_shift_3): Likewise.
7342         (cortex_a8_neon_vshl_ddd): Likewise.
7343         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7344         (cortex_a8_neon_vsra_vrsra): Likewise.
7345         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7346         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7347         (cortex_a8_neon_fp_vsum): Likewise.
7348         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7349         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7350         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7351         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7352         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7353         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7354         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7355         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7356         (cortex_a8_neon_bp_simple): Likewise.
7357         (cortex_a8_neon_bp_2cycle): Likewise.
7358         (cortex_a8_neon_bp_3cycle): Likewise.
7359         (cortex_a8_neon_ldr): Likewise.
7360         (cortex_a8_neon_str): Likewise.
7361         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7362         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7363         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7364         (cortex_a8_neon_vld2_4_regs): Likewise.
7365         (cortex_a8_neon_vld3_vld4): Likewise.
7366         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7367         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7368         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7369         (cortex_a8_neon_vst3_vst4): Likewise.
7370         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7371         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7372         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7373         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7374         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7375         (cortex_a8_neon_mcr): Likewise.
7376         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7377         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7378         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7379         change.
7380         (ca9_neon_mrrc): Likewise.
7381         (cortex_a9_neon_int_1): Likewise.
7382         (cortex_a9_neon_int_2): Likewise.
7383         (cortex_a9_neon_int_3): Likewise.
7384         (cortex_a9_neon_int_4): Likewise.
7385         (cortex_a9_neon_int_5): Likewise.
7386         (cortex_a9_neon_vqneg_vqabs): Likewise.
7387         (cortex_a9_neon_vmov): Likewise.
7388         (cortex_a9_neon_vaba): Likewise.
7389         (cortex_a9_neon_vaba_qqq): Likewise.
7390         (cortex_a9_neon_vsma): Likewise.
7391         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7392         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7393         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7394         long_scalar): Likewise.
7395         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7396         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7397         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7398         long_scalar_qdd_64_32_long): Likewise.
7399         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7400         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7401         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7402         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7403         (cortex_a9_neon_shift_1): Likewise.
7404         (cortex_a9_neon_shift_2): Likewise.
7405         (cortex_a9_neon_shift_3): Likewise.
7406         (cortex_a9_neon_vshl_ddd): Likewise.
7407         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7408         (cortex_a9_neon_vsra_vrsra): Likewise.
7409         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7410         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7411         (cortex_a9_neon_fp_vsum): Likewise.
7412         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7413         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7414         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7415         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7416         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7417         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7418         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7419         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7420         (cortex_a9_neon_bp_simple): Likewise.
7421         (cortex_a9_neon_bp_2cycle): Likewise.
7422         (cortex_a9_neon_bp_3cycle): Likewise.
7423         (cortex_a9_neon_ldr): Likewise.
7424         (cortex_a9_neon_str): Likewise.
7425         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7426         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7427         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7428         (cortex_a9_neon_vld2_4_regs): Likewise.
7429         (cortex_a9_neon_vld3_vld4): Likewise.
7430         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7431         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7432         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7433         (cortex_a9_neon_vst3_vst4): Likewise.
7434         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7435         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7436         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7437         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7438         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7439         (cortex_a9_neon_mcr): Likewise.
7440         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7441         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7442         (cortex_a9_fps): Likewise.
7443         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7444         change.
7445         (cortex_m4_fmuls): Likewise.
7446         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7447         change.
7448         (cortex_r4_mrc): Likewise.
7449         * config/arm/iterators.md: Update comment referring to neon_type.
7450         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7451         (iwmmxt_movsi_insn): Likewise.
7452         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7453         attribute change.
7454         (pj4_core_to_vfp): Likewise.
7455         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7456         attribute change.
7457         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7458         (vfp_fstore): Likewise.
7459         * doc/md.texi: Change references to neon_type to refer to type.
7461 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7463         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7465 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7467         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7468         lglobl if not weak.
7470 2013-09-04  Easwaran Raman  <eraman@google.com>
7472         PR middle-end/57370
7473         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7474         (build_and_add_sum): Use it.
7475         (appears_later_in_bb): Simplify code.
7477 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7479         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7481 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7483         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7485 2013-09-04  Jan Hubicka  <jh@suse.cz>
7487         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7488         * ipa-devirt.c: Include diganostic.h
7489         (odr_type_d): Add types and types_set.
7490         (hash_type_name): Work for types with vtables during LTO.
7491         (odr_hasher::remove): Fix comment; destroy types_set.
7492         (add_type_duplicate): New function,
7493         (get_odr_type): Use it.
7494         (dump_type_inheritance_graph): Dump type duplicates.
7495         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7496         graph.
7497         * tree.c (types_same_for_odr): Give exact answers on types with
7498         virtual tables.
7500 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7502         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7503         explaining their differences.
7505 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7507         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7509 2013-09-03  Jeff Law  <law@redhat.com>
7511         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7512         when not threading through a joiner block.  Pass joiner/no joiner
7513         state to register_jump_thread.
7514         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7515         state from argument rather than implying on path length.
7516         Dump the entire jump thread path into debugging dump.
7517         * tree-flow.h (register_jump_thread): Update prototype.
7519 2013-08-29  Xinliang David Li  <davidxl@google.com>
7521         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7522         Remove a trivial gcc_assert.
7524 2013-08-29  Xinliang David Li  <davidxl@google.com>
7526         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7527         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7528         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7529         Delay base decl alignment adjustment.
7530         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7531         * tree-vectorizer.h: New data structure.
7532         (set_dr_misalignment): New function.
7533         (dr_misalignment): Ditto.
7534         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7535         (vectorizable_load): Ditto.
7536         (ensure_base_align): New function.
7537         (vectorize_loops): Add dbg_cnt support.
7538         (execute_vect_slp): Ditto.
7539         * dbgcnt.def: New debug counter.
7540         * Makefile: New dependency.
7542 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7544         Revert:
7546         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7548         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7550 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7552         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7553         function descriptor.
7555 2013-09-03  Richard Biener  <rguenther@suse.de>
7557         * tree-affine.c (add_elt_to_tree): Fix association issue,
7558         avoid useless converts and make sure to always return a
7559         properly typed result.
7561 2013-09-03  Richard Biener  <rguenther@suse.de>
7563         PR middle-end/57656
7564         * fold-const.c (negate_expr_p): Fix division case.
7565         (negate_expr): Likewise.
7567 2013-09-03  Richard Biener  <rguenther@suse.de>
7569         PR lto/58285
7570         * tree-streamer-out.c: Include tm.h.
7571         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7573 2013-09-03  Jan Hubicka  <jh@suse.cz>
7575         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7577 2013-09-03  Alan Modra  <amodra@gmail.com>
7579         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7580         multilibs to big-endian target and vice versa.
7581         * config/rs6000/t-linux64: Use := assignment on all vars.
7582         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7583         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7584         * config/rs6000/t-linux64le: New file.
7585         * config/rs6000/t-linux64bele: New file.
7586         * config/rs6000/t-linux64lebe: New file.
7588 2013-09-02  Jan Hubicka  <jh@suse.cz>
7590         * ipa-inline-transform.c (inline_transform): Do not
7591         optimize_inline_calls when not optimizing.
7593 2013-09-02  Jan Hubicka  <jh@suse.cz>
7595         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7596         duplicated nodes for assembler names.
7597         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7598         hard registers from assembler name hash.
7600 2013-09-02  Jan Hubicka  <jh@suse.cz>
7602         * ipa-split.c (execute_split_functions): Split externally visible
7603         functions called once.
7605 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7607         PR ipa/58106
7608         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7609         linked list.  When finding the correct duplicate, also consider also
7610         the caller in additon to its inlined_to node.
7612 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7614         * config/aarch64/aarch64-simd-builtins.def
7615         (dup_lane_scalar): Remove.
7616         * config/aarch64/aarch64-simd.md
7617         (aarch64_simd_dup): Add 'w->w' alternative.
7618         (aarch64_dup_lane<mode>): Allow for VALL.
7619         (aarch64_dup_lane_scalar<mode>): Remove.
7620         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7621         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7622         (aarch64_get_lane_unsigned<mode>): Likewise.
7623         (aarch64_get_lane<mode>): Likewise.
7624         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7625         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7626         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7627         (VCON): Change container of V2SF.
7628         (vswap_width_name): Likewise.
7629         * config/aarch64/arm_neon.h
7630         (__aarch64_vdup_lane_any): New.
7631         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7632         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7633         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7635 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7637         PR middle-end/56382
7638         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7639         the source or the destination is a single hard register.
7641 2013-09-02  Richard Biener  <rguenther@suse.de>
7643         PR middle-end/57511
7644         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7645         non-linear SCEVs.
7647 2013-09-02  Richard Biener  <rguenther@suse.de>
7649         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7650         arithmetic to sizetype.
7652 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7654         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7655         Find auto-increment use both before and after candidate.
7657 2013-09-02  Marek Polacek  <polacek@redhat.com>
7659         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7661 2013-09-01  Jan Hubicka  <jh@suse.cz>
7663         * Makefile.in: Add ipa-profile.o
7664         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7665         * cgraph.c (struct cgraph_propagate_frequency_data,
7666         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7667         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7668         * cgraph.h (cgraph_propagate_frequency): Remove.
7669         * ipa-inline-analysis.c: Include ipa-utils.h;
7670         drop duplicated cfgloop.h.
7671         (inline_update_callee_summaries): Update.
7672         * ipa-profile.c: New file.
7673         * ipa-utils.h (ipa_propagate_frequency): Declare.
7674         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7675         data-streamer.h, value-prof.h.
7676         (symtab_remove_unreachable_nodes): Update profile.
7677         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7678         account_time_size, cmp_counts, dump_histogram,
7679         ipa_profile_generate_summary, ipa_profile_write_summary,
7680         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7681         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7682         Move to ipa-profile.c.
7684 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7686         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7688 2013-09-01  Jan Hubicka  <jh@suse.cz>
7690         * common.opt (fdevirtualize-speculatively): New function.
7691         * invoke.texi (fdevirtualize-speculatively): Document.
7692         * ipa-devirt.c: Include ipa-inline.h
7693         (likely_target_p): New function.
7694         (ipa_devirt): New function.
7695         (gate_ipa_devirt): New function.
7696         (pass_data_ipa_devirt): New static var.
7697         (pass_ipa_devirt): Likewise.
7698         (make_pass_ipa_devirt): New function.
7699         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7700         (common_handle_option): Disable devirtualization when
7701         value range profiling is available.
7702         * passes.def (pass_ipa_devirt): Add.
7703         * timever.def (TV_IPA_DEVIRT): New timevar.
7704         * tree-pass.h (make_pass_ipa_devirt):
7706 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7708         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7709         include sanitize(undefined).
7711 2013-08-31  Diego Novillo  <dnovillo@google.com>
7713         * Makefile.in (TREE_CORE_H): Define.
7714         (TREE_H): Use.
7715         (GTFILES): Add tree-core.h.
7716         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7717         size the array.
7718         * tree-core.h: New file.
7719         Move all data structures, enum, typedefs, global
7720         declarations and constants from ...
7721         * tree.h: ... here.
7723 2013-08-31  Jan Hubicka  <jh@suse.cz>
7725         * bulitins.c (expand_builtin): Do not early exit for gcov
7726         instrumented functions.
7728 2013-08-31  Marek Polacek  <polacek@redhat.com>
7730         * ubsan.c: Include tm_p.h.
7732 2013-08-31  Jan Hubicka  <jh@suse.cz>
7734         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7735         warning.
7737         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7738         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7739         * tree-cfg.c (verify_gimple_label): ... here.
7740         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7741         (ipa_merge_profiles): New function.
7742         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7743         (lto_input_function_body): Likewise.
7744         * ipa-utils.h (ipa_merge_profiles): Declare.
7745         * lto-streamer.h (lto_input_function_body): Update prototype.
7746         (emit_label_in_global_context_p): Remove.
7747         * lto-symtab.c: Include ipa-utils.h
7748         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7750 2013-08-31  Jan Hubicka  <jh@suse.cz>
7752         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7754 2013-08-31  Jan Hubicka  <jh@suse.cz>
7756         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7758 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7760         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7761         "cmp" RTX before signed_comparison_operator check to account
7762         for "code" changes.
7764 2013-08-30  Jan Hubicka  <jh@suse.cz>
7766         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7767         (detect_type_change_1): Rename to ...
7768         (detect_type_change): ... this one; early return on non-polymorphic
7769         types.
7770         (detect_type_change_ssa): Add comp_type parameter; update
7771         use of detect_type_change.
7772         (compute_complex_assign_jump_func): Add param_type parameter;
7773         update use of detect_type_change_ssa.
7774         (compute_complex_ancestor_jump_func): Likewise.
7775         (ipa_get_callee_param_type): New function.
7776         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7777         update calls to the jump function computation functions.
7779 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7780             Steven Bosscher  <steven@gcc.gnu.org>
7782         * cfgrtl.c (fixup_new_cold_bb): New routine.
7783         (commit_edge_insertions): Invoke fixup_partitions.
7784         (find_partition_fixes): New routine.
7785         (fixup_partitions): Ditto.
7786         (verify_hot_cold_block_grouping): Update comments.
7787         (rtl_verify_edges): Invoke find_partition_fixes.
7788         (rtl_verify_bb_pointers): Update comments.
7789         (rtl_verify_bb_layout): Ditto.
7790         * basic-block.h (probably_never_executed_edge_p): Declare.
7791         (fixup_partitions): Ditto.
7792         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7793         * bb-reorder.c (sanitize_hot_paths): New function.
7794         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7795         sanitize_hot_paths.
7796         * predict.c (probably_never_executed_edge_p): New routine.
7797         * cfg.c (check_bb_profile): Add partition insanity warnings.
7799 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7801         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7803 2013-08-30  Marek Polacek  <polacek@redhat.com>
7805         * Makefile.in (ubsan.o): Add.
7806         (c-family/c-ubsan.o): Add.
7807         (builtins.o): Add ubsan.h dependency.
7808         * ubsan.h: New file.
7809         * ubsan.c: New file.
7810         * common.opt: Add -fsanitize=undefined option.
7811         (flag_sanitize): Add variable.
7812         (fsanitize=): Add option.  Add Driver.
7813         (fsanitize=thread): Remove option.
7814         (fsanitize=address): Likewise.
7815         (static-libubsan): New option.
7816         * doc/invoke.texi: Document the new flag and -static-libubsan.
7817         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7818         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7819         * builtin-attrs.def (ATTR_COLD): Define.
7820         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7821         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7822         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7823         * flag-types.h (sanitize_code): New enum.
7824         * opts.c (common_handle_option): Parse command line arguments
7825         of -fsanitize=.  Add -fsanitize=unreachable option.
7826         * varasm.c (get_variable_section): Adjust.
7827         (assemble_noswitch_variable): Likewise.
7828         (assemble_variable): Likewise.
7829         (output_constant_def_contents): Likewise.
7830         (categorize_decl_for_section): Likewise.
7831         (place_block_symbol): Likewise.
7832         (output_object_block): Likewise.
7833         * builtins.def: Likewise.
7834         * toplev.c (compile_file): Likewise.
7835         (process_options): Likewise.
7836         * cppbuiltin.c: Likewise.
7837         * tsan.c (tsan_pass): Likewise.
7838         (tsan_gate): Likewise.
7839         (tsan_gate_O0): Likewise.
7840         * cfgexpand.c (partition_stack_vars): Likewise.
7841         (expand_stack_vars): Likewise.
7842         (defer_stack_allocation): Likewise.
7843         (expand_used_vars): Likewise.
7844         * cfgcleanup.c (old_insns_match_p): Likewise.
7845         * asan.c (asan_finish_file): Likewise.
7846         (asan_instrument): Likewise.
7847         (gate_asan): Likewise.
7848         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7849         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7850         (asan_global_struct): Use pointer_sized_int_node instead
7851         calling build_nonstandard_integer_type.
7852         (initialize_sanitizer_builtins): Likewise.
7853         (asan_finish_file): Likewise.
7854         * gcc.c: Document %{%:function(args):X}.
7855         (static_spec_functions): Add sanitize.
7856         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7857         store funcval != NULL there.
7858         (do_spec_1): Adjust handle_spec_function caller.
7859         (handle_braces): Allow %:function(args) as condition.
7860         (sanitize_spec_function): New function.
7861         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7862         (LIBUBSAN_SPEC): Likewise.
7863         (LIBUBSAN_EARLY_SPEC): Likewise.
7864         (SANITIZER_SPEC): Handle libubsan.
7865         (SANITIZER_EARLY_SPEC): Likewise.
7866         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7867         instead of fsanitize=address.
7868         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7869         instead of fsanitize=address*.
7870         * builtins.c: Include ubsan.h.
7871         (fold_builtin_0): Instrument __builtin_unreachable.
7872         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7873         instead of flag_asan.
7874         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7875         (pointer_sized_int_node): Define.
7876         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7878 2013-08-30  Mike Stump  <mikestump@comcast.net>
7880         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7881         with RE patterns.
7883 2013-08-29  Jan Hubicka  <jh@suse.cz>
7885         * cgraph.c (cgraph_function_body_availability): Handle weakref
7886         correctly.
7887         * passes.def: Remove pass_fixup_cfg.
7888         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7889         track when we need to remove functions.
7890         (gate_ipa_inline): Execute inlining always; add comment why.
7891         (pass_data_ipa_inline): Remove TODO_remove_functions.
7892         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7893         do not produce summaries.
7894         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7895         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7896         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7897         constant pool and vtable.
7899 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7901         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7902         New arm_neon.h's internal macros to specify 64-bit constants.
7903         Avoid using stdint.h's macros.
7905 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7907         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7908         before applying REG_P.
7910 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7912         PR tree-optimization/58277
7913         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7914         after seeing too many stmts with vdef in between dombb and current
7915         bb, invalidate everything.
7917 2013-08-30  Richard Biener  <rguenther@suse.de>
7919         * fold-const.c (fold_single_bit_test): Fix overflow test.
7921 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7923         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7924         and which can live in a register, always retrieve the value on entry.
7925         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7926         passed by invisible reference specially.
7927         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7928         (vt_add_function_parameter): Correctly deal with a parameter passed by
7929         invisible reference.
7931 2013-08-30  Jan Hubicka  <jh@suse.cz>
7933         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7935 2013-08-30  Richard Biener  <rguenther@suse.de>
7937         PR tree-optimization/58228
7938         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7939         allow invariant loads in nested loop vectorization.
7941 2013-08-30  Richard Biener  <rguenther@suse.de>
7943         PR tree-optimization/58223
7944         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7945         (has_anti_or_output_dependence): ... this and adjust to also
7946         look for output dependences.
7947         (mark_nodes_having_upstream_mem_writes): Adjust.
7948         (rdg_flag_uses): Likewise.
7950 2013-08-30  Richard Biener  <rguenther@suse.de>
7952         PR tree-optimization/58010
7953         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7954         assert that we have a loop-closed PHI.
7956 2013-08-29  Jan Hubicka  <jh@suse.cz>
7958         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7959         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7960         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7961         (lto_free_function_in_decl_state_for_node): New function.
7963 2013-08-29  Xinliang David Li  <davidxl@google.com>
7965         * loop-unroll.c (report_unroll_peel): Minor message change.
7966         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7967         Emit alignment peeling message with default -fopt-info.
7968         (vect_loop_versioning): Emit loop version info message.
7969         * tree-vectorizer.c (vectorize_loops): Minor message change.
7970         (execute_vect_slp): Ditto.
7972 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7974         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7975         of the clone from the DECL_NAME of the original function.
7977 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7979         * passes.c (register_pass): Add overload.
7980         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7982 2013-08-29  Jan Hubicka  <jh@suse.cz>
7984         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7985         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7986         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7987         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7988         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7989         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7990         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7991         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7993 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7995         * dumpfile.c (dump_loc): Output column number.
7996         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7997         * doc/invoke.texi: Document optall -fopt-info flag.
7998         * profile.c (read_profile_edge_counts): Use new dump framework.
7999         (compute_branch_probabilities): Ditto.
8000         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8001         when pass not in any opt group.
8002         * pass_manager.h (pass_manager::get_pass_profile): New method.
8003         * value-prof.c (check_counter): Use new dump framework.
8004         (check_ic_target): Ditto.
8005         * coverage.c (get_coverage_counts): Ditto.
8006         (coverage_init): Setup new dump framework.
8008 2013-08-29  Richard Biener  <rguenther@suse.de>
8010         PR tree-optimization/58246
8011         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8012         handle the dominance check inside a basic-block.
8014 2013-08-29  Richard Biener  <rguenther@suse.de>
8016         PR middle-end/57287
8017         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8018         of default defs that appear in abnormal PHI nodes.
8020 2013-08-29  Richard Biener  <rguenther@suse.de>
8022         PR tree-optimization/57685
8023         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8024         single-use operands to avoid exponential complexity.
8026 2013-08-28  Dehao Chen  <dehao@google.com>
8028         * ipa-inline.c (edge_badness): Fix integer underflow.
8030 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8032         * gtm-builtins.def (_ITM_free): Declare leaf.
8034 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8036         PR target/58067
8037         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8038         (*tls_local_dynamic_base_64_largepic): Likewise.
8039         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8040         Remove predicate from call operand.
8041         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8042         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8044 2013-08-28  Jeff Law  <law@redhat.com>
8046         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8047         checks for the number of predecessors and successors allowed.
8048         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8049         which require copying a joiner block if there is a request which
8050         is a subpath that requires no joiner block copying.
8052 2013-08-28  Jan Hubicka  <jh@suse.cz>
8054         * lto-streamer-out.c (DFS_write_tree_body): Drop
8055         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8056         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8057         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8058         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8059         Do not read DECL_ERROR_ISSUED.
8060         (unpack_ts_decl_with_vis_value_fields): Do not read
8061         DECL_DEFER_OUTPUT.
8062         (lto_input_ts_binfo_tree_pointers): Do not read
8063         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8064         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8065         write DECL_ERROR_ISSUED..
8066         (pack_ts_decl_with_vis_value_fields): Do not write
8067         DECL_DEFER_OUTPUT.
8068         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8069         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8070         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8071         * tree.h (tree_decl_common): Update comment.
8072         (DECL_ERROR_ISSUED): Remove.
8074 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8076         PR middle-end/58257
8077         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8079 2013-08-28  Jan Hubicka  <jh@suse.cz>
8081         * builtins.def (free): Declare leaf.
8083 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8085         * gdbhooks.py: New.
8086         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8087         * configure: Regenerate.
8089 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8091         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8092         (ipa_ancestor_jf_data): Likewise.
8093         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8094         (ipa_get_jf_pass_through_type_preserved): New function.
8095         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8096         (ipa_get_jf_ancestor_type_preserved): New function.
8097         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8098         (ipa_get_jf_ancestor_result): Likewise.
8099         (propagate_vals_accross_pass_through): Use
8100         ipa_get_jf_pass_through_result to do all the value mappings.
8101         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8102         type_preserved flag.
8103         (ipa_set_jf_cst_copy): New function.
8104         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8105         (ipa_set_jf_arith_pass_through): Likewise.
8106         (ipa_set_ancestor_jf): Likewise.
8107         (compute_complex_assign_jump_func): Set type_preserved instead of
8108         punting.
8109         (ipa_compute_jump_functions_for_edge): Likewise.
8110         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8111         (update_jump_functions_after_inlining): Update type_preserved.
8112         Explicitely create jump functions when combining one with pass_through.
8113         (ipa_write_jump_function): Stream the type_preserved flags.
8114         (ipa_read_jump_function): Likewise.
8116 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8117             Aldy Hernandez  <aldyh@redhat.com>
8119         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8120         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8121         * function.h (struct function): Add has_force_vect_loops and
8122         has_simduid_loops.
8123         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8124         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8125         handle it.
8126         * gimple.def: Update CLAUSES comments.
8127         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8128         (gimple_build_omp_for): Add argument to prototype.
8129         (gimple_omp_for_kind): New.
8130         (gimple_omp_for_set_kind): New.
8131         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8132         GOVD_DATA_SHARE_CLASS.
8133         (enum omp_region_type): Add ORT_SIMD.
8134         (gimple_add_tmp_var): Handle ORT_SIMD.
8135         (gimplify_var_or_parm_decl): Same.
8136         (is_gimple_stmt): Same.
8137         (omp_firstprivatize_variable): Same.
8138         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8139         (omp_notice_variable): Handle ORT_SIMD.
8140         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8141         (omp_check_private): Handle ORT_SIMD.
8142         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8143         OMP_CLAUSE_SAFELEN.
8144         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8145         Handle OMP_CLAUSE_LASTPRIVATE.
8146         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8147         OMP_CLAUSE_SAFELEN.
8148         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8149         (gimplify_expr): Handle OMP_SIMD.
8150         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8151         (expand_GOMP_SIMD_VF): New.
8152         (expand_GOMP_SIMD_LAST_LANE): New.
8153         * internal-fn.def (GOMP_SIMD_LANE): New.
8154         (GOMP_SIMD_VF): New.
8155         (GOMP_SIMD_LAST_LANE): New.
8156         * omp-low.c: Include target.h.
8157         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8158         OMP_CLAUSE_SAFELEN.
8159         (check_omp_nesting_restrictions): Same.
8160         (omp_max_vf): New.
8161         (lower_rec_simd_input_clauses): New.
8162         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8163         OMP_CLAUSE_LINEAR.
8164         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8165         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8166         (expand_omp_build_assign): New.
8167         (expand_omp_for_init_counts): New.
8168         (expand_omp_for_init_vars): New.
8169         (extract_omp_for_update_vars): New.
8170         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8171         and rewrite accordingly.
8172         (expand_omp_simd): New.
8173         (expand_omp_for): Use expand_omp_simd.
8174         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8175         (lower_omp_for): Do not lower the body.
8176         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8177         in their own loops.
8178         * tree-flow.h (find_omp_clause): Remove prototype.
8179         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8180         forcing vectorization of the loop, or if flag_tree_vectorize.
8181         (gate_tree_if_conversion): Similarly.
8182         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8183         gimple_build_omp_for.
8184         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8185         * tree-parloops (create_parallel_loop): Pass kind argument to
8186         gimple_build_omp_for.
8187         * tree-pretty-print.c (dump_omp_clause): Add cases for
8188         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8189         OMP_CLAUSE__SIMDUID_.
8190         (dump_generic_node): Handle OMP_SIMD.
8191         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8192         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8193         unroll OMP_SIMD loops here.
8194         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8195         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8196         loop->safelen.
8197         (vect_analyze_data_refs): Handle simd loops.
8198         * tree-vect-loop.c (vectorizable_live_operation): Handle
8199         IFN_GOMP_SIMD*.
8200         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8201         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8202         (vectorizable_load): Same.
8203         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8204         (struct simduid_to_vf): New.
8205         (simduid_to_vf::hash): New.
8206         (simduid_to-vf::equal): New.
8207         (struct simd_array_to_simduid): New.
8208         (simd_array_to_simduid::hash): New.
8209         (simd_array_to_simduid::equal): New.
8210         (adjust_simduid_builtins): New.
8211         (struct note_simd_array_uses_struct): New.
8212         (note_simd_array_uses_cb): New.
8213         (note_simd_array_uses): New.
8214         (vectorize_loops): Handle simd hints and adjust simd builtins
8215         accordingly.
8216         * tree-vectorizer.h (struct _stmt_vec_info): Add
8217         simd_lane_access_p field.
8218         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8219         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8220         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8221         (omp_clause_code_name): Same.
8222         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8223         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8224         * tree.def (OMP_SIMD): New entry.
8225         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8226         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8227         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8228         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8229         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8230         (OMP_CLAUSE_LINEAR_STEP): New.
8231         (OMP_CLAUSE_SAFELEN_EXPR): New.
8232         (OMP_CLAUSE__SIMDUID__DECL): New.
8233         (find_omp_clause): New prototype.
8235 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8237         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8238         Haswell processor detection.
8240 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8242         PR target/57927
8243         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8244         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8245         AVX2 capable processors.
8247 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8249         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8250         of vget_low_* with implementations in terms of other intrinsics.
8252 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8254         PR middle-end/57219
8255         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8256         values to -1, 0 and 1.
8258 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8260         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8261         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8262         (<optab><mode>3_insn): Remove.
8263         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8264         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8265         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8266         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8267         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8268         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8269         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8270         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8271         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8272         (ror<mode>3_insn): Likewise.
8273         * config/aarch64/predicates.md (aarch64_simd_register): New.
8275 2013-08-27  Richard Biener  <rguenther@suse.de>
8277         PR tree-optimization/57521
8278         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8279         one edge is non-critical.
8280         (find_phi_replacement_condition): Make sure to use a non-critical
8281         edge.  Cleanup and remove old bug workarounds.
8282         (bb_postdominates_preds): Remove.
8283         (if_convertible_loop_p_1): Do not compute post-dominators.
8284         (combine_blocks): Do not free post-dominators.
8285         (main_tree_if_conversion): Likewise.
8286         (pass_data_if_conversion): Add TODO_verify_ssa.
8288 2013-08-27  DJ Delorie  <dj@redhat.com>
8290         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8292 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8294         * function.c (assign_parm_find_data_types): Set passed_mode and
8295         nominal_mode to the TYPE_MODE of nominal_type for the built
8296         pointer type in case of the struct-pass-by-reference.
8298 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8300         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8301         (UINT16_TYPE): Change default to "unsigned int".
8303         * config/avr/avr.opt (mfract-convert-truncate): New option.
8304         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8305         is set, round negative fractional integers according to n1169
8306         when converting to integer types.
8308 2013-08-26  Jan Hubicka  <jh@suse.cz>
8310         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8311         methods can not be called indirectly when their address is not taken.
8313 2013-08-26  Jan Hubicka  <jh@suse.cz>
8315         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8316         ctor_for_folding.
8318 2013-08-26  Jan Hubicka  <jh@suse.cz>
8320         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8321         can be unshared.
8323 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8325         * reload.c (find_valid_class): Allow classes that do not include
8326         FIRST_PSEUDO_REGISTER - 1.
8328 2013-08-26  Jan Hubicka  <jh@suse.cz>
8330         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8331         fix edge count/frequency when speculation failed; fix type check
8332         for the direct call.
8334 2013-08-26  Jan Hubicka  <jh@suse.cz>
8336         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8338 2013-08-26  Jan Hubicka  <jh@suse.cz>
8340         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8341         to be changed by edge redirection.
8343 2013-08-26  Jan Hubicka  <jh@suse.cz>
8345         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8346         formating; add sanity check.
8347         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8348         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8349         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8350         (resolve_noninline_speculation): Update callee keys, too.
8352 2013-08-26  Jan Hubicka  <jh@suse.cz>
8354         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8355         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8357 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8359         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8360         into proper place.
8362 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8364         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8365         (x86_64_elf_select_section): Ditto.
8366         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8367         arguments.
8368         (ix86_pass_by_reference): Ditto.
8369         (output_set_got): Ditto.
8370         (ix86_unary_operator_ok): Ditto.
8371         (ix86_expand_builtin): Ditto.
8373 2013-08-23  Jan Hubicka  <jh@suse.cz>
8375         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8377 2013-08-23  Jan Hubicka  <jh@suse.cz>
8379         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8380         (tree_decl_with_vis): Add FINAL field.
8382 2013-08-23  Jeff Law  <law@redhat.com>
8384         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8385         the debugging dump when the expression is fully redundant.
8387 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8389         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8390         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8391         * pretty-print.c (pp_formatted_text_data): Likewise.
8392         (pp_write_text_to_stream): Likewise.
8393         (pp_write_text_as_dot_label_to_stream): Likewise.
8394         (pp_append_r): Likewise.
8395         (pp_format): Likewise.
8396         (pp_flush): Likewise.
8397         (pp_clear_output_area): Likewise.
8398         (pp_append_text): Likewise.
8399         (pp_formatted_text): Likewise.
8400         (pp_remaining_character_count_for_line): Likewise.
8401         (pp_newline): Likewise.
8402         (pp_character): Likewise.
8403         (output_buffer::~output_buffer): Define.
8404         (pretty_printer::~pretty_printer): Destruct output buffer.
8405         * pretty-print.h (output_buffer::~output_buffer): Declare.
8406         (pretty_printer::~pretty_printer): Declare virtual.
8408 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8410         PR other/57324
8411         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8412         HOST_WIDE_INT_M1U): New macros.
8413         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8414         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8415         unsigned -1 for lshift.
8416         * cse.c (cse_insn): Likewise.
8417         * double-int.c (rshift_double, lshift_double): Likewise.
8418         * builtins.c (fold_builtin_bitop): Likewise.
8419         * combine.c (force_to_mode): Likewise.
8420         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8421         * simplify-rtx.c (simplify_const_unary_operation,
8422         simplify_const_binary_operation): Likewise.
8423         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8424         check_va_list_escapes): Likewise.
8425         * rtlanal.c (nonzero_bits1): Likewise.
8426         * expmed.c (expand_smod_pow2): Likewise.
8427         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8429 2013-08-23  Jan Hubicka  <jh@suse.cz>
8431         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8432         as having address taken.
8434 2013-08-23  Jan Hubicka  <jh@suse.cz>
8436         * ipa-utils.h (method_class_type): Declare.
8437         * ipa-devirt.c (method_class_type): Export.
8439         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8440         do not walk base classes of anonymous types.
8442 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8444         PR rtl-optimization/58220
8445         PR regression/58221
8446         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8447         handle SEQUENCE insns properly.
8449 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8451         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8452         definition.
8453         (pp_newline_and_indent): Likewise.
8454         (pp_separate_with): Likewise.
8455         * pretty-print.c (pp_newline_and_flush): Define.
8456         (pp_newline_and_indent): Likewise.
8457         (pp_separate_with): Likewise.
8459 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8461         PR target/58218
8462         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8463         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8465 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8467         * config/i386/predicates.md (ext_sse_reg_operand): New.
8468         * config/i386/i386.md (*movti_internal): Use
8469         predicate to determine if EVEX is needed.
8470         (*movsi_internal): Ditto.
8471         (*movdf_internal): Ditto.
8472         (*movsf_internal): Ditto.
8473         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8475 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8477         PR tree-optimization/58209
8478         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8479         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8480         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8481         emit POINTER_PLUS_EXPR.
8482         (create_tailcall_accumulator): For pointer result type accumulate in
8483         sizetype type.
8485 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8487         * configure.ac: Add backslashes missing from the last change.
8488         * configure: Regenerate.
8490 2013-08-22  Jan Hubicka  <jh@susue.cz>
8492         * ipa.c (function_and_variable_visibility): First remember function
8493         was global and then make it local.
8495 2013-08-22  Julian Brown  <julian@codesourcery.com>
8497         * configure.ac: Add aarch64 to list of arches which use "nop" in
8498         debug_line test.
8499         * configure: Regenerate.
8501 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8503         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8504         gnu_libc_has_function.
8505         * config/s390/tpf.h: Likewise.
8507 2013-08-22  Jan Hubicka  <jh@susue.cz>
8509         * timevar.c (validate_phases): Add cast.
8511 2013-08-22  Jan Hubicka  <jh@susue.cz>
8513         * timevar.c (validate_phases): Use size_t for memory.
8514         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8516 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8518         * pretty-print.h (output_buffer::output_buffer): Declare.
8519         (pretty_printer::pretty_printer): Likewise.
8520         (pp_construct): Remove.
8521         * pretty-print.c (output_buffer::output_buffer): Define.
8522         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8523         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8524         (print_gimple_expr): Likewise.
8525         (print_gimple_seq): Likewise.
8526         (gimple_dump_bb): Likewise.
8527         * sched-vis.c (dump_value_slim): Likewise.
8528         (dump_insn_slim): Likewise.
8529         (dump_rtl_slim): Likewise.
8530         (str_pattern_slim): Likewise.
8531         * tree-mudflap.c (mf_varname_tree): Likewise.
8532         * graph.c (print_graph_cfg): Likewise.
8533         (start_graph_dump): Likewise.
8534         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8535         placement-new.
8536         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8537         pretty printer initialization.
8538         * coretypes.h (diagnostic_context): Remove superflous type alias
8539         declaration.
8540         (pretty_printer): Likewise.  Declare directly as a class.
8541         (pretty_print_info): Remove declaration as class.
8542         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8543         and pp_clear_output_area.
8544         (asan_add_global): Likewise.
8546 2013-08-22  Jan Hubicka  <jh@suse.cz>
8548         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8549         * ipa-utils.h (update_type_inheritance_graph): Declare.
8550         (possible_polymorphic_call_target_p): Declare.
8551         (possible_polymorphic_call_target_p): New.
8552         * ipa-devirt.c: Update toplevel comments.
8553         (cached_polymorphic_call_targets): Move up.
8554         (odr_type_d): Move ID down.
8555         (polymorphic_type_binfo_p): Update comment.
8556         (odr_hasher::remove): Likewise;
8557         (get_odr_type): Set anonymous_namespace.
8558         (dump_odr_type): Dump it.
8559         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8560         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8561         (record_binfo): Add comment.
8562         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8563         built.
8564         (devirt_node_removal_hook): Do not iCE when cache is freed.
8565         (possible_polymorphic_call_target_p): New predicate.
8566         (update_type_inheritance_graph): New function.
8568 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8569             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8570             Sergey Lega  <sergey.s.lega@intel.com>
8571             Anna Tikhonova  <anna.tikhonova@intel.com>
8572             Ilya Tocar  <ilya.tocar@intel.com>
8573             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8574             Ilya Verbin  <ilya.verbin@intel.com>
8575             Kirill Yukhin  <kirill.yukhin@intel.com>
8576             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8578         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8579         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8580         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8581         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8582         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8583         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8584         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8585         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8586         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8587         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8588         OPT_mavx512pf, OPT_mavx512er cases.
8589         * config/i386/constraints.md (v): New constraint.
8590         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8591         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8592         (bit_AVX512CD): New.
8593         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8594         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8595         * config/i386/i386-c.c (ix86_target_macros_internal):
8596         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8597         __AVX512PF__.
8598         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8599         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8600         * config/i386/i386.c (regclass_map, dbx_register_map)
8601         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8602         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8603         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8604         -mavx512pf options.
8605         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8606         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8607         -mavx512pf options.  Fix formatting.
8608         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8609         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8610         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8611         -mavx512cd, -mavx512pf options.
8612         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8613         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8614         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8615         ALL_SSE_REGS.
8616         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8617         (ix86_set_reg_reg_cost): Ditto.
8618         (x86_order_regs_for_local_alloc): Ditto.
8619         (MAX_VECT_LEN): Extend to 64-byte.
8620         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8621         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8622         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8623         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8624         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8625         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8626         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8627         (SSE_REG_MODE_P): Support new modes.
8628         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8629         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8630         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8631         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8632         (REG_CLASS_CONTENTS): Add new registers.
8633         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8634         (EXT_REX_SSE_REGNO_P): New.
8635         (HI_REGISTER_NAMES): Add new registers.
8636         * config/i386/i386.md: Define constants for new registers.
8637         (mode): Add new 512-bit modes.
8638         (prefix): Support evex prefix.
8639         (isa): Support avx512f, noavx512f, fma_avx512f.
8640         (ssemodesuffix): Add new 512-bit modes.
8641         (movxi): New.
8642         (*movxi_internal_avx512f): Ditto.
8643         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8644         Support MODE_XI.
8645         (*movsi_internal): Likewise.
8646         (*movdf_internal): Likewise.
8647         (*movsf_internal): Likewise.
8648         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8649         (<code><mode>3): Likewise.
8650         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8651         New.
8652         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8653         with the new constraint "v".
8654         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8655         modes.
8656         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8657         with the new constraint "v".
8658         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8659         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8660         (*<plusminus_insn><mode>3): Likewise.
8661         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8662         (*mul<mode>3): Likewise.
8663         (<sse>_vmmul<mode>3): Likewise.
8664         (<sse>_div<mode>3): Likewise.
8665         (<sse>_vmdiv<mode>3): Likewise.
8666         (<sse>_sqrt<mode>2): Likewise.
8667         (<sse>_vmsqrt<mode>2): Likewise.
8668         (*<code><mode>3_finite): Likewise.
8669         (*<code><mode>3) <smaxmin>: Likewise.
8670         (<sse>_vm<code><mode>3): Likewise.
8671         (*<code><mode>3) <any_logic>: Likewise.
8672         (*fma_fmadd_<mode>): Likewise.
8673         (*fma_fmsub_<mode>): Likewise.
8674         (*fma_fnmadd_<mode>): Likewise.
8675         (*fma_fnmsub_<mode>): Likewise.
8676         (*fma_fmaddsub_<mode>): Likewise.
8677         (*fma_fmsubadd_<mode>): Likewise.
8678         (*fmai_fmadd_<mode>): Likewise.
8679         (*fmai_fmsub_<mode>): Likewise.
8680         (*fmai_fnmadd_<mode>): Likewise.
8681         (*fmai_fnmsub_<mode>): Likewise.
8682         (sse_cvtsi2ss): Likewise.
8683         (sse_cvtsi2ssq): Likewise.
8684         (sse_cvtss2si): Likewise.
8685         (sse_cvtss2si_2): Likewise.
8686         (sse_cvtss2siq): Likewise.
8687         (sse_cvtss2siq_2): Likewise.
8688         (sse_cvttss2si): Likewise.
8689         (sse_cvtss2siq_2): Likewise.
8690         (float<sseintvecmodelower><mode>2): Likewise.
8691         (sse2_cvtsd2si_2): Likewise.
8692         (sse2_cvtsd2siq_2): Likewise.
8693         (*<plusminus_insn><mode>3): Likewise.
8694         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8695         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8696         (ashr<mode>3): Likewise.
8697         (<shift_insn><mode>3): Likewise.
8698         (avx2_<code><mode>3): Likewise.
8699         (*avx2_<code><mode>3): Likewise.
8700         (*andnot<mode>3): Likewise.
8701         (*<code><mode>3) <any_logic>: Likewise.
8702         (abs<mode>2): Likewise.
8703         (avx2_permvar<mode>): Likewise.
8704         (avx2_perm<mode>_1): Likewise.
8705         (*avx_vpermilp<mode>): Likewise.
8706         (avx_vpermilvar<mode>3): Likewise.
8707         (avx2_ashrv<mode>): Likewise.
8708         (avx2_<shift_insn>v<mode>): Likewise.
8709         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8710         -mavx512cd.
8711         * doc/rtl.texi: Document XImode.
8713 2013-08-21  Jeff Law  <law@redhat.com>
8715         * tree-flow.h (register_jump_thread): Pass vector of edges
8716         instead of each important edge.
8717         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8718         thread path into a vector and pass that to register_jump_thread.
8719         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8720         passed in edge vector to the current 3-edge form.
8722         Revert:
8723         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8725         * dce.c (fini_dce): Call df_analyze again just in case
8726         delete_unmarked_insns removed anything.
8728 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8730         * reload.h (struct reg_equivs): Rename to ..
8731         (struct reg_equivs_s): .. this.
8733 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8735         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8737 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8739         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8741 2013-08-21  Jeff Law  <law@redhat.com>
8743         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8744         simplify assignments too.  If the RHS collapses to a singleton
8745         range, then return the value for the range.
8747 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8749         * config/i386/sse.md (V16): Rename to...
8750         (VMOVE): this.
8751         (mov<mode>): Update iterator name.
8752         (*mov<mode>_internal): Ditto.
8753         (push<mode>1): Ditto.
8754         (movmisalign<mode>): Ditto.
8756 2013-08-20  Jan Hubicka  <jh@suse.cz>
8758         PR bootstrap/58186
8759         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8760         entry for direct edges.
8761         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8763 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8765         Revert my last two changes, r201865 and r201864:
8767         Revert r201865:
8768         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8770         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8771         instances can own GC refs.
8773         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8774         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8775         (gcc::context::gt_pch_nx): Likewise.
8776         (gcc::context::gt_pch_nx):  Likewise.
8777         * ggc.h (gt_ggc_mx <T>): New.
8778         (gt_pch_nx_with_op <T>): New.
8779         (gt_pch_nx <T>): New.
8780         * passes.c (opt_pass::gt_ggc_mx): New.
8781         (opt_pass::gt_pch_nx): New.
8782         (opt_pass::gt_pch_nx_with_op): New.
8783         (pass_manager::gt_ggc_mx): New.
8784         (pass_manager::gt_pch_nx): New.
8785         (pass_manager::gt_pch_nx_with_op): New.
8786         (pass_manager::operator new): Use
8787         ggc_internal_cleared_alloc_stat rather than xcalloc.
8788         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8789         (pass_manager::gt_ggc_mx): New.
8790         (pass_manager::gt_pch_nx): New.
8791         (pass_manager::gt_pch_nx_with_op): New.
8792         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8793         (opt_pass::operator new): New.
8794         (opt_pass::gt_ggc_mx): New.
8795         (opt_pass::gt_pch_nx): New.
8796         (opt_pass::gt_pch_nx_with_op): New.
8798         Revert r201864:
8799         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8801         * Makefile.in (GTFILES): Add context.h.
8802         * context.c (gcc::context::operator new): New.
8803         (gcc::context::gt_ggc_mx): New.
8804         (gcc::context::gt_pch_nx): New.
8805         (gcc::context::gt_pch_nx): New.
8806         * context.h (gcc::context): Add GTY((user)) marking.
8807         (gcc::context::operator new): New.
8808         (gcc::context::gt_ggc_mx): New.
8809         (gcc::context::gt_pch_nx): New.
8810         (gcc::context::gt_pch_nx): New.
8811         (g): Add GTY marking.
8812         (gt_ggc_mx (gcc::context *)): New.
8813         (gt_pch_nx (gcc::context *)): New.
8814         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8815         void *cookie)): New.
8816         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8818 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8820         * dce.c (fini_dce): Call df_analyze again just in case
8821         delete_unmarked_insns removed anything.
8823 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8825         PR rtl-optimizations/57451
8826         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8827         from crossing split section boundaries.
8829 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8831         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8832         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8833         with MULTLIB_DEFAULTS.
8835 2013-08-20  Nick Clifton  <nickc@redhat.com>
8837         * target.def (narrow_volatile_bitfield): Note that the default
8838         value is false, not !TARGET_STRICT_ALIGN.
8839         * doc/tm.texi: Regenerate.
8841 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8843         Fix LIB_SPEC for systems without libpthread.
8845         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8846         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8847         for Android.
8848         * config/i386/linux-common.h: Likewise.
8849         * config/mips/linux-common.h: Likewise.
8851 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8853         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8854         checks.
8856 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8858         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8859         instances can own GC refs.
8861         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8862         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8863         (gcc::context::gt_pch_nx): Likewise.
8864         (gcc::context::gt_pch_nx):  Likewise.
8865         * ggc.h (gt_ggc_mx <T>): New.
8866         (gt_pch_nx_with_op <T>): New.
8867         (gt_pch_nx <T>): New.
8868         * passes.c (opt_pass::gt_ggc_mx): New.
8869         (opt_pass::gt_pch_nx): New.
8870         (opt_pass::gt_pch_nx_with_op): New.
8871         (pass_manager::gt_ggc_mx): New.
8872         (pass_manager::gt_pch_nx): New.
8873         (pass_manager::gt_pch_nx_with_op): New.
8874         (pass_manager::operator new): Use
8875         ggc_internal_cleared_alloc_stat rather than xcalloc.
8876         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8877         (pass_manager::gt_ggc_mx): New.
8878         (pass_manager::gt_pch_nx): New.
8879         (pass_manager::gt_pch_nx_with_op): New.
8880         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8881         (opt_pass::operator new): New.
8882         (opt_pass::gt_ggc_mx): New.
8883         (opt_pass::gt_pch_nx): New.
8884         (opt_pass::gt_pch_nx_with_op): New.
8886 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8888         * Makefile.in (GTFILES): Add context.h.
8889         * context.c (gcc::context::operator new): New.
8890         (gcc::context::gt_ggc_mx): New.
8891         (gcc::context::gt_pch_nx): New.
8892         (gcc::context::gt_pch_nx): New.
8893         * context.h (gcc::context): Add GTY((user)) marking.
8894         (gcc::context::operator new): New.
8895         (gcc::context::gt_ggc_mx): New.
8896         (gcc::context::gt_pch_nx): New.
8897         (gcc::context::gt_pch_nx): New.
8898         (g): Add GTY marking.
8899         (gt_ggc_mx (gcc::context *)): New.
8900         (gt_pch_nx (gcc::context *)): New.
8901         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8902         void *cookie)): New.
8903         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8905 2013-08-20  Alan Modra  <amodra@gmail.com>
8907         PR target/57865
8908         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8909         (rs6000_emit_epilogue): Likewise.
8911 2013-08-19  Dehao Chen  <dehao@google.com>
8913         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8915 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8916             Jakub Jelinek  <jakub@redhat.com>
8918         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8919         (BUILT_IN_FABSD64): Likewise.
8920         (BUILT_IN_FABSD128): Likewise.
8921         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8922         (fold_builtin_1): Likewise.
8923         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8924         destination and source operands.
8925         (*abstd2_fpr): Likewise.
8926         (*nabstd2_fpr): Likewise.
8928 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8930         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8931         JUMP_P and INSN_P.
8933 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8935         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8936         incomplete.
8938 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8940         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8941         * builtins.c (default_libc_has_function): New.
8942         (gnu_libc_has_function): Ditto.
8943         (no_c99_libc_has_function): Ditto.
8944         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8945         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8946         (fold_builtin_sincos): Likewise.
8947         (fold_builtin_cexp): Likewise.
8948         * builtins.def (DEF_C94_BUILTIN): Likewise.
8949         (DEF_C99_BUILTIN): Likewise.
8950         (DEF_C99_C90RES_BUILTIN): Likewise.
8951         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8952         definitions to using this define.
8953         * config/darwin-protos.h (darwin_libc_has_function): New.
8954         * config/darwin.c (darwin_libc_has_function): Ditto.
8955         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8956         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8957         * config/darwin.h: Ditto.
8958         * config/elfos.h: Ditto.
8959         * config/freebsd.h: Ditto.
8960         * config/i386/cygming.h: Ditto.
8961         * config/i386/djgpp.h: Ditto.
8962         * config/i386/i386-interix.h: Ditto.
8963         * config/microblaze/microblaze.h: Ditto.
8964         * config/mmix/mmix.h: Ditto.
8965         * config/gnu-user.h: Ditto.
8966         * config/ia64/hpux.h: Ditto.
8967         * config/pa/pa-hpux.h: Ditto.
8968         * config/pdp11/pdp11.h: Ditto.
8969         * config/picochip/picochip.h: Ditto.
8970         * config/linux.h: Ditto.
8971         * config/netbsd.h: Ditto.
8972         * config/openbsd.h: Ditto.
8973         * config/rs6000/aix43.h: Ditto.
8974         * config/rs6000/aix51.h: Ditto.
8975         * config/rs6000/aix52.h: Ditto.
8976         * config/rs6000/aix53.h: Ditto.
8977         * config/rs6000/aix61.h: Ditto.
8978         * config/rs6000/darwin.h: Ditto.
8979         * config/rs6000/linux.h: Ditto.
8980         * config/rs6000/linux64.h: Ditto.
8981         * config/s390/tpf.h: Ditto.
8982         * config/sol2-10.h: Ditto.
8983         * config/sol2.h: Ditto.
8984         * config/vms/vms.h: Ditto.
8985         * config/vxworks.h: Ditto.
8986         * config/linux-android.c (linux_android_libc_has_function):
8987         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8988         * config/linux-protos.h (linux_android_libc_has_function):
8989         New declaration.
8990         * config/i386/i386.c (ix86_libc_has_function): New.
8991         * config/i386/i386-protos.h
8992         (ix86_libc_has_function): New declaration.
8993         * config/i386/i386.md
8994         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8995         ("isinf<mode>2): Likewise.
8996         * convert.c (convert_to_integer): Using new target hook
8997         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8998         TARGET_C99_FUNCTIONS.
8999         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9000         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9001         * coretypes.h (function_class): New enum for different
9002         classes of functions.
9003         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9004         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9005         (TARGET_HAS_SINCOS): Likewise.
9006         (TARGET_LIBC_HAS_FUNCTION): New.
9007         * doc/tm.texi: Regenerated.
9008         * targhooks.h (default_libc_has_function): New declaration.
9009         (no_c99_libc_has_function): Ditto.
9010         (gnu_libc_has_function): Ditto.
9011         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9012         and TARGET_HAS_SINCOS.
9014 2013-08-18  Jan Hubicka  <jh@suse.cz>
9016         * Makeifle-in (ipa-devirt.o): New.
9017         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9018         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9019         (analyze_functions): Look into possible targets of polymorphic call.
9020         * dumpfile.c (dump_files): Add type-inheritance dump.
9021         * dumpfile.h (TDI_inheritance): New.
9022         * ipa-devirt.c: New file.
9023         * ipa-utils.h (odr_type_d): Forward declare.
9024         (odr_type): New type.
9025         (build_type_inheritance_graph): Declare.
9026         (possible_polymorphic_call_targets): Declare and introduce inline
9027         variant when only edge is pased.
9028         (dump_possible_polymorphic_call_targets): Likewise.
9029         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9030         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9031         (types_same_for_odr): ... here.
9032         * tree.h (type_in_anonymous_namespace_p): Declare.
9034 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9036         PR tree-optimization/58006
9037         * tree-parloops.c (take_address_of): Don't ICE if get_name
9038         returns NULL.
9039         (eliminate_local_variables_stmt): Remove clobber stmts.
9041 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9043         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9044         error message is issued for an alias to undefined symbol.
9046 2013-08-18  Jan Hubicka  <jh@suse.cz>
9048         * cgraph.c (cgraph_create_indirect_edge): Discover
9049         polymorphic calls and record basic info into indirect_info.
9050         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9051         devirtualization, ignore objc function calls.
9052         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9053         call with no parm index info.
9054         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9055         * tree.c (virtual_method_call_p): New function.
9056         * tree.h (virtual_method_call_p): Declare.
9058 2013-08-16  Jan Hubicka  <jh@suse.cz>
9060         PR middle-end/58179
9061         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9063 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9065         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9066         attribute.
9068 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9070         * gengtype.c (type_for_name): Add special-case support for
9071         locating types within the "gcc::" namespace.
9072         (open_base_files): Emit a "using namespace gcc" directive.
9074 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9076         PR target/58160
9077         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9078         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9080         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9081         array instead of each individual operand as a separate argument.
9082         (emit_fusion_gpr_load): Likewise.
9083         (expand_fusion_gpr_load): Add new function declaration.
9085         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9086         signature to have the operands passed as an array, instead of as
9087         separate arguments.  Allow ZERO_EXTEND to be in the memory
9088         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9089         depend on the register live/dead flags when peepholes are run.
9090         (expand_fusion_gpr_load): New function to be called from the
9091         peephole2 pass, to change the register that addis sets to be the
9092         target register.
9093         (emit_fusion_gpr_load): Change the calling signature to have the
9094         operands passed as an array, instead of as separate arguments.
9095         Allow ZERO_EXTEND to be in the memory address, and also
9096         SIGN_EXTEND if -mpower8-fusion-sign.
9098         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9099         unspec enumeration.
9100         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9101         adjust the register addis loads up in the peephole2 pass.  Do not
9102         depend on the register live/dead state when the peephole pass is done.
9104 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9106         * gengtype.c (create_user_defined_type): Ensure that the kind
9107         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9108         declaration is seen before the GTY((user)) marking.
9110 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9112         PR target/58105
9113         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9115 2013-08-16  Jan Hubicka  <jh@suse.cz>
9117         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9118         arugment expected_type.
9119         (gimple_fold_call): Use it.
9120         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9121         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9122         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9123         (try_make_edge_direct_virtual_call): Likewise.
9124         * tree.c (obj_type_ref_class): New.
9125         * tree.h (obj_type_ref_class): Use it.
9127 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9129         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9130         (rtl_slim_pp): Likewise.
9131         (init_rtl_slim_pretty_print): Likewise.
9132         (dump_value_slim):  Don't call it.  Use local pretty printer.
9133         (dump_insn_slim): Likewise.
9134         (dump_rtl_slim): Likewise.
9135         (str_pattern_slim): Likewise.
9136         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9137         Simplify.
9139 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9141         PR tree-optimization/58164
9142         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9143         walk gimple_goto_dest of GIMPLE_GOTO.
9145         PR tree-optimization/58165
9146         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9147         bi_call must be the last stmt in a bb, don't split_block, instead
9148         use fallthru edge from it and give up if there is none.
9149         Release conds vector when returning early.
9151 2013-08-14  Xinliang David Li  <davidxl@google.com>
9153         * config/i386/i386.c (ix86_option_override_internal):
9154         Remove unused variable and field.
9156 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9158         PR target/57949
9159         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9160         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9161         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9162         and Linux, correct BLKmode alignment when 128-bit alignment is
9163         required and compatibility flag is not set.
9164         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9165         for zero-size arguments when compatibility flag is not set.
9167 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9169         PR tree-optimization/58145
9170         * tree-sra.c (build_ref_for_offset): If prev_base has
9171         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9173 2013-08-14  Xinliang David Li  <davidxl@google.com>
9175         * config/i386/i386.c (ix86_option_override_internal):
9176         Fix uninitialized variable error.
9178 2013-08-14  Xinliang David Li  <davidxl@google.com>
9180         * config/i386/i386.opt: Define two new options.
9181         * config/i386/x86-tune.def: Add arch selector field in macros.
9182         * config/i386/i386.h: Adjust macro definition.
9183         * config/i386/i386.c (ix86_option_override_internal):
9184         Refactor the code.
9185         (parse_mtune_ctrl_str): New function.
9186         (set_ix86_tune_features): New function.
9187         (ix86_function_specific_restore): Call the new helper function.
9189 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9191         PR rtl-optimization/57662
9192         * sel-sched.c (code_motion_process_successors): When the current insn
9193         is removed after the recursive traversal, break from the loop.
9194         Add comments and debug printouts.
9196 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9197             Alexandre Oliva  <aoliva@redhat.com>
9199         PR target/58067
9200         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9201         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9202         there also UNSPEC_PLTOFF.
9204 2013-08-14  Marek Polacek  <polacek@redhat.com>
9206         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9207         NUM_CONDITIONS bit positions.
9209 2013-08-13  Cary Coutant  <ccoutant@google.com>
9211         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9212         (attr_checksum): Hash vector contents instead of pointer.
9213         (attr_checksum_ordered): Likewise.
9215 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9217         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9218         when Pmode != word_mode.  Add length_address attribute.
9219         (sse3_monitor_<mode>): Merge from sse3_monitor and
9220         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9221         Pmode != word_mode.  Update insn length attribute.
9222         * config/i386/i386.c (ix86_option_override_internal): Update
9223         ix86_gen_monitor selection for merged sse3_monitor insn.
9225 2013-08-13  Julian Brown  <julian@codesourcery.com>
9227         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9228         perform invalid legitimization on greater-than-word-size modes for
9229         TARGET_E500_DOUBLE.
9231 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9233         * ira.c (setup_class_translate_array): Use aclass instead of cl
9234         for classes not fully covered by allocno classes.
9236 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9238         PR tree-optimization/57661
9239         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9240         * tree-inline.c (tree_function_versioning): Initialize it.
9241         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9242         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9243         that is not being copied.
9245         PR sanitizer/56417
9246         * asan.c (instrument_strlen_call): Fix typo in comment.
9247         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9249 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9251         * config/mips/mips.md (prefetch): Use lw instead of ld on
9252         loongson in 32bit mode.
9254 2013-08-13  Nick Clifton  <nickc@redhat.com>
9256         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9258 2013-08-13  Jan Hubicka  <jh@suse.cz>
9260         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9261         introduced edge; fix typo in sanity check.
9262         (cgraph_resolve_speculation): Export; improve diagnostic.
9263         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9264         speculation at type mismatch.
9265         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9266         (cgraph_resolve_speculation): Declare.
9267         (symtab_can_be_discarded): New function.
9268         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9269         * ipa-inline-transform.c (speculation_removed): New global var.
9270         (clone_inlined_nodes): See if speculation can be removed.
9271         (inline_call): If speculations was removed, we growths may not match.
9272         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9273         (speculation_useful_p): New function.
9274         (resolve_noninline_speculation): New function.
9275         (inline_small_functions): Resolve useless speculations.
9276         * ipa-inline.h (speculation_useful_p): Declare
9277         * ipa.c (can_replace_by_local_alias): Simplify.
9278         (ipa_profile): Produce speculative calls in non-lto, too;
9279         add simple cost model; produce local aliases.
9281 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9283         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9284         PASS_MANAGER_H.
9286 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9288         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9290 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9292         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9293         iterator instead of X87MODEF.
9295 2013-08-12  Perez Read  <netfirewall@gmail.com>
9297         PR target/58132
9298         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9299         operand 0 for intel asm alternative.
9300         (*movabs<mode>_2): Ditto for operand 1.
9302 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9304         * config/aarch64/arm_none.h
9305         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9307 2013-08-12  Nick Clifton  <nickc@redhat.com>
9309         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9311 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9313         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9314         expand for QI/HImode operand to produce more effictive code for
9315         unsigned char(short) --> float(double) conversion.
9317 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9319         * doc/invoke.texi: Mention that -ftls-model does not force the final
9320         model.
9322 2013-08-12  Marek Polacek  <polacek@redhat.com>
9323             Marc Glisse  <marc.glisse@inria.fr>
9325         PR tree-optimization/57980
9326         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9327         when creating -1 constant.
9329 2013-08-10  Jan Hubicka  <jh@suse.cz>
9331         Workaround binutils PR14342.
9332         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9333         (gimple_init_edge_profiler): Likewise.
9334         (gimple_gen_ic_func_profiler): Likewise.
9336 2013-08-09  Jan Hubicka  <jh@suse.cz>
9338         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9340 2013-08-09  Xinliang David Li  <davidxl@google.com>
9342         * config/i386/stringop.def: New file.
9343         * config/i386/stringop.opt: New file.
9344         * config/i386/i386-opts.h: Include stringopt.def.
9345         * config/i386/i386.opt: Include stringopt.opt.
9346         * config/i386/i386.c (ix86_option_override_internal):
9347         Override default size based stringop inline strategies with options.
9348         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9349         New function.
9351 2013-08-09  Jan Hubicka  <jh@suse.cz>
9353         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9355 2013-08-09  Jan Hubicka  <jh@suse.cz>
9357         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9358         CGRAPH_FREQ_MAX.
9359         (dump_cgraph_node): Dump profile-id.
9360         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9361         and common_target_probability.
9362         * lto-cgraph.c (lto_output_edge): Stream common targets.
9363         (lto_output_node): Stream profile ids.
9364         (input_node): Stream profile ids.
9365         (input_edge): Stream common targets.
9366         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9367         * ipa.c: Include value-prof.h
9368         (ipa_profile_generate_summary): Turn indirect call statement histograms
9369         into common targets.
9370         (ipa_profile): Turn common targets into speculative edges.
9372 2013-08-09  Jan Hubicka  <jh@suse.cz>
9374         * cgraph.h (cgraph_node): Add profile_id.
9375         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9376         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9377         (del_node_map): Update.
9378         (find_func_by_funcdef_no): Replace by ...
9379         (find_func_by_profile_id): ... this one.
9380         (gimple_ic_transform): Do not remove useful histograms when
9381         speculation is not done; dump info when indirect call removal
9382         can happen at LTO.
9383         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9384         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9385         (__gcov_indirect_call_profiler_v2): .. this one.
9386         * profile.h (init_node_map): Update.
9387         * coverage.c (coverage_compute_profile_id): New function.
9388         * coverage.h (coverage_compute_profile_id): Declare.
9389         * tree-profile.c (init_ic_make_global_vars): Make
9390         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9391         (gimple_init_edge_profiler): Update prototype of
9392         __gcov_indirect_call_profiler.
9393         (gimple_gen_ic_func_profiler): Simplify.
9394         (tree_profiling): Use init_node_map
9396 2013-08-09  Jan Hubicka  <jh@suse.cz>
9398         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9399         non-speculative refs.
9400         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9401         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9402         (cgraph_set_call_stmt): Likewise.
9403         (cgraph_create_edge_1): Fix release checking compilatoin;
9404         clear lto_stmt_uid.
9405         (cgraph_free_edge): Free indirect info.
9406         (cgraph_turn_edge_to_speculative): New function.
9407         (cgraph_speculative_call_info): New function.
9408         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9409         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9410         (dump_cgraph_node): Dump speculation.
9411         (verify_edge_count_and_frequency): Accept speculative edges.
9412         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9413         (verify_cgraph_node): Handle speculation.
9414         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9415         (cgraph_set_call_stmt): Update prototype.
9416         (cgraph_make_edge_direct): Update prototype.
9417         (cgraph_speculative_call_info): Declare.
9418         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9419         to change; update call of ipa_find_references.
9420         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9421         lto_stmt_uid and speculative flags.
9422         (ipa_dump_references): Dump speculation.
9423         (ipa_clone_references): Clone speculative flag.
9424         (ipa_clone_referring): Likewise.
9425         (ipa_clone_ref): New function.
9426         (ipa_find_reference): Look into lto_stmt_uids
9427         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9428         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9429         (ipa_find_reference): Update declaration.
9430         (ipa_clone_ref): Declare.
9431         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9432         stream speculative flag.
9433         (lto_output_ref): Stream statements uids and speculation.
9434         (input_ref): Likewise.
9435         (input_edge): Stream speuclation.
9436         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9437         (cgraph_set_call_stmt_including_clones): Handle speculation.
9438         * ipa-inline.c (heap_edge_removal_hook): New function.
9439         (inline_small_functions): Register it.
9440         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9441         also initialize refs.
9442         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9443         edge to change.
9444         (try_make_edge_direct_simple_call): Likewise.
9445         (try_make_edge_direct_simple_call): Likewise.
9446         (update_indirect_edges_after_inlining): Likewise.
9447         (remove_described_reference): Look proper lto_stmt_uid.
9448         (propagate_controlled_uses): Likewise.
9449         (propagate_controlled_uses): Liekwise.
9450         * tree-inline.c (copy_bb): Copy speculative edges.
9451         (redirect_all_calls): New function.
9452         (copy_cfg_body): Do redirection after loop info is updated.
9453         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9455 2013-08-09  Jan Hubicka  <jh@suse.cz>
9457         * lto-streamer-out.c (output_function): Renumber PHIs.
9458         * lto-streamer-in.c (input_function): Likewise.
9460 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9462         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9463         (get_lane_unsigned): Likewise.
9464         (dup_lane_scalar): Likewise.
9465         (get_lane): enable for VALL.
9466         * config/aarch64/aarch64-simd.md
9467         (aarch64_dup_lane_scalar<mode>): Remove.
9468         (aarch64_get_lane_signed<mode>): Likewise.
9469         (aarch64_get_lane_unsigned<mode>): Likewise.
9470         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9471         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9472         (aarch64_get_lane<mode>): Enable for all vector modes.
9473         (aarch64_get_lanedi): Remove misleading constraints.
9474         * config/aarch64/arm_neon.h
9475         (__aarch64_vget_lane_any): Define.
9476         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9477         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9478         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9479         * config/aarch64/iterators.md (VDQQH): New.
9480         (VDQQHS): Likewise.
9481         (vwcore): Likewise.
9483 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9485         * configure.ac: Add GAS check for LEON instructions on SPARC.
9486         * configure: Regenerate.
9487         * config.in: Likewise.
9488         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9489         sparc*-*-* block.
9490         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9491         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9492         for LEON or LEON3.
9493         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9494         (AS_LEON_FLAG): New macro.
9495         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9496         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9497         Deal with LEON and LEON3 for the memory model.
9498         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9499         (atomic_compare_and_swap<mode>_1): Likewise.
9500         (*atomic_compare_and_swap<mode>_1): Likewise.
9502 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9504         * config/arm/neon.md (vcond): Fix floating-point vector
9505         comparisons against 0.
9507 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9509         * lra-constraints.c (emit_spill_move): Remove assert.
9510         (process_alt_operands): Add more debugging
9511         output.  Increase reject for spilling into memory.  Decrease
9512         reject for reloading scratch.
9513         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9515 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9517         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9518         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9519         micromips incompatible.  Add nan2008.
9520         (MULTILIB_DIRNAMES): Add nan2008.
9521         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9522         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9523         and micromips incompatible.  Add nan2008.
9524         (MULTILIB_DIRNAMES): Add nan2008.
9525         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9527 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9529         PR rtl-optimization/58079
9530         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9531         simplify_comparison has widened a comparison with an integer.
9533 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9535         * config/arm/neon.md (movmisalign<mode>): Disable when we
9536         don't allow unaligned accesses.
9537         (*movmisalign<mode>_neon_store): Likewise.
9538         (*movmisalign<mode>_neon_load): Likewise.
9539         (*movmisalign<mode>_neon_store): Likewise.
9540         (*movmisalign<mode>_neon_load): Likewise.
9542 2013-08-08  Jan Hubicka  <jh@suse.cz>
9544         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9545         (make_pass_rebuild_cgraph_edges): Also clear references.
9546         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9547         * ipa-inline-transform.c (inline_transform): Remove all references
9548         after inlining.
9549         * cgraphunit.c (expand_function): Remove all references after
9550         expansion.
9551         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9552         (ipa_find_reference): Rewrite to iterator.
9553         (remove_stmt_references): Likewise.
9554         (ipa_clear_stmts_in_references): New function.
9555         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9556         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9557         clear references.
9558         * ipa-split.c (split_function): Remove references in split function.
9560 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9562         PR target/57431
9563         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9564         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9566 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9568         PR target/56979
9569         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9570         suggested mode for the assignment isn't compatible with the
9571         registers required.
9573 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9575         PR target/58065
9576         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9578 2013-08-07  Xinliang David Li  <davidxl@google.com>
9580         * config/i386/i386.opt: New option -mtune-ctrl=.
9581         * config/i386/x86-tune.def: New file.
9582         * config/i386/i386.h: include x86-tune.def.
9583         * config/i386/i386.c (ix86_option_override_internal):
9584         Parsing -mtune-ctrl= option and set tune features.
9586 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9588         PR other/12081
9589         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9590         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9591         to gen_2arg_fn_t.
9593 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9595         * rtl.h (update_alignments): Declare.
9596         * final.c (grow_label_align): New function extracted from...
9597         (shorten_branches): ...here.  Call it.
9598         (update_alignments): New function.
9599         * reorg.c (sibling_labels): New variable.
9600         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9601         the new label along with it onto the sibling_labels vector.
9602         (fill_simple_delay_slots): Adjust call to get_label_before.
9603         (fill_slots_from_thread): Likewise.
9604         (relax_delay_slots): Likewise.
9605         (make_return_insns): Likewise.
9606         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9608 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9610         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9611         document its semantics.
9612         (diagnostic_report_diagnostic): Adjust accordingly.
9614 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9616         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9617         (sparc_option_override): ...and port to new C++ pass API.
9618         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9620 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9622         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9624 2013-08-06  Caroline Tice  <cmtice@google.com>
9626         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9627         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9628         * tree-pass.h: Add pass_vtable_verify.
9629         * varasm.c (assemble_variable): Add code to properly set the comdat
9630         section and name for the .vtable_map_vars section.
9631         (assemble_vtyv_preinit_initializer): New function.
9632         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9633         LINK_ONCE flag.
9634         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9635         * vtable-verify.c: New file.
9636         * vtable-verify.h: New file.
9637         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9638         initialiation levels.
9639         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9640         * passes.def: Insert pass_vtable_verify.
9641         * aclocal.m4: Reorder includes.
9642         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9643         -fvtv-counts options.
9644         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9645         as appropriate, if -fvtable-verify=... is used.
9646         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9647         -fvtable-verify=... is used.
9648         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9649         (vtable-verify.o): Add new build rule.
9650         (GTFILES): Add vtable-verify.c to list.
9651         * common.opt (fvtable-verify=): New flag.
9652         (vtv_priority): Values for fvtable-verify= flag.
9653         (fvtv-counts): New flag.
9654         (fvtv-debug): New flag.
9655         * tree.h (save_vtable_map_decl): New extern function decl.
9657 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9659         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9660         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9661         (pass_data_rl78_devirt): ...new pass_data instance and...
9662         (make_pass_rl78_devirt): ...new function.
9663         (rl78_asm_file_start): Port pass registration to new C++ API.
9665 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9667         * coretypes.h (rtl_opt_pass): Add.
9668         (gcc::context): Add.
9669         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9670         (epiphany_init): Port to new C++ pass API.
9671         (epiphany_optimize_mode_switching): Likewise.
9672         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9673         (pass_manager::get_pass_mode_switching): New.
9674         (pass_manager::get_pass_peephole2): New.
9675         * mode-switching.c (pass_mode_switching): Add clone method.
9676         * recog.c (pass_peephole2): Add clone method.
9677         (pass_split_all_insns): Add clone method.
9679 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9681         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9682         (mips_option_override): ...here, porting to new C++ API for passes.
9684 2013-08-06  Jan Hubicka  <jh@suse.cz>
9686         * cgraph.c (cgraph_get_body): New function based on lto.c
9687         implementation.
9688         * cgraph.h (cgraph_get_body): Declare.
9689         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9690         LTO paths.
9691         * cgraphunit.c (expand_function): Get body prior expanding.
9692         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9693         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9694         really need.
9695         * passes.c (do_per_function_toporder): Get body.
9696         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9697         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9699 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9701         PR fortran/57987
9702         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9703         is not re-finalized.  Rename second parameter to no_collect.
9705 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9707         PR middle-end/58041
9708         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9709         MEM_REF has proper alignment information.
9711 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9713         PR other/12081
9714         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9715         class insn_gen_fn.
9716         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9717         rtx (*) (rtx, ...) with insn_gen_fn.
9718         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9719         insn_gen_fn::stored_funcptr.  Add initializer braces.
9721 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9723         Rewrite how instances of passes are cloned to remove assumptions
9724         about their sizes (thus allowing pass subclasses to have
9725         additional data fields, albeit non-GC-managed ones at this point).
9727         * passes.c (make_pass_instance): Now that passes have clone
9728         methods, rewrite this function to eliminate XNEW and memcpy
9729         calls that used hardcoded sizes.  Since this function no longer
9730         creates pass instances, rename it to...
9731         (add_pass_instance): ...this.  Document the old way that passes were
9732         numbered and flagged, and rework this function to continue using it.
9733         (next_pass_1): Add an initial_pass argument for use by
9734         add_pass_instance.
9735         (position_pass): When adding multiple instances of a pass, use
9736         the pass's clone method, rather than relying on the XNEW/memcpy
9737         within the former make_pass_instance (now add_pass_instance).
9738         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9739         the initial instance of the current pass within the pass manager.
9741 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9743         This is the automated part of the conversion of passes from C
9744         structs to C++ classes.
9746         Patch autogenerated by refactor_passes.py from
9747         https://github.com/davidmalcolm/gcc-refactoring-scripts
9748         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9750         * asan.c (pass_asan): Convert from a global struct to a subclass of
9751         gimple_opt_pass along with...
9752         (pass_data_asan): ...new pass_data instance and...
9753         (make_pass_asan): ...new function.
9754         (pass_asan_O0): Convert from a global struct to a subclass of
9755         gimple_opt_pass along with...
9756         (pass_data_asan_O0): ...new pass_data instance and...
9757         (make_pass_asan_O0): ...new function.
9758         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9759         subclass of rtl_opt_pass along with...
9760         (pass_data_inc_dec): ...new pass_data instance and...
9761         (make_pass_inc_dec): ...new function.
9762         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9763         a subclass of rtl_opt_pass along with...
9764         (pass_data_reorder_blocks): ...new pass_data instance and...
9765         (make_pass_reorder_blocks): ...new function.
9766         (pass_duplicate_computed_gotos): Convert from a global struct to a
9767         subclass of rtl_opt_pass along with...
9768         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9769         (make_pass_duplicate_computed_gotos): ...new function.
9770         (pass_partition_blocks): Convert from a global struct to a subclass of
9771         rtl_opt_pass along with...
9772         (pass_data_partition_blocks): ...new pass_data instance and...
9773         (make_pass_partition_blocks): ...new function.
9774         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9775         struct to a subclass of rtl_opt_pass along with...
9776         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9777         and...
9778         (make_pass_branch_target_load_optimize1): ...new function.
9779         (pass_branch_target_load_optimize2): Convert from a global struct to a
9780         subclass of rtl_opt_pass along with...
9781         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9782         and...
9783         (make_pass_branch_target_load_optimize2): ...new function.
9784         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9785         of rtl_opt_pass along with...
9786         (pass_data_jump): ...new pass_data instance and...
9787         (make_pass_jump): ...new function.
9788         (pass_jump2): Convert from a global struct to a subclass of
9789         rtl_opt_pass along with...
9790         (pass_data_jump2): ...new pass_data instance and...
9791         (make_pass_jump2): ...new function.
9792         * cfgexpand.c (pass_expand): Convert from a global struct to a
9793         subclass of rtl_opt_pass along with...
9794         (pass_data_expand): ...new pass_data instance and...
9795         (make_pass_expand): ...new function.
9796         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9797         of rtl_opt_pass along with...
9798         (pass_data_free_cfg): ...new pass_data instance and...
9799         (make_pass_free_cfg): ...new function.
9800         (pass_into_cfg_layout_mode): Convert from a global struct to a
9801         subclass of rtl_opt_pass along with...
9802         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9803         (make_pass_into_cfg_layout_mode): ...new function.
9804         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9805         subclass of rtl_opt_pass along with...
9806         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9807         (make_pass_outof_cfg_layout_mode): ...new function.
9808         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9809         struct to a subclass of gimple_opt_pass along with...
9810         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9811         (make_pass_build_cgraph_edges): ...new function.
9812         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9813         subclass of gimple_opt_pass along with...
9814         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9815         (make_pass_rebuild_cgraph_edges): ...new function.
9816         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9817         subclass of gimple_opt_pass along with...
9818         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9819         and...
9820         (make_pass_remove_cgraph_callee_edges): ...new function.
9821         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9822         struct to a subclass of rtl_opt_pass along with...
9823         (pass_data_stack_adjustments): ...new pass_data instance and...
9824         (make_pass_stack_adjustments): ...new function.
9825         * combine.c (pass_combine): Convert from a global struct to a subclass
9826         of rtl_opt_pass along with...
9827         (pass_data_combine): ...new pass_data instance and...
9828         (make_pass_combine): ...new function.
9829         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9830         global struct to a subclass of rtl_opt_pass along with...
9831         (pass_data_compare_elim_after_reload): ...new pass_data instance
9832         and...
9833         (make_pass_compare_elim_after_reload): ...new function.
9834         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9835         of rtl_opt_pass along with...
9836         (pass_data_rtl_cprop): ...new pass_data instance and...
9837         (make_pass_rtl_cprop): ...new function.
9838         * cse.c (pass_cse): Convert from a global struct to a subclass of
9839         rtl_opt_pass along with...
9840         (pass_data_cse): ...new pass_data instance and...
9841         (make_pass_cse): ...new function.
9842         (pass_cse2): Convert from a global struct to a subclass of
9843         rtl_opt_pass along with...
9844         (pass_data_cse2): ...new pass_data instance and...
9845         (make_pass_cse2): ...new function.
9846         (pass_cse_after_global_opts): Convert from a global struct to a
9847         subclass of rtl_opt_pass along with...
9848         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9849         (make_pass_cse_after_global_opts): ...new function.
9850         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9851         of rtl_opt_pass along with...
9852         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9853         (make_pass_ud_rtl_dce): ...new function.
9854         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9855         rtl_opt_pass along with...
9856         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9857         (make_pass_fast_rtl_dce): ...new function.
9858         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9859         a subclass of rtl_opt_pass along with...
9860         (pass_data_df_initialize_opt): ...new pass_data instance and...
9861         (make_pass_df_initialize_opt): ...new function.
9862         (pass_df_initialize_no_opt): Convert from a global struct to a
9863         subclass of rtl_opt_pass along with...
9864         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9865         (make_pass_df_initialize_no_opt): ...new function.
9866         (pass_df_finish): Convert from a global struct to a subclass of
9867         rtl_opt_pass along with...
9868         (pass_data_df_finish): ...new pass_data instance and...
9869         (make_pass_df_finish): ...new function.
9870         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9871         rtl_opt_pass along with...
9872         (pass_data_rtl_dse1): ...new pass_data instance and...
9873         (make_pass_rtl_dse1): ...new function.
9874         (pass_rtl_dse2): Convert from a global struct to a subclass of
9875         rtl_opt_pass along with...
9876         (pass_data_rtl_dse2): ...new pass_data instance and...
9877         (make_pass_rtl_dse2): ...new function.
9878         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9879         subclass of rtl_opt_pass along with...
9880         (pass_data_dwarf2_frame): ...new pass_data instance and...
9881         (make_pass_dwarf2_frame): ...new function.
9882         * except.c (pass_set_nothrow_function_flags): Convert from a global
9883         struct to a subclass of rtl_opt_pass along with...
9884         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9885         and...
9886         (make_pass_set_nothrow_function_flags): ...new function.
9887         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9888         subclass of rtl_opt_pass along with...
9889         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9890         and...
9891         (make_pass_convert_to_eh_region_ranges): ...new function.
9892         * final.c (pass_compute_alignments): Convert from a global struct to a
9893         subclass of rtl_opt_pass along with...
9894         (pass_data_compute_alignments): ...new pass_data instance and...
9895         (make_pass_compute_alignments): ...new function.
9896         (pass_final): Convert from a global struct to a subclass of
9897         rtl_opt_pass along with...
9898         (pass_data_final): ...new pass_data instance and...
9899         (make_pass_final): ...new function.
9900         (pass_shorten_branches): Convert from a global struct to a subclass of
9901         rtl_opt_pass along with...
9902         (pass_data_shorten_branches): ...new pass_data instance and...
9903         (make_pass_shorten_branches): ...new function.
9904         (pass_clean_state): Convert from a global struct to a subclass of
9905         rtl_opt_pass along with...
9906         (pass_data_clean_state): ...new pass_data instance and...
9907         (make_pass_clean_state): ...new function.
9908         * function.c (pass_instantiate_virtual_regs): Convert from a global
9909         struct to a subclass of rtl_opt_pass along with...
9910         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9911         (make_pass_instantiate_virtual_regs): ...new function.
9912         (pass_leaf_regs): Convert from a global struct to a subclass of
9913         rtl_opt_pass along with...
9914         (pass_data_leaf_regs): ...new pass_data instance and...
9915         (make_pass_leaf_regs): ...new function.
9916         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9917         subclass of rtl_opt_pass along with...
9918         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9919         and...
9920         (make_pass_thread_prologue_and_epilogue): ...new function.
9921         (pass_match_asm_constraints): Convert from a global struct to a
9922         subclass of rtl_opt_pass along with...
9923         (pass_data_match_asm_constraints): ...new pass_data instance and...
9924         (make_pass_match_asm_constraints): ...new function.
9925         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9926         subclass of rtl_opt_pass along with...
9927         (pass_data_rtl_fwprop): ...new pass_data instance and...
9928         (make_pass_rtl_fwprop): ...new function.
9929         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9930         rtl_opt_pass along with...
9931         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9932         (make_pass_rtl_fwprop_addr): ...new function.
9933         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9934         rtl_opt_pass along with...
9935         (pass_data_rtl_pre): ...new pass_data instance and...
9936         (make_pass_rtl_pre): ...new function.
9937         (pass_rtl_hoist): Convert from a global struct to a subclass of
9938         rtl_opt_pass along with...
9939         (pass_data_rtl_hoist): ...new pass_data instance and...
9940         (make_pass_rtl_hoist): ...new function.
9941         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9942         subclass of gimple_opt_pass along with...
9943         (pass_data_lower_cf): ...new pass_data instance and...
9944         (make_pass_lower_cf): ...new function.
9945         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9946         from a global struct to a subclass of gimple_opt_pass along with...
9947         (pass_data_strength_reduction): ...new pass_data instance and...
9948         (make_pass_strength_reduction): ...new function.
9949         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9950         of rtl_opt_pass along with...
9951         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9952         (make_pass_rtl_ifcvt): ...new function.
9953         (pass_if_after_combine): Convert from a global struct to a subclass of
9954         rtl_opt_pass along with...
9955         (pass_data_if_after_combine): ...new pass_data instance and...
9956         (make_pass_if_after_combine): ...new function.
9957         (pass_if_after_reload): Convert from a global struct to a subclass of
9958         rtl_opt_pass along with...
9959         (pass_data_if_after_reload): ...new pass_data instance and...
9960         (make_pass_if_after_reload): ...new function.
9961         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9962         a subclass of rtl_opt_pass along with...
9963         (pass_data_initialize_regs): ...new pass_data instance and...
9964         (make_pass_initialize_regs): ...new function.
9965         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9966         of ipa_opt_pass_d along with...
9967         (pass_data_ipa_cp): ...new pass_data instance and...
9968         (make_pass_ipa_cp): ...new function.
9969         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9970         global struct to a subclass of gimple_opt_pass along with...
9971         (pass_data_inline_parameters): ...new pass_data instance and...
9972         (make_pass_inline_parameters): ...new function.
9973         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9974         subclass of gimple_opt_pass along with...
9975         (pass_data_early_inline): ...new pass_data instance and...
9976         (make_pass_early_inline): ...new function.
9977         (pass_ipa_inline): Convert from a global struct to a subclass of
9978         ipa_opt_pass_d along with...
9979         (pass_data_ipa_inline): ...new pass_data instance and...
9980         (make_pass_ipa_inline): ...new function.
9981         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9982         struct to a subclass of gimple_opt_pass along with...
9983         (pass_data_local_pure_const): ...new pass_data instance and...
9984         (make_pass_local_pure_const): ...new function.
9985         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9986         ipa_opt_pass_d along with...
9987         (pass_data_ipa_pure_const): ...new pass_data instance and...
9988         (make_pass_ipa_pure_const): ...new function.
9989         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9990         to a subclass of ipa_opt_pass_d along with...
9991         (pass_data_ipa_reference): ...new pass_data instance and...
9992         (make_pass_ipa_reference): ...new function.
9993         * ipa-split.c (pass_split_functions): Convert from a global struct to
9994         a subclass of gimple_opt_pass along with...
9995         (pass_data_split_functions): ...new pass_data instance and...
9996         (make_pass_split_functions): ...new function.
9997         (pass_feedback_split_functions): Convert from a global struct to a
9998         subclass of gimple_opt_pass along with...
9999         (pass_data_feedback_split_functions): ...new pass_data instance and...
10000         (make_pass_feedback_split_functions): ...new function.
10001         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10002         global struct to a subclass of simple_ipa_opt_pass along with...
10003         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10004         instance and...
10005         (make_pass_ipa_function_and_variable_visibility): ...new function.
10006         (pass_ipa_free_inline_summary): Convert from a global struct to a
10007         subclass of simple_ipa_opt_pass along with...
10008         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10009         (make_pass_ipa_free_inline_summary): ...new function.
10010         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10011         subclass of ipa_opt_pass_d along with...
10012         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10013         and...
10014         (make_pass_ipa_whole_program_visibility): ...new function.
10015         (pass_ipa_profile): Convert from a global struct to a subclass of
10016         ipa_opt_pass_d along with...
10017         (pass_data_ipa_profile): ...new pass_data instance and...
10018         (make_pass_ipa_profile): ...new function.
10019         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10020         ipa_opt_pass_d along with...
10021         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10022         (make_pass_ipa_cdtor_merge): ...new function.
10023         * ira.c (pass_ira): Convert from a global struct to a subclass of
10024         rtl_opt_pass along with...
10025         (pass_data_ira): ...new pass_data instance and...
10026         (make_pass_ira): ...new function.
10027         (pass_reload): Convert from a global struct to a subclass of
10028         rtl_opt_pass along with...
10029         (pass_data_reload): ...new pass_data instance and...
10030         (make_pass_reload): ...new function.
10031         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10032         subclass of rtl_opt_pass along with...
10033         (pass_data_cleanup_barriers): ...new pass_data instance and...
10034         (make_pass_cleanup_barriers): ...new function.
10035         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10036         of rtl_opt_pass along with...
10037         (pass_data_loop2): ...new pass_data instance and...
10038         (make_pass_loop2): ...new function.
10039         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10040         rtl_opt_pass along with...
10041         (pass_data_rtl_loop_init): ...new pass_data instance and...
10042         (make_pass_rtl_loop_init): ...new function.
10043         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10044         rtl_opt_pass along with...
10045         (pass_data_rtl_loop_done): ...new pass_data instance and...
10046         (make_pass_rtl_loop_done): ...new function.
10047         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10048         subclass of rtl_opt_pass along with...
10049         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10050         (make_pass_rtl_move_loop_invariants): ...new function.
10051         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10052         rtl_opt_pass along with...
10053         (pass_data_rtl_unswitch): ...new pass_data instance and...
10054         (make_pass_rtl_unswitch): ...new function.
10055         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10056         subclass of rtl_opt_pass along with...
10057         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10058         and...
10059         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10060         (pass_rtl_doloop): Convert from a global struct to a subclass of
10061         rtl_opt_pass along with...
10062         (pass_data_rtl_doloop): ...new pass_data instance and...
10063         (make_pass_rtl_doloop): ...new function.
10064         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10065         a subclass of rtl_opt_pass along with...
10066         (pass_data_lower_subreg): ...new pass_data instance and...
10067         (make_pass_lower_subreg): ...new function.
10068         (pass_lower_subreg2): Convert from a global struct to a subclass of
10069         rtl_opt_pass along with...
10070         (pass_data_lower_subreg2): ...new pass_data instance and...
10071         (make_pass_lower_subreg2): ...new function.
10072         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10073         struct to a subclass of ipa_opt_pass_d along with...
10074         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10075         (make_pass_ipa_lto_gimple_out): ...new function.
10076         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10077         of ipa_opt_pass_d along with...
10078         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10079         (make_pass_ipa_lto_finish_out): ...new function.
10080         * mode-switching.c (pass_mode_switching): Convert from a global struct
10081         to a subclass of rtl_opt_pass along with...
10082         (pass_data_mode_switching): ...new pass_data instance and...
10083         (make_pass_mode_switching): ...new function.
10084         * modulo-sched.c (pass_sms): Convert from a global struct to a
10085         subclass of rtl_opt_pass along with...
10086         (pass_data_sms): ...new pass_data instance and...
10087         (make_pass_sms): ...new function.
10088         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10089         subclass of gimple_opt_pass along with...
10090         (pass_data_expand_omp): ...new pass_data instance and...
10091         (make_pass_expand_omp): ...new function.
10092         (pass_lower_omp): Convert from a global struct to a subclass of
10093         gimple_opt_pass along with...
10094         (pass_data_lower_omp): ...new pass_data instance and...
10095         (make_pass_lower_omp): ...new function.
10096         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10097         of gimple_opt_pass along with...
10098         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10099         (make_pass_diagnose_omp_blocks): ...new function.
10100         * passes.c (pass_early_local_passes): Convert from a global struct to
10101         a subclass of simple_ipa_opt_pass along with...
10102         (pass_data_early_local_passes): ...new pass_data instance and...
10103         (make_pass_early_local_passes): ...new function.
10104         (pass_all_early_optimizations): Convert from a global struct to a
10105         subclass of gimple_opt_pass along with...
10106         (pass_data_all_early_optimizations): ...new pass_data instance and...
10107         (make_pass_all_early_optimizations): ...new function.
10108         (pass_all_optimizations): Convert from a global struct to a subclass
10109         of gimple_opt_pass along with...
10110         (pass_data_all_optimizations): ...new pass_data instance and...
10111         (make_pass_all_optimizations): ...new function.
10112         (pass_all_optimizations_g): Convert from a global struct to a subclass
10113         of gimple_opt_pass along with...
10114         (pass_data_all_optimizations_g): ...new pass_data instance and...
10115         (make_pass_all_optimizations_g): ...new function.
10116         (pass_rest_of_compilation): Convert from a global struct to a subclass
10117         of rtl_opt_pass along with...
10118         (pass_data_rest_of_compilation): ...new pass_data instance and...
10119         (make_pass_rest_of_compilation): ...new function.
10120         (pass_postreload): Convert from a global struct to a subclass of
10121         rtl_opt_pass along with...
10122         (pass_data_postreload): ...new pass_data instance and...
10123         (make_pass_postreload): ...new function.
10124         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10125         subclass of rtl_opt_pass along with...
10126         (pass_data_gcse2): ...new pass_data instance and...
10127         (make_pass_gcse2): ...new function.
10128         * postreload.c (pass_postreload_cse): Convert from a global struct to
10129         a subclass of rtl_opt_pass along with...
10130         (pass_data_postreload_cse): ...new pass_data instance and...
10131         (make_pass_postreload_cse): ...new function.
10132         * predict.c (pass_profile): Convert from a global struct to a subclass
10133         of gimple_opt_pass along with...
10134         (pass_data_profile): ...new pass_data instance and...
10135         (make_pass_profile): ...new function.
10136         (pass_strip_predict_hints): Convert from a global struct to a subclass
10137         of gimple_opt_pass along with...
10138         (pass_data_strip_predict_hints): ...new pass_data instance and...
10139         (make_pass_strip_predict_hints): ...new function.
10140         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10141         of rtl_opt_pass along with...
10142         (pass_data_peephole2): ...new pass_data instance and...
10143         (make_pass_peephole2): ...new function.
10144         (pass_split_all_insns): Convert from a global struct to a subclass of
10145         rtl_opt_pass along with...
10146         (pass_data_split_all_insns): ...new pass_data instance and...
10147         (make_pass_split_all_insns): ...new function.
10148         (pass_split_after_reload): Convert from a global struct to a subclass
10149         of rtl_opt_pass along with...
10150         (pass_data_split_after_reload): ...new pass_data instance and...
10151         (make_pass_split_after_reload): ...new function.
10152         (pass_split_before_regstack): Convert from a global struct to a
10153         subclass of rtl_opt_pass along with...
10154         (pass_data_split_before_regstack): ...new pass_data instance and...
10155         (make_pass_split_before_regstack): ...new function.
10156         (pass_split_before_sched2): Convert from a global struct to a subclass
10157         of rtl_opt_pass along with...
10158         (pass_data_split_before_sched2): ...new pass_data instance and...
10159         (make_pass_split_before_sched2): ...new function.
10160         (pass_split_for_shorten_branches): Convert from a global struct to a
10161         subclass of rtl_opt_pass along with...
10162         (pass_data_split_for_shorten_branches): ...new pass_data instance
10163         and...
10164         (make_pass_split_for_shorten_branches): ...new function.
10165         * ree.c (pass_ree): Convert from a global struct to a subclass of
10166         rtl_opt_pass along with...
10167         (pass_data_ree): ...new pass_data instance and...
10168         (make_pass_ree): ...new function.
10169         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10170         subclass of rtl_opt_pass along with...
10171         (pass_data_stack_regs): ...new pass_data instance and...
10172         (make_pass_stack_regs): ...new function.
10173         (pass_stack_regs_run): Convert from a global struct to a subclass of
10174         rtl_opt_pass along with...
10175         (pass_data_stack_regs_run): ...new pass_data instance and...
10176         (make_pass_stack_regs_run): ...new function.
10177         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10178         subclass of rtl_opt_pass along with...
10179         (pass_data_cprop_hardreg): ...new pass_data instance and...
10180         (make_pass_cprop_hardreg): ...new function.
10181         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10182         subclass of rtl_opt_pass along with...
10183         (pass_data_reginfo_init): ...new pass_data instance and...
10184         (make_pass_reginfo_init): ...new function.
10185         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10186         of rtl_opt_pass along with...
10187         (pass_data_regmove): ...new pass_data instance and...
10188         (make_pass_regmove): ...new function.
10189         * regrename.c (pass_regrename): Convert from a global struct to a
10190         subclass of rtl_opt_pass along with...
10191         (pass_data_regrename): ...new pass_data instance and...
10192         (make_pass_regrename): ...new function.
10193         * reorg.c (pass_delay_slots): Convert from a global struct to a
10194         subclass of rtl_opt_pass along with...
10195         (pass_data_delay_slots): ...new pass_data instance and...
10196         (make_pass_delay_slots): ...new function.
10197         (pass_machine_reorg): Convert from a global struct to a subclass of
10198         rtl_opt_pass along with...
10199         (pass_data_machine_reorg): ...new pass_data instance and...
10200         (make_pass_machine_reorg): ...new function.
10201         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10202         of rtl_opt_pass along with...
10203         (pass_data_sched): ...new pass_data instance and...
10204         (make_pass_sched): ...new function.
10205         (pass_sched2): Convert from a global struct to a subclass of
10206         rtl_opt_pass along with...
10207         (pass_data_sched2): ...new pass_data instance and...
10208         (make_pass_sched2): ...new function.
10209         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10210         to a subclass of rtl_opt_pass along with...
10211         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10212         (make_pass_stack_ptr_mod): ...new function.
10213         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10214         to a subclass of rtl_opt_pass along with...
10215         (pass_data_rtl_store_motion): ...new pass_data instance and...
10216         (make_pass_rtl_store_motion): ...new function.
10217         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10218         of gimple_opt_pass along with...
10219         (pass_data_tracer): ...new pass_data instance and...
10220         (make_pass_tracer): ...new function.
10221         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10222         to a subclass of gimple_opt_pass along with...
10223         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10224         (make_pass_diagnose_tm_blocks): ...new function.
10225         (pass_lower_tm): Convert from a global struct to a subclass of
10226         gimple_opt_pass along with...
10227         (pass_data_lower_tm): ...new pass_data instance and...
10228         (make_pass_lower_tm): ...new function.
10229         (pass_tm_init): Convert from a global struct to a subclass of
10230         gimple_opt_pass along with...
10231         (pass_data_tm_init): ...new pass_data instance and...
10232         (make_pass_tm_init): ...new function.
10233         (pass_tm_mark): Convert from a global struct to a subclass of
10234         gimple_opt_pass along with...
10235         (pass_data_tm_mark): ...new pass_data instance and...
10236         (make_pass_tm_mark): ...new function.
10237         (pass_tm_edges): Convert from a global struct to a subclass of
10238         gimple_opt_pass along with...
10239         (pass_data_tm_edges): ...new pass_data instance and...
10240         (make_pass_tm_edges): ...new function.
10241         (pass_tm_memopt): Convert from a global struct to a subclass of
10242         gimple_opt_pass along with...
10243         (pass_data_tm_memopt): ...new pass_data instance and...
10244         (make_pass_tm_memopt): ...new function.
10245         (pass_ipa_tm): Convert from a global struct to a subclass of
10246         simple_ipa_opt_pass along with...
10247         (pass_data_ipa_tm): ...new pass_data instance and...
10248         (make_pass_ipa_tm): ...new function.
10249         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10250         subclass of gimple_opt_pass along with...
10251         (pass_data_call_cdce): ...new pass_data instance and...
10252         (make_pass_call_cdce): ...new function.
10253         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10254         subclass of gimple_opt_pass along with...
10255         (pass_data_build_cfg): ...new pass_data instance and...
10256         (make_pass_build_cfg): ...new function.
10257         (pass_split_crit_edges): Convert from a global struct to a subclass of
10258         gimple_opt_pass along with...
10259         (pass_data_split_crit_edges): ...new pass_data instance and...
10260         (make_pass_split_crit_edges): ...new function.
10261         (pass_warn_function_return): Convert from a global struct to a
10262         subclass of gimple_opt_pass along with...
10263         (pass_data_warn_function_return): ...new pass_data instance and...
10264         (make_pass_warn_function_return): ...new function.
10265         (pass_warn_function_noreturn): Convert from a global struct to a
10266         subclass of gimple_opt_pass along with...
10267         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10268         (make_pass_warn_function_noreturn): ...new function.
10269         (pass_warn_unused_result): Convert from a global struct to a subclass
10270         of gimple_opt_pass along with...
10271         (pass_data_warn_unused_result): ...new pass_data instance and...
10272         (make_pass_warn_unused_result): ...new function.
10273         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10274         a subclass of gimple_opt_pass along with...
10275         (pass_data_merge_phi): ...new pass_data instance and...
10276         (make_pass_merge_phi): ...new function.
10277         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10278         a subclass of gimple_opt_pass along with...
10279         (pass_data_lower_complex): ...new pass_data instance and...
10280         (make_pass_lower_complex): ...new function.
10281         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10282         gimple_opt_pass along with...
10283         (pass_data_lower_complex_O0): ...new pass_data instance and...
10284         (make_pass_lower_complex_O0): ...new function.
10285         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10286         subclass of gimple_opt_pass along with...
10287         (pass_data_lower_eh): ...new pass_data instance and...
10288         (make_pass_lower_eh): ...new function.
10289         (pass_refactor_eh): Convert from a global struct to a subclass of
10290         gimple_opt_pass along with...
10291         (pass_data_refactor_eh): ...new pass_data instance and...
10292         (make_pass_refactor_eh): ...new function.
10293         (pass_lower_resx): Convert from a global struct to a subclass of
10294         gimple_opt_pass along with...
10295         (pass_data_lower_resx): ...new pass_data instance and...
10296         (make_pass_lower_resx): ...new function.
10297         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10298         of gimple_opt_pass along with...
10299         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10300         (make_pass_lower_eh_dispatch): ...new function.
10301         (pass_cleanup_eh): Convert from a global struct to a subclass of
10302         gimple_opt_pass along with...
10303         (pass_data_cleanup_eh): ...new pass_data instance and...
10304         (make_pass_cleanup_eh): ...new function.
10305         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10306         to a subclass of simple_ipa_opt_pass along with...
10307         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10308         (make_pass_ipa_lower_emutls): ...new function.
10309         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10310         a subclass of gimple_opt_pass along with...
10311         (pass_data_if_conversion): ...new pass_data instance and...
10312         (make_pass_if_conversion): ...new function.
10313         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10314         subclass of gimple_opt_pass along with...
10315         (pass_data_build_ssa): ...new pass_data instance and...
10316         (make_pass_build_ssa): ...new function.
10317         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10318         global struct to a subclass of gimple_opt_pass along with...
10319         (pass_data_loop_distribution): ...new pass_data instance and...
10320         (make_pass_loop_distribution): ...new function.
10321         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10322         subclass of gimple_opt_pass along with...
10323         (pass_data_mudflap_1): ...new pass_data instance and...
10324         (make_pass_mudflap_1): ...new function.
10325         (pass_mudflap_2): Convert from a global struct to a subclass of
10326         gimple_opt_pass along with...
10327         (pass_data_mudflap_2): ...new pass_data instance and...
10328         (make_pass_mudflap_2): ...new function.
10329         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10330         subclass of gimple_opt_pass along with...
10331         (pass_data_mudflap_1): ...new pass_data instance and...
10332         (make_pass_mudflap_1): ...new function.
10333         (pass_mudflap_2): Convert from a global struct to a subclass of
10334         gimple_opt_pass along with...
10335         (pass_data_mudflap_2): ...new pass_data instance and...
10336         (make_pass_mudflap_2): ...new function.
10337         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10338         gimple_opt_pass along with...
10339         (pass_data_nrv): ...new pass_data instance and...
10340         (make_pass_nrv): ...new function.
10341         (pass_return_slot): Convert from a global struct to a subclass of
10342         gimple_opt_pass along with...
10343         (pass_data_return_slot): ...new pass_data instance and...
10344         (make_pass_return_slot): ...new function.
10345         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10346         to a subclass of gimple_opt_pass along with...
10347         (pass_data_object_sizes): ...new pass_data instance and...
10348         (make_pass_object_sizes): ...new function.
10349         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10350         global struct to a subclass of gimple_opt_pass along with...
10351         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10352         and...
10353         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10354         (pass_fixup_cfg): Convert from a global struct to a subclass of
10355         gimple_opt_pass along with...
10356         (pass_data_fixup_cfg): ...new pass_data instance and...
10357         (make_pass_fixup_cfg): ...new function.
10358         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10359         (make_pass_mudflap_1): ...new function.
10360         (pass_mudflap_2): Replace declaration with that of...
10361         (make_pass_mudflap_2): ...new function.
10362         (pass_asan): Replace declaration with that of...
10363         (make_pass_asan): ...new function.
10364         (pass_asan_O0): Replace declaration with that of...
10365         (make_pass_asan_O0): ...new function.
10366         (pass_tsan): Replace declaration with that of...
10367         (make_pass_tsan): ...new function.
10368         (pass_tsan_O0): Replace declaration with that of...
10369         (make_pass_tsan_O0): ...new function.
10370         (pass_lower_cf): Replace declaration with that of...
10371         (make_pass_lower_cf): ...new function.
10372         (pass_refactor_eh): Replace declaration with that of...
10373         (make_pass_refactor_eh): ...new function.
10374         (pass_lower_eh): Replace declaration with that of...
10375         (make_pass_lower_eh): ...new function.
10376         (pass_lower_eh_dispatch): Replace declaration with that of...
10377         (make_pass_lower_eh_dispatch): ...new function.
10378         (pass_lower_resx): Replace declaration with that of...
10379         (make_pass_lower_resx): ...new function.
10380         (pass_build_cfg): Replace declaration with that of...
10381         (make_pass_build_cfg): ...new function.
10382         (pass_early_tree_profile): Replace declaration with that of...
10383         (make_pass_early_tree_profile): ...new function.
10384         (pass_cleanup_eh): Replace declaration with that of...
10385         (make_pass_cleanup_eh): ...new function.
10386         (pass_sra): Replace declaration with that of...
10387         (make_pass_sra): ...new function.
10388         (pass_sra_early): Replace declaration with that of...
10389         (make_pass_sra_early): ...new function.
10390         (pass_early_ipa_sra): Replace declaration with that of...
10391         (make_pass_early_ipa_sra): ...new function.
10392         (pass_tail_recursion): Replace declaration with that of...
10393         (make_pass_tail_recursion): ...new function.
10394         (pass_tail_calls): Replace declaration with that of...
10395         (make_pass_tail_calls): ...new function.
10396         (pass_tree_loop): Replace declaration with that of...
10397         (make_pass_tree_loop): ...new function.
10398         (pass_tree_loop_init): Replace declaration with that of...
10399         (make_pass_tree_loop_init): ...new function.
10400         (pass_lim): Replace declaration with that of...
10401         (make_pass_lim): ...new function.
10402         (pass_tree_unswitch): Replace declaration with that of...
10403         (make_pass_tree_unswitch): ...new function.
10404         (pass_predcom): Replace declaration with that of...
10405         (make_pass_predcom): ...new function.
10406         (pass_iv_canon): Replace declaration with that of...
10407         (make_pass_iv_canon): ...new function.
10408         (pass_scev_cprop): Replace declaration with that of...
10409         (make_pass_scev_cprop): ...new function.
10410         (pass_empty_loop): Replace declaration with that of...
10411         (make_pass_empty_loop): ...new function.
10412         (pass_record_bounds): Replace declaration with that of...
10413         (make_pass_record_bounds): ...new function.
10414         (pass_graphite): Replace declaration with that of...
10415         (make_pass_graphite): ...new function.
10416         (pass_graphite_transforms): Replace declaration with that of...
10417         (make_pass_graphite_transforms): ...new function.
10418         (pass_if_conversion): Replace declaration with that of...
10419         (make_pass_if_conversion): ...new function.
10420         (pass_loop_distribution): Replace declaration with that of...
10421         (make_pass_loop_distribution): ...new function.
10422         (pass_vectorize): Replace declaration with that of...
10423         (make_pass_vectorize): ...new function.
10424         (pass_slp_vectorize): Replace declaration with that of...
10425         (make_pass_slp_vectorize): ...new function.
10426         (pass_complete_unroll): Replace declaration with that of...
10427         (make_pass_complete_unroll): ...new function.
10428         (pass_complete_unrolli): Replace declaration with that of...
10429         (make_pass_complete_unrolli): ...new function.
10430         (pass_parallelize_loops): Replace declaration with that of...
10431         (make_pass_parallelize_loops): ...new function.
10432         (pass_loop_prefetch): Replace declaration with that of...
10433         (make_pass_loop_prefetch): ...new function.
10434         (pass_iv_optimize): Replace declaration with that of...
10435         (make_pass_iv_optimize): ...new function.
10436         (pass_tree_loop_done): Replace declaration with that of...
10437         (make_pass_tree_loop_done): ...new function.
10438         (pass_ch): Replace declaration with that of...
10439         (make_pass_ch): ...new function.
10440         (pass_ccp): Replace declaration with that of...
10441         (make_pass_ccp): ...new function.
10442         (pass_phi_only_cprop): Replace declaration with that of...
10443         (make_pass_phi_only_cprop): ...new function.
10444         (pass_build_ssa): Replace declaration with that of...
10445         (make_pass_build_ssa): ...new function.
10446         (pass_build_alias): Replace declaration with that of...
10447         (make_pass_build_alias): ...new function.
10448         (pass_build_ealias): Replace declaration with that of...
10449         (make_pass_build_ealias): ...new function.
10450         (pass_dominator): Replace declaration with that of...
10451         (make_pass_dominator): ...new function.
10452         (pass_dce): Replace declaration with that of...
10453         (make_pass_dce): ...new function.
10454         (pass_dce_loop): Replace declaration with that of...
10455         (make_pass_dce_loop): ...new function.
10456         (pass_cd_dce): Replace declaration with that of...
10457         (make_pass_cd_dce): ...new function.
10458         (pass_call_cdce): Replace declaration with that of...
10459         (make_pass_call_cdce): ...new function.
10460         (pass_merge_phi): Replace declaration with that of...
10461         (make_pass_merge_phi): ...new function.
10462         (pass_split_crit_edges): Replace declaration with that of...
10463         (make_pass_split_crit_edges): ...new function.
10464         (pass_pre): Replace declaration with that of...
10465         (make_pass_pre): ...new function.
10466         (pass_profile): Replace declaration with that of...
10467         (make_pass_profile): ...new function.
10468         (pass_strip_predict_hints): Replace declaration with that of...
10469         (make_pass_strip_predict_hints): ...new function.
10470         (pass_lower_complex_O0): Replace declaration with that of...
10471         (make_pass_lower_complex_O0): ...new function.
10472         (pass_lower_complex): Replace declaration with that of...
10473         (make_pass_lower_complex): ...new function.
10474         (pass_lower_vector): Replace declaration with that of...
10475         (make_pass_lower_vector): ...new function.
10476         (pass_lower_vector_ssa): Replace declaration with that of...
10477         (make_pass_lower_vector_ssa): ...new function.
10478         (pass_lower_omp): Replace declaration with that of...
10479         (make_pass_lower_omp): ...new function.
10480         (pass_diagnose_omp_blocks): Replace declaration with that of...
10481         (make_pass_diagnose_omp_blocks): ...new function.
10482         (pass_expand_omp): Replace declaration with that of...
10483         (make_pass_expand_omp): ...new function.
10484         (pass_expand_omp_ssa): Replace declaration with that of...
10485         (make_pass_expand_omp_ssa): ...new function.
10486         (pass_object_sizes): Replace declaration with that of...
10487         (make_pass_object_sizes): ...new function.
10488         (pass_strlen): Replace declaration with that of...
10489         (make_pass_strlen): ...new function.
10490         (pass_fold_builtins): Replace declaration with that of...
10491         (make_pass_fold_builtins): ...new function.
10492         (pass_stdarg): Replace declaration with that of...
10493         (make_pass_stdarg): ...new function.
10494         (pass_early_warn_uninitialized): Replace declaration with that of...
10495         (make_pass_early_warn_uninitialized): ...new function.
10496         (pass_late_warn_uninitialized): Replace declaration with that of...
10497         (make_pass_late_warn_uninitialized): ...new function.
10498         (pass_cse_reciprocals): Replace declaration with that of...
10499         (make_pass_cse_reciprocals): ...new function.
10500         (pass_cse_sincos): Replace declaration with that of...
10501         (make_pass_cse_sincos): ...new function.
10502         (pass_optimize_bswap): Replace declaration with that of...
10503         (make_pass_optimize_bswap): ...new function.
10504         (pass_optimize_widening_mul): Replace declaration with that of...
10505         (make_pass_optimize_widening_mul): ...new function.
10506         (pass_warn_function_return): Replace declaration with that of...
10507         (make_pass_warn_function_return): ...new function.
10508         (pass_warn_function_noreturn): Replace declaration with that of...
10509         (make_pass_warn_function_noreturn): ...new function.
10510         (pass_cselim): Replace declaration with that of...
10511         (make_pass_cselim): ...new function.
10512         (pass_phiopt): Replace declaration with that of...
10513         (make_pass_phiopt): ...new function.
10514         (pass_forwprop): Replace declaration with that of...
10515         (make_pass_forwprop): ...new function.
10516         (pass_phiprop): Replace declaration with that of...
10517         (make_pass_phiprop): ...new function.
10518         (pass_tree_ifcombine): Replace declaration with that of...
10519         (make_pass_tree_ifcombine): ...new function.
10520         (pass_dse): Replace declaration with that of...
10521         (make_pass_dse): ...new function.
10522         (pass_nrv): Replace declaration with that of...
10523         (make_pass_nrv): ...new function.
10524         (pass_rename_ssa_copies): Replace declaration with that of...
10525         (make_pass_rename_ssa_copies): ...new function.
10526         (pass_sink_code): Replace declaration with that of...
10527         (make_pass_sink_code): ...new function.
10528         (pass_fre): Replace declaration with that of...
10529         (make_pass_fre): ...new function.
10530         (pass_check_data_deps): Replace declaration with that of...
10531         (make_pass_check_data_deps): ...new function.
10532         (pass_copy_prop): Replace declaration with that of...
10533         (make_pass_copy_prop): ...new function.
10534         (pass_vrp): Replace declaration with that of...
10535         (make_pass_vrp): ...new function.
10536         (pass_uncprop): Replace declaration with that of...
10537         (make_pass_uncprop): ...new function.
10538         (pass_return_slot): Replace declaration with that of...
10539         (make_pass_return_slot): ...new function.
10540         (pass_reassoc): Replace declaration with that of...
10541         (make_pass_reassoc): ...new function.
10542         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10543         (make_pass_rebuild_cgraph_edges): ...new function.
10544         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10545         (make_pass_remove_cgraph_callee_edges): ...new function.
10546         (pass_build_cgraph_edges): Replace declaration with that of...
10547         (make_pass_build_cgraph_edges): ...new function.
10548         (pass_local_pure_const): Replace declaration with that of...
10549         (make_pass_local_pure_const): ...new function.
10550         (pass_tracer): Replace declaration with that of...
10551         (make_pass_tracer): ...new function.
10552         (pass_warn_unused_result): Replace declaration with that of...
10553         (make_pass_warn_unused_result): ...new function.
10554         (pass_diagnose_tm_blocks): Replace declaration with that of...
10555         (make_pass_diagnose_tm_blocks): ...new function.
10556         (pass_lower_tm): Replace declaration with that of...
10557         (make_pass_lower_tm): ...new function.
10558         (pass_tm_init): Replace declaration with that of...
10559         (make_pass_tm_init): ...new function.
10560         (pass_tm_mark): Replace declaration with that of...
10561         (make_pass_tm_mark): ...new function.
10562         (pass_tm_memopt): Replace declaration with that of...
10563         (make_pass_tm_memopt): ...new function.
10564         (pass_tm_edges): Replace declaration with that of...
10565         (make_pass_tm_edges): ...new function.
10566         (pass_split_functions): Replace declaration with that of...
10567         (make_pass_split_functions): ...new function.
10568         (pass_feedback_split_functions): Replace declaration with that of...
10569         (make_pass_feedback_split_functions): ...new function.
10570         (pass_strength_reduction): Replace declaration with that of...
10571         (make_pass_strength_reduction): ...new function.
10572         (pass_ipa_lower_emutls): Replace declaration with that of...
10573         (make_pass_ipa_lower_emutls): ...new function.
10574         (pass_ipa_function_and_variable_visibility): Replace declaration with
10575         that of...
10576         (make_pass_ipa_function_and_variable_visibility): ...new function.
10577         (pass_ipa_tree_profile): Replace declaration with that of...
10578         (make_pass_ipa_tree_profile): ...new function.
10579         (pass_early_local_passes): Replace declaration with that of...
10580         (make_pass_early_local_passes): ...new function.
10581         (pass_ipa_whole_program_visibility): Replace declaration with that
10582         of...
10583         (make_pass_ipa_whole_program_visibility): ...new function.
10584         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10585         (make_pass_ipa_lto_gimple_out): ...new function.
10586         (pass_ipa_increase_alignment): Replace declaration with that of...
10587         (make_pass_ipa_increase_alignment): ...new function.
10588         (pass_ipa_inline): Replace declaration with that of...
10589         (make_pass_ipa_inline): ...new function.
10590         (pass_ipa_free_lang_data): Replace declaration with that of...
10591         (make_pass_ipa_free_lang_data): ...new function.
10592         (pass_ipa_free_inline_summary): Replace declaration with that of...
10593         (make_pass_ipa_free_inline_summary): ...new function.
10594         (pass_ipa_cp): Replace declaration with that of...
10595         (make_pass_ipa_cp): ...new function.
10596         (pass_ipa_reference): Replace declaration with that of...
10597         (make_pass_ipa_reference): ...new function.
10598         (pass_ipa_pure_const): Replace declaration with that of...
10599         (make_pass_ipa_pure_const): ...new function.
10600         (pass_ipa_pta): Replace declaration with that of...
10601         (make_pass_ipa_pta): ...new function.
10602         (pass_ipa_lto_finish_out): Replace declaration with that of...
10603         (make_pass_ipa_lto_finish_out): ...new function.
10604         (pass_ipa_tm): Replace declaration with that of...
10605         (make_pass_ipa_tm): ...new function.
10606         (pass_ipa_profile): Replace declaration with that of...
10607         (make_pass_ipa_profile): ...new function.
10608         (pass_ipa_cdtor_merge): Replace declaration with that of...
10609         (make_pass_ipa_cdtor_merge): ...new function.
10610         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10611         of...
10612         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10613         (pass_init_datastructures): Replace declaration with that of...
10614         (make_pass_init_datastructures): ...new function.
10615         (pass_fixup_cfg): Replace declaration with that of...
10616         (make_pass_fixup_cfg): ...new function.
10617         (pass_expand): Replace declaration with that of...
10618         (make_pass_expand): ...new function.
10619         (pass_instantiate_virtual_regs): Replace declaration with that of...
10620         (make_pass_instantiate_virtual_regs): ...new function.
10621         (pass_rtl_fwprop): Replace declaration with that of...
10622         (make_pass_rtl_fwprop): ...new function.
10623         (pass_rtl_fwprop_addr): Replace declaration with that of...
10624         (make_pass_rtl_fwprop_addr): ...new function.
10625         (pass_jump): Replace declaration with that of...
10626         (make_pass_jump): ...new function.
10627         (pass_jump2): Replace declaration with that of...
10628         (make_pass_jump2): ...new function.
10629         (pass_lower_subreg): Replace declaration with that of...
10630         (make_pass_lower_subreg): ...new function.
10631         (pass_cse): Replace declaration with that of...
10632         (make_pass_cse): ...new function.
10633         (pass_fast_rtl_dce): Replace declaration with that of...
10634         (make_pass_fast_rtl_dce): ...new function.
10635         (pass_ud_rtl_dce): Replace declaration with that of...
10636         (make_pass_ud_rtl_dce): ...new function.
10637         (pass_rtl_dce): Replace declaration with that of...
10638         (make_pass_rtl_dce): ...new function.
10639         (pass_rtl_dse1): Replace declaration with that of...
10640         (make_pass_rtl_dse1): ...new function.
10641         (pass_rtl_dse2): Replace declaration with that of...
10642         (make_pass_rtl_dse2): ...new function.
10643         (pass_rtl_dse3): Replace declaration with that of...
10644         (make_pass_rtl_dse3): ...new function.
10645         (pass_rtl_cprop): Replace declaration with that of...
10646         (make_pass_rtl_cprop): ...new function.
10647         (pass_rtl_pre): Replace declaration with that of...
10648         (make_pass_rtl_pre): ...new function.
10649         (pass_rtl_hoist): Replace declaration with that of...
10650         (make_pass_rtl_hoist): ...new function.
10651         (pass_rtl_store_motion): Replace declaration with that of...
10652         (make_pass_rtl_store_motion): ...new function.
10653         (pass_cse_after_global_opts): Replace declaration with that of...
10654         (make_pass_cse_after_global_opts): ...new function.
10655         (pass_rtl_ifcvt): Replace declaration with that of...
10656         (make_pass_rtl_ifcvt): ...new function.
10657         (pass_into_cfg_layout_mode): Replace declaration with that of...
10658         (make_pass_into_cfg_layout_mode): ...new function.
10659         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10660         (make_pass_outof_cfg_layout_mode): ...new function.
10661         (pass_loop2): Replace declaration with that of...
10662         (make_pass_loop2): ...new function.
10663         (pass_rtl_loop_init): Replace declaration with that of...
10664         (make_pass_rtl_loop_init): ...new function.
10665         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10666         (make_pass_rtl_move_loop_invariants): ...new function.
10667         (pass_rtl_unswitch): Replace declaration with that of...
10668         (make_pass_rtl_unswitch): ...new function.
10669         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10670         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10671         (pass_rtl_doloop): Replace declaration with that of...
10672         (make_pass_rtl_doloop): ...new function.
10673         (pass_rtl_loop_done): Replace declaration with that of...
10674         (make_pass_rtl_loop_done): ...new function.
10675         (pass_web): Replace declaration with that of...
10676         (make_pass_web): ...new function.
10677         (pass_cse2): Replace declaration with that of...
10678         (make_pass_cse2): ...new function.
10679         (pass_df_initialize_opt): Replace declaration with that of...
10680         (make_pass_df_initialize_opt): ...new function.
10681         (pass_df_initialize_no_opt): Replace declaration with that of...
10682         (make_pass_df_initialize_no_opt): ...new function.
10683         (pass_reginfo_init): Replace declaration with that of...
10684         (make_pass_reginfo_init): ...new function.
10685         (pass_inc_dec): Replace declaration with that of...
10686         (make_pass_inc_dec): ...new function.
10687         (pass_stack_ptr_mod): Replace declaration with that of...
10688         (make_pass_stack_ptr_mod): ...new function.
10689         (pass_initialize_regs): Replace declaration with that of...
10690         (make_pass_initialize_regs): ...new function.
10691         (pass_combine): Replace declaration with that of...
10692         (make_pass_combine): ...new function.
10693         (pass_if_after_combine): Replace declaration with that of...
10694         (make_pass_if_after_combine): ...new function.
10695         (pass_ree): Replace declaration with that of...
10696         (make_pass_ree): ...new function.
10697         (pass_partition_blocks): Replace declaration with that of...
10698         (make_pass_partition_blocks): ...new function.
10699         (pass_match_asm_constraints): Replace declaration with that of...
10700         (make_pass_match_asm_constraints): ...new function.
10701         (pass_regmove): Replace declaration with that of...
10702         (make_pass_regmove): ...new function.
10703         (pass_split_all_insns): Replace declaration with that of...
10704         (make_pass_split_all_insns): ...new function.
10705         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10706         (make_pass_fast_rtl_byte_dce): ...new function.
10707         (pass_lower_subreg2): Replace declaration with that of...
10708         (make_pass_lower_subreg2): ...new function.
10709         (pass_mode_switching): Replace declaration with that of...
10710         (make_pass_mode_switching): ...new function.
10711         (pass_sms): Replace declaration with that of...
10712         (make_pass_sms): ...new function.
10713         (pass_sched): Replace declaration with that of...
10714         (make_pass_sched): ...new function.
10715         (pass_ira): Replace declaration with that of...
10716         (make_pass_ira): ...new function.
10717         (pass_reload): Replace declaration with that of...
10718         (make_pass_reload): ...new function.
10719         (pass_clean_state): Replace declaration with that of...
10720         (make_pass_clean_state): ...new function.
10721         (pass_branch_prob): Replace declaration with that of...
10722         (make_pass_branch_prob): ...new function.
10723         (pass_value_profile_transformations): Replace declaration with that
10724         of...
10725         (make_pass_value_profile_transformations): ...new function.
10726         (pass_postreload_cse): Replace declaration with that of...
10727         (make_pass_postreload_cse): ...new function.
10728         (pass_gcse2): Replace declaration with that of...
10729         (make_pass_gcse2): ...new function.
10730         (pass_split_after_reload): Replace declaration with that of...
10731         (make_pass_split_after_reload): ...new function.
10732         (pass_branch_target_load_optimize1): Replace declaration with that
10733         of...
10734         (make_pass_branch_target_load_optimize1): ...new function.
10735         (pass_thread_prologue_and_epilogue): Replace declaration with that
10736         of...
10737         (make_pass_thread_prologue_and_epilogue): ...new function.
10738         (pass_stack_adjustments): Replace declaration with that of...
10739         (make_pass_stack_adjustments): ...new function.
10740         (pass_peephole2): Replace declaration with that of...
10741         (make_pass_peephole2): ...new function.
10742         (pass_if_after_reload): Replace declaration with that of...
10743         (make_pass_if_after_reload): ...new function.
10744         (pass_regrename): Replace declaration with that of...
10745         (make_pass_regrename): ...new function.
10746         (pass_cprop_hardreg): Replace declaration with that of...
10747         (make_pass_cprop_hardreg): ...new function.
10748         (pass_reorder_blocks): Replace declaration with that of...
10749         (make_pass_reorder_blocks): ...new function.
10750         (pass_branch_target_load_optimize2): Replace declaration with that
10751         of...
10752         (make_pass_branch_target_load_optimize2): ...new function.
10753         (pass_leaf_regs): Replace declaration with that of...
10754         (make_pass_leaf_regs): ...new function.
10755         (pass_split_before_sched2): Replace declaration with that of...
10756         (make_pass_split_before_sched2): ...new function.
10757         (pass_compare_elim_after_reload): Replace declaration with that of...
10758         (make_pass_compare_elim_after_reload): ...new function.
10759         (pass_sched2): Replace declaration with that of...
10760         (make_pass_sched2): ...new function.
10761         (pass_stack_regs): Replace declaration with that of...
10762         (make_pass_stack_regs): ...new function.
10763         (pass_stack_regs_run): Replace declaration with that of...
10764         (make_pass_stack_regs_run): ...new function.
10765         (pass_df_finish): Replace declaration with that of...
10766         (make_pass_df_finish): ...new function.
10767         (pass_compute_alignments): Replace declaration with that of...
10768         (make_pass_compute_alignments): ...new function.
10769         (pass_duplicate_computed_gotos): Replace declaration with that of...
10770         (make_pass_duplicate_computed_gotos): ...new function.
10771         (pass_variable_tracking): Replace declaration with that of...
10772         (make_pass_variable_tracking): ...new function.
10773         (pass_free_cfg): Replace declaration with that of...
10774         (make_pass_free_cfg): ...new function.
10775         (pass_machine_reorg): Replace declaration with that of...
10776         (make_pass_machine_reorg): ...new function.
10777         (pass_cleanup_barriers): Replace declaration with that of...
10778         (make_pass_cleanup_barriers): ...new function.
10779         (pass_delay_slots): Replace declaration with that of...
10780         (make_pass_delay_slots): ...new function.
10781         (pass_split_for_shorten_branches): Replace declaration with that of...
10782         (make_pass_split_for_shorten_branches): ...new function.
10783         (pass_split_before_regstack): Replace declaration with that of...
10784         (make_pass_split_before_regstack): ...new function.
10785         (pass_convert_to_eh_region_ranges): Replace declaration with that
10786         of...
10787         (make_pass_convert_to_eh_region_ranges): ...new function.
10788         (pass_shorten_branches): Replace declaration with that of...
10789         (make_pass_shorten_branches): ...new function.
10790         (pass_set_nothrow_function_flags): Replace declaration with that of...
10791         (make_pass_set_nothrow_function_flags): ...new function.
10792         (pass_dwarf2_frame): Replace declaration with that of...
10793         (make_pass_dwarf2_frame): ...new function.
10794         (pass_final): Replace declaration with that of...
10795         (make_pass_final): ...new function.
10796         (pass_rtl_seqabstr): Replace declaration with that of...
10797         (make_pass_rtl_seqabstr): ...new function.
10798         (pass_release_ssa_names): Replace declaration with that of...
10799         (make_pass_release_ssa_names): ...new function.
10800         (pass_early_inline): Replace declaration with that of...
10801         (make_pass_early_inline): ...new function.
10802         (pass_inline_parameters): Replace declaration with that of...
10803         (make_pass_inline_parameters): ...new function.
10804         (pass_update_address_taken): Replace declaration with that of...
10805         (make_pass_update_address_taken): ...new function.
10806         (pass_convert_switch): Replace declaration with that of...
10807         (make_pass_convert_switch): ...new function.
10808         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10809         to a subclass of simple_ipa_opt_pass along with...
10810         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10811         (make_pass_ipa_tree_profile): ...new function.
10812         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10813         subclass of gimple_opt_pass along with...
10814         (pass_data_sra_early): ...new pass_data instance and...
10815         (make_pass_sra_early): ...new function.
10816         (pass_sra): Convert from a global struct to a subclass of
10817         gimple_opt_pass along with...
10818         (pass_data_sra): ...new pass_data instance and...
10819         (make_pass_sra): ...new function.
10820         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10821         gimple_opt_pass along with...
10822         (pass_data_early_ipa_sra): ...new pass_data instance and...
10823         (make_pass_early_ipa_sra): ...new function.
10824         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10825         subclass of gimple_opt_pass along with...
10826         (pass_data_ccp): ...new pass_data instance and...
10827         (make_pass_ccp): ...new function.
10828         (pass_fold_builtins): Convert from a global struct to a subclass of
10829         gimple_opt_pass along with...
10830         (pass_data_fold_builtins): ...new pass_data instance and...
10831         (make_pass_fold_builtins): ...new function.
10832         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10833         subclass of gimple_opt_pass along with...
10834         (pass_data_copy_prop): ...new pass_data instance and...
10835         (make_pass_copy_prop): ...new function.
10836         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10837         global struct to a subclass of gimple_opt_pass along with...
10838         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10839         (make_pass_rename_ssa_copies): ...new function.
10840         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10841         subclass of gimple_opt_pass along with...
10842         (pass_data_dce): ...new pass_data instance and...
10843         (make_pass_dce): ...new function.
10844         (pass_dce_loop): Convert from a global struct to a subclass of
10845         gimple_opt_pass along with...
10846         (pass_data_dce_loop): ...new pass_data instance and...
10847         (make_pass_dce_loop): ...new function.
10848         (pass_cd_dce): Convert from a global struct to a subclass of
10849         gimple_opt_pass along with...
10850         (pass_data_cd_dce): ...new pass_data instance and...
10851         (make_pass_cd_dce): ...new function.
10852         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10853         subclass of gimple_opt_pass along with...
10854         (pass_data_dominator): ...new pass_data instance and...
10855         (make_pass_dominator): ...new function.
10856         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10857         gimple_opt_pass along with...
10858         (pass_data_phi_only_cprop): ...new pass_data instance and...
10859         (make_pass_phi_only_cprop): ...new function.
10860         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10861         subclass of gimple_opt_pass along with...
10862         (pass_data_dse): ...new pass_data instance and...
10863         (make_pass_dse): ...new function.
10864         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10865         a subclass of gimple_opt_pass along with...
10866         (pass_data_forwprop): ...new pass_data instance and...
10867         (make_pass_forwprop): ...new function.
10868         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10869         struct to a subclass of gimple_opt_pass along with...
10870         (pass_data_tree_ifcombine): ...new pass_data instance and...
10871         (make_pass_tree_ifcombine): ...new function.
10872         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10873         subclass of gimple_opt_pass along with...
10874         (pass_data_ch): ...new pass_data instance and...
10875         (make_pass_ch): ...new function.
10876         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10877         subclass of gimple_opt_pass along with...
10878         (pass_data_tree_loop): ...new pass_data instance and...
10879         (make_pass_tree_loop): ...new function.
10880         (pass_tree_loop_init): Convert from a global struct to a subclass of
10881         gimple_opt_pass along with...
10882         (pass_data_tree_loop_init): ...new pass_data instance and...
10883         (make_pass_tree_loop_init): ...new function.
10884         (pass_lim): Convert from a global struct to a subclass of
10885         gimple_opt_pass along with...
10886         (pass_data_lim): ...new pass_data instance and...
10887         (make_pass_lim): ...new function.
10888         (pass_tree_unswitch): Convert from a global struct to a subclass of
10889         gimple_opt_pass along with...
10890         (pass_data_tree_unswitch): ...new pass_data instance and...
10891         (make_pass_tree_unswitch): ...new function.
10892         (pass_predcom): Convert from a global struct to a subclass of
10893         gimple_opt_pass along with...
10894         (pass_data_predcom): ...new pass_data instance and...
10895         (make_pass_predcom): ...new function.
10896         (pass_vectorize): Convert from a global struct to a subclass of
10897         gimple_opt_pass along with...
10898         (pass_data_vectorize): ...new pass_data instance and...
10899         (make_pass_vectorize): ...new function.
10900         (pass_graphite): Convert from a global struct to a subclass of
10901         gimple_opt_pass along with...
10902         (pass_data_graphite): ...new pass_data instance and...
10903         (make_pass_graphite): ...new function.
10904         (pass_graphite_transforms): Convert from a global struct to a subclass
10905         of gimple_opt_pass along with...
10906         (pass_data_graphite_transforms): ...new pass_data instance and...
10907         (make_pass_graphite_transforms): ...new function.
10908         (pass_check_data_deps): Convert from a global struct to a subclass of
10909         gimple_opt_pass along with...
10910         (pass_data_check_data_deps): ...new pass_data instance and...
10911         (make_pass_check_data_deps): ...new function.
10912         (pass_iv_canon): Convert from a global struct to a subclass of
10913         gimple_opt_pass along with...
10914         (pass_data_iv_canon): ...new pass_data instance and...
10915         (make_pass_iv_canon): ...new function.
10916         (pass_scev_cprop): Convert from a global struct to a subclass of
10917         gimple_opt_pass along with...
10918         (pass_data_scev_cprop): ...new pass_data instance and...
10919         (make_pass_scev_cprop): ...new function.
10920         (pass_record_bounds): Convert from a global struct to a subclass of
10921         gimple_opt_pass along with...
10922         (pass_data_record_bounds): ...new pass_data instance and...
10923         (make_pass_record_bounds): ...new function.
10924         (pass_complete_unroll): Convert from a global struct to a subclass of
10925         gimple_opt_pass along with...
10926         (pass_data_complete_unroll): ...new pass_data instance and...
10927         (make_pass_complete_unroll): ...new function.
10928         (pass_complete_unrolli): Convert from a global struct to a subclass of
10929         gimple_opt_pass along with...
10930         (pass_data_complete_unrolli): ...new pass_data instance and...
10931         (make_pass_complete_unrolli): ...new function.
10932         (pass_parallelize_loops): Convert from a global struct to a subclass
10933         of gimple_opt_pass along with...
10934         (pass_data_parallelize_loops): ...new pass_data instance and...
10935         (make_pass_parallelize_loops): ...new function.
10936         (pass_loop_prefetch): Convert from a global struct to a subclass of
10937         gimple_opt_pass along with...
10938         (pass_data_loop_prefetch): ...new pass_data instance and...
10939         (make_pass_loop_prefetch): ...new function.
10940         (pass_iv_optimize): Convert from a global struct to a subclass of
10941         gimple_opt_pass along with...
10942         (pass_data_iv_optimize): ...new pass_data instance and...
10943         (make_pass_iv_optimize): ...new function.
10944         (pass_tree_loop_done): Convert from a global struct to a subclass of
10945         gimple_opt_pass along with...
10946         (pass_data_tree_loop_done): ...new pass_data instance and...
10947         (make_pass_tree_loop_done): ...new function.
10948         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10949         struct to a subclass of gimple_opt_pass along with...
10950         (pass_data_cse_reciprocals): ...new pass_data instance and...
10951         (make_pass_cse_reciprocals): ...new function.
10952         (pass_cse_sincos): Convert from a global struct to a subclass of
10953         gimple_opt_pass along with...
10954         (pass_data_cse_sincos): ...new pass_data instance and...
10955         (make_pass_cse_sincos): ...new function.
10956         (pass_optimize_bswap): Convert from a global struct to a subclass of
10957         gimple_opt_pass along with...
10958         (pass_data_optimize_bswap): ...new pass_data instance and...
10959         (make_pass_optimize_bswap): ...new function.
10960         (pass_optimize_widening_mul): Convert from a global struct to a
10961         subclass of gimple_opt_pass along with...
10962         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10963         (make_pass_optimize_widening_mul): ...new function.
10964         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10965         subclass of gimple_opt_pass along with...
10966         (pass_data_phiopt): ...new pass_data instance and...
10967         (make_pass_phiopt): ...new function.
10968         (pass_cselim): Convert from a global struct to a subclass of
10969         gimple_opt_pass along with...
10970         (pass_data_cselim): ...new pass_data instance and...
10971         (make_pass_cselim): ...new function.
10972         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10973         subclass of gimple_opt_pass along with...
10974         (pass_data_phiprop): ...new pass_data instance and...
10975         (make_pass_phiprop): ...new function.
10976         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10977         subclass of gimple_opt_pass along with...
10978         (pass_data_pre): ...new pass_data instance and...
10979         (make_pass_pre): ...new function.
10980         (pass_fre): Convert from a global struct to a subclass of
10981         gimple_opt_pass along with...
10982         (pass_data_fre): ...new pass_data instance and...
10983         (make_pass_fre): ...new function.
10984         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10985         subclass of gimple_opt_pass along with...
10986         (pass_data_reassoc): ...new pass_data instance and...
10987         (make_pass_reassoc): ...new function.
10988         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10989         subclass of gimple_opt_pass along with...
10990         (pass_data_sink_code): ...new pass_data instance and...
10991         (make_pass_sink_code): ...new function.
10992         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10993         subclass of gimple_opt_pass along with...
10994         (pass_data_strlen): ...new pass_data instance and...
10995         (make_pass_strlen): ...new function.
10996         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10997         struct to a subclass of gimple_opt_pass along with...
10998         (pass_data_build_alias): ...new pass_data instance and...
10999         (make_pass_build_alias): ...new function.
11000         (pass_build_ealias): Convert from a global struct to a subclass of
11001         gimple_opt_pass along with...
11002         (pass_data_build_ealias): ...new pass_data instance and...
11003         (make_pass_build_ealias): ...new function.
11004         (pass_ipa_pta): Convert from a global struct to a subclass of
11005         simple_ipa_opt_pass along with...
11006         (pass_data_ipa_pta): ...new pass_data instance and...
11007         (make_pass_ipa_pta): ...new function.
11008         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11009         subclass of gimple_opt_pass along with...
11010         (pass_data_uncprop): ...new pass_data instance and...
11011         (make_pass_uncprop): ...new function.
11012         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11013         global struct to a subclass of gimple_opt_pass along with...
11014         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11015         (make_pass_late_warn_uninitialized): ...new function.
11016         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11017         to a subclass of gimple_opt_pass along with...
11018         (pass_data_init_datastructures): ...new pass_data instance and...
11019         (make_pass_init_datastructures): ...new function.
11020         (pass_early_warn_uninitialized): Convert from a global struct to a
11021         subclass of gimple_opt_pass along with...
11022         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11023         (make_pass_early_warn_uninitialized): ...new function.
11024         (pass_update_address_taken): Convert from a global struct to a
11025         subclass of gimple_opt_pass along with...
11026         (pass_data_update_address_taken): ...new pass_data instance and...
11027         (make_pass_update_address_taken): ...new function.
11028         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11029         struct to a subclass of gimple_opt_pass along with...
11030         (pass_data_release_ssa_names): ...new pass_data instance and...
11031         (make_pass_release_ssa_names): ...new function.
11032         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11033         subclass of gimple_opt_pass along with...
11034         (pass_data_stdarg): ...new pass_data instance and...
11035         (make_pass_stdarg): ...new function.
11036         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11037         global struct to a subclass of gimple_opt_pass along with...
11038         (pass_data_convert_switch): ...new pass_data instance and...
11039         (make_pass_convert_switch): ...new function.
11040         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11041         to a subclass of gimple_opt_pass along with...
11042         (pass_data_tail_recursion): ...new pass_data instance and...
11043         (make_pass_tail_recursion): ...new function.
11044         (pass_tail_calls): Convert from a global struct to a subclass of
11045         gimple_opt_pass along with...
11046         (pass_data_tail_calls): ...new pass_data instance and...
11047         (make_pass_tail_calls): ...new function.
11048         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11049         struct to a subclass of gimple_opt_pass along with...
11050         (pass_data_lower_vector): ...new pass_data instance and...
11051         (make_pass_lower_vector): ...new function.
11052         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11053         gimple_opt_pass along with...
11054         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11055         (make_pass_lower_vector_ssa): ...new function.
11056         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11057         to a subclass of gimple_opt_pass along with...
11058         (pass_data_slp_vectorize): ...new pass_data instance and...
11059         (make_pass_slp_vectorize): ...new function.
11060         (pass_ipa_increase_alignment): Convert from a global struct to a
11061         subclass of simple_ipa_opt_pass along with...
11062         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11063         (make_pass_ipa_increase_alignment): ...new function.
11064         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11065         gimple_opt_pass along with...
11066         (pass_data_vrp): ...new pass_data instance and...
11067         (make_pass_vrp): ...new function.
11068         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11069         subclass of simple_ipa_opt_pass along with...
11070         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11071         (make_pass_ipa_free_lang_data): ...new function.
11072         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11073         gimple_opt_pass along with...
11074         (pass_data_tsan): ...new pass_data instance and...
11075         (make_pass_tsan): ...new function.
11076         (pass_tsan_O0): Convert from a global struct to a subclass of
11077         gimple_opt_pass along with...
11078         (pass_data_tsan_O0): ...new pass_data instance and...
11079         (make_pass_tsan_O0): ...new function.
11080         * var-tracking.c (pass_variable_tracking): Convert from a global
11081         struct to a subclass of rtl_opt_pass along with...
11082         (pass_data_variable_tracking): ...new pass_data instance and...
11083         (make_pass_variable_tracking): ...new function.
11084         * web.c (pass_web): Convert from a global struct to a subclass of
11085         rtl_opt_pass along with...
11086         (pass_data_web): ...new pass_data instance and...
11087         (make_pass_web): ...new function.
11088         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11089         declaration with that of...
11090         (make_pass_mode_switch_use): ...new function.
11091         (pass_resolve_sw_modes): Replace declaration with that of...
11092         (make_pass_resolve_sw_modes): ...new function.
11093         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11094         from a global struct to a subclass of rtl_opt_pass along with...
11095         (pass_data_mode_switch_use): ...new pass_data instance and...
11096         (make_pass_mode_switch_use): ...new function.
11097         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11098         from a global struct to a subclass of rtl_opt_pass along with...
11099         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11100         (make_pass_resolve_sw_modes): ...new function.
11101         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11102         struct to a subclass of rtl_opt_pass along with...
11103         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11104         (make_pass_insert_vzeroupper): ...new function.
11105         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11106         global struct to a subclass of rtl_opt_pass along with...
11107         (pass_data_work_around_errata): ...new pass_data instance and...
11108         (make_pass_work_around_errata): ...new function.
11109         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11110         struct to a subclass of rtl_opt_pass along with...
11111         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11112         (make_pass_mips_machine_reorg2): ...new function.
11114 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11116         * passes.c (pass_manager::operator new): New.
11118 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11120         Handwritten part of conversion of passes to C++ classes.
11122         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11123         (toplev.o): Add dep on PASS_MANAGER_H.
11124         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11125         of early local pases to reflect this moving from a global to a
11126         member of gcc::pass_manager.
11127         (cgraph_add_new_function): Likewise.
11128         * lto-cgraph.c (lto_output_node): Update for conversion of
11129         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11130         * passes.c (opt_pass::clone): New.
11131         (opt_pass::gate): New.
11132         (opt_pass::execute): New.
11133         (opt_pass::opt_pass): New.
11134         (pass_manager::execute_early_local_passes): New.
11135         (pass_manager::execute_pass_mode_switching): new.
11136         (finish_optimization_passes): Convert to...
11137         (pass_manager::finish_optimization_passes): ...this.
11138         (finish_optimization_passes): Update for conversion of passes to
11139         C++ classes.
11140         (register_dump_files_1): Use has_gate since we cannot portably
11141         check a vtable entry against NULL.
11142         (dump_one_pass): Likewise.
11143         (ipa_write_summaries_2): Likewise.
11144         (ipa_write_optimization_summaries_1): Likewise.
11145         (ipa_read_summaries_1): Likewise.
11146         (ipa_read_optimization_summaries_1): Likewise.
11147         (execute_ipa_stmt_fixups): Likewise.
11148         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11149         pass-creation functions rather than wiring up globals, and
11150         storing the results in fields of pass_manager generated using
11151         pass-instances.def.
11152         (pass_manager::dump_profile_report): Update for conversion of
11153         passes to C++ classes.
11154         (pass_manager::execute_ipa_summary_passes): Likewise.
11155         (execute_one_ipa_transform_pass): Likewise.
11156         (execute_one_pass): Use has_gate and has_execute since we cannot
11157         portably check a vtable entry against NULL.
11158         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11159         (pass_manager): Use pass-instances.def to add fields for the
11160         various pass instances.
11161         * toplev.c (finalize): Update for move of
11162         finish_optimization_passes to a method of gcc::pass_manager.
11163         * toplev.h (finish_optimization_passes): Move to method of class
11164         pass_manager.
11165         * tree-pass.h (struct pass_data): New.
11166         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11167         (opt_pass::gate): Convert to virtual function.
11168         (opt_pass::~opt_pass): New.
11169         (opt_pass::clone): New.
11170         (opt_pass::execute): Convert to virtual function.
11171         (opt_pass::opt_pass): New.
11172         (opt_pass::ctxt_): new.
11173         (gimple_opt_pass): Convert to subclass of opt_pass.
11174         (gimple_opt_pass::gimple_opt_pass): New.
11175         (rtl_opt_pass): Convert to subclass of opt_pass.
11176         (rtl_opt_pass::rtl_opt_pass): New.
11177         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11178         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11179         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11180         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11181         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11182         invocation of pass_mode_switching to reflect this moving from a
11183         global to a member of gcc::pass_manager.
11184         (ix86_option_override): Rework how pass_insert_vzeroupper is
11185         added to the pass_manager to reflect autogenerated changes.
11186         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11188 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11190         PR rtl-optimization/57708
11191         * recog.c (peep2_find_free_register): Validate all regs in a
11192         multi-reg mode.
11194 2013-08-05  Jan Hubicka  <jh@suse.cz>
11196         PR lto/57602
11197         * cgraph.c (verify_cgraph_node): Accept local flags from other
11198         partitions.
11199         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11200         (function_and_variable_visibility): Likewise.
11201         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11203 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11205         * graph.c (init_graph_slim_pretty_print): Remove.
11206         (print_graph_cfg): Do not call it.  Use local pretty printer.
11207         (start_graph_dump): Likewise.
11209 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11211         * gimple-pretty-print.c (buffer): Remove.
11212         (initialized): Likewise.
11213         (maybe_init_pretty_print): Likewise.
11214         (print_gimple_stmt): Do not call it.  Use non-static local
11215         pretty_printer variable.
11216         (print_gimple_expr): Likewise.
11217         (print_gimple_seq): Likewise.
11218         (gimple_dump_bb): Likewise.
11220 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11222         * asan.c (asan_pp): Remove.
11223         (asan_pp_initialized): Likewise.
11224         (asan_pp_initialize): Likewise.
11225         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11226         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11227         (asan_add_global): Likewise.
11229 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11231         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11232         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11233         * pretty-print.c (pp_formatted_text_data): Likewise.
11234         (pp_indent): Rename from pp_base_indent.
11235         (pp_format): Rename from pp_base_format.
11236         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11237         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11238         (pp_flush): Rename from pp_base_flush.
11239         (pp_set_line_maximum_length): Rename from
11240         pp_base_set_line_maximum_length.
11241         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11242         (pp_set_prefix): Rename from pp_base_set_prefix.
11243         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11244         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11245         (pp_append_text): Rename from pp_base_append_text.
11246         (pp_formatted_text): Rename from pp_base_formatted_text.
11247         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11248         (pp_remaining_character_count_for_line): Rename from
11249         pp_base_remaining_character_count_for_line.
11250         (pp_newline): Rename from pp_base_newline.
11251         (pp_character): Rename from pp_base_character.
11252         (pp_string): Rename from pp_base_string.
11253         (pp_maybe_space): Rename from pp_base_maybe_space.
11254         * asan.c (asan_pp_string): Adjust.
11255         (asan_emit_stack_protection): Likewise.
11256         (asan_add_global): Likewise.
11257         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11258         * tree-mudflap.c (mf_varname_tree): Likewise.
11259         * tree-pretty-print.c (pp_tree_identifier): Rename from
11260         pp_base_tree_identifier.
11261         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11262         Declare as function.
11264 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11266         * pretty-print.h (pp_bar_bar): New.
11267         (pp_ampersand_ampersand): Likewise.
11268         (pp_less_equal): Likewise.
11269         (pp_greater_equal): Likewise.
11270         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11271         printer functions instead of pp_string or operators and punctuators.
11272         (dump_gimple_call): Likewise.
11273         (dump_gimple_omp_for): Likewise.
11274         (dump_gimple_transaction): Likewise.
11275         (dump_gimple_phi): Likewise.
11276         (pp_gimple_stmt_1): Likewise.
11277         * sched-vis.c (print_insn): Likewise.
11278         * tree-mudflap.c (mf_varname_tree): Likewise.
11279         * tree-pretty-print.c (dump_block_node): Likewise.
11280         (dump_generic_node): Likewise.
11282 2013-08-02  Jan Hubicka  <jh@suse.cz>
11284         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11285         boundaries.
11286         * lto-streamer-out.c (tree_is_indexable): Results decls and
11287         parm decls are not indexable.
11288         (DFS_write_tree_body): Do not follow args and results.
11289         (hash_tree): Likewise.
11290         (output_functions): Rearrange so struct function is needed
11291         only when real body is output; be able to also ouptut abstract
11292         functions; output DECL_ARGUMENTS and DECL_RESULT.
11293         (lto_output): When not in WPA, ale store abstract functions.
11294         (write_symbol): Do not care about RESULT_DECL.
11295         (output_symbol_p): Handle correctly sbtract decls.
11296         * lto-streamer-in.c (input_function): Rearrange so struct
11297         function can be NULL at entry; allow streaming of
11298         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11299         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11300         sanity check during LTO.
11301         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11302         RESULT_DECl and DECL_ARGUMENTS.
11303         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11304         Likewise.
11306 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11308         * pretty-print.h (pp_underscore): New.
11309         (pp_comma): Tidy.
11310         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11311         printer functions instead of pp_character.
11312         (dump_binary_rhs): Likewise.
11313         (dump_ternary_rhs): Likewise.
11314         (dump_gimple_call_args): Likewise.
11315         (pp_points_to_solution): Likewise.
11316         (dump_gimple_call): Likewise.
11317         (dump_gimple_switch): Likewise.
11318         (dump_gimple_cond): Likewise.
11319         (dump_gimple_bind): Likewise.
11320         (dump_gimple_try): Likewise.
11321         (dump_gimple_omp_for): Likewise.
11322         (dump_gimple_omp_continue): Likewise.
11323         (dump_gimple_omp_single): Likewise.
11324         (dump_gimple_omp_sections): Likewise.
11325         (dump_gimple_omp_block): Likewise.
11326         (dump_gimple_omp_critical): Likewise.
11327         (dump_gimple_transaction): Likewise.
11328         (dump_gimple_asm): Likewise.
11329         (dump_gimple_phi): Likewise.
11330         (dump_gimple_omp_parallel): Likewise.
11331         (dump_gimple_omp_task): Likewise.
11332         (dump_gimple_omp_atomic_load): Likewise.
11333         (dump_gimple_omp_atomic_store): Likewise.
11334         (dump_gimple_mem_ops): Likewise.
11335         (pp_gimple_stmt_1): Likewise.
11336         (pp_cfg_jump): Likewise.
11337         (dump_implicit_edges): Likewise.
11338         (gimple_dump_bb_for_graph): Likewise.
11339         * graph.c (draw_cfg_node): Likewise.
11340         * langhooks.c (lhd_print_error_function): Likewise.
11341         * sched-vis.c (print_exp): Likewise.
11342         (print_value): Likewise.
11343         (print_pattern): Likewise.
11344         (print_insn): Likewise.
11345         (rtl_dump_bb_for_graph): Likewise.
11346         * tree-pretty-print.c (dump_function_declaration): Likewise.
11347         (dump_array_domain): Likewise.
11348         (dump_omp_clause): Likewise.
11349         (dump_location): Likewise.
11350         (dump_generic_node): Likewise.
11351         (print_struct_decl): Likewise.
11352         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11354 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11356         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11357         candidate table when replacing a candidate statement.
11358         (replace_rhs_if_not_dup): Likewise.
11359         (replace_one_candidate): Likewise.
11361 2013-08-02  Jan Hubicka  <jh@suse.cz>
11362             Martin Liska  <marxin.liska@gmail.com>
11364         * cgraphunit.c (add_new_function): Fix logic when adding from
11365         late IPA pass.
11366         (assemble_thunk): Rename to ...
11367         (expand_thunk); .. this one; export; get it working with
11368         general functions; make produced gimple valid.
11369         * cgraph.h (expand_thunk): Declare.
11371 2013-08-02  Jan Hubicka  <jh@suse.cz>
11373         * ipa-cp.c (gather_context_independent_values): Use
11374         ipa_get_param_move_cost.
11375         (get_replacement_map): Remove PARAM; move parameter folding
11376         into tree-inline.c
11377         (create_specialized_node): Update.
11378         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11379         assert that we have gimple body; update move_cost.
11380         (count_formal_params): Assert that we have gimple body.
11381         (ipa_dump_param): New function.
11382         (ipa_alloc_node_params): Break out from ...
11383         (ipa_initialize_node_params): ... here.
11384         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11385         (ipa_write_node_info): Stream move costs.
11386         (ipa_read_node_info): Read move costs.
11387         (ipa_update_after_lto_read): Do not recompute node params.
11388         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11389         (ipa_get_param): Check we are not in WPA.
11390         (ipa_get_param_move_cost): New.
11391         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11392         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11393         parm numbers to be present.
11395 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11397         PR rtl-optimization/58048
11398         * lra-constraints.c (process_alt_operands): Don't check asm
11399         operand on register.
11401 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11403         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11404         the implied StoreLoad barrier for atomic operations if before.
11406 2013-08-02  Jan Hubicka  <jh@suse.cz>
11407             Martin Liska  <marxin.liska@gmail.com>
11409         * cgraph.c (cgraph_function_body_availability): Do not check
11410         cgraph flags.
11411         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11412         symtab_node_availability): Declare.
11413         * ipa.c (can_replace_by_local_alias): New.
11414         (function_and_variable_visibility): Use it.
11415         * symtab.c (symtab_for_node_and_aliases,
11416         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11418 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11420         PR rtl-optimization/57963
11421         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11422         (lra_constraints): Use them.
11424 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11426         * config/arm/types.md (define_attr "type"): Add "load_acq"
11427         and "store_rel".
11428         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11429         changes.
11430         (cortex_a53_store1): Likewise.
11432 2013-08-01  Jan Hubicka  <jh@suse.cz>
11434         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11435         partitions are not needed.
11437 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11439         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11440         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11441         MAYBE_NON_Q_CLASS_P where appropriate.
11443 2013-08-01  Jan Hubicka  <jh@suse.cz>
11445         * cgraph.h (release_function_body): Declare.
11446         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11447         of unused delcarations.
11449 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11451         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11452         RTL form when subtracting a constant.
11454 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11456         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11457         Generate canonical plus rtx with negated immediate instead of minus
11458         where appropriate.
11459         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11461 2013-08-01  Jan Hubicka  <jh@suse.cz>
11463         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11464         (cgraph_release_function_body): Likewise.
11465         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11466         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11467         to used_as_abstract_origin.
11468         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11469         symbols used as abstract origins.
11470         * cgraphunit.c (analyze_functions): Update.
11471         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11472         used_as_abstract_origin.
11473         * tree-inline.c (tree_function_versioning): Update
11474         used_as_abstract_origin; be ready for DECL_RESULT and
11475         DECL_ARGUMENTS to be NULL.
11477         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11478         for abstract functions.
11479         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11480         real symbols.
11482 2013-08-01  Jan Hubicka  <jh@suse.cz>
11484         * profile.c (compute_value_histograms): Fix thinko.
11486 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11488         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11489         aarch-common-protos.h to extra_headers.
11490         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11491         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11492         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11494 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11496         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11497         Include "../arm/types.md".  Define "type" attribute for all patterns.
11498         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11499         attribute changes.
11501 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11503         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11504         to support power8 load fusion.
11505         (fusion_gpr_mem_load): Likewise.
11507         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11509         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11510         declarations for power8 load fusion.
11511         (emit_fusion_gpr_load): Likewise.
11513         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11514         tuning for power8, turn on fusion mode by default.  Turn on sign
11515         extending fusion mode if normal fusion mode is on, and we are at
11516         -O2 or -O3.
11517         (fusion_gpr_load_p): New function, return true if we can fuse an
11518         addis instruction with a dependent load to a GPR.
11519         (emit_fusion_gpr_load): Emit the instructions for power8 load
11520         fusion to GPRs.
11522         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11523         (VSX load fusion peepholes): New peepholes to fuse together an
11524         addi instruction with a VSX load instruction.
11526         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11527         peepholes to fuse an addis instruction with a load to a GPR base
11528         register.  If we are supporting sign extending fusions, convert
11529         sign extending loads to zero extending loads and add an explicit
11530         sign extension.
11532 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11534         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11535         aarch-common-protos.h to extra_headers.
11536         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11537         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11538         (arm_early_store_addr_dep): Likewise.
11539         (arm_no_early_alu_shift_dep): Likewise.
11540         (arm_no_early_alu_shift_value_dep): Likewise.
11541         (arm_no_early_mul_dep): Likewise.
11542         (arm_no_early_store_addr_dep): Likewise.
11543         (arm_mac_accumulator_is_mul_result): Likewise.
11544         (arm_mac_accumulator_is_result): Likewise.
11545         * config/arm/aarch-common.c: ... here.  New file.
11546         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11547         here to ...
11548         (arm_early_store_addr_dep): Likewise.
11549         (arm_no_early_alu_shift_dep): Likewise.
11550         (arm_no_early_alu_shift_value_dep): Likewise.
11551         (arm_no_early_mul_dep): Likewise.
11552         (arm_no_early_store_addr_dep): Likewise.
11553         (arm_mac_accumulator_is_mul_result): Likewise.
11554         (arm_mac_accumulator_is_result): Likewise.
11555         * config/arm/aarch-common-protos.h: ... here.  New file.
11556         * config/arm/t-arm (aarch-common.o): Define.
11558 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11560         * config/arm/arm.md: Include new file "types.md".
11561         (define_attr "type"): Move from here to ...
11562         (define_attr "mul32"): Likewise.
11563         (define_attr "mul64"): Likewise.
11564         * config/arm/types.md: ... here.  New file.
11566 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11568         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11569         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11571 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11573         * gen-pass-instances.awk: Fix offset of substr().
11575 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11577         * Makefile.in (pass-instances.def): New.
11578         (passes.o): Replace dependency on passes.def with one on
11579         pass-instances.def
11581         * gen-pass-instances.awk: New.
11583         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11584         rather than passes.def, updating local definition of NEXT_PASS
11585         macro to add an extra NUM parameter (currently unused).
11587 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11589         * Makefile.in (PASS_MANAGER_H): New.
11590         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11591         (passes.o): Likewise.
11592         (statistics.o): Likewise.
11593         (cgraphunit.o): Likewise.
11594         (context.o): Depend on PASS_MANAGER_H.
11596         * pass_manager.h: New.
11598         * cgraphunit.c (cgraph_add_new_function): Update for moves
11599         of globals to fields of pass_manager.
11600         (analyze_function): Likewise.
11601         (expand_function): Likewise.
11602         (ipa_passes): Likewise.
11603         (compile): Likewise.
11605         * context.c (context::context): New.
11606         * context.h  (context::context): New.
11607         (context::get_passes): New.
11608         (context::passes_): New.
11610         * lto-cgraph.c (input_node): Update for moves of globals to
11611         fields of pass_manager.
11613         * passes.c (all_passes): Remove, in favor of a field of the
11614         same name within the new class pass_manager.
11615         (all_small_ipa_passes): Likewise.
11616         (all_lowering_passes): Likewise.
11617         (all_regular_ipa_passes): Likewise.
11618         (all_late_ipa_passes): Likewise.
11619         (all_lto_gen_passes): Likewise.
11620         (passes_by_id): Likewise.
11621         (passes_by_id_size): Likewise.
11622         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11623         the new class pass_manager.
11624         (set_pass_for_id): Convert to...
11625         (pass_manager::set_pass_for_id): ...method.
11626         (get_pass_for_id): Convert to...
11627         (pass_manager::get_pass_for_id): ...method.
11628         (register_one_dump_file): Move body of implementation into...
11629         (pass_manager::register_one_dump_file): ...here.
11630         (register_dump_files_1): Convert to...
11631         (pass_manager::register_dump_files_1): ...method.
11632         (register_dump_files): Convert to...
11633         (pass_manager::register_dump_files): ...method.
11634         (create_pass_tab): Update for moves of globals to fields of
11635         pass_manager.
11636         (dump_passes): Move body of implementation into...
11637         (pass_manager::dump_passes): ...here.
11638         (register_pass): Move body of implementation into...
11639         (pass_manager::register_pass): ...here.
11640         (init_optimization_passes): Convert into...
11641         (pass_manager::pass_manager): ...constructor for new
11642         pass_manager class, and initialize the pass_lists array.
11643         (check_profile_consistency): Update for moves of globals to
11644         fields of pass_manager.
11645         (dump_profile_report): Move body of implementation into...
11646         (pass_manager::dump_profile_report): ...here.
11647         (ipa_write_summaries_1): Update for moves of pass lists from
11648         being globals to fields of pass_manager.
11649         (ipa_write_optimization_summaries): Likewise.
11650         (ipa_read_summaries):  Likewise.
11651         (ipa_read_optimization_summaries): Likewise.
11652         (execute_all_ipa_stmt_fixups): Likewise.
11654         * statistics.c (statistics_fini): Update for moves of globals to
11655         fields of pass_manager.
11657         * toplev.c (general_init): Replace call to
11658         init_optimization_passes with construction of the pass_manager
11659         instance.
11661         * tree-pass.h (all_passes): Remove, in favor of a field of the
11662         same name within the new class pass_manager.
11663         (all_small_ipa_passes): Likewise.
11664         (all_lowering_passes): Likewise.
11665         (all_regular_ipa_passes): Likewise.
11666         (all_lto_gen_passes): Likewise.
11667         (all_late_ipa_passes): Likewise.
11668         (passes_by_id): Likewise.
11669         (passes_by_id_size): Likewise.
11670         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11671         the new class pass_manager.
11672         (get_pass_for_id): Remove.
11674 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11676         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11677         configs.
11679 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11681         * arm.md (mulhi3): New expand pattern.
11683 2013-07-30  Jan Hubicka  <jh@suse.cz>
11684             Martin Liska  <marxin.liska@gmail.com>
11686         * profile.c (compute_value_histograms): Do not ICE when
11687         there is mismatch only on some counters.
11689 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11691         PR rtl-optimization/57637
11692         * function.c (move_insn_for_shrink_wrap): Also check the
11693         GEN set of the LIVE problem for the liveness analysis
11694         if it exists, otherwise give up.
11696 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11698         PR tree-optimization/57993
11699         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11700         replaced statement in the candidate table.
11701         (phi_add_costs): Return infinite cost when the hidden basis does
11702         not dominate all phis on which the candidate is dependent.
11703         (replace_one_candidate): Record replaced statement in the
11704         candidate table.
11706 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11708         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11709         (ashlv2si3): New expander.
11710         (*ashlv2si3_i): New define_insn_and_split.
11711         * predicates.md (float_operation): Allow patterns with three
11712         basic sub-patterns.
11714         PR rtl-optimization/58021
11715         * mode-switching.c (create_pre_exit): Always split off preceding
11716         insns if we are not at the basic block head.
11718 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11720         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11721         (UCLIBC_DYNAMIC_LINKER): New macro.
11722         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11723         `-mnan=2008'.
11724         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11725         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11726         `-mnan=2008'.
11727         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11728         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11729         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11730         for SF and DF modes.  Use ieee_quad_format for TF mode.
11731         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11732         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11733         (mips_option_override): Handle `-mnan=legacy'.
11734         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11735         `-mabs=2008' and `-mnan=2008'.
11736         (OPTION_DEFAULT_SPECS): Add "nan" default.
11737         (ASM_SPEC): Handle `-mnan='.
11738         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11739         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11740         comment accordingly.
11741         (neg<mode>2): Likewise.
11742         * config/mips/mips.opt (mabs, mnan): New options.
11743         * doc/install.texi (Configuration): Document `--with-nan=' option.
11744         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11745         `-mnan=' options.
11746         (MIPS Options): Document them.
11747         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11748         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11749         * configure: Regenerate.
11750         * config.in: Regenerate.
11752 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11754         * config/i386/i386.md (float post-reload splitters): Do not check
11755         for subregs of SSE registers.
11757 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11758             H.J. Lu  <hongjiu.lu@intel.com>
11760         PR target/57954
11761         PR target/57988
11762         * config/i386/i386.md (post-reload splitter
11763         to avoid partial SSE reg dependency stalls): New pattern.
11765 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11767         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11769 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11771         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11772         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11774 2013-07-26  Cary Coutant  <ccoutant@google.com>
11776         * dwarf2out.c (die_checksum_ordered): Don't include template
11777         instantiations in signature.
11778         (is_template_parameter): New function.
11779         (is_template_instantiation): New function.
11780         (generate_skeleton_bottom_up): Don't include template instantiations
11781         in type unit DIE.
11782         (generate_skeleton): Likewise.
11783         (break_out_comdat_types): Move recursive call to break out nested
11784         types earlier.
11785         (prune_unused_types_mark_generic_parms_dies): Call
11786         is_template_parameter.
11788 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11790         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11791         uses vector registers.
11792         * config/aarch64/iterators.md: Add attributes rtn and vas.
11794 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11795             Richard Earnshaw  <richard.earnshaw@arm.com>
11797         * combine.c (simplify_comparison): Re-canonicalize operands
11798         where appropriate.
11799         * config/arm/arm.md (movcond_addsi): New splitter.
11801 2013-07-25  Sterling Augustine  <saugustine@google.com>
11803         * dwarf2out.c (size_of_pubnames): Move code to...
11804         (include_pubname_in_output): ...here.  New.
11805         (want_pubnames): Rearrange.
11806         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11808 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11810         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11812 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11814         PR target/38836
11815         * doc/extend.texi: Remove obsolete builtins. Fix
11816         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11818 2013-07-25  Jan Hubicka  <jh@suse.cz>
11820         * cgraph.c (release_function_body): Break out from ...
11821         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11822         and DECL_ARGUMENTS.
11823         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11824         old_tree in the map.
11825         (create_specialized_node): Update.
11826         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11827         into index.
11828         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11829         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11830         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11831         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11832         DECL_RESULT.
11834 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11836         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11837         addsi3_carryin_alt2_<optab>): Correct output template.
11839 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11841         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11842         Adjust for arm_restrict_it.
11843         Remove trailing whitespace.
11845 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11847         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11848         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11850         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11852 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11854         PR rtl-optimization/57960
11855         * lra-constraints.c (process_alt_operands): Use the right mode
11856         when checking strict_low.
11858 2013-07-25  Jan Hubicka  <jh@suse.cz>
11860         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11861         * cgraph.c (cgraph_remove_node): Do not release function body
11862         when in cgraph streaming.
11863         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11864         in other partitions are not considered reachable; fix handling of
11865         clones.
11867 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11869         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11871 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11873         PR target/19599
11874         PR target/57731
11875         PR target/57837
11876         * config/arm/arm.md ("*sibcall_insn): Replace use of
11877         Ss with US. Adjust output for v5 and v4t.
11878         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11880         * config/arm/constraints.md ("Ss"): Rename to US.
11882 2013-07-25  Terry Guo  <terry.guo@arm.com>
11884         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11885         shift_add/shift_sub0/shift_sub1 RTXs.
11887 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11888             Anton Blanchard  <anton@au1.ibm.com>
11890         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11891         (altivec_vpks<VI_char>ss): Likewise.
11892         (altivec_vpks<VI_char>us): Likewise.
11893         (altivec_vpku<VI_char>us): Likewise.
11894         (altivec_vpku<VI_char>um): Likewise.
11896 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11898         Introduce context class.
11900         * Makefile.in (CONTEXT_H): New.
11901         (OBJS): Add context.o.
11902         (toplev.o): Add CONTEXT_H to dependencies.
11903         (context.o): New.
11905         * toplev.c (general_init): Create the singleton gcc::context instance.
11907         * context.c: New.
11909         * context.h: New.
11911 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11913         PR rtl-optimization/57968
11914         * mode-switching.c (create_pre_exit): Allow instructions that
11915         don't set a return register to need a non-exit mode.
11917 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11918             Anton Blanchard  <anton@au1.ibm.com>
11920         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11921         operands to vperm for little endian.
11922         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11923         of lvsl to create the control mask for a vperm for little endian.
11925 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11926             Anton Blanchard  <anton@au1.ibm.com>
11928         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11929         two operands for little-endian.
11931 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11933         * config/mips/mips.c (mips_case_values_threshold): New.
11934         (TARGET_CASE_VALUES_THRESHOLD): Define.
11936 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11937             Anton Blanchard  <anton@au1.ibm.com>
11939         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11940         selection of field for vector splat in little endian mode.
11942 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11944         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11945         expanders to rs6000.md.
11946         (ior<mode>3): Likewise.
11947         (and<mode>3): Likewise.
11948         (one_cmpl<mode>2): Likewise.
11949         (nor<mode>3): Likewise.
11950         (andc<mode>3): Likewise.
11951         (eqv<mode>3): Likewise.
11952         (nand<mode>3): Likewise.
11953         (orc<mode>3): Likewise.
11955         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11956         declaration.
11958         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11959         to split multi-word logical operations.
11960         (rs6000_split_logical_di): Likewise.
11961         (rs6000_split_logical): Likewise.
11963         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11964         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11965         and allow TImode operations in 32-bit.
11966         (vsx_and<mode>3_64bit): Likewise.
11967         (vsx_ior<mode>3_32bit): Likewise.
11968         (vsx_ior<mode>3_64bit): Likewise.
11969         (vsx_xor<mode>3_32bit): Likewise.
11970         (vsx_xor<mode>3_64bit): Likewise.
11971         (vsx_one_cmpl<mode>2_32bit): Likewise.
11972         (vsx_one_cmpl<mode>2_64bit): Likewise.
11973         (vsx_nor<mode>3_32bit): Likewise.
11974         (vsx_nor<mode>3_64bit): Likewise.
11975         (vsx_andc<mode>3_32bit): Likewise.
11976         (vsx_andc<mode>3_64bit): Likewise.
11977         (vsx_eqv<mode>3_32bit): Likewise.
11978         (vsx_eqv<mode>3_64bit): Likewise.
11979         (vsx_nand<mode>3_32bit): Likewise.
11980         (vsx_nand<mode>3_64bit): Likewise.
11981         (vsx_orc<mode>3_32bit): Likewise.
11982         (vsx_orc<mode>3_64bit): Likewise.
11984         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11985         logical types in GPRs.
11987         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11988         logical insns to rs6000.md, and allow TImode operations in
11989         32-bit.
11990         (altivec_ior<mode>3): Likewise.
11991         (altivec_xor<mode>3): Likewise.
11992         (altivec_one_cmpl<mode>2): Likewise.
11993         (altivec_nor<mode>3): Likewise.
11994         (altivec_andc<mode>3): Likewise.
11996         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11997         attributes for moving the 128-bit logical operations into
11998         rs6000.md.
11999         (BOOL_REGS_OUTPUT): Likewise.
12000         (BOOL_REGS_OP1): Likewise.
12001         (BOOL_REGS_OP2): Likewise.
12002         (BOOL_REGS_UNARY): Likewise.
12003         (BOOL_REGS_AND_CR0): Likewise.
12004         (one_cmpl<mode>2): Add support for DI logical operations on
12005         32-bit, splitting the operations to 32-bit.
12006         (anddi3): Likewise.
12007         (iordi3): Likewise.
12008         (xordi3): Likewise.
12009         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12010         changes to combine the 32/64-bit code, allow logical operations on
12011         TI mode in 32-bit, and to use similar match_operator patterns like
12012         scalar mode uses.  Combine the Altivec and VSX code for logical
12013         operations, and move it here.
12014         (ior<mode>3, 128-bit types): Likewise.
12015         (xor<mode>3, 128-bit types): Likewise.
12016         (one_cmpl<mode>3, 128-bit types): Likewise.
12017         (nor<mode>3, 128-bit types): Likewise.
12018         (andc<mode>3, 128-bit types): Likewise.
12019         (eqv<mode>3, 128-bit types): Likewise.
12020         (nand<mode>3, 128-bit types): Likewise.
12021         (orc<mode>3, 128-bit types): Likewise.
12022         (and<mode>3_internal): Likewise.
12023         (bool<mode>3_internal): Likewise.
12024         (boolc<mode>3_internal1): Likewise.
12025         (boolc<mode>3_internal2): Likewise.
12026         (boolcc<mode>3_internal1): Likewise.
12027         (boolcc<mode>3_internal2): Likewise.
12028         (eqv<mode>3_internal1): Likewise.
12029         (eqv<mode>3_internal2): Likewise.
12030         (one_cmpl1<mode>3_internal): Likewise.
12032 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12034         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12035         Rename flag_stack_usage to flag_stack_usage_info.
12037 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12039         * config/microblaze/sync.md: New file.
12040         * config/microblaze/microblaze.md: Include sync.md
12041         * config/microblaze/microblaze.c: Add print_operand 'y'.
12042         * config/microblaze/constraints.md: Add memory_contraint
12043         'Q' which is a single register.
12045 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12047         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12049 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12051         * reload.c (find_reloads): Exit loop once we find this operand
12052         cannot be reloaded somehow for this alternative.
12054         * reload.c (find_reloads): Exit loop once we find a hard register.
12056         * rtlanal.c (computed_jump_p): Exit loop once we find label
12057         reference is used.
12059         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12061         * cfgloopmanip.c (remove_path): Exit loop after setting
12062         irred_invalidated.
12064         * gensupport.c (subst_dup): Avoid loop if code is not
12065         MATCH_DUP nor MATCH_OP_DUP.
12067 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12069         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12071 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12073         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12074         true for SP_REGNUM if mode == ptr_mode.
12075         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12076         with value R0_REGNUM + 31.
12078 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12080         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12081         pad pointer-typed argument downward.
12083 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12085         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12086         and __ILP32__ when the ILP32 model is in use.
12088 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12090         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12091         (aarch64_load_symref_appropriately): In the case of
12092         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12093         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12094         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12095         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12096         if 'mode' doesn't equal to 'ptr_mode'.
12097         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12098         'vcall_offset'; change to call aarch64_emit_move differently depending
12099         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12100         to calculate the upper bound of 'vcall_offset'.
12101         (aarch64_cannot_force_const_mem): Change to also return true if
12102         mode != ptr_mode.
12103         (aarch64_legitimize_reload_address): In the case of large
12104         displacements, add new local variable 'xmode' and an assertion
12105         based on it; change to use 'xmode' to generate the new rtx and
12106         reload.
12107         (aarch64_asm_trampoline_template): Change to generate the template
12108         differently depending on TARGET_ILP32 or not; change to use
12109         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12110         (aarch64_trampoline_size): Removed.
12111         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12112         and replace immediate literals with it.  Change to use 'ptr_mode'
12113         instead of 'DImode' and call convert_memory_address if the mode
12114         of 'fnaddr' doesn't equal to 'ptr_mode'.
12115         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12116         to output symbol.
12117         (aarch64_elf_asm_destructor): Likewise.
12118         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12119         on TARGET_ILP32 instead of aarch64_trampoline_size.
12120         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12121         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12122         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12123         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12124         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12125         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12126         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12127         depending on the value of 'mode'.
12128         (add_losym_<mode>): New.
12129         (ldr_got_small_<mode>): New, based on ldr_got_small.
12130         (ldr_got_small): Remove.
12131         (ldr_got_small_sidi): New.
12132         * config/aarch64/iterators.md (P): New.
12133         (PTR): Change to 'ptr_mode' in the condition.
12135 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12137         * config.gcc (aarch64*-*-*): Support --with-abi.
12138         (aarch64*-*-elf): Support --with-multilib-list.
12139         (aarch64*-*-linux*): Likewise.
12140         (supported_defaults): Add abi to aarch64*-*-*.
12141         * configure.ac: Mention AArch64 for --with-multilib-list.
12142         * configure: Re-generated.
12143         * config/aarch64/biarchilp32.h: New file.
12144         * config/aarch64/biarchlp64.h: New file.
12145         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12146         (ABI_SPEC): Ditto.
12147         (MULTILIB_DEFAULTS): Ditto.
12148         (DRIVER_SELF_SPECS): Ditto.
12149         (ASM_SPEC): Update to also substitute -mabi.
12150         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12151         file whose name depends on -mabi= and -mbig-endian.
12152         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12153         TARGET_ILP32.
12154         (POINTER_SIZE): New define.
12155         (POINTERS_EXTEND_UNSIGNED): Ditto.
12156         (enum aarch64_abi_type): New enumeration tag.
12157         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12158         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12159         (TARGET_ILP32): New define.
12160         * config/aarch64/aarch64.opt (mabi): New.
12161         (aarch64_abi): New.
12162         (ilp32, lp64): New values for -mabi.
12163         * config/aarch64/t-aarch64 (comma): New define.
12164         (MULTILIB_OPTIONS): Ditto.
12165         (MULTILIB_DIRNAMES): Ditto.
12166         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12167         * doc/invoke.texi: Document -mabi for AArch64.
12169 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12171         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12173 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12174             Anton Blanchard  <anton@au1.ibm.com>
12176         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12177         endianness when selecting field to splat.
12179 2013-07-22  Eric Christopher  <echristo@gmail.com>
12181         * dwarf2out.c (die_odr_checksum): New function to use
12182         CHECKSUM_ macros and ULEB128 for DIE tag.
12183         (generate_type_signature): Use.
12185 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12187         * config.gcc (sparc*-*-*): Accept leon3 processor.
12188         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12189         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12190         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12191         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12192         (mfix-ut699): Adjust comment.
12193         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12194         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12195         (CPP_CPU_SPEC): Likewise.
12196         (ASM_CPU_SPEC): Likewise.
12197         * config/sparc/sparc.c (leon3_cost): New constant.
12198         (sparc_option_override): Add leon3 support.
12199         (mem_ref): New function.
12200         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12201         (sparc_do_work_around_errata): Look into the instruction in the delay
12202         slot and adjust accordingly.  Add fix for the data cache nullify issues
12203         of the UT699.  Change insertion position for the NOP.
12204         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12205         (leon3_load): New reservation.
12206         (leon_store): Bump latency to 2.
12207         (grfpu): New automaton.
12208         (grfpu_alu): New unit.
12209         (grfpu_ds): Likewise.
12210         (leon_fp_alu): Adjust.
12211         (leon_fp_mult): Delete.
12212         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12213         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12214         * config/sparc/sparc.md (cpu): Add leon3.
12215         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12216         (swapsi): Likewise.
12217         (atomic_test_and_set): Likewise.
12218         (ldstub): Likewise.
12220 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12222         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12223         default for R5900 targets.
12224         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12225         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12226         * config/mips/mips.c (mips_option_override): Report an error for
12227         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12228         for -march=r5900 -mhard-float.
12230 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12232         * df-problems.c (can_move_insns_across): Exit loop once we
12233         find a non-fixed, non-global register.
12235         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12236         setting can_throw.
12238         * omega.c (omega_eliminate_red): Break after setting red_found.
12239         (omega_problem_has_red_equations): Similarly after setting found.
12240         (omega_query_variable): Similarly after setting coupled.
12242 2013-07-22  Marek Polacek  <polacek@redhat.com>
12244         * gimplify.c: Don't include gimple.h twice.
12246 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12248         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12249         instead of TARGET_THUMB1.
12250         (Pz): New constraint.
12251         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12252         encodings.
12253         (compare_negsi_si): Likewise.
12254         (compare_addsi2_op0): Likewise.
12255         (compare_addsi2_op1): Likewise.
12256         (addsi3_carryin_<optab>): Likewise.
12257         (addsi3_carryin_alt2_<optab>): Likewise.
12258         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12259         for arm_restrict_it.
12260         (subsi3_carryin): Likewise.
12261         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12262         (minmax_arithsi): Disable for arm_restrict_it.
12263         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12264         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12265         (satsi_<SAT:code>_shift): Likewise.
12266         (arm_shiftsi3): Add alternative for 16-bit encoding.
12267         (arm32_movhf): Disable for arm_restrict_it.
12268         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12269         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12271 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12273         * config/arm/arm.md (attribute "insn"): Delete.
12274         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12275         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12276         (not_shiftsi): Update for attribute change.
12277         (not_shiftsi_compare0): Likewise.
12278         (not_shiftsi_compare0_scratch): Likewise.
12279         (arm_one_cmplsi2): Likewise.
12280         (thumb1_one_cmplsi2): Likewise.
12281         (notsi_compare0): Likewise.
12282         (notsi_compare0_scratch): Likewise.
12283         (thumb1_movdi_insn): Likewise.
12284         (arm_movsi_insn): Likewise.
12285         (movhi_insn_arch4): Likewise.
12286         (movhi_bytes): Likewise.
12287         (arm_movqi_insn): Likewise.
12288         (thumb1_movqi_insn): Likewise.
12289         (arm32_movhf): Likewise.
12290         (thumb1_movhf): Likewise.
12291         (arm_movsf_soft_insn): Likewise.
12292         (thumb1_movsf_insn): Likewise.
12293         (thumb_movdf_insn): Likewise.
12294         (movsicc_insn): Likewise.
12295         (movsfcc_soft_insn): Likewise.
12296         (and_scc): Likewise.
12297         (cond_move): Likewise.
12298         (if_move_not): Likewise.
12299         (if_not_move): Likewise.
12300         (if_shift_move): Likewise.
12301         (if_move_shift): Likewise.
12302         (if_shift_shift): Likewise.
12303         (if_not_arith): Likewise.
12304         (if_arith_not): Likewise.
12305         (cond_move_not): Likewise.
12306         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12307         (neon_mov<mode>): Likewise.
12308         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12309         (thumb2_movsi_vfp): Likewise.
12310         (movsf_vfp): Likewise.
12311         (thumb2_movsf_vfp): Likewise.
12312         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12313         change.
12314         (cortexa7_older_only): Likewise.
12315         (cortexa7_younger): Likewise.
12316         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12317         (1020alu_shift_op): Likewise.
12318         (1020alu_shift_reg_op): Likewise.
12319         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12320         (alu_shift_op): Likewise.
12321         (alu_shift_reg_op): Likewise.
12322         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12323         (11_alu_shift_op): Likewise.
12324         (11_alu_shift_reg_op): Likewise.
12325         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12326         (9_alu_shift_reg_op): Likewise.
12327         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12328         change.
12329         (cortex_a15_alu_shift): Likewise.
12330         (cortex_a15_alu_shift_reg): Likewise.
12331         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12332         (cortex_a5_alu_shift): Likewise.
12333         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12334         change.
12335         (cortex_a53_alu_shift): Likewise.
12336         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12337         change.
12338         (cortex_a7_alu_reg): Likewise.
12339         (cortex_a7_alu_shift): Likewise.
12340         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12341         (cortex_a8_alu_shift): Likewise.
12342         (cortex_a8_alu_shift_reg): Likewise.
12343         (cortex_a8_mov): Likewise.
12344         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12345         (cortex_a9_dp_shift): Likewise.
12346         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12347         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12348         (cortex_r4_mov): Likewise.
12349         (cortex_r4_alu_shift): Likewise.
12350         (cortex_r4_alu_shift_reg): Likewise.
12351         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12352         (526_alu_shift_op): Likewise.
12353         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12354         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12355         (626te_alu_shift_op): Likewise.
12356         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12357         (726te_alu_op): Likewise.
12358         (726te_alu_shift_op): Likewise.
12359         (726te_alu_shift_reg_op): Likewise.
12360         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12361         (mp626_alu_shift_op): Likewise.
12362         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12363         (pj4_alu_e1_conds): Likewise.
12364         (pj4_alu): Likewise.
12365         (pj4_alu_conds): Likewise.
12366         (pj4_shift): Likewise.
12367         (pj4_shift_conds): Likewise.
12368         (pj4_alu_shift): Likewise.
12369         (pj4_alu_shift_conds): Likewise.
12371 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12373         * config/arm/predicates.md (shiftable_operator_strict_it):
12374         New predicate.
12375         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12376         Disable cond_exec version for arm_restrict_it.
12377         (thumb2_smaxsi3): Convert to generate cond_exec.
12378         (thumb2_sminsi3): Likewise.
12379         (thumb32_umaxsi3): Likewise.
12380         (thumb2_uminsi3): Likewise.
12381         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12382         (thumb2_neg_abssi2): Likewise.
12383         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12384         (thumb2_movsicc_insn): Adjust alternatives.
12385         (thumb2_mov_negscc): Disable for arm_restrict_it.
12386         (thumb2_mov_negscc_strict_it): New pattern.
12387         (thumb2_mov_notscc_strict_it): New pattern.
12388         (thumb2_mov_notscc): Disable for arm_restrict_it.
12389         (thumb2_ior_scc): Likewise.
12390         (thumb2_ior_scc_strict_it): New pattern.
12391         (thumb2_cond_move): Adjust for arm_restrict_it.
12392         (thumb2_cond_arith): Disable for arm_restrict_it.
12393         (thumb2_cond_arith_strict_it): New pattern.
12394         (thumb2_cond_sub): Adjust for arm_restrict_it.
12395         (thumb2_movcond): Likewise.
12396         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12397         (thumb2_zero_extendhisi2_v6): Likewise.
12398         (thumb2_zero_extendqisi2_v6): Likewise.
12399         (orsi_notsi_si): Likewise.
12400         (orsi_not_shiftsi_si): Likewise.
12402 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12404         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12405         instruction sequence is 1 byte shorter.
12407 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12409         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12410         it is not needed after split.
12412 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12414         PR target/51784
12415         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12416         second label for nonlocal goto receivers. Don't output pic base labels
12417         unless we're producing PIC; mark that action unreachable().
12418         (ix86_save_reg): If the function contains a nonlocal label, save the
12419         PIC base reg.
12420         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12421         * config/darwin.c (emitted_pic_label_num): New GTY.
12422         (update_pic_label_number_if_needed): New.
12423         (machopic_output_function_base_name): Adjust for nonlocal receiver
12424         case.
12425         (machopic_should_output_picbase_label): New.
12426         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12427         (nonlocal_goto_receiver): New insn and split.
12429 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12431         * config/aarch64/aarch64-builtins.c
12432         (aarch64_fold_builtin): Fold abs in all modes.
12433         * config/aarch64/aarch64-simd-builtins.def
12434         (abs): Enable for all modes.
12435         * config/aarch64/arm_neon.h
12436         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12437         (vabs_f64): Add missing intrinsic.
12439 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12441         * config/aarch64/arm_neon.h (vabs_s64): New function
12443 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12445         PR target/57516
12446         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12447         * config/avr/avr.md (adjust_len): Add `round'.
12448         * config/avr/avr-protos.h (avr_out_round): New prototype.
12449         (avr_out_plus): Add `out_label' argument.
12450         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12451         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12452         Handle the case where `insn' is just a pattern.
12453         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12454         (avr_out_round): New function.
12455         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12457 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12459         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12460         Add check for flag_stack_usage to handle -fstack-usage support
12462 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12464         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12465         interaction for new Power8 flags and VSX.
12467 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12469         PR middle-end/57698
12470         * tree-inline.c (expand_call_inline): Emit errors during
12471         early_inlining only if optimization is not turned on.
12473 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12475         * passes.def: New.
12477         * passes.c (init_optimization_passes): Move the construction of
12478         the pass hierarchy into a new passes.def file.
12480         * Makefile.in (passes.o): Add dependency on passes.def.
12482 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12484         * passes.c (init_optimization_passes): Introduce macros for
12485         constructing the tree of passes (INSERT_PASSES_AFTER,
12486         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12487         TERMINATE_PASS_LIST).
12489 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12490             Wei Mi  <wmi@google.com>
12492         PR rtl-optimization/57878
12493         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12494         top.
12495         (reload_pseudo_compare_func): Check nregs first for reload
12496         pseudos.
12498 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12500         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12502 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12504         * read-rtl.c (validate_const_int): Once an invalid character is
12505         seen, quit the loop.
12507         * gengtype.c (write_roots): Similarly once we find the "deletable"
12508         or "if_marked" option.
12510 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12512         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12513         "xtab" and "sat".  Move value "clz" from here to ...
12514         (attriubte "type"): ... here.
12515         (satsi_<SAT:code>): Delete "insn" attribute.
12516         (satsi_<SAT:code>_shift): Likewise.
12517         (arm_zero_extendqisi2addsi): Likewise.
12518         (arm_extendqisi2addsi): Likewise.
12519         (clzsi2): Update for attribute changes.
12520         (rbitsi2): Likewise.
12521         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12522         attribute.
12523         (arm_usatsihi): Likewise.
12524         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12526 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12528         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12529         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12530         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12531         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12532         in alphabetical order.
12533         (attribute "core_cycles"): Update for attribute changes.
12534         (arm_addsi3): Likewise.
12535         (addsi3_compare0): Likewise.
12536         (addsi3_compare0_scratch): Likewise.
12537         (addsi3_compare_op1): Likewise.
12538         (addsi3_compare_op2): Likewise.
12539         (compare_addsi2_op0): Likewise.
12540         (compare_addsi2_op1): Likewise.
12541         (addsi3_carryin_shift_<optab>): Likewise.
12542         (subsi3_carryin_shift): Likewise.
12543         (rsbsi3_carryin_shift): Likewise.
12544         (arm_subsi3_insn): Likewise.
12545         (subsi3_compare0): Likewise.
12546         (subsi3_compare): Likewise.
12547         (arm_andsi3_insn): Likewise.
12548         (thumb1_andsi3_insn): Likewise.
12549         (andsi3_compare0): Likewise.
12550         (andsi3_compare0_scratch): Likewise.
12551         (zeroextractsi_compare0_scratch
12552         (andsi_not_shiftsi_si): Likewise.
12553         (iorsi3_insn): Likewise.
12554         (iorsi3_compare0): Likewise.
12555         (iorsi3_compare0_scratch): Likewise.
12556         (arm_xorsi3): Likewise.
12557         (thumb1_xorsi3_insn): Likewise.
12558         (xorsi3_compare0): Likewise.
12559         (xorsi3_compare0_scratch): Likewise.
12560         (satsi_<SAT:code>_shift): Likewise.
12561         (rrx): Likewise.
12562         (arm_shiftsi3): Likewise.
12563         (shiftsi3_compare0): Likewise.
12564         (not_shiftsi): Likewise.
12565         (not_shiftsi_compare0): Likewise.
12566         (not_shiftsi_compare0_scratch): Likewise.
12567         (arm_one_cmplsi2): Likewise.
12568         (thumb_one_complsi2): Likewise.
12569         (notsi_compare0): Likewise.
12570         (notsi_compare0_scratch): Likewise.
12571         (thumb1_zero_extendhisi2): Likewise.
12572         (arm_zero_extendhisi2): Likewise.
12573         (arm_zero_extendhisi2_v6): Likewise.
12574         (arm_zero_extendhisi2addsi): Likewise.
12575         (thumb1_zero_extendqisi2): Likewise.
12576         (thumb1_zero_extendqisi2_v6): Likewise.
12577         (arm_zero_extendqisi2): Likewise.
12578         (arm_zero_extendqisi2_v6): Likewise.
12579         (arm_zero_extendqisi2addsi): Likewise.
12580         (thumb1_extendhisi2): Likewise.
12581         (arm_extendhisi2): Likewise.
12582         (arm_extendhisi2_v6): Likewise.
12583         (arm_extendqisi): Likewise.
12584         (arm_extendqisi_v6): Likewise.
12585         (arm_extendqisi2addsi): Likewise.
12586         (thumb1_extendqisi2): Likewise.
12587         (thumb1_movdi_insn): Likewise.
12588         (arm_movsi_insn): Likewise.
12589         (movsi_compare0): Likewise.
12590         (movhi_insn_arch4): Likewise.
12591         (movhi_bytes): Likewise.
12592         (arm_movqi_insn): Likewise.
12593         (thumb1_movqi_insn): Likewise.
12594         (arm32_movhf): Likewise.
12595         (thumb1_movhf): Likewise.
12596         (arm_movsf_soft_insn): Likewise.
12597         (thumb1_movsf_insn): Likewise.
12598         (movdf_soft_insn): Likewise.
12599         (thumb_movdf_insn): Likewise.
12600         (arm_cmpsi_insn): Likewise.
12601         (cmpsi_shiftsi): Likewise.
12602         (cmpsi_shiftsi_swp): Likewise.
12603         (arm_cmpsi_negshiftsi_si): Likewise.
12604         (movsicc_insn): Likewise.
12605         (movsfcc_soft_insn): Likewise.
12606         (arith_shiftsi): Likewise.
12607         (arith_shiftsi_compare0
12608         (arith_shiftsi_compare0_scratch
12609         (sub_shiftsi): Likewise.
12610         (sub_shiftsi_compare0
12611         (sub_shiftsi_compare0_scratch
12612         (and_scc): Likewise.
12613         (cond_move): Likewise.
12614         (if_plus_move): Likewise.
12615         (if_move_plus): Likewise.
12616         (if_move_not): Likewise.
12617         (if_not_move): Likewise.
12618         (if_shift_move): Likewise.
12619         (if_move_shift): Likewise.
12620         (if_shift_shift): Likewise.
12621         (if_not_arith): Likewise.
12622         (if_arith_not): Likewise.
12623         (cond_move_not): Likewise.
12624         (thumb1_ashlsi3): Set type attribute.
12625         (thumb1_ashrsi3): Likewise.
12626         (thumb1_lshrsi3): Likewise.
12627         (thumb1_rotrsi3): Likewise.
12628         (shiftsi3_compare0_scratch): Likewise.
12629         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12630         (neon_mov<mode>): Likewise.
12631         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12632         attribute changes.
12633         (thumb2_movsi_insn): Likewise.
12634         (thumb2_cmpsi_neg_shiftsi): Likewise.
12635         (thumb2_extendqisi_v6): Likewise.
12636         (thumb2_zero_extendhisi2_v6): Likewise.
12637         (thumb2_zero_extendqisi2_v6): Likewise.
12638         (thumb2_shiftsi3_short): Likewise.
12639         (thumb2_addsi3_compare0_scratch): Likewise.
12640         (orsi_not_shiftsi_si): Likewise.
12641         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12642         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12643         changes.
12644         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12645         (1020alu_shift_op): Likewise.
12646         (1020alu_shift_reg_op): Likewise.
12647         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12648         (alu_shift_op): Likewise.
12649         (alu_shift_reg_op): Likewise.
12650         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12651         (11_alu_shift_op): Likewise.
12652         (11_alu_shift_reg_op): Likewise.
12653         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12654         (9_alu_shift_reg_op): Likewise.
12655         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12656         changes.
12657         (cortex_a15_alu_shift): Likewise.
12658         (cortex_a15_alu_shift_reg): Likewise.
12659         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12660         changes.
12661         (cortex_a5_alu_shift): Likewise.
12662         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12663         changes.
12664         (cortex_a53_alu_shift): Likewise.
12665         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12666         changes.
12667         (cortex_a7_alu_reg): Likewise.
12668         (cortex_a7_alu_shift): Likewise.
12669         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12670         changes.
12671         (cortex_a8_alu_shift): Likewise.
12672         (cortex_a8_alu_shift_reg): Likewise.
12673         (cortex_a8_mov): Likewise.
12674         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12675         (cortex_a9_dp_shift): Likewise.
12676         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12677         changes.
12678         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12679         changes.
12680         (cortex_r4_mov): Likewise.
12681         (cortex_r4_alu_shift): Likewise.
12682         (cortex_r4_alu_shift_reg): Likewise.
12683         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12684         (526_alu_shift_op): Likewise.
12685         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12686         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12687         (626te_alu_shift_op): Likewise.
12688         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12689         (726te_alu_op): Likewise.
12690         (726te_alu_shift_op): Likewise.
12691         (726te_alu_shift_reg_op): Likewise.
12692         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12693         (mp626_alu_shift_op): Likewise.
12694         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12695         (pj4_alu_e1_conds): Likewise.
12696         (pj4_alu): Likewise.
12697         (pj4_alu_conds): Likewise.
12698         (pj4_shift): Likewise.
12699         (pj4_shift_conds): Likewise.
12700         (pj4_alu_shift): Likewise.
12701         (pj4_alu_shift_conds): Likewise.
12702         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12703         changes.
12704         (cortexa7_older_only): Likewise.
12705         (cortexa7_younger): Likewise.
12707 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12709         * ipa-pure-const.c (generate_summary): Rename to...
12710         (pure_const_generate_summary): ... this.
12712 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12714         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12716 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12718         PR target/57909
12719         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12720         usage in HI mode.
12722 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12724         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12725         enabled without -march=zEC12.
12726         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12727         flags to be set.
12729 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12731         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12732         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12733         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12734         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12735         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12736         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12737         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12738         ISA_HAS_FP4.
12739         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12740         and ISA_HAS_NMADD3_NMSUB3.
12741         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12742         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12743         (nmsub4<mode>, nmsub3<mode>): Likewise.
12744         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12746 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12748         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12749         TARGET_MIPS5400 checking.
12751 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12752             Peter Bergner  <bergner@vnet.ibm.com>
12754         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12755         registers in the comment.
12756         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12757         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12758         rather than FIRST_PSEUDO_REGISTERS.
12760 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12762         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12763         enable extra ISA flags with TARGET_HTM.
12765 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12767         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12768         Fix comment typos.
12770 2013-07-15  Cong Hou  <congh@google.com>
12772         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12773         in compare function for sorting.
12775 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12777         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12778         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12779         * config/rs6000/rs6000.opt: Add -mhtm option.
12780         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12781         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12782         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12783         __HTM__ if the HTM instructions are available.
12784         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12785         htm_spr_reg_operand): New define_predicates.
12786         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12787         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12788         Include htm.md.
12789         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12790         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12791         HTM builtin functions.
12792         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12793         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12794         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12795         (rs6000_builtin_mask_calculate): Likewise.
12796         (rs6000_option_override_internal): Likewise.
12797         (bdesc_htm): Add new HTM builtin support.
12798         (htm_spr_num): New function.
12799         (htm_spr_regno): Likewise.
12800         (rs6000_htm_spr_icode): Likewise.
12801         (htm_expand_builtin): Likewise.
12802         (htm_init_builtins): Likewise.
12803         (rs6000_expand_builtin): Add support for HTM builtin functions.
12804         (rs6000_init_builtins): Likewise.
12805         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12806         option.
12807         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12808         (TARGET_HTM, MASK_HTM): Define macros.
12809         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12810         (FIXED_REGISTERS): Likewise.
12811         (CALL_USED_REGISTERS): Likewise.
12812         (CALL_REALLY_USED_REGISTERS): Likewise.
12813         (REG_ALLOC_ORDER): Likewise.
12814         (enum reg_class): Likewise.
12815         (REG_CLASS_NAMES): Likewise.
12816         (REG_CLASS_CONTENTS): Likewise.
12817         (REGISTER_NAMES): Likewise.
12818         (ADDITIONAL_REGISTER_NAMES): Likewise.
12819         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12820         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12821         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12822         * config/rs6000/htm.md: New file.
12823         * config/rs6000/htmintrin.h: New file.
12824         * config/rs6000/htmxlintrin.h: New file.
12826 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12828         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12829         Define SYMBOL_TINY_GOT, update comment.
12830         * config/aarch64/aarch64.c
12831         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12832         (aarch64_expand_mov_immediate): Likewise.
12833         (aarch64_print_operand): Likewise.
12834         (aarch64_classify_symbol): Likewise.
12835         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12836         (ldr_got_tiny): Define.
12838 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12840         PR tree-optimization/54094
12841         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12842           scheduling dimension for the parallelism check from the polyhedral
12843           information in the AST.
12844         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12845           in 2D + 1 form.
12847 2013-07-13  Jason Merrill  <jason@redhat.com>
12849         * print-tree.c (debug_vec_tree): Use debug_raw.
12850         (debug_raw (vec<tree, va_gc> &)): New.
12851         (debug_raw (vec<tree, va_gc> *)): New.
12852         * tree.h: Declare them.
12854 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12856         * ifcvt.c (ifcvt_after_combine): New static variable.
12857         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12858         for size.
12859         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12860         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12861         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12863 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12865         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12867 2013-07-12  Michael Matz  <matz@suse.de>
12869         PR middle-end/55771
12870         * convert.c (convert_to_real): Reject non-float inner types.
12872 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12874         * config/aarch64/aarch64-protos.h
12875         (aarch64_simd_immediate_valid_for_move): Remove.
12876         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12877         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12878         cases.
12879         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12881 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12883         * config/mips/mips.c (mips_conditional_register_usage): Do not
12884         use t[0-7] registers in MIPS16 mode when optimizing for size.
12886 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12888         * config/i386/i386.c (dispatch_function_versions): Fix array
12889         indexing of function_version_info to match actual_versions.
12891 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12893         * vec.h (struct va_gc): Move release out-of-line.
12894         (va_gc::release): Call ggc_free on released vec.
12896 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12898         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12899         Require GOT register as additional operand in UNSPEC.
12900         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12901         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12902         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12903         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12904         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12905         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12906         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12908 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12910         PR target/57631
12911         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12912         name seen by assembler/linker rather if available.
12914 2013-07-11  Andreas Schwab  <schwab@suse.de>
12916         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12918 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12920         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12922 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12924         * doc/tm.texi.in: Move hook documentation to ....
12925         * target.def: ... here.
12927         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12928         text on @hook line.
12929         * doc/tm.texi: Regenerate.
12931 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12933         PR c++/57869
12934         * doc/invoke.texi: Document -Wconditionally-supported.
12936 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12938         PR target/57844
12939         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12940         of my_fp.
12942 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12944         PR target/57506
12945         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12946         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12947         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12948         Remove duplicate devices.
12949         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12950         * config/avr/t-multilib: Regenerate.
12951         * config/avr/avr-tables.opt: Regenerate.
12952         * doc/avr-mmcu.texi: Regenerate.
12954 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12956         PR target/56987
12957         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12959 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12961         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12962         the cost of MULT when optimizing for size.
12964 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12966         * config/cr16/cr16-protos.h: Don't include target.h.
12968 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12970         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12971         adjust register size for TDmode and TFmode for VSX registers.
12973 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12975         PR target/56892
12976         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12977         hook_bool_const_tree_true.
12979 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12981         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12982         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12983         * config/s390/s390.md: Define FPR*_REGNUM constants.
12984         Fix FPR2_REGNUM constant (18 -> 17).
12985         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12986         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12987         ("trunc<BFP:mode><DFP_ALL:mode>2")
12988         ("trunc<DFP_ALL:mode><BFP:mode>2")
12989         ("*extend<BFP:mode><DFP_ALL:mode>2")
12990         ("*extend<DFP_ALL:mode><BFP:mode>2")
12991         ("extend<BFP:mode><DFP_ALL:mode>2")
12992         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12993         FPR4_REGNUM.
12995 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12997         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12999 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13001         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13002         and cfun_fpr_bit_p to cfun_fpr_save_p.
13003         (s390_frame_area, s390_register_info, s390_frame_info)
13004         (s390_emit_prologue, s390_emit_epilogue)
13005         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13006         register numbers.
13007         * config/s390/s390.h: Define *_REGNUM macros for floating point
13008         register numbers.
13010 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13012         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13014 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13016         PR rtl-optimization/57786
13017         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13018         and break out of the loop when it is set to false.
13020 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13022         PR target/57819
13023         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13024         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13025         (const_int 63)) 0)).
13026         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13027         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13028         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13030         PR rtl-optimization/57829
13031         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13032         mask bits outside of mode are just sign-extension from mode to HWI.
13034 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13036         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13037         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13038         adjust_address instead of change_address to keep info about alignment.
13039         (emit_strmov): Remove.
13040         (emit_memmov): New function.
13041         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13042         (expand_movmem_epilogue): Likewise and return updated rtx for
13043         destination.
13044         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13045         destination and source.
13046         (decide_alignment): Refactor, handle vector_loop.
13047         (ix86_expand_movmem): Likewise.
13048         (ix86_expand_setmem): Likewise.
13049         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13051 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13053         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13054         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13056 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13058         * config/i386/sse.md (sse_movlhps): Change alternative 3
13059         of operand 2 to "m".
13061 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13063         PR target/57807
13064         * config/i386/sse.md (iptr): New mode attribute.
13065         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13066         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13067         (<sse>_vmmul<mode>3): Ditto.
13068         (<sse>_vmdiv<mode>3): Ditto.
13069         (sse_vmrcpv4sf2): Ditto.
13070         (<sse>_vmsqrt<mode>2): Ditto.
13071         (sse_vmrsqrtv4sf2): Ditto.
13072         (<sse>_vm<code><mode>3): Ditto.
13073         (avx_vmcmp<mode>3): Ditto.
13074         (<sse>_vmmaskcmp<mode>3): Ditto.
13075         (<sse>_comi): Ditto.
13076         (<sse>_ucomi): Ditto.
13077         (*xop_vmfrcz_<mode>): Ditto.
13078         (*fmai_fmadd_<mode>): Ditto.
13079         (*fmai_fmsub_<mode>): Ditto.
13080         (*fmai_fnmadd_<mode>): Ditto.
13081         (*fmai_fnmsub_<mode>): Ditto.
13082         (*fma4i_vmfmadd_<mode>): Ditto.
13083         (*fma4i_vmfmsub_<mode>): Ditto.
13084         (*fma4i_vmfnmadd_<mode>): Ditto.
13085         (*fma4i_vmfnmsub_<mode>): Ditto.
13086         (*xop_vmfrcz_<mode>): Ditto.
13087         (sse_cvtps2pi): Ditto.
13088         (sse_cvttps2pi): Ditto.
13089         (sse_cvtss2si): Ditto.
13090         (sse_cvtss2si_2): Ditto.
13091         (sse_cvtss2siq_2): Ditto.
13092         (sse_cvttss2si): Ditto.
13093         (sse_cvttss2siq): Ditto.
13094         (sse_cvtsd2si): Ditto.
13095         (sse_cvtsd2si_2): Ditto.
13096         (sse_cvtsd2siq_2): Ditto.
13097         (sse_cvttsd2si): Ditto.
13098         (sse_cvttsd2siq): Ditto.
13099         (sse_cvtsd2ss): Ditto.
13100         (sse_cvtss2sd): Ditto.
13101         (avx2_pbroadcast<mode>): Ditto.
13102         (avx2_pbroadcast<mode>_1): Ditto.
13103         (*avx_vperm_broadcast_v4sf): Ditto.
13105         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13106         (sse_movlhps): Ditto.
13107         (sse_storehps): Ditto.
13108         (sse_loadhps): Ditto.
13109         (sse_storelps): Ditto.
13110         (sse_loadlps): Ditto.
13111         (*vec_concatv4sf): Ditto.
13112         (*vec_interleave_highv2df): Ditto.
13113         (*vec_interleave_lowv2df): Ditto.
13114         (*vec_extractv2df_1_sse): Ditto.
13115         (*vec_extractv2df_0_sse): Ditto.
13116         (sse2_storelpd): Ditto.
13117         (sse2_loadlpd): Ditto.
13118         (sse2_movsd): Ditto.
13119         (*vec_concatv4si): Ditto.
13120         (vec_concatv2di): Ditto.
13122         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13123         for Intel asm dialect.
13124         (mmx_punpcklwd): Ditto.
13125         (mmx_punpckldq): Ditto.
13127         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13128         for intel assembler dialect.
13130 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13132         PR target/29776
13133         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13134         for BUILT_IN_C{LZ,LRSB}*.
13135         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13136         * tree-vrp.c (extract_range_basic): Handle
13137         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13138         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13139         fall thru to code calling set_value*.
13140         * builtins.c (expand_builtin): Remove *IMAX cases.
13141         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13142         if width is bigger than 2*HWI.
13144 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13146         PR rtl-optimization/55342
13147         * lra-int.h (lra_subreg_reload_pseudos): New.
13148         * lra.c: Add undoing optional reloads to the block diagram.
13149         (lra_subreg_reload_pseudos): New.
13150         (lra_optional_reload_pseudos): Change comments.
13151         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13152         lra_optional_reload_pseudos after undo transformations.
13153         * lra-assigns.c (pseudo_prefix_title): New.
13154         (lra_setup_reg_renumber): Use it.
13155         (spill_for): Ditto.  Check subreg reload pseudos too.
13156         (assign_by_spills): Consider subreg reload pseudos too.
13157         * lra-constraints.c (simplify_operand_subreg): Use
13158         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13159         (curr_insn_transform): Recognize and do optional reloads.
13160         (undo_optional_reloads): New.
13161         (lra_undo_inheritance): Call undo_optional_reloads.
13163 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13165         * tree-complex.c (expand_complex_operations_1): Fix typo.
13167 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13169         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13170         (tune_params): New member 'const vec_costs'.
13171         * config/aarch64/aarch64.c (generic_vector_cost): New.
13172         (generic_tunings): New member 'generic_vector_cost'.
13173         (aarch64_builtin_vectorization_cost): New.
13174         (aarch64_add_stmt_cost): New.
13175         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13176         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13178 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13180         PR target/57777
13181         * config/i386/predicates.md (vsib_address_operand): Disallow
13182         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13184 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13186         PR middle-end/55030
13187         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13188         expand_builtin_setjmp_receiver.
13189         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13190         with NULL for the label parameter.
13191         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13192         the frame-pointer.  Adjust comments.
13193         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13194         only if LABEL is non-NULL.
13196 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13198         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13199         (ARM_ABI_AAPCS64): Ditto.
13200         (arm_abi): Ditto.
13201         (ARM_DEFAULT_ABI): Ditto.
13203 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13205         * config/aarch64/aarch64-builtins.c
13206         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13207         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13208         (st1): Likewise.
13209         * config/aarch64/aarch64-simd.md
13210         (aarch64_ld1<VALL:mode>): New.
13211         (aarch64_st1<VALL:mode>): Likewise.
13212         * config/aarch64/arm_neon.h
13213         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13215 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13217         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13218         target ISA is AVX.
13219         (ix86_option_override_internal):Turn on all -mavx target flags by
13220         default as they are dependent on AVX anyway.
13222 2013-07-02  Cary Coutant  <ccoutant@google.com>
13224         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13225         deterministic hash.
13226         (loc_checksum_ordered): Likewise.
13227         (hash_loc_operands): Remove inline keyword.
13229 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13231         PR tree-optimization/57741
13232         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13233         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13234         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13235         Allow REAL_CST step_exprs if flag_associative_math.
13236         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13238 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13240         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13242 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13244         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13246 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13248         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13249         encoding.
13250         (iorsi3_insn): Likewise.
13251         (arm_xorsi3): Likewise.
13253 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13255         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13256         to ...
13257         (attribute "type"): ... here, and prefix with "wmmx_".
13258         (attribute "core_cycles"): Update for attribute changes.
13259         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13260         (tbcstv4hi): Likewise.
13261         (tbcstv2si): Likewise.
13262         (iwmmxt_iordi3): Likewise.
13263         (iwmmxt_xordi3): Likewise.
13264         (iwmmxt_anddi3): Likewise.
13265         (iwmmxt_nanddi3): Likewise.
13266         (iwmmxt_arm_movdi): Likewise.
13267         (iwmmxt_movsi_insn): Likewise.
13268         (mov<mode>_internal): Likewise.
13269         (and<mode>3_iwmmxt): Likewise.
13270         (ior<mode>3_iwmmxt): Likewise.
13271         (xor<mode>3_iwmmxt): Likewise.
13272         (add<mode>3_iwmmxt): Likewise.
13273         (ssaddv8qi3): Likewise.
13274         (ssaddv4hi3): Likewise.
13275         (ssaddv2si3): Likewise.
13276         (usaddv8qi3): Likewise.
13277         (usaddv4hi3): Likewise.
13278         (usaddv2si3): Likewise.
13279         (sub<mode>3_iwmmxt): Likewise.
13280         (sssubv8qi3): Likewise.
13281         (sssubv4hi3): Likewise.
13282         (sssubv2si3): Likewise.
13283         (ussubv8qi3): Likewise.
13284         (ussubv4hi3): Likewise.
13285         (ussubv2si3): Likewise.
13286         (mulv4hi3_iwmmxt): Likewise.
13287         (smulv4hi3_highpart): Likewise.
13288         (umulv4hi3_highpart): Likewise.
13289         (iwmmxt_wmacs): Likewise.
13290         (iwmmxt_wmacsz): Likewise.
13291         (iwmmxt_wmacu): Likewise.
13292         (iwmmxt_wmacuz): Likewise.
13293         (iwmmxt_clrdi): Likewise.
13294         (iwmmxt_clrv8qi): Likewise.
13295         (iwmmxt_clr4hi): Likewise.
13296         (iwmmxt_clr2si): Likewise.
13297         (iwmmxt_uavgrndv8qi3): Likewise.
13298         (iwmmxt_uavgrndv4hi3): Likewise.
13299         (iwmmxt_uavgv8qi3): Likewise.
13300         (iwmmxt_uavgv4hi3): Likewise.
13301         (iwmmxt_tinsrb): Likewise.
13302         (iwmmxt_tinsrh): Likewise.
13303         (iwmmxt_tinsrw): Likewise.
13304         (iwmmxt_textrmub): Likewise.
13305         (iwmmxt_textrmsb): Likewise.
13306         (iwmmxt_textrmuh): Likewise.
13307         (iwmmxt_textrmsh): Likewise.
13308         (iwmmxt_textrmw): Likewise.
13309         (iwmxxt_wshufh): Likewise.
13310         (eqv8qi3): Likewise.
13311         (eqv4hi3): Likewise.
13312         (eqv2si3): Likewise.
13313         (gtuv8qi3): Likewise.
13314         (gtuv4hi3): Likewise.
13315         (gtuv2si3): Likewise.
13316         (gtv8qi3): Likewise.
13317         (gtv4hi3): Likewise.
13318         (gtv2si3): Likewise.
13319         (smax<mode>3_iwmmxt): Likewise.
13320         (umax<mode>3_iwmmxt): Likewise.
13321         (smin<mode>3_iwmmxt): Likewise.
13322         (umin<mode>3_iwmmxt): Likewise.
13323         (iwmmxt_wpackhss): Likewise.
13324         (iwmmxt_wpackwss): Likewise.
13325         (iwmmxt_wpackdss): Likewise.
13326         (iwmmxt_wpackhus): Likewise.
13327         (iwmmxt_wpackwus): Likewise.
13328         (iwmmxt_wpackdus): Likewise.
13329         (iwmmxt_wunpckihb): Likewise.
13330         (iwmmxt_wunpckihh): Likewise.
13331         (iwmmxt_wunpckihw): Likewise.
13332         (iwmmxt_wunpckilb): Likewise.
13333         (iwmmxt_wunpckilh): Likewise.
13334         (iwmmxt_wunpckilw): Likewise.
13335         (iwmmxt_wunpckehub): Likewise.
13336         (iwmmxt_wunpckehuh): Likewise.
13337         (iwmmxt_wunpckehuw): Likewise.
13338         (iwmmxt_wunpckehsb): Likewise.
13339         (iwmmxt_wunpckehsh): Likewise.
13340         (iwmmxt_wunpckehsw): Likewise.
13341         (iwmmxt_wunpckelub): Likewise.
13342         (iwmmxt_wunpckeluh): Likewise.
13343         (iwmmxt_wunpckeluw): Likewise.
13344         (iwmmxt_wunpckelsb): Likewise.
13345         (iwmmxt_wunpckelsh): Likewise.
13346         (iwmmxt_wunpckelsw): Likewise.
13347         (ror<mode>3): Likewise.
13348         (ashr<mode>3_iwmmxt): Likewise.
13349         (lshr<mode>3_iwmmxt): Likewise.
13350         (ashl<mode>3_iwmmxt): Likewise.
13351         (ror<mode>3_di): Likewise.
13352         (ashr<mode>3_di): Likewise.
13353         (lshr<mode>3_di): Likewise.
13354         (ashl<mode>3_di): Likewise.
13355         (iwmmxt_wmadds): Likewise.
13356         (iwmmxt_wmaddu): Likewise.
13357         (iwmmxt_tmia): Likewise.
13358         (iwmmxt_tmiaph): Likewise.
13359         (iwmmxt_tmiabb): Likewise.
13360         (iwmmxt_tmiatb): Likewise.
13361         (iwmmxt_tmiabt): Likewise.
13362         (iwmmxt_tmiatt): Likewise.
13363         (iwmmxt_tmovmskb): Likewise.
13364         (iwmmxt_tmovmskh): Likewise.
13365         (iwmmxt_tmovmskw): Likewise.
13366         (iwmmxt_waccb): Likewise.
13367         (iwmmxt_wacch): Likewise.
13368         (iwmmxt_waccw): Likewise.
13369         (iwmmxt_waligni): Likewise.
13370         (iwmmxt_walignr): Likewise.
13371         (iwmmxt_walignr0): Likewise.
13372         (iwmmxt_walignr1): Likewise.
13373         (iwmmxt_walignr2): Likewise.
13374         (iwmmxt_walignr3): Likewise.
13375         (iwmmxt_wsadb): Likewise.
13376         (iwmmxt_wsadh): Likewise.
13377         (iwmmxt_wsadbz): Likewise.
13378         (iwmmxt_wsadhz): Likewise.
13379         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13380         (iwmmxt_wabsdiffb): Likewise.
13381         (iwmmxt_wabsdiffh): Likewise.
13382         (iwmmxt_wabsdiffw): Likewise.
13383         (iwmmxt_waddsubhx): Likewise
13384         (iwmmxt_wsubaddhx): Likewise.
13385         (addc<mode>3): Likewise.
13386         (iwmmxt_avg4): Likewise.
13387         (iwmmxt_avg4r): Likewise.
13388         (iwmmxt_wmaddsx): Likewise.
13389         (iwmmxt_wmaddux): Likewise.
13390         (iwmmxt_wmaddsn): Likewise.
13391         (iwmmxt_wmaddun): Likewise.
13392         (iwmmxt_wmulwsm): Likewise.
13393         (iwmmxt_wmulwum): Likewise.
13394         (iwmmxt_wmulsmr): Likewise.
13395         (iwmmxt_wmulumr): Likewise.
13396         (iwmmxt_wmulwsmr): Likewise.
13397         (iwmmxt_wmulwumr): Likewise.
13398         (iwmmxt_wmulwl): Likewise.
13399         (iwmmxt_wqmulm): Likewise.
13400         (iwmmxt_wqmulwm): Likewise.
13401         (iwmmxt_wqmulmr): Likewise.
13402         (iwmmxt_wqmulwmr): Likewise.
13403         (iwmmxt_waddbhusm): Likewise.
13404         (iwmmxt_waddbhusl): Likewise.
13405         (iwmmxt_wqmiabb): Likewise.
13406         (iwmmxt_wqmiabt): Likewise.
13407         (iwmmxt_wqmiatb): Likewise.
13408         (iwmmxt_wqmiatt): Likewise.
13409         (iwmmxt_wqmiabbn): Likewise.
13410         (iwmmxt_wqmiabtn): Likewise.
13411         (iwmmxt_wqmiatbn): Likewise.
13412         (iwmmxt_wqmiattn): Likewise.
13413         (iwmmxt_wmiabb): Likewise.
13414         (iwmmxt_wmiabt): Likewise.
13415         (iwmmxt_wmiatb): Likewise.
13416         (iwmmxt_wmiatt): Likewise.
13417         (iwmmxt_wmiabbn): Likewise.
13418         (iwmmxt_wmiabtn): Likewise.
13419         (iwmmxt_wmiatbn): Likewise.
13420         (iwmmxt_wmiattn): Likewise.
13421         (iwmmxt_wmiawbb): Likewise.
13422         (iwmmxt_wmiawbt): Likewise.
13423         (iwmmxt_wmiawtb): Likewise.
13424         (iwmmxt_wmiawtt): Likewise.
13425         (iwmmxt_wmiawbbn): Likewise.
13426         (iwmmxt_wmiawbtn): Likewise.
13427         (iwmmxt_wmiawtbn): Likewise.
13428         (iwmmxt_wmiawttn): Likewise.
13429         (iwmmxt_wmerge): Likewise.
13430         (iwmmxt_tandc<mode>3): Likewise.
13431         (iwmmxt_torc<mode>3): Likewise.
13432         (iwmmxt_torvsc<mode>3): Likewise.
13433         (iwmmxt_textrc<mode>3): Likewise.
13434         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13435         (wmmxt_pack): Likewise.
13436         (wmmxt_mult_c1): Likewise.
13437         (wmmxt_mult_c2): Likewise.
13438         (wmmxt_alu_c1): Likewise.
13439         (wmmxt_alu_c2): Likewise.
13440         (wmmxt_alu_c3): Likewise.
13441         (wmmxt_transfer_c1): Likewise.
13442         (wmmxt_transfer_c2): Likewise.
13443         (wmmxt_transfer_c3): Likewise.
13444         (marvell_f_iwmmxt_wstr): Likewise.
13445         (marvell_f_iwmmxt_wldr): Likewise.
13447 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13449         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13451 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13453         Revert:
13454         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13455         * lra-constraints.c (need_for_split_p): Check call used hard regs
13456         living through calls.
13458         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13459         call used regs for call insn.
13461 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13463         PR target/57736
13464         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13465         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13467 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13469         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13471 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13473         * lra-constraints.c (need_for_split_p): Check call used hard regs
13474         living through calls.
13476 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13478         PR target/57744
13479         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13480         to tie with any other modes.  Eliminate Altivec vector mode tests,
13481         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13482         code, to return 0 if testing MODE2 for a condition, if we've
13483         already tested MODE1 for the same condition.
13485 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13487         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13488         layout.
13490 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13492         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13493         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13495 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13497         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13498         Define.
13499         (aarch64_symbolic_constant_p): Remove.
13500         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13501         static.  Fix line length and white space.
13502         (aarch64_symbolic_constant_p): Remove.
13503         * config/aarch64/predicates.md (aarch64_valid_symref):
13504         Use aarch64_classify_symbol_expression.
13506 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13508         * config/arm/constraints.md (Ts): New constraint.
13509         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13510         16-bit encodings.
13511         (compare_scc): Use "Ts" constraint for operand 0.
13512         (ior_scc_scc): Likewise.
13513         (and_scc_scc): Likewise.
13514         (and_scc_scc_nodom): Likewise.
13515         (ior_scc_scc_cmp): Likewise for operand 7.
13516         (and_scc_scc_cmp): Likewise.
13517         * config/arm/thumb2.md (thumb2_movsi_insn):
13518         Add alternatives for 16-bit encodings.
13519         (thumb2_movhi_insn): Likewise.
13520         (thumb2_movsicc_insn): Likewise.
13521         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13522         (thumb2_negscc): Use "Ts" constraint.
13523         Move mvn instruction outside cond_exec block.
13524         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13525         for 16-bit encodings.
13527 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13529         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13530         encoding.
13531         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13532         (mulsi3subsi): Likewise.
13533         (mulsidi3adddi): Likewise.
13534         (mulsidi3_v6): Likewise.
13535         (umulsidi3_v6): Likewise.
13536         (umulsidi3adddi_v6): Likewise.
13537         (smulsi3_highpart_v6): Likewise.
13538         (umulsi3_highpart_v6): Likewise.
13539         (mulhisi3tb): Likewise.
13540         (mulhisi3bt): Likewise.
13541         (mulhisi3tt): Likewise.
13542         (maddhisi4): Likewise.
13543         (maddhisi4tb): Likewise.
13544         (maddhisi4tt): Likewise.
13545         (maddhidi4): Likewise.
13546         (maddhidi4tb): Likewise.
13547         (maddhidi4tt): Likewise.
13548         (zeroextractsi_compare0_scratch): Likewise.
13549         (insv_zero): Likewise.
13550         (insv_t2): Likewise.
13551         (anddi_notzesidi_di): Likewise.
13552         (anddi_notsesidi_di): Likewise.
13553         (andsi_notsi_si): Likewise.
13554         (iordi_zesidi_di): Likewise.
13555         (xordi_zesidi_di): Likewise.
13556         (andsi_iorsi3_notsi): Likewise.
13557         (smax_0): Likewise.
13558         (smax_m1): Likewise.
13559         (smin_0): Likewise.
13560         (not_shiftsi): Likewise.
13561         (unaligned_loadsi): Likewise.
13562         (unaligned_loadhis): Likewise.
13563         (unaligned_loadhiu): Likewise.
13564         (unaligned_storesi): Likewise.
13565         (unaligned_storehi): Likewise.
13566         (extv_reg): Likewise.
13567         (extzv_t2): Likewise.
13568         (divsi3): Likewise.
13569         (udivsi3): Likewise.
13570         (arm_zero_extendhisi2addsi): Likewise.
13571         (arm_zero_extendqisi2addsi): Likewise.
13572         (compareqi_eq0): Likewise.
13573         (arm_extendhisi2_v6): Likewise.
13574         (arm_extendqisi2addsi): Likewise.
13575         (arm_movt): Likewise.
13576         (thumb2_ldrd): Likewise.
13577         (thumb2_ldrd_base): Likewise.
13578         (thumb2_ldrd_base_neg): Likewise.
13579         (thumb2_strd): Likewise.
13580         (thumb2_strd_base): Likewise.
13581         (thumb2_strd_base_neg): Likewise.
13582         (arm_negsi2): Add alternative for 16-bit encoding.
13583         (arm_one_cmplsi2): Likewise.
13585 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13587         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13588         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13589         (movdfcc): Likewise.
13590         * config/arm/vfp.md (*thumb2_movsf_vfp):
13591         Disable predication for arm_restrict_it.
13592         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13593         (*thumb2_movdfcc_vfp): Likewise.
13594         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13595         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13596         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13597         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13598         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13599         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13600         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13601         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13602         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13603         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13604         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13605         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13606         Disable predication for arm_restrict_it.
13608 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13610         * config/i386/bmiintrin.h (_bextr_u32): New.
13611         (_bextr_u64): Ditto.
13613 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13615         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13616         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13617         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13618         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13619         * config/mips/n32-elf.h: ...this new file.
13621 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13623         PR target/57224
13624         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13625         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13627 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13629         * config/mips/mips-tables.opt: Regenerate.
13630         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13631         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13632         * doc/invoke.texi: Add -m14kc.
13634 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13636         PR target/57623
13637         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13638         constraints of operand 1 and 2.
13640         PR target/57623
13641         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13642         to match RTL canonicalization.  Swap predicates and
13643         constraints of operand 1 and 2.
13645 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13647         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13648         Process OP_INOUT regs for splitting too.
13650 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13652         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13653         decl before the loop, initialize to NULL.
13654         (vectorizable_load): Initialize ptr_incr to NULL.
13656 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13658         PR lto/57208
13659         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13660         * ipa-ref.c (ipa_maybe_record_reference): New function.
13661         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13662         * ipa-cp.c (create_specialized_node): Record potential references from
13663         aggvals.
13664         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13666 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13668         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13669         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13670         'mode' to force_reg.
13671         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13672         (aarch64_expand_mov_immediate): Likewise.
13674 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13676         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13677         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13679 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13681         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13682         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13683         (struct machine_function): Add tbegin_p.
13684         (s390_canonicalize_comparison): Fold CC mode compares to
13685         conditional jump if possible.
13686         (s390_emit_jump): Return the emitted jump.
13687         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13688         Handle CCRAWmode compares.
13689         (s390_option_override): Default to -mhtm if available.
13690         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13691         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13692         FPRs instead of df_regs_ever_live_p.
13693         (s390_optimize_nonescaping_tx): New function.
13694         (s390_init_frame_layout): Extend clobbered_regs array to cover
13695         FPRs as well.
13696         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13697         (s390_expand_tbegin): New function.
13698         (enum s390_builtin): New enum definition.
13699         (code_for_builtin): New array definition.
13700         (s390_init_builtins): New function.
13701         (s390_expand_builtin): New function.
13702         (TARGET_INIT_BUILTINS): Define.
13703         (TARGET_EXPAND_BUILTIN): Define.
13704         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13705         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13706         (s390_alc_comparison): Likewise.
13707         * config/s390/s390-modes.def: Add CCRAWmode.
13708         * config/s390/s390.h (processor_flags): Add PF_TX.
13709         (TARGET_CPU_HTM): Define macro.
13710         (TARGET_HTM): Define macro.
13711         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13712         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13713         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13714         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13715         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13716         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13717         ("*cc_to_int"): Move up.
13718         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13719         constants other than 0.
13720         ("*ccraw_to_int"): New insn and splitter definition.
13721         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13722         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13723         ("tx_assist"): New expander.
13724         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13725         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13726         * config/s390/s390.opt: Add -mhtm option.
13727         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13728         * config/s390/htmxlintrin.h: New file.
13729         * config/s390/htmintrin.h: New file.
13730         * config/s390/s390intrin.h: New file.
13731         * doc/extend.texi: Document htm builtins.
13732         * config.gcc: Add the new header files to extra_headers.
13734 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13736         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13737         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13739 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13740             Pat Haugen  <pthaugen@us.ibm.com>
13741             Peter Bergner  <bergner@vnet.ibm.com>
13743         * config/rs6000/power8.md: New.
13744         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13745         setting for power8 entry.
13746         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13747         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13748         test for Power4/Power5 only.
13749         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13750         support.
13751         (force_new_group): Adjust comment.
13752         * config/rs6000/rs6000.md: Include power8.md.
13754 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13756         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13757         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13758         declaration.
13759         (tune_params): Update comment.
13760         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13761         (arm_max_conditional_execute): New function.
13762         (thumb2_final_prescan_insn): Use max_insn_skipped and
13763         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13765 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13767         PR tree-optimization/57705
13768         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13769         SSA_NAME step, provided that it is not defined inside the loop.
13770         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13771         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13773 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13775         PR middle-end/57670
13776         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13777         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13778         calls in the dump.
13779         (ipa_note_param_call): Initialize member_ptr flag.
13780         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13781         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13782         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13783         (ipa_read_indirect_edge_info): Likewise.
13785 2013-06-25  Richard Biener  <rguenther@suse.de>
13787         PR middle-end/56977
13788         * passes.c (init_optimization_passes): Move pass_fold_builtins
13789         and pass_dce earlier with -Og.
13791 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13793         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13794         <BIT_FIELD_REF>: Remove trailing TAB.
13795         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13796         remove blank line.
13798 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13800         PR tree-optimization/57358
13801         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13802         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13803         (ipa_analyze_params_uses): Generate pessimistic info when true.
13805 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13807         PR tree-optimization/57539
13808         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13809         global.inlined_to of the new node to it.  All callers changed.
13810         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13811         inlining_into, pass it to cgraph_clone_node.
13812         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13813         ipa_free_edge_args_substructures.
13814         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13815         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13816         Assert we have found an rdesc in the rdesc list.
13818 2013-06-24  Richard Biener  <rguenther@suse.de>
13820         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13821         (pointer_set_lookup): Declare.
13822         (class pointer_map): New template class implementing a
13823         generic pointer to T map.
13824         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13825         pointer_map<T>::contains, pointer_map<T>::insert,
13826         pointer_map<T>::traverse): New functions.
13827         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13828         (pointer_set_lookup): New function.
13829         (pointer_set_contains): Use pointer_set_lookup.
13830         (pointer_set_insert): Likewise.
13831         (insert_aux): Remove.
13832         (struct pointer_map_t): Embed a pointer_set_t.
13833         (pointer_map_create): Adjust.
13834         (pointer_map_destroy): Likewise.
13835         (pointer_map_contains): Likewise.
13836         (pointer_map_insert): Likewise.
13837         (pointer_map_traverse): Likewise.
13838         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13839         pointer_map<unsigned> instead of a pointer_map_t.
13840         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13841         (streamer_tree_cache_lookup): Likewise.
13842         (streamer_tree_cache_create): Likewise.
13843         (streamer_tree_cache_delete): Likewise.
13844         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13845         pointer_map<unsigned> instead of a pointer_map_t.
13846         (lto_init_tree_ref_encoder): Adjust.
13847         (lto_destroy_tree_ref_encoder): Likewise.
13848         * lto-section-out.c (lto_output_decl_index): Likewise.
13849         (lto_record_function_out_decl_state): Likewise.
13850         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13852 2013-06-24  Richard Biener  <rguenther@suse.de>
13854         PR tree-optimization/57488
13855         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13857 2013-06-24  Alan Modra  <amodra@gmail.com>
13859         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13860         (gen_easy_altivec_constant): Likewise.
13861         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13862         easy_vector_constant_msb): Likewise.
13864 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13866         PR target/57688
13867         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13868         add missing return true.
13870 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13872         PR target/52483
13873         * config/sh/predicates.md (general_extend_operand): Invoke
13874         general_movsrc_operand for memory operands.
13875         (general_movsrc_operand): Allow reg+reg addressing, do not use
13876         general_operand for memory operands.
13878 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13880         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13881         when current target options does not apply.
13882         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13883         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13884         * config/i386/bmiintrin.h: Pass appropriate target
13885         attributes to header.
13886         * config/i386/mmintrin.h: Ditto.
13887         * config/i386/nmmintrin.h: Ditto.
13888         * config/i386/avx2intrin.h: Ditto.
13889         * config/i386/fxsrintrin.h: Ditto.
13890         * config/i386/tbmintrin.h: Ditto.
13891         * config/i386/xsaveintrin.h: Ditto.
13892         * config/i386/f16cintrin.h: Ditto.
13893         * config/i386/xtestintrin.h: Ditto.
13894         * config/i386/xsaveoptintrin.h: Ditto.
13895         * config/i386/bmi2intrin.h: Ditto.
13896         * config/i386/lzcntintrin.h: Ditto.
13897         * config/i386/smmintrin.h: Ditto.
13898         * config/i386/wmmintrin.h: Ditto.
13899         * config/i386/x86intrin.h: Remove all header include guards.
13900         * config/i386/prfchwintrin.h: Ditto.
13901         * config/i386/pmmintrin.h: Ditto.
13902         * config/i386/tmmintrin.h: Ditto.
13903         * config/i386/xmmintrin.h: Ditto.
13904         * config/i386/popcntintrin.h: Ditto.
13905         * config/i386/rdseedintrin.h: Ditto.
13906         * config/i386/ammintrin.h: Ditto.
13907         * config/i386/emmintrin.h: Ditto.
13908         * config/i386/immintrin.h: Remove all header include guards.
13909         * config/i386/fma4intrin.h: Ditto.
13910         * config/i386/lwpintrin.h: Ditto.
13911         * config/i386/xopintrin.h: Ditto.
13912         * config/i386/ia32intrin.h: Ditto.
13913         * config/i386/avxintrin.h: Ditto.
13914         * config/i386/rtmintrin.h: Ditto.
13915         * config/i386/fmaintrin.h: Ditto.
13916         * config/i386/mm3dnow.h: Ditto.
13918 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13920         * common/config/i386/i386-common.c: Handle LZCNT.
13922 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13924         * doc/extend.texi: Use __atomic_store_n instead of
13925         __atomic_store in HLE example.
13927 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13929         * config/sh/sh.c: Remove <cstdlib> workaround.
13931 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13933         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13935 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13937         * doc/extend.texi: Document that __atomic_clear and
13938           __atomic_test_and_set should only be used with bool.
13940 2013-06-20  Jan Hubicka  <jh@suse.cz>
13942         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13943         types_same_for_odr.
13944         * tree.c (decls_same_for_odr): New function.
13945         (same_for_edr): New function.
13946         (types_same_for_odr): New function.
13947         (get_binfo_at_offset): Use it.
13948         * tree.h (types_same_for_odr): Declare.
13950 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13951             Jason Merrill  <jason@redhat.com>
13953         * system.h: Include <cstdlib> as well as <stdlib.h>.
13955 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13957         PR target/57655
13958         * config/i386/i386.c (construct_container): Report error if
13959         long double is used with disabled x87 float returns.
13961 2013-06-20  Jan Hubicka  <jh@suse.cz>
13963         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13965 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13967         PR rtl-optimization/57425
13968         PR rtl-optimization/57569
13969         * alias.c (write_dependence_p): Remove parameters mem_mode and
13970         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13971         Changed all callers.
13972         (canon_anti_dependence): Get comments and semantics in sync.
13973         Add parameter mem_canonicalized.  Changed all callers.
13974         * rtl.h (canon_anti_dependence): Update prototype.
13976 2013-06-20  Richard Biener  <rguenther@suse.de>
13978         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13979         byte case, inline streamer_read_uchar and defer section
13980         overrun check.
13982 2013-06-20  Richard Biener  <rguenther@suse.de>
13984         PR tree-optimization/57584
13985         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13986         SSA names into the expanded expression that take part in
13987         abnormal coalescing.
13989 2013-06-19  Sharad Singhai  <singhai@google.com>
13991         * gcov.c (print_usage): Handle new option.
13992         (process_args): Ditto.
13993         (get_gcov_intermediate_filename): New function.
13994         (output_intermediate_file): New function.
13995         (output_gcov_file): New function
13996         (generate_results): Handle new option.
13997         (release_function): Relase demangled name.
13998         (read_graph_file): Handle demangled name.
13999         (output_lines): Ditto.
14000         * doc/gcov.texi: Document gcov intermediate format.
14002 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14004         PR bootstrap/57604
14005         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14006         (lra_emit_add): Use the functions.  Add comment about Y as an
14007         address segment.
14009 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14011         PR driver/57652
14012         * collect2.c (collect_atexit): New.
14013         (collect_exit): Delete.
14014         (main): Register collect_atexit with atexit.
14015         (collect_wait): Change collect_exit to exit.
14016         (do_wait): Same.
14017         * collect2.h (collect_exit): Delete.
14018         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14020 2013-06-19  Wei Mi  <wmi@google.com>
14022         PR rtl-optimization/57518
14023         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14024         if regno is used in paradoxical subreg.
14025         (update_equiv_regs): Check pdx_subregs[regno] before
14026         set a reg to be equivalent with a mem.
14028 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14030         PR driver/57651
14031         * file-find.h (find_a_file): Add a mode parameter.
14032         * file-find.c (find_a_file): Likewise.
14033         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14034         with X_OK for the executables.
14035         * collect2.c (main): Call find_a_file with X_OK.
14037 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14039         PR target/56942
14040         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14041         Use NEXT_INSN instead of next_real_insn.
14043 2013-06-19  Jan Hubicka  <jh@suse.cz>
14045         * cgraph.h (const_value_known_p): Replace by ...
14046         (ctor_for_folding): .. this one.
14047         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14048         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14049         * expr.c (expand_expr_real_1): Likewise.
14050         (string_constant): Likewise.
14051         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14052         * ipa.c (process_references): Likewise.
14053         (symtab_remove_unreachable_nodes): Likewise.
14054         * ipa-inline-analysis.c (param_change_prob): Likewise.
14055         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14056         (get_base_constructor): Likwise.
14057         * varpool.c (varpool_remove_node): Likewise.
14058         (varpool_remove_initializer): LIkewise.
14059         (dump_varpool_node): LIkwise.
14060         (const_value_known_p): Rewrite to ...
14061         (ctor_for_folding): ... this one.
14063 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14065         PR driver/57651
14066         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14067         PERSONALITY in $PATH derived prefixes.
14069 2013-06-19  Jeff Law  <law@redhat.com>
14071         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14072         in comment.
14074         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14075         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14076         booleans.
14078 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14080         * config/arm/vfp.md: Move VFP instruction classification documentation
14081         to ...
14082         * config/arm/arm.md: ... here.  Update instruction classification
14083         documentation.
14085 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14087         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14088         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14089         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14091 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14093         PR target/57609
14094         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14095         with NEXT_INSN.  Use tablejump_p to check for jump table data
14096         insns.
14098 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14100         PR c++/56544
14101         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14102         that now in C++ the value is correct per the C++ standards.
14104 2013-06-19  Richard Biener  <rguenther@suse.de>
14106         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14107         for global context.
14109 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14111         Revert:
14112         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14114         PR target/57609
14115         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14116         with next_active_insn.
14118 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14120         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14121         functions are inlined during failures to flag an error.
14122         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14123         in early inline pass.
14125 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14127         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14128         in comments.
14130 2013-06-18  Julian Brown  <julian@codesourcery.com>
14132         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14133         Permit virtual register pre-reload if !strict.
14134         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14135         change.
14136         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14137         prototype.
14138         * config/arm/neon.md (movmisalign<mode>): Use
14139         neon_perm_struct_or_reg_operand instead of
14140         neon_struct_or_register_operand.
14141         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14142         neon_permissive_struct_operand instead of neon_struct_operand.
14143         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14144         neon_vector_mem_operand.
14145         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14146         neon_vector_mem_operand.
14147         (neon_permissive_struct_operand): New.
14148         (neon_struct_or_register_operand): Rename to...
14149         (neon_perm_struct_or_reg_operand): This. Adjust call to
14150         neon_vector_mem_operand.
14152 2013-06-18  Richard Biener  <rguenther@suse.de>
14154         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14155         * lto-streamer.h: Include pointer-set.h.
14156         (struct lto_decl_slot): Remove.
14157         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14158         Remove next_index entry.
14159         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14160         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14161         (lto_init_tree_ref_encoder): Adjust.
14162         (lto_destroy_tree_ref_encoder): Likewise.
14163         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14164         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14165         (lto_output_decl_index): Adjust.
14166         (lto_new_out_decl_state): Likewise.
14167         (lto_record_function_out_decl_state): Likewise.
14168         * lto-streamer-out.c (copy_function): Likewise.
14170 2013-06-18  Richard Biener  <rguenther@suse.de>
14172         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14173         * cgraphunit.c: Include cfgloop.h.
14174         (init_lowered_empty_function): Initialize the loop tree.
14175         (assemble_thunk): Insert new BBs into loops.
14177 2013-06-18  Richard Biener  <rguenther@suse.de>
14179         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14180         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14181         the map from cache entry to cache index optional.
14182         (streamer_tree_cache_replace_tree): Adjust accordingly.
14183         (streamer_tree_cache_append): Likewise.
14184         (streamer_tree_cache_delete): Likewise.
14185         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14186         streamer cache map from cache entry to cache index.
14187         * lto-streamer-out.c (create_output_block): Adjust.
14189 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14191         * config/arm/arm.md (attribute "insn"): Move multiplication and
14192         division attributes to...
14193         (attribute "type"): ... here.  Remove mult.
14194         (attribute "mul32"): New attribute.
14195         (attribute "mul64"): Add umaal.
14196         (*arm_mulsi3): Update attributes.
14197         (*arm_mulsi3_v6): Likewise.
14198         (*thumb_mulsi3): Likewise.
14199         (*thumb_mulsi3_v6): Likewise.
14200         (*mulsi3_compare0): Likewise.
14201         (*mulsi3_compare0_v6): Likewise.
14202         (*mulsi_compare0_scratch): Likewise.
14203         (*mulsi_compare0_scratch_v6): Likewise.
14204         (*mulsi3addsi): Likewise.
14205         (*mulsi3addsi_v6): Likewise.
14206         (*mulsi3addsi_compare0): Likewise.
14207         (*mulsi3addsi_compare0_v6): Likewise.
14208         (*mulsi3addsi_compare0_scratch): Likewise.
14209         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14210         (*mulsi3subsi): Likewise.
14211         (*mulsidi3adddi): Likewise.
14212         (*mulsi3addsi_v6): Likewise.
14213         (*mulsidi3adddi_v6): Likewise.
14214         (*mulsidi3_nov6): Likewise.
14215         (*mulsidi3_v6): Likewise.
14216         (*umulsidi3_nov6): Likewise.
14217         (*umulsidi3_v6): Likewise.
14218         (*umulsidi3adddi): Likewise.
14219         (*umulsidi3adddi_v6): Likewise.
14220         (*smulsi3_highpart_nov6): Likewise.
14221         (*smulsi3_highpart_v6): Likewise.
14222         (*umulsi3_highpart_nov6): Likewise.
14223         (*umulsi3_highpart_v6): Likewise.
14224         (mulhisi3): Likewise.
14225         (*mulhisi3tb): Likewise.
14226         (*mulhisi3bt): Likewise.
14227         (*mulhisi3tt): Likewise.
14228         (maddhisi4): Likewise.
14229         (*maddhisi4tb): Likewise.
14230         (*maddhisi4tt): Likewise.
14231         (maddhidi4): Likewise.
14232         (*maddhidi4tb): Likewise.
14233         (*maddhidi4tt): Likewise.
14234         (divsi3): Likewise.
14235         (udivsi3): Likewise.
14236         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14237         (thumb2_mulsi_short_compare0): Likewise.
14238         (thumb2_mulsi_short_compare0_scratch): Likewise.
14239         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14240         (1020mult2): Likewise.
14241         (1020mult3): Likewise.
14242         (1020mult4): Likewise.
14243         (1020mult5): Likewise.
14244         (1020mult6): Likewise.
14245         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14246         change.
14247         (cortex_a15_mult64): Likewise.
14248         (cortex_a15_sdiv): Likewise.
14249         (cortex_a15_udiv): Likewise.
14250         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14251         (mult2): Likewise.
14252         (mult3): Likewise.
14253         (mult4): Likewise.
14254         (mult5): Likewise.
14255         (mult6): Likewise.
14256         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14257         (pj4_ir_div): Likewise.
14258         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14259         (11_mult2): Likewise.
14260         (11_mult3): Likewise.
14261         (11_mult4): Likewise.
14262         (11_mult5): Likewise.
14263         (11_mult6): Likewise.
14264         (11_mult7): Likewise.
14265         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14266         (cortex_a8_mla): Likewise.
14267         (cortex_a8_mull): Likewise.
14268         (cortex_a8_smulwy): Likewise.
14269         (cortex_a8_smlald): Likewise.
14270         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14271         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14272         (cortex_r4_mul_3): Likewise.
14273         (cortex_r4_mla_4): Likewise.
14274         (cortex_r4_mla_3): Likewise.
14275         (cortex_r4_smlald): Likewise.
14276         (cortex_r4_mull): Likewise.
14277         (cortex_r4_sdiv): Likewise.
14278         (cortex_r4_udiv): Likewise.
14279         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14280         (cortex_a7_idiv): Likewise.
14281         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14282         (9_mult2): Likewise.
14283         (9_mult3): Likewise.
14284         (9_mult4): Likewise.
14285         (9_mult5): Likewise.
14286         (9_mult6): Likewise.
14287         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14288         (cortex_a53_sdiv): Likewise.
14289         (cortex_a53_udiv): Likewise.
14290         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14291         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14292         (mp626_mult2): Likewise.
14293         (mp626_mult3): Likewise.
14294         (mp626_mult4): Likewise.
14295         * config/arm/fa526.md (526_mult1): Update attribute change.
14296         (526_mult2): Likewise.
14297         * config/arm/arm-generic.md (mult): Update attribute change.
14298         (mult_ldsched_strongarm): Likewise.
14299         (mult_ldsched): Likewise.
14300         (multi_cycle): Likewise.
14301         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14302         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14303         (606te_mult2): Likewise.
14304         (606te_mult3): Likewise.
14305         (606te_mult4): Likewise.
14306         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14307         (cortex_a9_mac16): Likewise.
14308         (cortex_a9_multiply): Likewise.
14309         (cortex_a9_mac): Likewise.
14310         (cortex_a9_multiply_long): Likewise.
14311         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14312         (626te_mult2): Likewise.
14313         (626te_mult3): Likewise.
14314         (626te_mult4): Likewise.
14316 2013-06-18  Richard Biener  <rguenther@suse.de>
14318         PR lto/57334
14319         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14321 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14323         PR target/57609
14324         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14325         with next_active_insn.
14327 2013-06-18  Alan Modra  <amodra@gmail.com>
14329         * config/rs6000/rs6000.h (enum data_align): New.
14330         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14331         (DATA_ABI_ALIGNMENT): Define.
14332         (CONSTANT_ALIGNMENT): Correct comment.
14333         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14334         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14336 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14338         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14339         ATTRIBUTE_UNUSED marking.
14341 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14343         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14344         alternative and update.
14345         (aarch64_dup_lanedi): Delete.
14346         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14347         * config/aarch64/aarch64-simd-builtins.def: Update.
14349 2013-06-17  Richard Biener  <rguenther@suse.de>
14351         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14352         (lto_input_scc): Declare.
14353         (lto_input_tree_1): Likewise.
14354         (struct lto_stats_d): Add num_tree_bodies_output and
14355         num_pickle_refs_output.
14356         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14357         (lto_read_tree_1): Split out from ...
14358         (lto_read_tree): ... this.
14359         (lto_input_scc): New function.
14360         (lto_input_tree_1): Split out from ...
14361         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14362         (lto_data_in_create): Create the streamer cache without hashes.
14363         * lto-streamer-out.c (create_output_block): Create the streamer
14364         cache with hashes when not doing WPA.
14365         (lto_write_tree_1): Split out from ...
14366         (lto_write_tree): ... this.
14367         (get_symbol_initial_value): New function.
14368         (lto_output_tree_1): Split out from ...
14369         (lto_output_tree): ... this.  Write trees as series of SCCs
14370         using a DFS walk via DFS_write_tree.
14371         (struct sccs, struct scc_entry): New types.
14372         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14373         (DFS_write_tree_body): New function.
14374         (DFS_write_tree): Likewise.
14375         (hash_tree): Likewise.
14376         (scc_entry_compare): Likewise.
14377         (hash_scc): Likewise.
14378         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14379         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14380         TREE_CHAIN as regular reference.
14381         (streamer_read_integer_cst): Remove.
14382         (streamer_get_pickled_tree): Adjust.
14383         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14384         of DECL_EXTERNALs in BLOCK_VARS for now.
14385         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14386         reference.
14387         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14388         Add hash value argument and record that if hashes are recorded
14389         in the cache.
14390         (streamer_tree_cache_insert_1): Adjust.
14391         (streamer_tree_cache_insert): Likewise.
14392         (streamer_tree_cache_insert_at): Rename to ...
14393         (streamer_tree_cache_replace_tree): ... this and adjust.
14394         (streamer_tree_cache_append): Adjust.
14395         (record_common_node): Likewise.
14396         (streamer_tree_cache_create): Add argument whether to
14397         record hash values together with trees.
14398         (streamer_tree_cache_delete): Adjust.
14399         * tree-streamer.h (struct streamer_tree_cache_d): Add
14400         vector of hashes.
14401         (streamer_read_integer_cst): Remove.
14402         (streamer_tree_cache_insert): Adjust.
14403         (streamer_tree_cache_append): Likewise.
14404         (streamer_tree_cache_insert_at): Rename to ...
14405         (streamer_tree_cache_replace_tree): ... this and adjust.
14406         (streamer_tree_cache_create): Add argument whether to record hashes.
14407         (streamer_tree_cache_get): Rename to ...
14408         (streamer_tree_cache_get_tree): ... this.
14409         (streamer_tree_cache_get_hash): New function.
14410         * tree.c (cache_integer_cst): New function.
14411         * tree.h (cache_integer_cst): Declare.
14412         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14413         * lto-symtab.c (lto_varpool_replace_node): Only release
14414         DECL_INITIAL of non-prevailing decls.
14415         * varpool.c (varpool_remove_initializer): Do not release
14416         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14418 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14420         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14421         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14422         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14423         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14424         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14425         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14426         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14427         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14428         instead of TARGET_64BIT.
14429         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14430         Require ISA_HAS_<D>DIV.
14432 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14434         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14435         (mips*-*-linux*): Move default with_llsc setting to where other
14436         defaults are set.
14437         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14438         with_arch block.
14439         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14440         Likewise.  Remove default with_tune setting.  Move default float
14441         setting to its own block.  Handle with_llsc in the same block as above.
14443 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14445         PR rtl-optimization/57425
14446         PR rtl-optimization/57569
14447         * alias.c (write_dependence_p): Add new parameters mem_mode,
14448         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14449         Changed all callers.
14450         (canon_anti_dependence): New function.
14451         * cse.c (check_dependence): Use canon_anti_dependence.
14452         * cselib.c (cselib_invalidate_mem): Likewise.
14453         * rtl.h (canon_anti_dependence): Declare.
14455 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14457         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14458         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14459         ".set mips3" for 64-bit targets.
14461 2013-06-15  Dehao Chen  <dehao@google.com>
14463         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14464         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14465         (gimple_check_call_args): Likewise.
14466         * value-prof.c (check_ic_target): Likewise.
14467         * ipa-inline.c (early_inliner): Likewise.
14468         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14469         * cgraph.c (cgraph_create_edge_1): Likewise.
14470         (cgraph_make_edge_direct): Likewise.
14472 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14474         PR target/57615
14475         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14476         rs6000_output_move_128bit to handle emitting quad memory
14477         operations.  Set attribute length to 8 bytes.
14479 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14481         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14482         New pattern.
14483         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14484         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14485         (aarch64_<su>mlsl<mode>): Likewise.
14487 2013-06-14  Mike Stump  <mikestump@comcast.net>
14489         * Makefile.in (TARGET_H): Add insn-codes.h.
14491 2013-06-14  Alan Modra  <amodra@gmail.com>
14493         PR middle-end/57134
14494         PR middle-end/57586
14495         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14496         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14497         bitfield expansion when EXPAND_MEMORY.
14498         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14500 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14502         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14503         test for clearing quad memory on 32-bit later.
14505 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14507         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14508         (fold_negate_expr): Likewise.
14509         (fold_real_zero_addition_p): Handle vectors.
14510         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14512 2013-06-14  Alan Modra  <amodra@gmail.com>
14514         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14516 2013-06-13  Jan Hubicka  <jh@suse.cz>
14518         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14519         Local comdats are not externally visible.
14520         * symtab.c (dump_symtab_base): Dump externally visible.
14521         (verify_symtab_base): Verify back links in the symtab hash.
14523 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14525         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14526         CONVERT_EXPR as equal nodes.
14528 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14530         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14532 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14534         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14535         Generalize to complex and vector.
14536         * tree.c (build_all_ones_cst): New function.
14537         * tree.h (build_all_ones_cst): Declare it.
14539 2013-06-13  Alan Modra  <amodra@gmail.com>
14541         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14542         * config/rs6000/rs6000.md (signbittf2): New insn.
14543         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14544         (abstf2_internal, cmptf_internal2): Likewise.
14545         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14547 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14548             Pat Haugen  <pthaugen@us.ibm.com>
14549             Peter Bergner  <bergner@vnet.ibm.com>
14551         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14552         power8 byte, half-word, and quad-word atomic instructions.
14553         (emit_store_conditional): Likewise.
14554         (rs6000_expand_atomic_compare_and_swap): Likewise.
14555         (rs6000_expand_atomic_op): Likewise.
14557         * config/rs6000/sync.md (larx): Add new modes for power8.
14558         (stcx): Likewise.
14559         (AINT): New mode iterator to include TImode as well as normal
14560         integer modes on power8.
14561         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14562         that VSX registers are not considered.  Use AINT mode iterator
14563         instead of INT1 to allow inclusion of quad word atomic operations
14564         on power8.
14565         (load_locked<mode>): Likewise.
14566         (store_conditional<mode>): Likewise.
14567         (atomic_compare_and_swap<mode>): Likewise.
14568         (atomic_exchange<mode>): Likewise.
14569         (atomic_nand<mode>): Likewise.
14570         (atomic_fetch_<fetchop_name><mode>): Likewise.
14571         (atomic_nand_fetch<mode>): Likewise.
14572         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14573         each type.
14574         (ATOMIC): On power8, add QImode, HImode modes.
14575         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14576         modes that promote to SImode.
14577         (load_lockedti): Convert TImode arguments to PTImode, so that we
14578         get a guaranteed even/odd register pair.
14579         (load_lockedpti): Likewise.
14580         (store_conditionalti): Likewise.
14581         (store_conditionalpti): Likewise.
14583         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14584         atomic load/store instructions.
14585         (HSI): Likewise.
14587 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14589         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14590         loads.
14591         (insn_count): New attribute, with most cases extracted from...
14592         (length): ...here.  Redefine most cases in terms of insn_count.
14593         (single_insn): Delete.
14594         (can_delay): Use insn_count to check for single instructions.
14595         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14596         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14597         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14598         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14599         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14600         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14601         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14602         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14603         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14604         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14605         rather than "length".
14606         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14607         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14608         Use "insn_count" rather than "length".
14609         * config/mips/mips-dsp.md
14610         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14611         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14612         length attributes.
14614 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14616         PR tree-optimization/57361
14617         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14619 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14621         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14622         to split.
14623         (aarch64_simd_combine<mode>): New instruction expansion.
14624         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14625         function prototype.
14626         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14627         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14629 2013-06-12  Jan Hubicka  <jh@suse.cz>
14631         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14632         decl has when in streaming stage.
14633         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14634         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14636 2013-06-12  Roland Stigge  <stigge@antcom.de>
14638         PR target/57578
14639         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14641 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14643         PR tree-optimization/57537
14644         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14645         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14647 2013-06-12  Richard Biener  <rguenther@suse.de>
14649         * data-streamer.h (streamer_write_char_stream): CSE
14650         obs->current_pointer.
14651         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14652         streamer_write_char_stream manually and optimize the resulting loop.
14653         (streamer_write_hwi_stream): Likewise.
14655 2013-06-12  Jan Hubicka  <jh@suse.cz>
14657         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14658         * cgraph.h (varpool_create_empty_node): Declare.
14659         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14660         duplicated nodes.
14661         * symtab.c (symtab_unregister_node): Be lax about missin entries
14662         in node hash.
14663         (symtab_get_node): Update comment.
14664         * varpool.c (varpool_create_empty_node): Break out from ...
14665         (varpool_node_for_decl): ... here.
14666         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14668 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14670         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14671         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14672         part.  Use straight-line flow at the end.
14673         <COMPONENT_REF>: Remove superfluous else.
14674         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14676 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14678         PR target/56564
14679         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14680         target hook even for !TREE_PUBLIC decls.  If no resolution info
14681         is available, return false for common and external decls.
14683 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14685         * config/rl78/constraints.md (U): New constraint.
14686         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14687         valloc attribute.
14689 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14691         PR target/57589
14692         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14693         to allow returning address to AT_PLATFORM name.
14695 2013-06-11  Jan Hubicka  <jh@suse.cz>
14697         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14698         * cgraph.h (symtab_node_base): Add weakref flag.
14699         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14700         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14701         (output_weakrefs): Use weakref flag.
14702         * fold-const.c (simple_operand_p): Handle WEAK.
14703         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14704         * ipa.c (varpool_externally_visible_p): Drop weakref.
14705         (function_and_variable_visibility): Update comment; fix weakref
14706         sanity checks; do not clear DECL_WEAK on them.
14707         * lto-cgraph.c (lto_output_node): update.
14708         (lto_output_varpool_node): Update.
14709         (input_overwrite_node): Update.
14710         (input_node): Update.
14711         (input_varpool_node): Update.
14712         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14713         (lto_symtab_merge_symbols): Add sanity check.
14714         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14715         * passes.c (rest_of_decl_compilation): Set static flag, too.
14716         * symtab.c (dump_symtab_base): Dump weakref.
14717         (verify_symtab_base): Sanity check weakrefs.
14718         (symtab_make_decl_local): Remove duplicated code.
14719         (symtab_alias_ultimate_target): Simplify.
14720         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14722 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14724         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14725         sequence_vect == NULL.
14727 2013-06-11  DJ Delorie  <dj@redhat.com>
14729         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14730         (rl78_unwind_word_mode): New.
14732 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14734         * final.c (debug_prefix_maps): Make static.
14736 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14738         * function.c (initial_trampoline): Remove stray copy.
14740 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14742         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14744 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14746         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14747         within bounds at the beginning of the function.
14749 2013-06-11  Alan Modra  <amodra@gmail.com>
14751         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14752         reporting.
14753         (get_named_section): Don't NULL !DECL_P decl.
14755 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14757         * doc/invoke.texi (core-avx2): Document.
14758         (slm): Likewise.
14759         (atom): Updated with MOVBE.
14761 2013-06-11  Richard Biener  <rguenther@suse.de>
14763         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14765 2013-06-11  Anton Blanchard  <anton@samba.org>
14767         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14768         correct shift value in little-endian mode.
14770 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14772         PR target/56564
14773         * varasm.c (get_variable_align): Move #endif to the right place.
14775 2013-06-10  Cary Coutant  <ccoutant@google.com>
14777         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14778         for hash so that hash table traversal order is deterministic.
14780 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14781             Pat Haugen  <pthaugen@us.ibm.com>
14782             Peter Bergner  <bergner@vnet.ibm.com>
14784         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14785         of vectors in GPRS if they are direct moves or quad word load or
14786         store moves.
14788         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14789         declaration.
14790         (direct_move_p): Likewise.
14791         (quad_load_store_p): Likewise.
14793         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14794         classes into bins based on the physical register type.
14795         (reg_class_to_reg_type): Likewise.
14796         (IS_STD_REG_TYPE): Likewise.
14797         (IS_FP_VECT_REG_TYPE): Likewise.
14798         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14799         use direct move instructions.
14800         (reload_gpr_vsx): Likewise.
14801         (reload_vsx_gpr): Likewise.
14802         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14803         information that is a simplification of register classes.  Also
14804         precalculate direct move reload helpers.
14805         (direct_move_p): New function to return true if the operation can
14806         be done as a direct move instruciton.
14807         (quad_load_store_p): New function to return true if the operation
14808         is a quad memory operation.
14809         (rs6000_legitimize_address): If quad memory, only allow register
14810         indirect for TImode addresses.
14811         (rs6000_legitimate_address_p): Likewise.
14812         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14813         (rs6000_reload_register_type): Likewise.
14814         (register_to_reg_type): Return register type.
14815         (rs6000_secondary_reload_simple_move): New helper function for
14816         secondary reload and secondary memory needed to identify anything
14817         that is a simple move, and does not need reloading.
14818         (rs6000_secondary_reload_direct_move): New helper function for
14819         secondary reload to identify cases that can be done with several
14820         instructions via the direct move instructions.
14821         (rs6000_secondary_reload_move): New helper function for secondary
14822         reload to identify moves between register types that can be done.
14823         (rs6000_secondary_reload): Add support for quad memory operations
14824         and for direct move.
14825         (rs6000_secondary_memory_needed): Likewise.
14826         (rs6000_debug_secondary_memory_needed): Change argument names.
14827         (rs6000_output_move_128bit): New function to return the move to
14828         use for 128-bit moves, including knowing about the various
14829         limitations of quad memory operations.
14831         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14832         memory operations.  call rs6000_output_move_128bit for the actual
14833         instruciton(s) to generate.
14834         (vsx_movti_64bit): Likewise.
14836         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14837         (UNSPEC_P8V_MTVSRWZ): Likewise.
14838         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14839         (UNSPEC_P8V_MTVSRD): Likewise.
14840         (UNSPEC_P8V_XXPERMDI): Likewise.
14841         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14842         (UNSPEC_FUSION_GPR): Likewise.
14843         (FMOVE128_GPR): New iterator for direct move.
14844         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14845         (f32_sv): Likewise.
14846         (f32_dm): Likewise.
14847         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14848         loads and direct move instructions.
14849         (zero_extendsidi2_lfiwzx): Likewise.
14850         (extendsidi2_lfiwax): Likewise.
14851         (extendsidi2_nocell): Likewise.
14852         (floatsi<mode>2_lfiwax): Likewise.
14853         (lfiwax): Likewise.
14854         (floatunssi<mode>2_lfiwzx): Likewise.
14855         (lfiwzx): Likewise.
14856         (fix_trunc<mode>_stfiwx): Likewise.
14857         (fixuns_trunc<mode>_stfiwx): Likewise.
14858         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14859         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14860         (parity<mode>2_cmpb): Set length/type attr.
14861         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14862         for 'mr.' to fast_compare.
14863         (bpermd_<mode>): Change type attr to popcnt.
14864         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14865         (p8_mtvsrwz_1): Likewise.
14866         (p8_mtvsrwz_2): Likewise.
14867         (reload_fpr_from_gpr<mode>): Likewise.
14868         (p8_mtvsrd_1): Likewise.
14869         (p8_mtvsrd_2): Likewise.
14870         (p8_xxpermdi_<mode>): Likewise.
14871         (reload_vsx_from_gpr<mode>): Likewise.
14872         (reload_vsx_from_gprsf): Likewise.
14873         (p8_mfvsrd_3_<mode>): LIkewise.
14874         (reload_gpr_from_vsx<mode>): Likewise.
14875         (reload_gpr_from_vsxsf): Likewise.
14876         (p8_mfvsrd_4_disf): Likewise.
14877         (multi-word GPR splits): Do not split direct moves or quad memory
14878         operations.
14880 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14882         * tree-into-ssa.c (interesting_blocks): Make static.
14884 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14886         PR target/56564
14887         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14888         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14889         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14890         (get_variable_align): New function.
14891         (get_variable_section, emit_bss, emit_common,
14892         assemble_variable_contents, place_block_symbol): Use
14893         get_variable_align instead of DECL_ALIGN.
14894         (assemble_noswitch_variable): Add align argument, use it
14895         instead of DECL_ALIGN.
14896         (assemble_variable): Adjust caller.  Use get_variable_align
14897         instead of DECL_ALIGN.
14898         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14899         caller.
14900         (DATA_ABI_ALIGNMENT): Define.
14901         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14902         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14903         opt is false, only return the psABI mandated alignment increase.
14904         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14905         (DATA_ABI_ALIGNMENT): ... this.
14906         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14907         (DATA_ABI_ALIGNMENT): ... this.
14908         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14909         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14910         (DATA_ABI_ALIGNMENT): ... this.
14911         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14912         * doc/tm.texi: Regenerated.
14914 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14916         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14917         cmp_code to construct REG_EQUAL note.
14919 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14921         PR target/57568
14922         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14923         that operands[2] doesn't overlap with operands[0].
14925 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14926             Jan Hubicka  <jh@suse.cz>
14928         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14929         hack to mark symbols as used.
14931 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14933         PR rtl-optimization/57559
14934         * lra-constraints.c (process_alt_operands): Don't discourage
14935         memory with known offset for offsetable memory constraint.
14936         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14938 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14940         * varasm.c (struct oc_local_state): Reorder fields.
14941         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14942         and adjust accordingly.
14943         (output_constructor): Reorder initialization code and adjust call to
14944         output_constructor_bitfield.
14946 2013-06-07  Jan Hubicka  <jh@suse.cz>
14948         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14950 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14952         * tree-object-size.c (unknown): Make const.
14954 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14956         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14957         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14958         for last alternative in the cpu_facility attribute.
14960 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14962         PR target/56315
14963         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14964         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14965         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14966         * config/arm/constraints.md (Dg): New constraint.
14967         * config/arm/neon.md (xordi3_neon): Remove.
14968         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14969         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14971 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14973         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14974         Clean up alternatives.
14976 2013-06-07  Alan Modra  <amodra@gmail.com>
14978         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14979         va_list_gpr_size.
14981 2013-06-07  Alan Modra  <amodra@gmail.com>
14983         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14985 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14987         * config/arm/constraints.md (Df): New constraint.
14988         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14989         Correct length attribute for last two alternatives.
14991 2013-06-07  Alan Modra  <amodra@gmail.com>
14993         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14994         override user -mfp-in-toc.
14995         (offsettable_ok_by_alignment): Consider just the current access
14996         rather than the whole object, unless BLKmode.  Handle
14997         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14998         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14999         for -mcmodel=medium.
15000         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15001         override user -mfp-in-toc or -msum-in-toc.  Default to
15002         -mno-fp-in-toc for -mcmodel=medium.
15004 2013-06-06  DJ Delorie  <dj@redhat.com>
15006         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15007         TARGET_VALID_POINTER_MODE.
15009 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15010             Pat Haugen  <pthaugen@us.ibm.com>
15011             Peter Bergner  <bergner@vnet.ibm.com>
15013         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15014         Document new power8 builtins.
15016         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15017         condition code register, to allow 128-bit logical operations to be
15018         done in the VSX or GPR registers.
15019         (nor<mode>3): Use the canonical form for nor.
15020         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15021         vclz*, and vpopcnt* vector instructions.
15022         (nand<mode>3): Likewise.
15023         (orc<mode>3): Likewise.
15024         (clz<mode>2): LIkewise.
15025         (popcount<mode>2): Likewise.
15027         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15028         that only the GPRs are recognized.
15030         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15031         support for new power8 builtins.
15033         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15034         builtin functions.
15035         (xscvdpspn): Likewise.
15036         (vclz): Likewise.
15037         (vclzb): Likewise.
15038         (vclzh): Likewise.
15039         (vclzw): Likewise.
15040         (vclzd): Likewise.
15041         (vpopcnt): Likewise.
15042         (vpopcntb): Likewise.
15043         (vpopcnth): Likewise.
15044         (vpopcntw): Likewise.
15045         (vpopcntd): Likewise.
15046         (vgbbd): Likewise.
15047         (vmrgew): Likewise.
15048         (vmrgow): Likewise.
15049         (eqv): Likewise.
15050         (eqv_v16qi3): Likewise.
15051         (eqv_v8hi3): Likewise.
15052         (eqv_v4si3): Likewise.
15053         (eqv_v2di3): Likewise.
15054         (eqv_v4sf3): Likewise.
15055         (eqv_v2df3): Likewise.
15056         (nand): Likewise.
15057         (nand_v16qi3): Likewise.
15058         (nand_v8hi3): Likewise.
15059         (nand_v4si3): Likewise.
15060         (nand_v2di3): Likewise.
15061         (nand_v4sf3): Likewise.
15062         (nand_v2df3): Likewise.
15063         (orc): Likewise.
15064         (orc_v16qi3): Likewise.
15065         (orc_v8hi3): Likewise.
15066         (orc_v4si3): Likewise.
15067         (orc_v2di3): Likewise.
15068         (orc_v4sf3): Likewise.
15069         (orc_v2df3): Likewise.
15071         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15072         allow power8 quad mode in 64-bit.
15073         (rs6000_builtin_vectorized_function): Add support to vectorize
15074         ISA 2.07 count leading zeros, population count builtins.
15075         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15076         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15077         (builtin_function_type): Add vgbbd builtin function which takes an
15078         unsigned argument.
15079         (altivec_expand_vec_perm_const): Add support for new power8 merge
15080         instructions.
15082         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15083         that does not include TImdoe for use with 32-bit.
15084         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15085         instructions.
15086         (UNSPEC_VSX_CVDPSPN): Likewise.
15087         (vsx_xscvdpspn): Likewise.
15088         (vsx_xscvspdpn): Likewise.
15089         (vsx_xscvdpspn_scalar): Likewise.
15090         (vsx_xscvspdpn_directmove): Likewise.
15091         (vsx_and<mode>3): Split logical operations into 32-bit and
15092         64-bit. Add support to do logical operations on TImode as well as
15093         VSX vector types.  Allow logical operations to be done in either
15094         VSX registers or in general purpose registers in 64-bit mode.  Add
15095         splitters if GPRs were used. For AND, add clobber of CCmode to
15096         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15097         encoding.
15098         (vsx_and<mode>3_32bit): Likewise.
15099         (vsx_and<mode>3_64bit): Likewise.
15100         (vsx_ior<mode>3): Likewise.
15101         (vsx_ior<mode>3_32bit): Likewise.
15102         (vsx_ior<mode>3_64bit): Likewise.
15103         (vsx_xor<mode>3): Likewise.
15104         (vsx_xor<mode>3_32bit): Likewise.
15105         (vsx_xor<mode>3_64bit): Likewise.
15106         (vsx_one_cmpl<mode>2): Likewise.
15107         (vsx_one_cmpl<mode>2_32bit): Likewise.
15108         (vsx_one_cmpl<mode>2_64bit): Likewise.
15109         (vsx_nor<mode>3): Likewise.
15110         (vsx_nor<mode>3_32bit): Likewise.
15111         (vsx_nor<mode>3_64bit): Likewise.
15112         (vsx_andc<mode>3): Likewise.
15113         (vsx_andc<mode>3_32bit): Likewise.
15114         (vsx_andc<mode>3_64bit): Likewise.
15115         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15116         and xxlorc instructions.
15117         (vsx_eqv<mode>3_64bit): Likewise.
15118         (vsx_nand<mode>3_32bit): Likewise.
15119         (vsx_nand<mode>3_64bit): Likewise.
15120         (vsx_orc<mode>3_32bit): Likewise.
15121         (vsx_orc<mode>3_64bit): Likewise.
15123         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15125         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15126         instruction.
15127         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15128         (p8_vmrgow): Likewise.
15129         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15130         GPRs to be split under VSX.
15131         (p8v_clz<mode>2): Add power8 count leading zero support.
15132         (p8v_popcount<mode>2): Add power8 population count support.
15133         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15134         support.
15136         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15137         instruction.
15139         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15140         builtin functions.
15141         (vec_nand): Likewise.
15142         (vec_vclz): Likewise.
15143         (vec_vclzb): Likewise.
15144         (vec_vclzd): Likewise.
15145         (vec_vclzh): Likewise.
15146         (vec_vclzw): Likewise.
15147         (vec_vgbbd): Likewise.
15148         (vec_vmrgew): Likewise.
15149         (vec_vmrgow): Likewise.
15150         (vec_vpopcnt): Likewise.
15151         (vec_vpopcntb): Likewise.
15152         (vec_vpopcntd): Likewise.
15153         (vec_vpopcnth): Likewise.
15154         (vec_vpopcntw): Likewise.
15156 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15158         PR rtl-optimization/57468
15159         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15160         spilled pseudos.
15162 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15164         PR rtl-optimization/57459
15165         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15166         type when setting live regs.
15168 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15170         * config/s390/s390.opt (mlra): New option.
15171         * config/s390/s390.c (s390_decompose_address): Check displacement
15172         for all registers for LRA.
15173         (s390_secondary_reload): Don't used secondary reloads for LRA.
15174         (s390_lra_p): New function.
15175         (TARGET_LRA_P): Define.
15176         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15177         of attribute cpu_facility to zarch for the last alternative.
15178         (*cmpmem_short): Ditto.
15180 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15182         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15183         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15184         (arm_expand_prologue): Likewise.
15186 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15188         PR c++/53743
15189         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15190         as this is now done by redirect_edge_and_branch_force.
15191         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15192         barriers, and fix interaction with splitting.
15193         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15194         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15195         reflect changes made in the routine.
15196         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15197         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15198         since this is called in cfglayout mode, and replace partition fixup
15199         with assert as that is now done by force_nonfallthru_and_redirect.
15200         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15201         already be marked with region crossing note.
15202         (insert_section_boundary_note): Make non-static, gate on flag
15203         has_bb_partition, rewrite to also check for multiple partitions.
15204         (rest_of_handle_reorder_blocks): Remove call to
15205         insert_section_boundary_note, now done later during free_cfg.
15206         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15207         * bb-reorder.h (insert_section_boundary_note): Declare.
15208         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15209         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15210         invoke insert_section_boundary_note.
15211         (try_redirect_by_replacing_jump): Remove unnecessary
15212         check for region crossing note.
15213         (fixup_partition_crossing): New function.
15214         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15215         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15216         in non-cfglayout mode.
15217         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15218         remove old code that tried to do this. Emit barrier correctly
15219         when we are in cfglayout mode.
15220         (last_bb_in_partition): New function.
15221         (rtl_split_edge): Correctly fixup partition boundaries.
15222         (commit_one_edge_insertion): Remove old code that tried to
15223         fixup region crossing edge since this is now handled in
15224         split_block, and set up insertion point correctly since
15225         block may now end in a jump.
15226         (verify_hot_cold_block_grouping): Guard against checking when not in
15227         linearized RTL mode.
15228         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15229         notes.
15230         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15231         rtl_verify_flow_info, so not called in cfglayout mode.
15232         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15233         (fixup_reorder_chain): Remove old code that attempted to fixup region
15234         crossing note as this is now handled in force_nonfallthru_and_redirect.
15235         (duplicate_insn_chain): Don't duplicate switch section notes.
15236         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15237         note.
15238         * basic-block.h (emit_barrier_after_bb): Declare.
15240 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15242         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15243         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15244         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15246 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15248         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15249         where appropriate.
15250         * config/arm/ldmstm.md: Regenerate.
15252 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15254         * config/arm/sync.md (atomic_loaddi_1):
15255         Disable predication for arm_restrict_it.
15256         (arm_load_exclusive<mode>): Likewise.
15257         (arm_load_exclusivesi): Likewise.
15258         (arm_load_exclusivedi): Likewise.
15259         (arm_load_acquire_exclusive<mode>): Likewise.
15260         (arm_load_acquire_exclusivesi): Likewise.
15261         (arm_load_acquire_exclusivedi): Likewise.
15262         (arm_store_exclusive<mode>): Likewise.
15263         (arm_store_exclusive<mode>): Likewise.
15264         (arm_store_release_exclusivedi): Likewise.
15265         (arm_store_release_exclusive<mode>): Likewise.
15267 2013-06-06  Richard Biener  <rguenther@suse.de>
15269         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15270         after LTO_null.
15271         (lto_tag_is_tree_code_p): Adjust.
15272         (lto_tag_is_gimple_code_p): Likewise.
15273         (lto_gimple_code_to_tag): Likewise.
15274         (lto_tag_to_gimple_code): Likewise.
15275         (lto_tree_code_to_tag): Likewise.
15276         (lto_tag_to_tree_code): Likewise.
15277         * data-streamer.h (streamer_write_hwi_in_range): Use
15278         uhwi streaming to stream the normalized range.
15279         (streamer_read_hwi_in_range): Likewise.
15281 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15283         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15284         (predicable_short_it): Likewise.
15285         (predicated): Likewise.
15286         (enabled): Handle above.
15287         (define_cond_exec): Set predicated attribute to yes.
15289 2013-06-05  Mike Stump  <mikestump@comcast.net>
15291         * gdbinit.in (__FUNCTION__): Add.
15293 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15295         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15296         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15298 2013-06-05  Jan Hubicka  <jh@suse.cz>
15300         * varasm.c (mark_decl_referenced): Revert the removal until targets
15301         are fixed.
15303 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15305         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15306         instead of mark_decl_referenced.
15308 2013-06-05  Jan Hubicka  <jh@suse.cz>
15310         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15311         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15312         and symtab_used_from_object_file_p.
15313         (cgraph_make_node_local_1): Clear forced_by_abi.
15314         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15315         * cgraph.h (symtab_node_base): Add forced_by_abi.
15316         (decide_is_variable_needed): Remove.
15317         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15318         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15319         (decide_is_symbol_needed): ... this one; handle symbols in general;
15320         always analyze virtuals; honnor forced_by_abi.
15321         (cgraph_finalize_function): Update.
15322         (varpool_finalize_decl): Update.
15323         (symbol_defined_and_needed): Remove.
15324         (analyze_functions): Update.
15325         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15326         output_refs, input_overwrite_node): Handle forced_by_abi.
15327         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15328         (address_taken_from_non_vtable_p): ... this one.
15329         (comdat_can_be_unshared_p_1): New function.
15330         (cgraph_comdat_can_be_unshared_p): Rename to ...
15331         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15332         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15333         (function_and_variable_visibility): Clear forced_by_abi as needed.
15334         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15335         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15336         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15337         * varpool.c (decide_is_variable_needed): Remove.
15339 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15341         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15342         (arm_option_override): Override arm_restrict_it where appropriate.
15343         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15344         * config/arm/arm.opt (mrestrict-it): New command-line option.
15345         * doc/invoke.texi: Document -mrestrict-it.
15347 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15349         * tsan.c (tsan_atomic_table): Make const.
15351 2013-06-05  Richard Biener  <rguenther@suse.de>
15353         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15354         index associated with the tree we are supposed to replace.
15355         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15356         TREE_ASM_WRITTEN as zero for everything but SSA names.
15358 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15360         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15362 2013-06-04  Jan Hubicka  <jh@suse.cz>
15364         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15365         (input_node, input_varpool_node): Handle correctly external same
15366         body aliases.
15367         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15368         nodes at ltrans stage.
15370 2013-06-04  Jan Hubicka  <jh@suse.cz>
15372         * ipa-inline.c (update_caller_keys): Fix availability test.
15373         (update_callee_keys): Likewise.
15374         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15375         to follow ELF standard.
15377 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15379         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15380         (mips64r5900el-*-elf*): New configurations.
15381         * config/mips/mips-cpus.def (r5900): New processor.
15382         * config/mips/mips-tables.opt: Regenerate.
15383         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15384         (mips_issue_rate): Handle PROCESSOR_R5900.
15385         (mips_reorg_process_insns): Force reorder mode for the R5900.
15386         * config/mips/mips.h (TARGET_MIPS5900): Define.
15387         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15388         TARGET_MIPS5900.
15389         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15390         TARGET_MIPS5900.
15391         * config/mips/mips.md (processor): Add r5900.
15392         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15394 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15396         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15397         into function to generate MOVI instruction.
15398         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15399         (aarch64_preferred_simd_mode): Turn into wrapper.
15400         (aarch64_output_scalar_simd_mov_immediate): New function.
15401         * config/aarch64/aarch64-protos.h: Add prototype for above.
15403 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15405         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15406         element_char member.
15407         (sizetochar): Return signed char.
15408         (aarch64_simd_valid_immediate): Remove elchar and other
15409         unnecessary variables.
15410         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15411         Calculate element_char as required.
15412         * config/aarch64/aarch64-protos.h: Update and move prototype
15413         for aarch64_output_simd_mov_immediate.
15414         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15415         Update arguments.
15417 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15419         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15420         information completed by aarch64_simd_valid_immediate.
15421         (aarch64_legitimate_constant_p): Update arguments.
15422         (aarch64_simd_valid_immediate): Work with struct rather than many
15423         pointers.
15424         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15425         (aarch64_simd_make_constant): Update arguments.
15426         (aarch64_output_simd_mov_immediate): Work with struct rather than
15427         many pointers.  Output immediate directly rather than as operand.
15428         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15429         Update prototype.
15430         * config/aarch64/constraints.md (Dn): Update arguments.
15432 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15434         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15435         longer static.
15436         (aarch64_simd_immediate_valid_for_move): Remove.
15437         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15438         (aarch64_simd_make_constant): Update call.
15439         (aarch64_output_simd_mov_immediate): Update call.
15440         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15441         Add prototype.
15442         * config/aarch64/constraints.md (Dn): Update call.
15444 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15446         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15447         return type to bool for prototype.
15448         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15449         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15450         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15452 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15454         * config/mips/mips.opt (meva): New.
15455         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15456         (ASM_SPEC): Handle -meva.
15457         * doc/invoke.texi (meva):  Document.
15459 2013-06-04  Alan Modra  <amodra@gmail.com>
15461         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15462         constant output.
15464 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15466         * rtl.def: Add extra fourth optional field to define_cond_exec.
15467         * gensupport.c (process_one_cond_exec): Process attributes from
15468         define_cond_exec.
15469         * doc/md.texi: Document fourth field in define_cond_exec.
15471 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15473         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15474         out the processing order as in store_bit_field_1.
15476 2013-06-04  Jan Hubicka  <jh@suse.cz>
15478         PR middle-end/57500
15479         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15480         non-VAR_DECL node if it does not exist yet.
15482 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15484         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15485         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15486         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15487         target_cpu_default setting.
15489 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15491         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15492         default to -fopt-info=optimized instead of all.
15493         * doc/invoke.texi: Ditto.
15494         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15495         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15496         (execute_vect_slp): Emit BB vectorization success under
15497         MSG_OPTIMIZED_LOCATIONS.
15498         * tree-vect-slp.c (vect_slp_transform_bb): Change
15499         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15500         * tree-vect-loop.c (vect_transform_loop): Ditto.
15502 2013-06-03  Jason Merrill  <jason@redhat.com>
15504         PR c++/57415
15505         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15506         Use TARGET_EXPR for C++.
15508 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15510         PR rtl-optimization/57268
15511         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15512         if DEBUG_INSN_P (insn).
15514         Reapply
15515         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15517         PR rtl-optimization/57268
15518         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15519         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15521 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15523         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15524         (ix86_avoid_lea_for_addr): Likewise.
15525         (exact_dependency_1): Likewise.
15526         (ix86_adjust_cost): Likewise.
15527         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15528         removed.
15529         (do_reorder_for_imul): Fix typo, formatting and
15530         !reload_completed check removed.
15531         (ix86_sched_reorder): Fix typo and formatting.
15532         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15533         list.
15535 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15537         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15539 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15541         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15542         <STRING_CST>: Likewise.
15543         <VECTOR_CST>: Likewise.
15545 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15546             Mikael Morin  <mikael@gcc.gnu.org>
15548         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15549         * config.in: Regenerated.
15550         * configure: Regenerated.
15552 2013-06-01  Jan Hubicka  <jh@suse.cz>
15554         PR middle-end/57366
15555         * cgraphunit.c (compile): When weakref is not supported,
15556         set up transparent aliases before final output pass.
15557         * varasm.c (assemble_alias): Do not try to do it here.
15559 2013-06-01  Jan Hubicka  <jh@suse.cz>
15561         PR middle-end/57467
15562         * passes.c (for_per_function): Skip unanalyzed functions.
15564 2013-06-01  Jan Hubicka  <jh@suse.cz>
15566         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15567         (lto_symtab_merge_symbols_1): ... this one.
15568         (lto_symtab_merge_cgraph_nodes): Rename to ...
15569         (lto_symtab_merge_symbols): ... this one; simplify.
15570         * cgraph.c (same_body_aliases_done): Rename to ...
15571         (cpp_implicit_aliases_done): ... this one.
15572         (cgraph_create_function_alias): Update.
15573         (cgraph_same_body_alias): Update.
15574         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15575         (verify_edge_corresponds_to_fndecl): Simplify.
15576         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15577         (cgraph_node): Remove same_body_alias.
15578         (varpool_node): Remove alias_of and extra_name_alias.
15579         (same_body_aliases_done): Rename to ..
15580         (cpp_implicit_aliases_done): ... this one.
15581         (symtab_alias_ultimate_target): Add default parameter.
15582         (symtab_resolve_alias): New function.
15583         (fixup_same_cpp_alias_visibility): Declare.
15584         (cgraph_function_node): Add default parameter.
15585         (cgraph_node_asm_name): Likewise.
15586         (cgraph_function_or_thunk_node): Add default parameter; do
15587         not ICE when it is NULL.
15588         (varpool_variable_node): Likewise.
15589         * tree-emutls.c (create_emultls_var): Update.
15590         (ipa_lower_emutls): Update.
15591         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15592         (cgraph_reset_node): Reset alias info.
15593         (cgraph_finalize_function): Update.
15594         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15595         (analyze_function): Simplify.
15596         (cgraph_process_same_body_aliases): Simplify.
15597         (analyze_functions): Fixup same body aliases.
15598         (handle_alias_pairs): Simplify.
15599         (assemble_thunk): Update.
15600         (assemble_thunks_and_aliases): Update.
15601         (output_weakrefs): Rewrite.
15602         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15603         (lto_output_varpool_node): Likewise.
15604         (compute_ltrans_boundary): Remve assert.
15605         (get_alias_symbol): New functoin.
15606         (input_node): Rewrite alias handling.
15607         (input_varpool_node): Likewise.
15608         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15609         * ipa.c (process_references): Handle weakrefs correctly.
15610         (symtab_remove_unreachable_nodes): Likewise.
15611         * trans-mem.c (get_cg_data): Update.
15612         (ipa_tm_create_version_alias): Update.
15613         (ipa_tm_execute): Update.
15614         * symtab.c (dump_symtab_base): Dump aliases.
15615         (verify_symtab_base): Verify aliases.
15616         (symtab_node_availability): New function.
15617         (symtab_alias_ultimate_target): Simplify.
15618         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15619         handle all the fixup cases.
15620         (symtab_resolve_alias): New function.
15621         * passes.c (ipa_write_summaries): Handle weakrefs.
15622         * varpool.c (varpool_analyze_node): Simplify.
15623         (assemble_aliases): Update.
15624         (varpool_create_variable_alias): Simplify.
15625         (varpool_extra_name_alias): Simplify.
15626         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15627         (lto_symtab_merge_symbols): ... this one.
15629 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15631         Revert
15632         PR rtl-optimization/57268
15633         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15634         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15636 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15638         Partially reverted:
15639         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15641         PR middle-end/57073
15642         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15643         further up.
15645 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15647         PR rtl-optimization/57268
15648         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15649         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15651 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15653         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15654         unordered comparison operators when -fno-trapping-math is in effect
15655         on the e500.
15656         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15657         and implement unordered comparison operators properly on the e500.
15659 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15661         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15662         for constant scalar integers.
15663         (simplify_relational_operation_1): Likewise.
15665 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15667         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15668         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15669         Fix comment.
15671 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15672             Igor Zamyatin  <igor.zamyatin@intel.com>
15674         Silvermont (SLM) architecture performance tuning.
15675         * config/i386/i386.h (enum ix86_tune_indices): Add
15676         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15677         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15679         * config/i386/i386.c (initial_ix86_tune_features)
15680         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15681         (ix86_lea_outperforms): Handle Silvermont tuning.
15682         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15683         call.
15684         (ix86_use_lea_for_mov): Likewise.
15685         (ix86_avoid_lea_for_addr): Likewise.
15686         (ix86_lea_for_add_ok): Likewise.
15687         (exact_dependency_1): New function.
15688         (exact_store_load_dependency): Likewise.
15689         (ix86_adjust_cost): Handle Silvermont tuning.
15690         (do_reoder_for_imul): Likewise.
15691         (swap_top_of_ready_list): New function.
15692         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15694         * config/i386/i386.md (peepholes that split memory operand in fp
15695         converts): New.
15697 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15699         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15700         Remove un-necessary braces.
15702 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15704         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15705         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15707 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15709         PR middle-end/57073
15710         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15712 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15714         PR target/56315
15715         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15716         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15717         * config/arm/neon.md (iordi3_neon): Remove.
15718         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15719         * config/arm/predicates.md (imm_for_neon_logic_operand):
15720         Move to earlier in the file.
15721         (neon_logic_op2): Likewise.
15722         (arm_iordi_operand_neon): New predicate.
15724 2013-05-31  Richard Biener  <rguenther@suse.de>
15726         PR tree-optimization/57478
15727         PR tree-optimization/57453
15728         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15729         are life as well.
15731 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15733         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15734         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15736 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15737             Thomas Koenig  <tkoenig@gcc.gnu.org>
15739         PR middle-end/57073
15740         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15741         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15743 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15745         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15747 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15749         * target.def (register_usage_leveling_p): New hook.
15750         * targhooks.c (default_register_usage_leveling_p): New.
15751         * targhooks.h (default_register_usage_leveling_p): New prototype.
15752         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15753         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15754         * doc/tm.texi: Update.
15755         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15757 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15759         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15760         (*insv_reg<mode>): New define_insn.
15762 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15764         PR rtl-optimization/57439
15765         * postreload.c (move2add_valid_value_p): Check that we have
15766         a zero subreg_regno_offset when accessing the register in
15767         the requested mode.
15769 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15770             Igor Zamyatin  <igor.zamyatin@intel.com>
15772         Silvermont (SLM) architecture pipeline model, tuning and
15773         insn selection.
15774         * config.gcc: Add slm config options and target.
15776         * config/i386/slm.md: New.
15778         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15780         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15781         PROCESSOR_SLM.
15782         (ix86_target_macros_internal): Likewise.
15784         * config/i386/i386.c (slm_cost): New cost.
15785         (m_SLM): New macro flag.
15786         (initial_ix86_tune_features): Set m_SLM.
15787         (x86_accumulate_outgoing_args): Likewise.
15788         (x86_arch_always_fancy_math_387): Likewise.
15789         (processor_target_table): Add slm cost.
15790         (cpu_names): Add slm cpu name.
15791         (x86_option_override_internal): Set SLM ISA.
15792         (ix86_issue_rate): New case PROCESSOR_SLM.
15793         (ia32_multipass_dfa_lookahead): Likewise.
15794         (fold_builtin_cpu): Add slm.
15796         * config/i386/i386.h (TARGET_SLM): New target macro.
15797         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15798         (processor_type): Add PROCESSOR_SLM.
15800         * config/i386/i386.md (cpu): Add new value "slm".
15801         (slm.md): Include slm.md.
15803 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15804             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15806         * config/arm/arm-protos.h: Add and update function protos.
15807         * config/arm/arm.c (use_simple_return_p): New added.
15808         (thumb2_expand_return): Check simple_return flag.
15809         * config/arm/arm.md: Add simple_return and conditional simple_return.
15810         * config/arm/iterators.md: Add iterator for return and simple_return.
15812 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15814         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15815         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15816         (arm_emit_vfp_multi_reg_pop): Likewise.
15817         (thumb2_emit_ldrd_pop): Likewise.
15818         (arm_expand_epilogue): Add misc REG_CFA notes.
15819         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15821 2013-05-29  Lawrence Crowl  <crowl@google.com>
15823         * config/arm/t-arm: Update for below.
15825         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15826         Change type to hash_table.  Update dependent calls and types.
15828         * config/i386/t-cygming: Update for below.
15830         * config/i386/t-interix: Update for below.
15832         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15833         Change type to hash_table.  Update dependent calls and types.
15834         (i386_find_on_wrapper_list::wrappers): Likewise.
15836         * config/ia64/t-ia64: Update for below.
15838         * config/ia64/ia64.c (bundle_state_table):
15839         Change type to hash_table.  Update dependent calls and types.
15841         * config/mips/mips.c (mips_reorg_process_insns::htab):
15842         Change type to hash_table.  Update dependent calls and types.
15844         * config/sol2.c (solaris_comdat_htab):
15845         Change type to hash_table.  Update dependent calls and types.
15847         * config/t-sol2: Update for above.
15849 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15851         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15852         functions are not yet marked as defined.
15854 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15855             Pat Haugen  <pthaugen@us.ibm.com>
15856             Peter Bergner  <bergner@vnet.ibm.com>
15858         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15859         instructions.
15860         (VEC_A): Likewise.
15861         (VEC_C): Likewise.
15862         (vrotl<mode>3): Likewise.
15863         (vashl<mode>3): Likewise.
15864         (vlshr<mode>3): Likewise.
15865         (vashr<mode>3): Likewise.
15867         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15868         support for power8 V2DI builtins.
15870         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15871         power8 V2DI builtins.
15872         (vupkhsw): Likewise.
15873         (vupklsw): Likewise.
15874         (vaddudm): Likewise.
15875         (vminsd): Likewise.
15876         (vmaxsd): Likewise.
15877         (vminud): Likewise.
15878         (vmaxud): Likewise.
15879         (vpkudum): Likewise.
15880         (vpksdss): Likewise.
15881         (vpkudus): Likewise.
15882         (vpksdus): Likewise.
15883         (vrld): Likewise.
15884         (vsld): Likewise.
15885         (vsrd): Likewise.
15886         (vsrad): Likewise.
15887         (vsubudm): Likewise.
15888         (vcmpequd): Likewise.
15889         (vcmpgtsd): Likewise.
15890         (vcmpgtud): Likewise.
15891         (vcmpequd_p): Likewise.
15892         (vcmpgtsd_p): Likewise.
15893         (vcmpgtud_p): Likewise.
15894         (vupkhsw): Likewise.
15895         (vupklsw): Likewise.
15896         (vaddudm): Likewise.
15897         (vmaxsd): Likewise.
15898         (vmaxud): Likewise.
15899         (vminsd): Likewise.
15900         (vminud): Likewise.
15901         (vpksdss): Likewise.
15902         (vpksdus): Likewise.
15903         (vpkudum): Likewise.
15904         (vpkudus): Likewise.
15905         (vrld): Likewise.
15906         (vsld): Likewise.
15907         (vsrad): Likewise.
15908         (vsrd): Likewise.
15909         (vsubudm): Likewise.
15911         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15912         support for power8 V2DI instructions.
15914         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15915         power8 V2DI instructions.  Combine pack and unpack insns to use an
15916         iterator for each mode.  Check whether a particular mode supports
15917         Altivec instructions instead of just checking TARGET_ALTIVEC.
15918         (UNSPEC_VPKUWUM): Likewise.
15919         (UNSPEC_VPKSHSS): Likewise.
15920         (UNSPEC_VPKSWSS): Likewise.
15921         (UNSPEC_VPKUHUS): Likewise.
15922         (UNSPEC_VPKSHUS): Likewise.
15923         (UNSPEC_VPKUWUS): Likewise.
15924         (UNSPEC_VPKSWUS): Likewise.
15925         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15926         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15927         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15928         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15929         (UNSPEC_VUPKHSB): Likewise.
15930         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15931         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15932         (UNSPEC_VUPKHSH): Likewise.
15933         (UNSPEC_VUPKLSB): Likewise.
15934         (UNSPEC_VUPKLSH): Likewise.
15935         (VI2): Likewise.
15936         (VI_char): Likewise.
15937         (VI_scalar): Likewise.
15938         (VI_unit): Likewise.
15939         (VP): Likewise.
15940         (VP_small): Likewise.
15941         (VP_small_lc): Likewise.
15942         (VU_char): Likewise.
15943         (add<mode>3): Likewise.
15944         (altivec_vaddcuw): Likewise.
15945         (altivec_vaddu<VI_char>s): Likewise.
15946         (altivec_vadds<VI_char>s): Likewise.
15947         (sub<mode>3): Likewise.
15948         (altivec_vsubcuw): Likewise.
15949         (altivec_vsubu<VI_char>s): Likewise.
15950         (altivec_vsubs<VI_char>s): Likewise.
15951         (altivec_vavgs<VI_char>): Likewise.
15952         (altivec_vcmpbfp): Likewise.
15953         (altivec_eq<mode>): Likewise.
15954         (altivec_gt<mode>): Likewise.
15955         (altivec_gtu<mode>): Likewise.
15956         (umax<mode>3): Likewise.
15957         (smax<mode>3): Likewise.
15958         (umin<mode>3): Likewise.
15959         (smin<mode>3): Likewise.
15960         (altivec_vpkuhum): Likewise.
15961         (altivec_vpkuwum): Likewise.
15962         (altivec_vpkshss): Likewise.
15963         (altivec_vpkswss): Likewise.
15964         (altivec_vpkuhus): Likewise.
15965         (altivec_vpkshus): Likewise.
15966         (altivec_vpkuwus): Likewise.
15967         (altivec_vpkswus): Likewise.
15968         (altivec_vpks<VI_char>ss): Likewise.
15969         (altivec_vpks<VI_char>us): Likewise.
15970         (altivec_vpku<VI_char>us): Likewise.
15971         (altivec_vpku<VI_char>um): Likewise.
15972         (altivec_vrl<VI_char>): Likewise.
15973         (altivec_vsl<VI_char>): Likewise.
15974         (altivec_vsr<VI_char>): Likewise.
15975         (altivec_vsra<VI_char>): Likewise.
15976         (altivec_vsldoi_<mode>): Likewise.
15977         (altivec_vupkhsb): Likewise.
15978         (altivec_vupkhs<VU_char>): Likewise.
15979         (altivec_vupkls<VU_char>): Likewise.
15980         (altivec_vupkhsh): Likewise.
15981         (altivec_vupklsb): Likewise.
15982         (altivec_vupklsh): Likewise.
15983         (altivec_vcmpequ<VI_char>_p): Likewise.
15984         (altivec_vcmpgts<VI_char>_p): Likewise.
15985         (altivec_vcmpgtu<VI_char>_p): Likewise.
15986         (abs<mode>2): Likewise.
15987         (vec_unpacks_hi_v16qi): Likewise.
15988         (vec_unpacks_hi_v8hi): Likewise.
15989         (vec_unpacks_lo_v16qi): Likewise.
15990         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15991         (vec_unpacks_lo_v8hi): Likewise.
15992         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15993         (vec_pack_trunc_v8h): Likewise.
15994         (vec_pack_trunc_v4si): Likewise.
15995         (vec_pack_trunc_<mode>): Likewise.
15997         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15998         V2DI builtins.
15999         (vec_vmaxsd): Likewise.
16000         (vec_vmaxud): Likewise.
16001         (vec_vminsd): Likewise.
16002         (vec_vminud): Likewise.
16003         (vec_vpksdss): Likewise.
16004         (vec_vpksdus): Likewise.
16005         (vec_vpkudum): Likewise.
16006         (vec_vpkudus): Likewise.
16007         (vec_vrld): Likewise.
16008         (vec_vsld): Likewise.
16009         (vec_vsrad): Likewise.
16010         (vec_vsrd): Likewise.
16011         (vec_vsubudm): Likewise.
16012         (vec_vupkhsw): Likewise.
16013         (vec_vupklsw): Likewise.
16015 2013-05-29  Jan Hubicka  <jh@suse.cz>
16017         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16018         flags; reorder rest of fields in more consistent way.
16019         (varpool_node): Remove analyzed, finalized and alias.
16020         (cgraph_ndoe): Likewise.
16021         (symtab_alias_ultimate_target): New function.
16022         (cgraph_function_node): Move offline.
16023         (cgraph_reset_node): Declare.
16024         (cgraph_comdat_can_be_unshared_p): Remove.
16025         (varpool_remove_initializer): Declare.
16026         (varpool_first_defined_variable, varpool_next_defined_variable
16027         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16028         (cgraph_function_with_gimple_body_p): Update.
16029         (varpool_all_refs_explicit_p): Update.
16030         (symtab_alias_target): New function.
16031         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16032         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16033         (cgraph_function_or_thunk_node): Simplify using
16034         symtab_alias_ultimate_target.
16035         (varpool_variable_node): Likewise.
16036         * cgraph.c (cgraph_create_function_alias): Update.
16037         (cgraph_add_thunk): Update.
16038         (cgraph_remove_node): Update.
16039         (dump_cgraph_node): Do not dump removed flags.
16040         (cgraph_function_body_availability): Update.
16041         (cgraph_propagate_frequency): Update.
16042         (verify_cgraph_node): Check sanity of local flag.
16043         (cgraph_function_node): Move here from cgraph.h; revamp for
16044         cgraph_function_or_thunk_node.
16045         * lto-symtab.c (lto_varpool_replace_node): Update.
16046         (lto_symtab_resolve_can_prevail_p): Update.
16047         (lto_symtab_merge_cgraph_nodes): Update.
16048         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16049         propagate_constants_accross_call, devirtualization_time_bonus,
16050         ipcp_propagate_stage): Update.
16051         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16052         * ipa-inline-transform.c (clone_inlined_nodes,
16053         preserve_function_body_p): Update.
16054         * ipa-reference.c (propagate): Update.
16055         (write_node_summary_p): Update.
16056         * toplev.c (wrapup_global_declaration_2): Update.
16057         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16058         (analyze_function) ... this one.
16059         (cgraph_process_new_functions): Update.
16060         (cgraph_reset_node): Export.
16061         (cgraph_finalize_function): Update.
16062         (cgraph_add_new_function): Update.
16063         (process_function_and_variable_attributes): Update.
16064         (varpool_finalize_decl): Update.
16065         (symbol_finalized): Remove.
16066         (symbol_finalized_and_needed): Rename to ...
16067         (symbol_defined_and_needed): ... update.
16068         (cgraph_analyze_functions): Update.
16069         (handle_alias_pairs): Update.
16070         (mark_functions_to_output): Update.
16071         (assemble_thunk): Update.
16072         (output_in_order): Update.
16073         (output_weakrefs): Update.
16074         (finalize_compilation_unit): Update.
16075         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16076         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16077         input_node, input_varpool_node): Update.
16078         * dbxout.c (dbxout_expand_expr): Update.
16079         * cgraphclones.c (cgraph_clone_node): Update.
16080         (cgraph_copy_node_for_versioning): Update.
16081         (cgraph_materialize_clone): Update.
16082         (cgraph_materialize_all_clones): Update.
16083         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16084         propagate_pure_const, propagate_nothrow): Update.
16085         * lto-streamer-out.c (lto_output, write_symbol): Update.
16086         * ipa-utils.c (ipa_reverse_postorder): Update.
16087         * ipa-inline.c (can_inline_edge_p): Update.
16088         (update_caller_keys, ipa_inline): Update.
16089         * dwarf2out.c (reference_to_unused,
16090         premark_types_used_by_global_vars_helper): Update.
16091         * tree-eh.c (tree_could_trap_p): Update.
16092         * ipa-split.c (consider_split, execute_split_functions): Update.
16093         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16094         has_addr_references_p): Update; move ahead in file for better
16095         readability.
16096         (process_references): Simplify.
16097         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16098         bodies are removed.
16099         (cgraph_comdat_can_be_unshared_p): Make static.
16100         (cgraph_externally_visible_p): Update.
16101         (varpool_externally_visible_p): Update.
16102         (function_and_variable_visibility): Update.
16103         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16104         ipa_tm_mark_force_output_node): Update.
16105         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16106         estimate_edge_devirt_benefit, inline_generate_summary,
16107         inline_write_summary): Update.
16108         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16109         * ipa-prop.c (ipa_compute_jump_functions): Update.
16110         (ipa_print_node_params, ipa_prop_read_section,
16111         ipa_update_after_lto_read, read_replacements_section): Update.
16112         * varasm.c (mark_decl_referenced): Update.
16113         (assemble_alias, dump_tm_clone_pairs): Update.
16114         * tree-inline.c (copy_bb): Update.
16115         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16116         Update.
16117         * symtab.c (dump_symtab_base): Print new flags.
16118         (verify_symtab_base): Verify new flags.
16119         (symtab_alias_ultimate_target): New function.
16120         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16121         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16122         Update.
16123         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16124         Update.
16125         * i386.c (ix86_get_function_versions_dispatcher,
16126         ix86_generate_version_dispatcher_body): Update.
16127         (fold_builtin_cpu): Use varpool_add_new_variable.
16128         * varpool.c (varpool_remove_initializer): Break out from ...
16129         (varpool_remove_node): ... this one.
16130         (dump_varpool_node, varpool_node_for_asm,
16131         cgraph_variable_initializer_availability, varpool_analyze_node,
16132         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16133         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16134         Update.
16136 2013-05-29  Jan Hubicka  <jh@suse.cz>
16138         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16140 2013-05-29  Easwaran Raman  <eraman@google.com>
16142         PR tree-optimization/57442
16143         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16144         when control exits the main loop.
16146 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16148         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16149         and RX600.
16150         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16151         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16152         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16153         option.
16155 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16157         PR tree-optimization/57441
16158         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16159         Don't limit size of incr_vec to number of candidates.
16161 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16163         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16164         and mips16 directories.
16165         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16166         (MULTILIB_DIRNAMES): Ditto.
16167         (MULTILIB_EXCEPTIONS): Add new exceptions.
16168         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16169         (MULTILIB_DIRNAMES): Ditto.
16170         (MULTILIB_EXCEPTIONS): Add new exceptions.
16172 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16173             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16175         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16176         SYMBOL_TINY_ABSOLUTE.
16177         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16178         SYMBOL_TINY_ABSOLUTE.
16179         (aarch64_expand_mov_immediate): Likewise.
16180         (aarch64_classify_symbol): Likewise.
16181         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16182         Permit SYMBOL_TINY_ABSOLUTE.
16183         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16185 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16186             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16188         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16189         Refactor if/switch.  Replace gcc_assert with if.
16191 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16193         * config/i386/i386.c (initial_ix86_tune_features): Enable
16194         FP Reassociation for AMD bdver1 and bdver2.
16196 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16198         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16199         and IMAGPART_EXPR do not occur within other handled_components.
16201 2013-05-29  Richard Biener  <rguenther@suse.de>
16203         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16204         access on whether the use is in the BB we currently try to
16205         vectorize.
16206         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16207         vectorize to vect_bb_slp_scalar_cost.
16209 2013-05-29  Richard Biener  <rguenther@suse.de>
16211         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16212         computing scalar cost offsetted by stmts that are kept live
16213         by scalar uses.
16214         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16215         for computation of scalar cost.
16217 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16219         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16221 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16223         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16224         array notation built-in reduction functions.
16225         * doc/passes.texi (Passes): Added documentation about changes done
16226         for Cilk Plus.
16227         * doc/invoke.texi (C Dialect Options): Added documentation about
16228         the -fcilkplus flag.
16229         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16230         (BUILTINS_DEF): Depend on cilkplus.def.
16231         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16232         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16233         * cilkplus.def: New file.
16235 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16237         PR rtl-optimization/57439
16238         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16240 2013-05-28  Easwaran Raman  <eraman@google.com>
16242         PR tree-optimization/57337
16243         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16244         (find_insert_point): Correctly identify the insertion point
16245         when two statements with the same UID is compared.
16247 2013-05-28  Richard Biener  <rguenther@suse.de>
16249         PR tree-optimization/56787
16250         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16251         from the list of data references.
16252         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16253         clobbers.
16254         (vect_analyze_loop_operations): Likewise.
16255         (vect_transform_loop): Remove clobbers.
16257 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16259         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16260         and REALPART_EXPRs have scalar type.
16262 2013-05-28  Richard Biener  <rguenther@suse.de>
16264         PR tree-optimization/57411
16265         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16266         virtual operands.
16267         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16268         virtual operand propagation.
16270 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16272         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16273         destination register for bmasksi_vis.
16274         (vector_init_bshuffle): Likewise.
16275         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16277 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16279         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16280         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16281         mode if the instruction isn't available in the original mode.
16282         * config/sparc/sparc.opt (mfix-ut699): New option.
16283         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16284         (divdf3): Turn into expander.
16285         (divdf3_nofix): New insn.
16286         (divdf3_fix): Likewise.
16287         (divsf3): Disable if -mfix-ut699.
16288         (sqrtdf2): Turn into expander.
16289         (sqrtdf2_nofix): New insn.
16290         (sqrtdf2_fix): Likewise.
16291         (sqrtsf2): Disable if -mfix-ut699.
16293 2013-05-27  Richard Biener  <rguenther@suse.de>
16295         PR middle-end/57412
16296         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16297         block for the new loop.
16299 2013-05-27  Richard Biener  <rguenther@suse.de>
16301         PR tree-optimization/57343
16302         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16303         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16304         (number_of_iterations_cond): Do not build the folded tree.
16306 2013-05-27  Richard Biener  <rguenther@suse.de>
16308         Revert
16309         PR middle-end/57381
16310         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16311         OEP_CONSTANT_ADDRESS_OF retained.
16313         PR tree-optimization/57417
16314         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16315         for unchanged base.
16316         (set_ssa_val_to): Compare addresses using
16317         get_addr_base_and_unit_offset.
16319 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16321         PR rtl-optimization/56833
16322         * postreload.c (move2add_record_mode): New function.
16323         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16324         (move2add_use_add2_insn): Use move2add_record_sym_value.
16325         (move2add_use_add3_insn): Likewise.
16326         (reload_cse_move2add): Use move2add_valid_value_p and
16327         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16328         affected regs by setting reg_mode to VOIDmode.
16329         (move2add_note_store): Don't pretend the inside of a SUBREG is
16330         the actual destination.  Invalidate single/leading registers by
16331         setting reg_mode to VOIDmode.
16332         Use move2add_record_sym_value, move2add_valid_value_p and
16333         move2add_record_mode.
16335 2013-05-27  Richard Biener  <rguenther@suse.de>
16337         PR tree-optimization/57396
16338         * tree-affine.c (double_int_constant_multiple_p): Properly
16339         return false for val == 0 and div != 0.
16341 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16343         * config/mips/mips.h: Use #elif in preprocessor conditions.
16345 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16347         PR target/53916
16348         * config/mips/constraints.md (kl): New constraint.
16349         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16350         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16351         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16352         from LO for MIPS16.
16353         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16355 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16357         PR target/55777
16358         * config/mips/mips.c (mips_can_inline_p): New function.
16359         (TARGET_CAN_INLINE_P): Define.
16361 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16363         * sched-int.h (ds_t, dw_t): Make unsigned int.
16364         Fix documentation that describes how all the ds_t bits are used.
16365         Reserve the last bit for delayed-branch scheduling.
16366         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16367         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16368         (gen_dep_weak_1): Remove prototype.
16369         * sched-deps.c (get_dep_weak_1): Make static.
16370         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16371         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16372         * doc/tm.texi: Regenerate.
16373         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16375 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16377         PR debug/56950
16378         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16380 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16381             Sandra Loosemore  <sandra@codesourcery.com>
16383         * config.gcc (powerpc-*): Allow native for with-cpu.
16385 2013-05-24  Jeff Law  <law@redhat.com>
16387         PR tree-optimization/57124
16388         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16389         conversion feeding a condition if the range has an overflow
16390         if -fstrict-overflow.  Add warnings for when we do make the
16391         transformation.
16393 2013-05-24  Dehao Chen  <dehao@google.com>
16395         * tree-cfg.c (locus_discrim_map): Fix the typo.
16396         (locus_discrim_hasher): Likewise.
16397         (locus_discrim_hasher::hash): Likewise.
16398         (locus_discrim_hasher::equal): Likewise.
16400 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16402         PR tree-optimization/57294
16403         * cgraph.h (ipa_record_stmt_references): Declare.
16404         * cgraphbuild.c (ipa_record_stmt_references): New function.
16405         (build_cgraph_edges): Use ipa_record_stmt_references.
16406         (rebuild_cgraph_edges): Likewise.
16407         (cgraph_rebuild_references): Likewise.
16408         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16409         associated with the old statement and build references from the
16410         newly built statements.
16411         * ipa-ref.c (ipa_remove_stmt_references): New function.
16412         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16414 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16416         * lra-constraints.c (emit_spill_move): Use smaller mode for
16417         mem-mem moves.
16418         (check_and_process_move): Consider mem-reg moves for secondary
16419         too.
16420         (curr_insn_transform): Don't lose insns emitted before for
16421         secondary memory moves.
16422         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16423         reg set up in the current insn.
16425 2013-05-24  Dehao Chen  <dehao@google.com>
16427         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16428         hash function.
16429         (locus_descrim_hasher::equal): Likewise.
16430         (build_gimple_cfg): New discriminator assignment algorithm.
16431         (make_edges): Likewise.
16432         (next_discriminator_for_locus): Likewise.
16433         (same_line_p): Likewise.
16434         (assign_discriminators): Likewise.
16435         (make_cond_expr_edges): Likewise.
16436         (make_gimple_switch_edges): Likewise.
16437         (make_goto_expr_edges): Likewise.
16438         (make_gimple_asm_edges): Likewise.
16440 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16442         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16443         X format specifier to only display bottom 16 bits.
16444         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16445         immediate to match for operand 2, since it will be masked.
16447 2013-05-24  Richard Biener  <rguenther@suse.de>
16449         PR tree-optimization/57287
16450         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16451         all SSA names that occur in abnormal PHIs.
16453 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16455         PR tree-ssa/57385
16456         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16457         that index is not negative.
16459 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16461         PR rtl-optimization/55177
16462         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16463         (simplify_byte_swapping_operation): New.
16464         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16465         (simplify_relational_operation_1): Deal with BSWAP.
16467 2013-05-23  Richard Henderson  <rth@redhat.com>
16469         PR target/56742
16470         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16471         (ix86_reorg): Call it.
16473 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16475         PR target/57379
16476         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16477         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16478         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16480 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16482         PR debug/57351
16483         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16485 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16486             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16488         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16489         * config/aarch64/constraints.md (Usa): Remove.
16490         * doc/md.texi (AArch64 Usa): Remove.
16492 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16493             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16495         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16496         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16497         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16498         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16500 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16502         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16503         instruction (AdvSIMD).
16504         * config/aarch64/aarch64-builtins.c
16505         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16506         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16508 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16510         PR middle-end/57347
16511         * tree.h (contains_bitfld_component_ref_p): Declare.
16512         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16513         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16514         caller.
16515         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16516         not access a bit-field.  Assert all final offsets are byte-aligned.
16518 2013-05-23  Richard Biener  <rguenther@suse.de>
16520         PR tree-optimization/57380
16521         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16522         least one invariant or re-used load.
16523         * passes.c (init_optimization_passes): Move pass_phiprop before
16524         pass_forwprop.
16526 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16528         * config/aarch64/aarch64-simd.md
16529         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16531 2013-05-23  Richard Biener  <rguenther@suse.de>
16533         PR middle-end/57381
16534         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16535         OEP_CONSTANT_ADDRESS_OF retained.
16537 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16539         PR middle-end/57344
16540         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16541         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16543 2013-05-23  Richard Biener  <rguenther@suse.de>
16545         PR rtl-optimization/57341
16546         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16547         instead of true_dependence.
16549 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16551         * bb-reorder.c (branch_threshold): Make const.
16552         (exec_threshold): Ditto.
16554 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16555             Pat Haugen  <pthaugen@us.ibm.com>
16556             Peter Bergner  <bergner@vnet.ibm.com>
16558         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16559         documentation for the power8 crypto builtins.
16561         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16563         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16564         macros for defining power8 builtin functions.
16565         (BU_P8V_AV_2): Likewise.
16566         (BU_P8V_AV_P): Likewise.
16567         (BU_P8V_VSX_1): Likewise.
16568         (BU_P8V_OVERLOAD_1): Likewise.
16569         (BU_P8V_OVERLOAD_2): Likewise.
16570         (BU_CRYPTO_1): Likewise.
16571         (BU_CRYPTO_2): Likewise.
16572         (BU_CRYPTO_3): Likewise.
16573         (BU_CRYPTO_OVERLOAD_1): Likewise.
16574         (BU_CRYPTO_OVERLOAD_2): Likewise.
16575         (XSCVSPDP): Fix typo, point to the correct instruction.
16576         (VCIPHER): Add power8 crypto builtins.
16577         (VCIPHERLAST): Likewise.
16578         (VNCIPHER): Likewise.
16579         (VNCIPHERLAST): Likewise.
16580         (VPMSUMB): Likewise.
16581         (VPMSUMH): Likewise.
16582         (VPMSUMW): Likewise.
16583         (VPERMXOR_V2DI): Likewise.
16584         (VPERMXOR_V4SI: Likewise.
16585         (VPERMXOR_V8HI: Likewise.
16586         (VPERMXOR_V16QI: Likewise.
16587         (VSHASIGMAW): Likewise.
16588         (VSHASIGMAD): Likewise.
16589         (VPMSUM): Likewise.
16590         (VPERMXOR): Likewise.
16591         (VSHASIGMA): Likewise.
16593         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16594         __CRYPTO__ if the crypto instructions are available.
16595         (altivec_overloaded_builtins): Add support for overloaded power8
16596         builtins.
16598         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16599         support for power8 crypto builtins.
16600         (builtin_function_type): Likewise.
16601         (altivec_init_builtins): Add support for builtins that take vector
16602         long long (V2DI) arguments.
16604         * config/rs6000/crypto.md: New file, define power8 crypto
16605         instructions.
16607 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16608             Pat Haugen  <pthaugen@us.ibm.com>
16609             Peter Bergner  <bergner@vnet.ibm.com>
16611         * doc/invoke.texi (Option Summary): Add power8 options.
16612         (RS/6000 and PowerPC Options): Likewise.
16614         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16615         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16616         wm, wn, wr documentation.
16618         * config/rs6000/constraints.md (wm): New constraint for VSX
16619         registers if direct move instructions are enabled.
16620         (wn): New constraint for no registers.
16621         (wq): New constraint for quad word even GPR registers.
16622         (wr): New constraint if 64-bit instructions are enabled.
16623         (wv): New constraint if power8 vector instructions are enabled.
16624         (wQ): New constraint for quad word memory locations.
16626         * config/rs6000/predicates.md (const_0_to_15_operand): New
16627         constraint for 0..15 for crypto instructions.
16628         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16629         as GPR and floating point registers.
16630         (int_reg_operand): New predicate to match only GPR registers.
16631         (base_reg_operand): New predicate to match base registers.
16632         (quad_int_reg_operand): New predicate to match even GPR registers
16633         for quad memory operations.
16634         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16635         operations in both GPR and VSX registers.
16636         (quad_memory_operand): New predicate for quad memory operations.
16637         (reg_or_indexed_operand): New predicate for direct move support.
16639         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16640         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16641         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16642         (POWERPC_MASKS): Add power8 options.
16643         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16644         various options.
16646         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16647         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16649         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16650         (-mpower8-fusion): New power8 options.
16651         (-mpower8-fusion-sign): Likewise.
16652         (-mpower8-vector): Likewise.
16653         (-mcrypto): Likewise.
16654         (-mdirect-move): Likewise.
16655         (-mquad-memory): Likewise.
16657         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16658         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16659         registers.
16660         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16661         (rs6000_debug_vector_unit): Add p8_vector.
16662         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16663         definitions.  Also print fusion state.
16664         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16665         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16666         (rs6000_option_override_internal): Add support for power8.
16667         (rs6000_common_init_builtins): Add debugging for skipped builtins
16668         if -mdebug=builtin.
16669         (rs6000_adjust_cost): Add power8 support.
16670         (rs6000_issue_rate): Likewise.
16671         (insn_must_be_first_in_group): Likewise.
16672         (insn_must_be_last_in_group): Likewise.
16673         (force_new_group): Likewise.
16674         (rs6000_register_move_cost): Likewise.
16675         (rs6000_opt_masks): Likewise.
16677         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16678         power8 capable assembler, default to power7 options.
16679         (TARGET_DIRECT_MOVE): Likewise.
16680         (TARGET_CRYPTO): Likewise.
16681         (TARGET_P8_VECTOR): Likewise.
16682         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16683         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16684         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16685         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16686         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16687         (TARGET_XSCVDPSPN): Likewise.
16688         (TARGET_XSCVSPDPN): Likewsie.
16689         (TARGET_SYNC_HI_QI): Likewise.
16690         (TARGET_SYNC_TI): Likewise.
16691         (MASK_CRYPTO): Likewise.
16692         (MASK_DIRECT_MOVE): Likewise.
16693         (MASK_P8_FUSION): Likewise.
16694         (MASK_P8_VECTOR): Likewise.
16695         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16696         temporary used by some of the direct move instructions to get two FP
16697         temporary registers does not force creation of a stack frame.
16698         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16699         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16700         that any VSX registers are tieable, even if they are also an
16701         Altivec vector mode.
16702         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16703         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16704         (RS6000_BTM_CRYPTO): Likewise.
16705         (RS6000_BTM_COMMON): Likewise.
16707         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16708         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16709         (enum rs6000_vector): Add power8 vector support.
16711 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16713         PR target/19599
16714         PR target/57340
16715         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16716         (any_sibcall_could_use_r3): this and handle indirect calls.
16717         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16719 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16721         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16723 2013-05-22  Richard Biener  <rguenther@suse.de>
16725         PR middle-end/57349
16726         * profile.c (branch_prob): Do not split blocks that are
16727         abnormally receiving from ECF_RETURNS_TWICE functions.
16729 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16731         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16732         address mode.  Move pointer mode initialization to the same place.
16734 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16736         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16737         while it has any effect.
16739 2013-05-21  Easwaran Raman  <eraman@google.com>
16741         PR tree-optimization/57322
16742         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16743         UID of the statement added to the BB to be 1.
16745 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16747         PR tree-optimization/57331
16748         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16749         of conversion from pointer type to integral type with integer.
16751 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16753         PR lto/57289
16754         * ipa-prop.c (ipa_read_node_info): Process param_used and
16755         controlled_uses in the same order as when writing.
16757 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16759         PR plugins/56754
16760         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16762 2013-05-21  Richard Biener  <rguenther@suse.de>
16764         PR tree-optimization/57318
16765         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16766         estimate stmts with side-effects as likely eliminated.
16768 2013-05-21  Richard Biener  <rguenther@suse.de>
16770         PR tree-optimization/57330
16771         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16772         preserve the call stmts fntype.
16774 2013-05-21  Richard Biener  <rguenther@suse.de>
16776         PR tree-optimization/57303
16777         * tree-ssa-sink.c (statement_sink_location): Improve killing
16778         stmt detection and properly handle self-assignments.
16780 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16782         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16783         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16784         registers. Set register size out of the PARALLEL loop.
16786 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16788         PR target/56547
16789         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16790         (*fmasf4, *fmasf4_media): New insns.
16792 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16794         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16795         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16796         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16797         (mips_idiv_insns): Update the comments to say that the returned
16798         instruction counts are in units of BASE_INSN_LENGTH.
16799         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16800         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16801         using 2 rather than 4 as the length of indirect MIPS16 and
16802         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16803         length of a NOP.  Don't divide MIPS16 lengths by 2.
16804         (mips16_split_long_branches): Assume a branch is long if the
16805         length is greater than 4 rather than 8.
16806         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16807         rather than multiplying them by 2.  Multiply instruction counts
16808         by BASE_INSN_LENGTH rather than 4.
16809         (*jump_mips16, tls_get_tp_mips16_<mode>)
16810         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16812 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16814         * config/mips/mips.md (extended_mips16): Remove branch case.
16815         (length): Remove duplicated extended_mips16 test.
16817 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16819         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16821 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16823         * recog.h (Recog_data): Rename to...
16824         (recog_data_d): ...this.
16825         (recog_data): Update accordingly.
16826         * recog.c (recog_data): Likewise.
16827         * reload.c (save_recog_data): Likewise.
16828         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16829         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16831 2013-05-17  Julian Brown  <julian@codesourcery.com>
16833         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16834         found in a REG_EQUAL note, invalidate it.
16836 2013-05-17  Easwaran Raman  <eraman@google.com>
16838         * tree-ssa-reassoc.c (find_insert_point): New function.
16839         (insert_stmt_after): Likewise.
16840         (get_def_stmt): Likewise.
16841         (ensure_ops_are_available): Likewise.
16842         (not_dominated_by): Likewise.
16843         (rewrite_expr_tree): Do not move statements beyond what is
16844         necessary. Remove call to swap_ops_for_binary_stmt...
16845         (reassociate_bb): ... and move it here.
16846         (build_and_add_sum): Assign UIDs for new statements.
16847         (linearize_expr): Likewise.
16848         (do_reassoc): Renumber gimple statement UIDs.
16850 2013-05-17  Jan Hubicka  <jh@suse.cz>
16852         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16853         weakrefs.
16854         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16855         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16856         weakrefs.
16857         (output_weakrefs): Update.
16859 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16860             Martin Jambor  <mjambor@suse.cz>
16862         PR middle-end/57276
16863         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16864         value that corresponds to the given aggval is found in values vector.
16866 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16868         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16869         sse, sse2, sse3, ssse3 and sse4a flags to options.
16871 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16873         * gengtype-state.c: (s_expr_writer): New class, to handle
16874         prettifying of output layout of s-expressions.
16875         (state_writer): New class, to write out gtype.state.
16876         (state_written_type_count): Move this variable into member data of
16877         state_writer.
16878         (s_expr_writer::s_expr_writer): New code: constructor for new class
16879         (state_writer::state_writer(): ditto
16880         (s_expr_writer::write_new_line): New function
16881         (s_expr_writer::write_any_indent): ditto
16882         (s_expr_writer::begin_s_expr): ditto
16883         (s_expr_writer::end_s_expr): ditto
16884         (write_state_fileloc): convert to method of state_writer...
16885         (state_writer:: write_state_fileloc): ...and use methods of
16886         s_expr_writer to write indentation into the gtype.state output file
16887         to visually represent the hierarchical structure of the list
16888         structures
16889         (write_state_fields): ditto, renaming to...
16890         (state_writer::write_state_fields)
16891         (write_state_a_string): ditto, renaming to...
16892         (state_writer::write_state_a_string)
16893         (write_state_string_option): ditto, renaming to...
16894         (state_writer::write_state_string_option)
16895         (write_state_type_option): ditto, renaming to...
16896         (state_writer::write_state_type_option)
16897         (write_state_nested_option): ditto, renaming to...
16898         (state_writer::write_state_nested_option)
16899         (write_state_option): ditto, renaming to...
16900         (state_writer::write_state_option)
16901         (write_state_options): ditto, renaming to...
16902         (state_writer::write_state_options)
16903         (write_state_lang_bitmap): ditto, renaming to...
16904         (state_writer::write_state_lang_bitmap)
16905         (write_state_version): ditto, renaming to...
16906         (state_writer::write_state_version)
16907         (write_state_scalar_type): ditto, renaming to...
16908         (state_writer::write_state_scalar_type)
16909         (write_state_string_type): ditto, renaming to...
16910         (state_writer::write_state_string_type)
16911         (write_state_undefined_type): ditto, renaming to...
16912         (state_writer::write_state_undefined_type)
16913         (write_state_struct_union_type): ditto, renaming to...
16914         (state_writer::write_state_struct_union_type)
16915         (write_state_struct_type): ditto, renaming to...
16916         (state_writer::write_state_struct_type)
16917         (write_state_user_struct_type): ditto, renaming to...
16918         (state_writer::write_state_user_struct_type)
16919         (write_state_lang_struct_type): ditto, renaming to...
16920         (state_writer::write_state_lang_struct_type)
16921         (write_state_param_struct_type): ditto, renaming to...
16922         (state_writer::write_state_param_struct_type)
16923         (write_state_pointer_type): ditto, renaming to...
16924         (state_writer::write_state_pointer_type)
16925         (write_state_array_type): ditto, renaming to...
16926         (state_writer::write_state_array_type)
16927         (write_state_gc_used): ditto, renaming to...
16928         (state_writer::write_state_gc_used)
16929         (write_state_common_type_content): ditto, renaming to...
16930         (state_writer::write_state_common_type_content)
16931         (write_state_type): ditto, renaming to...
16932         (state_writer::write_state_type)
16933         (write_state_pair_list): ditto, renaming to...
16934         (state_writer::write_state_pair_list)
16935         (write_state_pair): ditto, renaming to...
16936         (state_writer::write_state_pair)
16937         (write_state_typedefs): ditto, renaming to...
16938         (state_writer::write_state_typedefs)
16939         (write_state_structures): ditto, renaming to...
16940         (state_writer::write_state_structures)
16941         (write_state_param_structs): ditto, renaming to...
16942         (state_writer::write_state_param_structs)
16943         (write_state_variables): ditto, renaming to...
16944         (state_writer::write_state_variables)
16945         (write_state_srcdir): ditto, renaming to...
16946         (state_writer::write_state_srcdir)
16947         (write_state_files_list): ditto, renaming to...
16948         (state_writer::write_state_files_list)
16949         (write_state_languages): ditto, renaming to...
16950         (state_writer::write_state_languages)
16951         (write_state): create a state_writer instance and use it when
16952         writing out the state file
16954 2013-05-17  Mike Stump  <mikestump@comcast.net>
16956         PR rtl-optimization/57304
16957         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16958         accessing DF_REF_REAL_LOC.
16960 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16962         PR rtl-optimization/57281
16963         PR rtl-optimization/57300
16964         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16965         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16966         what the other splitter did if the registers are dead.
16968 2013-05-17  Richard Biener  <rguenther@suse.de>
16970         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16971         MEM_REF offsets.
16973 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16975         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16976         linking.
16978 2013-05-17  Marek Polacek  <polacek@redhat.com>
16980         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16981         length when doing non-zero store of storing '\0' to '\0'.
16983 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16985         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16986         vect_external_def oprnd1 with loop_vinfo, try to emit
16987         optional cast, negation and and stmts on the loop preheader
16988         edge instead of into the pattern def seq.
16990         PR tree-optimization/57051
16991         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16992         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16994 2013-05-16  Nick Clifton  <nickc@redhat.com>
16996         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16997         (rl78_is_naked_func): New function.
16998         (rl78_expand_prologue): Skip prologue generation for naked functions.
16999         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17000         * doc/extend.texi (naked): Add RL78 to the list of processors
17001         that supports this attribute.
17003 2013-05-16  Jeff Law  <law@redhat.com>
17005         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17007 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17009         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17010         cache parameters using detect_caches_amd also for CYRIX,
17011         NSC and TM2 signatures.
17013 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17014             Dzianis Kahanovich  <mahatma@eu.by>
17016         PR target/45359
17017         PR target/46396
17018         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17019         VIA/Centaur processors and determine their cache parameters
17020         using detect_caches_amd.
17022 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17024         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17025         (rtl_verify_edges): New function.
17026         (rtl_verify_bb_insns): Ditto.
17027         (rtl_verify_bb_pointers): Ditto.
17028         (rtl_verify_bb_insn_chain): Ditto.
17029         (rtl_verify_fallthru): Ditto.
17030         (rtl_verify_bb_layout): Ditto.
17031         (rtl_verify_flow_info_1): Outline checks into new functions.
17032         (rtl_verify_flow_info): Ditto.
17034 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17036         * cfghooks.c (copy_bbs): Add update_dominance argument.
17037         * cfghooks.h (copy_bbs): Update prototype.
17038         * tree-cfg.c (gimple_duplicate_sese_region):
17039         Add update_dominance argument.
17040         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17041         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17042         gimple_duplicate_sese_region call.
17043         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17044         Update copy_bbs call.
17045         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17046         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17048 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17050         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17051         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17052         vect_recog_rotate_pattern.
17053         (vect_recog_rotate_pattern): New function.
17055 2013-05-16  Jason Merrill  <jason@redhat.com>
17057         * Makefile.in (LLINKER): New variable.
17058         (mostlyclean): Remove link mutex.
17059         * configure.ac: Handle --enable-link-mutex.
17060         * lock-and-run.sh: New script.
17062 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17064         PR target/19599
17065         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17066         for NULL decl.
17068 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17070         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17072 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17074         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17075         * config/arm/arm.c (next_consecutive_mem): New function.
17076         (gen_movmem_ldrd_strd): Likewise.
17077         * config/arm/arm.md (movmemqi): Update condition and code.
17078         (unaligned_loaddi, unaligned_storedi): New patterns.
17080 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17082         * config.gcc: Obsolete *-*-solaris2.9*.
17083         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17085 2013-05-16  Richard Biener  <rguenther@suse.de>
17087         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17088         earlier, after GRAPHITE transforms and IV canonicalization.
17090 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17092         * omp-low.c (extract_omp_for_data): For collapsed loops,
17093         if at least one of the loops is known at compile time to
17094         iterate zero times, set count to 0.
17095         (expand_omp_regimplify_p): New function.
17096         (expand_omp_for_generic): For collapsed loops, if at least
17097         one of the loops isn't known to iterate at least once,
17098         add runtime check with setting count to 0.
17099         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17100         For unsigned types if it isn't known at compile time that
17101         the loop will iterate at least once, add runtime check to bypass
17102         the whole loop if initial condition isn't true.
17104 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17106         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17108 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17110         PR middle-end/57286
17111         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17112         transformations to avoid an infinite loop.
17114 2013-05-16  Marek Polacek  <polacek@redhat.com>
17116         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17118 2013-05-15  Leif Ekblad  <leif@rdos.net>
17120         * config/i386/i386.c (ix86_decompose_address): Use
17121         DEFAULT_TLS_SEG_REG to access TLS segment register.
17122         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17123         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17124         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17126 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17128         PR target/57260
17129         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17130         sibling calls to functions that would normally be lazily bound,
17131         unless $gp is call-clobbered.
17133 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17135         * config/i386/i386.c (ix86_option_override_internal): Update
17136         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17137         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17138         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17139         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17140         of TARGET_3DNOW.
17141         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17143 2013-05-15  Andreas Schwab  <schwab@suse.de>
17145         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17146         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17147         third operand.
17149 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17151         * loop-unroll.c (report_unroll_peel): Check decision before
17152         emitting unroll/peel message.
17154 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17156         * function.h (has_bb_partition): New rtl_data flag.
17157         (bb_reorder_complete): Ditto.
17158         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17159         instead of flag_reorder_blocks_and_partition.
17160         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17161         with some enhancements.
17162         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17163         * bb-reorder.c (connect_traces): Check for has_bb_partition
17164         instead of flag_reorder_blocks_and_partition.
17165         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17166         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17167         verify_hot_cold_block_grouping.
17168         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17170 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17172         PR target/19599
17173         * config/arm/predicates.md (call_insn_operand): New predicate.
17174         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17175         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17176         if insn is not a tail call.
17177         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17178         registers.
17179         * config/arm/arm.h (enum reg_class): New caller save register class.
17180         (REG_CLASS_NAMES): Likewise.
17181         (REG_CLASS_CONTENTS): Likewise.
17182         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17183         without decls.
17185 2013-05-15  Richard Biener  <rguenther@suse.de>
17187         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17188         of MSG_OPTIMIZED_LOCATIONS.
17189         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17190         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17191         message.
17192         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17193         of MSG_OPTIMIZED_LOCATIONS.
17194         (execute_vect_slp): Likewise.
17195         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17196         (vect_create_cond_for_alias_checks): Likewise.
17197         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17198         (vect_recog_widen_mult_pattern): Likewise.
17199         (vect_recog_widen_sum_pattern): Likewise.
17200         (vect_recog_over_widening_pattern): Likewise.
17201         (vect_recog_widen_shift_pattern): Likewise.
17202         (vect_recog_vector_vector_shift_pattern): Likewise.
17203         (vect_recog_divmod_pattern): Likewise.
17204         (vect_recog_mixed_size_cond_pattern): Likewise.
17205         (vect_recog_bool_pattern): Likewise.
17206         (vect_pattern_recog_1): Likewise.
17208 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17210         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17211         non-functions to builtin_unreachable.
17212         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17213         correct when new direct edges were discovered.
17215 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17217         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17218         header, print symbol order instead of node uid, print more information
17219         about indirect edge targets.
17220         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17221         uids.
17222         (ipa_make_edge_direct_to_target): Likewise.
17223         (remove_described_reference): Likewise.
17224         (propagate_controlled_uses): Likewise.
17225         (ipa_print_node_params): Also print symbol order.
17226         (ipcp_transform_function): Print symbol order instead of node uids.
17227         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17228         (cgraph_get_create_real_symbol_node): Likewise.
17229         * ipa-cp.c (print_lattice): Likewise.
17230         (print_all_lattices): Likewise.
17231         (determine_versionability): Likewise.
17232         (initialize_node_lattices): Likewise.
17233         (estimate_local_effects): Likewise.
17234         (update_profiling_info): Likewise.
17235         (create_specialized_node): Likewise.
17236         (perhaps_add_new_callers): Likewise.
17237         (decide_about_value): Likewise.
17238         (decide_whether_version_node): Likewise.
17239         (identify_dead_nodes): Likewise.
17240         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17241         (dump_inline_summary): Likewise.
17242         (estimate_node_size_and_time): Likewise.
17243         (inline_analyze_function): Likewise.
17244         * ipa-inline.c (report_inline_failed_reason): Likewise.
17245         (want_early_inline_function_p): Likewise.
17246         (edge_badness): Likewise.
17247         (update_edge_key): Likewise.
17248         (inline_small_functions): Likewise.  Add dumping of order to two other
17249         dumps.
17250         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17251         instead of node uids.
17252         (propagate_pure_const): Likewise.
17253         (propagate_pure_const): Likewise.
17254         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17255         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17256         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17257         of node uids.
17258         * tree-pretty-print.c (dump_function_header): Likewise.
17259         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17260         Print symbol order instead of node uids.
17262 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17264         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17265         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17267 2013-05-15  Richard Biener  <rguenther@suse.de>
17269         PR tree-optimization/57275
17270         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17271         return value for fail to do runtime alias checks for gather loads.
17273 2013-05-15  Jan Hubicka  <jh@suse.cz>
17275         PR lto/57038
17276         PR lto/47375
17277         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17278         weakrefs are not external.
17279         (lto_symtab_merge_decls): Fix thinko when dealing with
17280         non-lto_symtab decls.
17281         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17282         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17283         * varpool.c (dump_varpool_node): Dump more flags.
17285 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17287         * config/i386/i386.c (processor_alias_table): Add instruction
17288         FSGSBASE for AMD bdver3 architecture.
17290 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17292         * tree.c (warn_deprecated_use): Print file:line using locus color.
17293         * diagnostic.c (diagnostic_report_current_module): Print file:line
17294         and file:line:column using locus color.
17296 2013-05-14  Mike Stump  <mikestump@comcast.net>
17298         * gdbinit.in: Add __null.
17300 2013-05-14  Mike Stump  <mikestump@comcast.net>
17302         * recog.h: Rename struct recog_data to Recog_data.
17303         * recog.c: Likewise.
17304         * reload.c (can_reload_into): Likewise.
17305         * config/picochip/picochip.c: Likewise.
17307 2013-05-14  Mike Stump  <mikestump@comcast.net>
17309         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17311 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17313         * resource.h (struct resources): Remove unch_memory member.
17314         (CLEAR_RESOURCE): Don't clear unch_memory.
17315         * resource.c (mark_referenced_resources): Don't set it.
17316         (mark_set_resources): Likewise.
17317         (mark_target_live_regs): Don't clear it.
17318         (init_resource_info): Likewise.
17319         * reorg.c (resource_conflicts_p): Don't compare it.
17320         (redundant_insn): Don't set it.
17322         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17323         Remove prototypes.
17324         * emit-rtl.c (next_label): Remove unused function.
17325         (skip_consecutive_labels, link_cc0_insns): Move to ...
17326         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17327         only place where these functions are used, and make them static.
17329 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17331         * fold-const.c (fold_negate_expr): Handle vectors.
17332         (fold_truth_not_expr): Make it static.
17333         (fold_invert_truthvalue): New static function.
17334         (invert_truthvalue_loc): Handle vectors. Do not call
17335         fold_truth_not_expr directly.
17336         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17337         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17338         (fold_comparison): Handle vector constants.
17339         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17340         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17341         * tree.h (fold_truth_not_expr): Remove declaration.
17343 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17345         * config/aarch64/aarch64-simd.md
17346         (aarch64_vcond_internal<mode>): Rename to...
17347         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17348         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17349         float modes. Clarify all iterator modes.
17350         (vcond<mode><mode>): Use new name for vcond expanders.
17351         (vcond<v_cmp_result><mode>): Likewise.
17352         (vcondu<mode><mode>: Likewise.
17353         * config/aarch64/iterators.md (VDQF_COND): New.
17355 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17357         PR bootstrap/57266
17358         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17359         variable for the shift amount. Check that we shift by non-negative
17360         amounts.
17362 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17364         PR target/42017
17365         * config/arm/arm.h (EPILOGUE_USES): Only return true
17366         for LR_REGNUM after epilogue_completed.
17368 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17370         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17371         is error_mark_node.
17373 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17375         PR target/57261
17376         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17377         and Solaris 11+/x86 with gld.
17378         * configure: Regenerate.
17380 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17382         * expmed.c (expand_shift_1): Canonicalize rotates by
17383         constant bitsize / 2 to bitsize - 1.
17384         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17385         case ROTATERT>: Likewise.
17387         Revert:
17388         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17390         * config/i386/i386.md (rotateinv): New code attr.
17391         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17392         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17393         roll $31, %eax, etc.
17395 2013-05-14  Richard Biener  <rguenther@suse.de>
17397         PR middle-end/57235
17398         * tree-eh.c (sink_clobbers): Give up for successors with
17399         multiple predecessors and no virtual uses.
17401 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17403         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17404         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17406 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17408         PR middle-end/57251
17409         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17410         the case when both op0 and op1 have VOIDmode.
17412 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17414         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17415         in multiply-accumulate mode.
17417 2013-05-13  Guozhi Wei  <carrot@google.com>
17419         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17421 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17423         PR target/56975
17424         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17425         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17426         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17427         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17428         (ix86_expand_prologue): Likewise.
17429         (ix86_expand_split_stack_prologue): Likewise.
17430         (legitimate_pic_address_disp_p): Likewise.
17431         (legitimize_pic_address): Likewise.
17432         (legitimize_tls_address): Likewise.
17433         (legitimize_pe_coff_symbol): Likewise.
17434         (output_pic_addr_const): Likewise.
17435         (construct_plt_address): Likewise.
17436         (ix86_expand_call): Likewise.
17437         (x86_output_mi_thunk): Likewise.
17438         (x86_function_profiler): Likewise.
17440 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17442         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17443         similar switch cases.
17444         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17445         (aarch64_simd_mov_to_<mode>low): Delete.
17446         (aarch64_simd_mov_to_<mode>high): Delete.
17447         (move_lo_quad_<mode>): Add w<-r alternative.
17448         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17449         (aarch64_simd_mov_from_*): Update type attribute.
17450         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17451         statement.
17453 2013-05-13  Jan Hubicka  <jh@suse.cz>
17455         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17456         * config/i386/i386.c (ix86_compute_frame_layout,
17457         ix86_expand_epilogue, emit_i387_cw_initialization,
17458         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17459         ix86_local_alignment): Fix use of size/speed predicates.
17461 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17463         PR tree-optimization/45216
17464         PR tree-optimization/57157
17465         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17466         the (-Y) & (B - 1) variant if OP is |.
17467         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17468         return shifted.  Use (-op1) & (prec - 1) as other_amount
17469         instead of prec - op1.
17471 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17473         PR middle-end/42371
17474         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17475         (ipa_constant_data): New type.
17476         (ipa_jump_func): Use ipa_constant_data to hold information about
17477         constant jump functions.
17478         (ipa_get_jf_constant): Adjust to jump function type changes.
17479         (ipa_get_jf_constant_rdesc): New function.
17480         (ipa_param_descriptor): New field controlled_uses.
17481         (ipa_get_controlled_uses): New function.
17482         (ipa_set_controlled_uses): Likewise.
17483         * ipa-ref.h (ipa_find_reference): Declare.
17484         * ipa-prop.c (ipa_cst_ref_desc): New type.
17485         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17486         changes.
17487         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17488         New parameter cs.  Adjust all callers.
17489         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17490         (remove_described_reference): New function.
17491         (jfunc_rdesc_usable): Likewise.
17492         (try_make_edge_direct_simple_call): Decrement controlled use count,
17493         attempt to remove reference if it hits zero.
17494         (combine_controlled_uses_counters): New function.
17495         (propagate_controlled_uses): Likewise.
17496         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17497         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17498         (ipa_print_node_params): Print described use counter.
17499         (ipa_write_jump_function): Adjust to jump function type changes.
17500         (ipa_read_jump_function): New parameter CS, pass it to
17501         ipa_set_jf_constant.  Adjust caller.
17502         (ipa_write_node_info): Stream controlled use count
17503         (ipa_read_node_info): Likewise.
17504         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17505         asserting.
17506         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17507         count.  Remove cloning-added reference if it reaches zero.
17508         * ipa-ref.c (ipa_find_reference): New function.
17510 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17512         * config/i386/i386.c (processor_target_table): Modified default
17513         alignment values for AMD BD and BT architectures.
17515 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17517         * tree-vect-generic.c (uniform_vector_p): Move ...
17518         * tree.c (uniform_vector_p): ... here.
17519         * tree.h (uniform_vector_p): Declare it.
17520         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17521         into a scalar.
17523 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17525         PR tree-optimization/57230
17526         * tree-ssa-strlen.c (handle_char_store): Record length for
17527         array store from STRING_CST.
17529         PR tree-optimization/57230
17530         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17531         check.
17533 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17535         * config/epiphany/epiphany.c (epiphany_init): Check size of
17536         NUM_MODES_FOR_MODE_SWITCHING.
17537         (epiphany_expand_prologue):
17538         Remove CONFIG_REGNUM initial value handling code.
17539         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17540         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17541         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17542         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17543         Don't return 1 for FP_MODE_NONE.
17544         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17545         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17546         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17547         * config/epiphany/epiphany.md (save_config): New pattern.
17549 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17551         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17553 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17555         * config/i386/i386.md (memory): Handle sseishft1.
17556         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17557         (*vec_extractv2di_1): Ditto.
17559 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17561         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17562         saved registers.
17564 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17566         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17567         Add mthumb/march=armv7-a multilib.
17568         Add mthumb/march=armv7-r multilib.
17569         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17571 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17573         * config/v850/t-rtems: Add more multilibs.
17575 2013-05-10  Richard Biener  <rguenther@suse.de>
17577         PR tree-optimization/57214
17578         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17579         not propagate from SSA names that occur in abnormal PHI nodes.
17581 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17583         * stor-layout.c (element_precision): New function.
17584         * machmode.h (element_precision): Declare it.
17585         * tree.c (build_minus_one_cst): New function.
17586         (element_precision): Likewise.
17587         * tree.h (build_minus_one_cst): Declare new function.
17588         (element_precision): Likewise.
17589         * fold-const.c (operand_equal_p): Use element_precision.
17590         (fold_binary_loc): Handle vector types.
17591         * convert.c (convert_to_integer): Use element_precision.
17592         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17593         separately.
17595 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17597         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17598         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17599         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17600         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17601         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17602         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17603         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17604         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17605         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17606         (Uuw8): New constraints.
17607         (Usb4): Move into alphabetical order.
17608         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17609         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17610         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17611         previously unnamed patterns.
17612         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17613         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17614         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17615         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17616         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17617         of specific lengths.
17619 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17621         * config/i386/i386.md (rotateinv): New code attr.
17622         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17623         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17624         roll $31, %eax, etc.
17626         PR tree-optimization/45216
17627         PR tree-optimization/57157
17628         * tree-ssa-forwprop.c (simplify_rotate): New function.
17629         (ssa_forward_propagate_and_combine): Call it.
17631 2013-05-10  Richard Biener  <rguenther@suse.de>
17633         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17634         disable peeling when we version for aliasing.
17635         (vector_alignment_reachable_p): Honor explicit user alignment.
17636         (vect_supportable_dr_alignment): Likewise.
17637         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17638         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17639         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17640         then peeling to arrange for the cost-model check to come first.
17642 2013-05-10  Alan Modra  <amodra@gmail.com>
17644         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17645         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17646         * configure: Regenerate.
17648 2013-05-10  Alan Modra  <amodra@gmail.com>
17650         PR target/55033
17651         * varasm.c (default_elf_select_section): Move !DECL_P check..
17652         (get_named_section): ..to here before calling get_section_name.
17653         Adjust assertion.
17654         (default_section_type_flags): Add DECL_P check.
17655         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17656         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17658 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17660         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17661         When using gen_stack_adjust_str with a register offset, add a
17662         REG_FRAME_RELATED_EXPR note.
17664 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17666         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17667         (*vec_extractv4si_zext_mem): Ditto.
17668         (*vec_extractv2di): Add 0->x and x->x alternatives.
17669         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17670         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17672 2013-05-09  Jason Merrill  <jason@redhat.com>
17674         N3639 C++1y VLA support
17675         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17676         DECL_VALUE_EXPR.
17678         * tree.c (build_constructor_va): New.
17679         * tree.h: Declare it.
17681 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17683         PR lto/57084
17684         * gimple-fold.c (canonicalize_constructor_val): Call
17685         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17687 2013-05-09  Jan Hubicka  <jh@suse.cz>
17688             Richard Biener  <rguenther@suse.de>
17690         PR lto/54095
17691         * symtab.c (symtab_make_decl_local): Do not add private names.
17693 2013-05-09  Jan Hubicka  <jh@suse.cz>
17695         PR lto/54095
17696         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17697         (unlink_from_assembler_name_hash): Likewise.
17698         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17699         symtab_unregister_node, symtab_initialize_asm_name_hash,
17700         change_decl_assembler_name): Update.
17702 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17704         * config/aarch64/aarch64.md: New movtf split.
17705         (*movtf_aarch64): Update.
17706         (aarch64_movdi_tilow): Handle TF modes and rename to
17707         aarch64_movdi_<mode>low.
17708         (aarch64_movdi_tihigh): Handle TF modes and rename to
17709         aarch64_movdi_<mode>high
17710         (aarch64_movtihigh_di): Handle TF modes and rename to
17711         aarch64_mov<mode>high_di
17712         (aarch64_movtilow_di): Handle TF modes and rename to
17713         aarch64_mov<mode>low_di
17714         (aarch64_movtilow_tilow): Remove spurious whitespace.
17715         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17716         splits.
17717         (aarch64_print_operand): Update.
17719 2013-05-09  Alan Modra  <amodra@gmail.com>
17721         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17722         powerpc64le.
17723         * configure: Regenerate.
17725 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17727         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17728         splitter preparation statements.
17729         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17730         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17731         adjust_address_nv.
17733 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17735         * gimple-ssa-strength-reduction.c (count_candidates): Change
17736         return value to int.
17737         (analyze_candidates_and_replace): Change type of length to int.
17739 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17741         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17742         (*vec_extract<mode>): Use VI12_128 mode iterator.
17743         (*vec_extract<mode>_mem): Ditto.
17744         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17745         attribute.
17747 2013-05-08  Diego Novillo  <dnovillo@google.com>
17749         PR bootstrap/54659
17751         Revert:
17752         2012-08-17  Diego Novillo  <dnovillo@google.com>
17754         PR bootstrap/54281
17755         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17756         * config.in: Regenerate.
17757         * configure: Regenerate.
17758         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17760 2013-05-08  Jan Hubicka  <jh@suse.cz>
17762         PR lto/54095
17763         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17764         * cgraph.h (symtab_node_base): Add unique_name.
17765         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17766         input_overwrite_node, input_varpool_node): Stream unique_name.
17767         * cgraphclones.c (cgraph_create_virtual_clone,
17768         cgraph_function_versioning): Set unique_name.
17769         * ipa.c (function_and_variable_visibility): Set unique_name.
17771 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17773         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17774         (alloc_cand_and_find_basis): Restrict conditional candidate
17775         processing to CAND_MULTs.
17777 2013-05-08  Jan Hubicka  <jh@suse.cz>
17779         PR lto/54095
17780         lto-symtab.c (lto_symtab_symbol_p): New function.
17781         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17782         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17783         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17784         Skip static symbols.
17786 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17788         PR tree-optimization/57200
17789         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17790         Only call inform if the preceding warning_at returns true.
17792 2013-05-07  Han Shen  <shenhan@google.com>
17794         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17795         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17796         * common.opt (fstack-protector-strong): New option.
17797         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17798         * doc/invoke.texi (Optimization Options): Document
17799         "-fstack-protector-strong".
17800         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17802 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17804         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17806 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17808         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17809         it is not lhs of the insn.
17810         (setup_reg_equiv): Remove insn having equiv memory which it is not
17811         lhs of the insn.
17812         * lra-constraints.c (process_address): Try to improve generation
17813         code for address base + disp.
17814         (lra_constraints): Make correct the code for checking insn setting
17815         up backward equivalence.  Remove insn only if it is in the init
17816         insn list.
17817         * lra-eliminations.c (update_reg_eliminate): Change return value.
17818         (lra_eliminate): Use the result.
17820 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17822         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17823         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17824         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17825         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17826         register target operands.
17827         (*vec_extractv8hi_sse2): New pattern.
17828         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17829         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17830         (*vec_extract<mode>_mem): New insn and split pattern.
17832 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17834         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17835         (TARGET_ASAN_SHADOW_OFFSET): Define.
17836         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17837         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17839 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17841         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17842         (incr_vec_index): Return -1 if increment not found.
17843         (create_add_on_incoming_edge): Assert if increment not found.
17844         (record_increment): Limit number of increments recorded.
17845         (all_phi_incrs_profitable): Return false if an increment not found.
17846         (replace_profitable_candidates): Don't process increments that were
17847         not recorded.
17848         (analyze_candidates_and_replace): Limit size of incr_vec.
17850 2013-05-07  Richard Biener  <rguenther@suse.de>
17852         * calls.c (special_function_p): setjmp-like functions are leaf.
17853         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17854         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17856 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17858         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17859         (aarch64_simd_mov<mode>): New expander.
17860         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17861         (aarch64_simd_mov_to_<mode>high): Likewise.
17862         (aarch64_simd_mov_from_<mode>low): Likewise.
17863         (aarch64_simd_mov_from_<mode>high): Likewise.
17864         (aarch64_dup_lane<mode>): Update.
17865         (aarch64_dup_lanedi): New instruction pattern.
17866         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17867         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17869 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17871         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17872         (replace_mult_candidate): Remove unnecessary argument; remove
17873         unnecessary parameter from call to introduce_cast_before_cand.
17874         (replace_unconditional_candidate): Remove unnecessary parameter
17875         from call to replace_mult_candidate.
17876         (replace_conditional_candidate): Likewise.
17877         (insert_initializers): Use make_temp_ssa_name.
17878         (introduce_cast_before_cand): Remove unnecessary argument; use
17879         make_temp_ssa_name.
17880         (replace_one_candidate): Remove unnecessary argument; remove
17881         unnecessary parameter from calls to introduce_cast_before_cand.
17882         (replace_profitable_candidates): Remove unnecessary parameters
17883         from calls to replace_one_candidate.
17885 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17887         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17888         phi def as possibly hiding a basis for a CAND_ADD whose operands
17889         have been commuted in the analysis.
17890         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17892 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17894         * config/aarch64/aarch64.md
17895         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17896         shift value between 0-4.
17898 2013-05-07  Richard Biener  <rguenther@suse.de>
17900         * double-int.h (rshift): New overload.
17901         * double-int.c (rshift): New function.
17902         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17903         (create_reference_ops_from_ref): Remove.
17904         (vn_reference_insert): Use shared ops for constructing the
17905         reference and copy it.
17907 2013-05-07  Richard Biener  <rguenther@suse.de>
17909         PR middle-end/57190
17910         * tree-eh.c (sink_clobbers): Properly propagate
17911         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17913 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17915         PR tree-optimization/57149
17916         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17917         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17918         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17919         uninit_undefined_value_p instead of ssa_undefined_value_p.
17921         PR debug/57184
17922         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17923         for modifier == EXPAND_INITIALIZER.
17925 2013-05-07  Anton Blanchard  <anton@samba.org>
17927         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17928         for powerpc64 little endian.
17929         * configure: Regenerate.
17931 2013-05-06  Graham Stott  <grahams@btinternet.com>
17933         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17934         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17935         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17936         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17938 2013-05-06  Graham Stott  <grahams@btinternet.com>
17940         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17941         codes which allow non-lvalues.
17943 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17945         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17946         components are all 1s.
17947         (integer_minus_onep): New function.
17948         * tree.h (integer_minus_onep): Declare it.
17949         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17950         integer_minus_onep instead of integer_all_onesp.
17952 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17954         PR target/52933
17955         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17956         variations of these patterns.
17958 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17960         * config/i386/i386.md (isa): Add x64_sse4 member.
17961         (enabled): Handle x64_sse4.
17962         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17963         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17964         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17965         instruction for SSE4_1 targets.  Update insn attributes.
17966         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17967         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17968         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17969         const_1 selector.
17970         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17971         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17972         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17974 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17976         PR target/57108
17977         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17979 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17981         * final.c (do_assembler_dialects): Don't handle curly braces and
17982         vertical bar escaped by % as dialect delimiters.
17983         (output_asm_insn): Print curly braces and vertical bar if escaped
17984         by % and ASSEMBLER_DIALECT defined.
17985         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17986         * doc/tm.texi: Regenerated.
17988 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17990         * config/mips/mips.c: Include tree-pass.h.
17991         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17992         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17993         (pass_mips_machine_reorg2): New machine specific pass.
17994         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17995         (mips_option_override): Register the new pass.
17996         * rtl.h (cleanup_barriers): Remove prototype.
17997         (dbr_schedule): Likewise.
17998         * jump.c (cleanup_barriers): Make static.
17999         * reorg.c (dbr_schedule): Likewise.
18001 2013-05-06  Richard Biener  <rguenther@suse.de>
18003         PR tree-optimization/57185
18004         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18005         SSA names properly.
18007 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18009         PR target/57106
18010         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18011         between "first_arg" and "insn", not "last" and "insn".
18013 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18015         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18016         (find_candidates_in_block): Re-enable slsr_process_phi.
18017         (create_phi_basis): Fix double counting of candidate adjustment.
18019 2013-05-06  Richard Biener  <rguenther@suse.de>
18021         PR middle-end/57147
18022         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18023         the edge is also fallthru, preserve it and just clear the
18024         abnormal flag.
18025         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18026         also complex, preserve that and just clear the fallthru flag.
18027         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18028         update virtual operands.
18030 2013-05-06  Alan Modra  <amodra@gmail.com>
18032         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18033         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18034         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18035         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18036         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18037         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18039 2013-05-06  Alan Modra  <amodra@gmail.com>
18041         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18042         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18043         (DEFAULT_ASM_ENDIAN): Define.
18044         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18045         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18046         Update -K PIC clause from sysv4.h.
18047         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18048         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18050 2013-05-06  Alan Modra  <amodra@gmail.com>
18052         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18053         twice for little-endian.
18054         (ashrdi3_no_power, ashrdi3): Support little-endian.
18056 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18058         PR target/55303
18059         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18060         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18061         related expanders.
18062         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18063         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18064         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18065         New predicates.
18067 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18068             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18070         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18071         * config/pa/pa.opt: Make mbig-switch a no-op.
18072         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18073         (CASE_VECTOR_MODE): Always return SImode.
18074         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18075         for the !TARGET_BIG_SWITCH case.
18076         * config/pa/pa-linux.h: Likewise.
18077         * config/pa/pa-openbsd.h: Likewise.
18078         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18079         * config/pa/pa.md (short_jump): Remove define_insn.
18080         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18081         (casesi0): Remove define_insn.
18082         (type): Remove btable_branch.
18083         (pa_combine_type): Likewise.
18084         (in_nullified_branch_delay): Likewise.
18085         (in_call_delay): Likewise.
18086         (define_delay): Likewise.
18087         (define_insn_reservation "Z3"): Likewise.
18088         (define_insn_reservation "Z4"): Likewise.
18089         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18090         (pa_adjust_insn_length): Remove adjustment for btable branches.
18091         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18092         and mno-big-switch
18094 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18096         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18097         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18098         Add m->r,x alternatives.
18099         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18100         splitters using SWI48x mode iterator.
18101         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18102         TARGET_64BIT.  Add m->x alternative.
18103         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18104         Add o->x alternative.  Enable for TARGET_SSE.
18105         (sse_storeq): Remove expander.
18106         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18107         with memory input operand.
18108         (*vec_extractv2di_1 splitter): New.
18109         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18110         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18112 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18114         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18115         (INT_LOWPART): Delete.
18116         (extract_MB): Adjust.
18117         (extract_ME): Adjust.
18118         (print_operand): Adjust.
18120 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18122         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18123         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18124         (reg_or_logical_cint_operand, easy_fp_constant,
18125         logical_const_operand): Delete "CONST_DOUBLE" case.
18126         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18127         "HOST_BITS_PER_WIDE_INT == 64" test.
18128         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18129         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18130         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18131         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18132         test.
18133         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18134         CONST_DOUBLE DImode/VOIDmode case.
18135         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18136         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18137         CONST_DOUBLE VOIDmode case.
18138         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18139         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18140         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18141         Delete CONST_DOUBLE case.
18142         (splitters for mov FMOVE64 const_double): Delete
18143         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18144         "HOST_BITS_PER_WIDE_INT >= 64" test.
18145         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18146         case.
18147         (mov DI const_double): Delete.
18149 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18151         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18152         on op shows all bits zero in mode of a lowpart subreg, return zero.
18154 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18156         PR target/57150
18157         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18158         to save TFmode registers and DImode to save TImode registers for
18159         caller save operations.
18160         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18161         mark being partially clobbered since they only use the first
18162         double word.
18164         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18165         and TDmode only use the upper 64-bits of each VSX register.
18167 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18169         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18170         (find_candidates_in_block): Disable slsr_process_phi.
18172 2013-05-03  Guozhi Wei  <carrot@google.com>
18174         * coverage.c (coverage_obj_init): Move the construction of gcov
18175         constructor to ...
18176         (build_init_ctor): ... here.
18178 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18180         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18181         (slsr_cand_d): Redefine def_phi.
18182         (stride_status, phi_adjust_status, count_phis_status): New enums.
18183         (find_phi_def): New.
18184         (find_basis_for_base_expr): New.
18185         (find_basis_for_candidate): Handle hidden bases.
18186         (alloc_cand_and_find_basis): Handle phi candidates.
18187         (slsr_process_phi): New.
18188         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18189         (create_mul_imm_cand): Likewise.
18190         (create_add_ssa_cand): Exclude phi base candidates.
18191         (create_add_imm_cand): Likewise.
18192         (slsr_process_cast): Likewise.
18193         (slsr_process_copy): Likewise.
18194         (find_candidates_in_block): Handle phi candidates.
18195         (dump_candidate): Likewise.
18196         (unconditional_cands): Delete.
18197         (unconditional_cands_with_known_stride_p): Delete.
18198         (phi_dependent_cand_p): New.
18199         (cand_increment): Handle phi-dependent candidates.
18200         (replace_dependent): Delete.
18201         (replace_mult_candidate): New.
18202         (replace_unconditional_candidate): New.
18203         (incr_vec_index): Move to avoid forward reference.
18204         (create_add_on_incoming_edge): New.
18205         (create_phi_basis): New.
18206         (replace_dependents): Delete.
18207         (replace_conditional_candidate): New.
18208         (phi_add_costs): New.
18209         (replace_uncond_cands_and_profitable_phis): New.
18210         (record_increment): Handle phi adjustments.
18211         (record_phi_increments): New.
18212         (record_increments): Handle phi adjustments.
18213         (phi_incr_cost): New.
18214         (lowest_cost_path): Handle phis.
18215         (total_savings): Likewise.
18216         (analyze_increments): Likewise.
18217         (ncd_with_phi): New.
18218         (ncd_of_cand_and_phis): New.
18219         (nearest_common_dominator_for_cands): Handle phi increments.
18220         (all_phi_incrs_profitable): New.
18221         (replace_profitable_candidates): Handle phi-dependent candidates.
18222         (analyze_candidates_and_replace): Likewise.
18224 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18226         PR bootstrap/57154
18227         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18228         do not exceed REG_BR_PROB_BASE.
18230 2013-05-03  Jeff Law  <law@redhat.com>
18232         PR tree-optimization/57144
18233         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18234         operand of the condition will bit into the new type when eliminating
18235         a cast feeding a condition.
18237 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18239         PR rtl-optimization/57130
18240         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18241         of COMPARE as in_code to the recursive call if needed.
18243 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18245         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18246         (enabled): Handle new members.
18247         * config/i386/sse.md (*vec_concatv2si): Merge from
18248         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18249         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18251 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18253         PR tree-optimization/57027
18254         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18255         for fnms opportunity, check we got the prerequisite kind
18256         of tree / gimple before using accessor functions.
18258 2013-05-03  Richard Biener  <rguenther@suse.de>
18260         * double-int.h (lshift): New overload without precision
18261         and arith argument.
18262         (operator *=, operator +=, operator -=): Move ...
18263         * double-int.c (operator *=, operator +=, operator -=): ... here
18264         and implement more efficiently.
18265         (mul_double_with_sign): Remove.
18266         (lshift_double): Adjust to take unsinged shift argument, push
18267         dispatching code to callers.
18268         (mul_double_wide_with_sign): Add early out for callers that
18269         are not interested in high parts or overflow.
18270         (lshift): New function.
18271         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18272         dispatch code here.
18273         (lrotate, rrotate): Use logical shifts.
18274         * expr.c (get_inner_reference): Use lshift.
18275         * fixed-value.c (do_fixed_divide): Likewise.
18276         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18277         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18278         (indirect_refs_may_alias_p): Likewise.
18279         (stmt_kills_ref_p_1): Likewise.
18281 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18283         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18285 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18287         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18288         scalar form of FABD instruction.
18290 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18292         * lra-constraints.c (process_alt_operands): Add checking alt
18293         number to choose the best alternative.
18295 2013-05-02  Richard Biener  <rguenther@suse.de>
18297         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18298         bitmap and its handling.
18299         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18301 2013-05-02  Richard Biener  <rguenther@suse.de>
18303         PR middle-end/57140
18304         * tree-inline.c (copy_loops): Properly handle removed loops.
18305         (copy_cfg_body): Mark destination loops for fixup if source
18306         loops needed fixup.
18308 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18310         PR target/56732
18311         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18312         generating simple_return for naked functions.
18314 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18316         PR middle-end/56988
18317         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18318         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18319         flags match.
18320         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18321         ipa_agg_replacement_value structures.
18322         (known_aggs_to_agg_replacement_list): Likewise.
18323         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18324         (read_agg_replacement_chain): Likewise.
18325         (ipcp_transform_function): Also check that by_ref flags match.
18327 2013-05-02  Richard Biener  <rguenther@suse.de>
18329         * graphds.h (struct graph): Add obstack member.
18330         * graphds.c (new_graph): Initialize obstack and allocate
18331         vertices from it.
18332         (add_edge): Allocate edge from the obstack.
18333         (free_graph): Free the obstack instead of all edges and vertices.
18335 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18337         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18338         divides.
18339         * cfg.c (update_bb_profile_for_threading): Ditto.
18340         * tree-inline.c (copy_bb): Ditto.
18341         (copy_edges_for_bb): Ditto.
18342         (initialize_cfun): Ditto.
18343         (copy_cfg_body): Ditto.
18344         (expand_call_inline): Ditto.
18345         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18346         (estimate_node_size_and_time): Ditto.
18347         (inline_merge_summary): Ditto.
18348         * cgraphclones.c (cgraph_clone_edge): Ditto.
18349         (cgraph_clone_node): Ditto.
18350         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18351         (compute_trg_info): Ditto.
18353 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18355         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18356         S reg when fp attribute set.
18357         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18359 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18361         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18362         New pattern.
18363         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18364         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18365         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18367 2013-05-02  Richard Biener  <rguenther@suse.de>
18369         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18370         (struct instantiate_cache_entry): New type.
18371         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18372         (struct instantiate_cache_type): New type.
18373         (set_instantiated_value, get_instantiated_value): Remove.
18374         (get_instantiated_value_entry): New function.
18375         (instantiate_scev_name): Use the new cache and adjust.
18376         (instantiate_scev_poly): Adjust.
18377         (instantiate_scev_binary): Likewise.
18378         (instantiate_array_ref): Likewise.
18379         (instantiate_scev_convert): Likewise.
18380         (instantiate_scev_not): Likewise.
18381         (instantiate_scev_3): Likewise.
18382         (instantiate_scev_2): Likewise.
18383         (instantiate_scev_r): Likewise.
18384         (instantiate_scev): Likewise.
18385         (resolve_mixers): Likewise.
18387 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18389         PR target/57091
18390         * lra-constraints.c (best_small_class_operands_num): Remove.
18391         (process_alt_operands): Remove small_class_operands_num.  Take
18392         small classes operands into losers and only if the operand is not
18393         matched.  Modify debugging output.
18394         (curr_insn_transform): Remove best_small_class_operands_num.
18395         Print insn name.
18397 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18399         * config/aarch64/aarch64-builtins.c
18400         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18401         * config/aarch64/aarch64-simd-builtins.def
18402         (reduc_splus_): Add new modes.
18403         (reduc_uplus_): New.
18404         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18405         (reduc_uplus_v4sf): Likewise.
18406         (reduc_splus_v4sf): Likewise.
18407         (aarch64_addv<mode>): Likewise.
18408         (reduc_uplus_<mode>): Likewise.
18409         (reduc_splus_<mode>): Likewise.
18410         (aarch64_addvv2di): Likewise.
18411         (reduc_uplus_v2di): Likewise.
18412         (reduc_splus_v2di): Likewise.
18413         (aarch64_addvv2si): Likewise.
18414         (reduc_uplus_v2si): Likewise.
18415         (reduc_splus_v2si): Likewise.
18416         (reduc_<sur>plus_<mode>): New.
18417         (reduc_<sur>plus_v2di): Likewise.
18418         (reduc_<sur>plus_v2si): Likewise.
18419         (reduc_<sur>plus_v4sf): Likewise.
18420         (aarch64_addpv4sf): Likewise.
18421         * config/aarch64/arm_neon.h
18422         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18423         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18424         add UNSPEC_SADDV, UNSPEC_UADDV.
18425         (SUADDV): New.
18426         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18428 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18430         * config/aarch64/arm_neon.h
18431         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18433 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18435         * config/aarch64/aarch64-builtins
18436         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18438 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18440         * config/aarch64/aarch64-simd-builtins.def
18441         (reduc_smax_): New.
18442         (reduc_smin_): Likewise.
18443         (reduc_umax_): Likewise.
18444         (reduc_umin_): Likewise.
18445         (reduc_smax_nan_): Likewise.
18446         (reduc_smin_nan_): Likewise.
18447         (fmax): Remove.
18448         (fmin): Likewise.
18449         (smax): Update for V2SF, V4SF and V2DF modes.
18450         (smin): Likewise.
18451         (smax_nan): New.
18452         (smin_nan): Likewise.
18453         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18454         (<su><maxmin><mode>3): ...This, refactor.
18455         (s<maxmin><mode>3): New.
18456         (<maxmin_uns><mode>3): Likewise.
18457         (reduc_<maxmin_uns>_<mode>): Refactor.
18458         (reduc_<maxmin_uns>_v4sf): Likewise.
18459         (reduc_<maxmin_uns>_v2si): Likewise.
18460         (aarch64_<fmaxmin><mode>: Remove.
18461         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18462         new builtin names.
18463         (vmin<q>_f<32,64>): Likewise.
18464         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18465         (FMAXMIN): New.
18466         (su): Add mappings for smax, smin, umax, umin.
18467         (maxmin): New.
18468         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18469         (FMAXMIN): Rename as...
18470         (FMAXMIN_UNS): ...This.
18471         (maxminv): Remove.
18472         (fmaxminv): Likewise.
18473         (fmaxmin): Likewise.
18474         (maxmin_uns): New.
18475         (maxmin_uns_op): Likewise.
18477 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18479         * config/aarch64/arm_neon.h
18480         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18481         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18482         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18484 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18486         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18487         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18489 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18491         * config/aarch64/aarch64-simd.md
18492         (vcond<mode>_internal): Handle special cases for constant masks.
18493         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18494         (vcondu<mode><mode>): Likewise.
18495         (vcond<v_cmp_result><mode>): New.
18497 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18499         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18500         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18501         * config/aarch64/aarch64-simd-builtins.def
18502         (cmeq): Update to BUILTIN_VALLDI.
18503         (cmgt): Likewise.
18504         (cmge): Likewise.
18505         (cmle): Likewise.
18506         (cmlt): Likewise.
18507         * config/aarch64/arm_neon.h
18508         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18509         to builtins or C as appropriate.
18511 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18513         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18514         (cmgeu): ...This.
18515         (cmhi): Rename to...
18516         (cmgtu): ...This.
18517         * config/aarch64/aarch64-simd.md
18518         (simd_mode): Add SF.
18519         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18520         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18521         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18522         (cstore<mode>_neg): ...This.
18523         * config/aarch64/iterators.md
18524         (VALLF): new.
18525         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18526         (COMPARISONS): New.
18527         (UCOMPARISONS): Likewise.
18528         (optab): Add missing comparisons.
18529         (n_optab): New.
18530         (cmp_1): Likewise.
18531         (cmp_2): Likewise.
18532         (CMP): Likewise.
18533         (cmp): Remove.
18534         (VCMP_S): Likewise.
18535         (VCMP_U): Likewise.
18536         (V_cmp_result): Add DF, SF modes.
18537         (v_cmp_result): Likewise.
18538         (v): Likewise.
18539         (vmtype): Likewise.
18540         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18542 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18544         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18545         define_insn to define_insn_and_split.
18546         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18547         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18548         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18549         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18550         (thumb2_negscc): Likewise.
18552 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18554         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18556 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18558         * config/arm/thumb2.md: Remove trailing whitespaces.
18560 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18562         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18563         Use gen_int_mode rather than GEN_INT.
18565 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18567         * value-prof.c (stream_in_histogram_value): Remove the strayed
18568         debug_gimple_stmt.
18570 2013-04-30  Richard Biener  <rguenther@suse.de>
18572         PR middle-end/57122
18573         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18575 2013-04-30  Richard Biener  <rguenther@suse.de>
18577         PR middle-end/57107
18578         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18580 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18582         PR rtl-optimization/56957
18583         PR rtl-optimization/57105
18584         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18585         variable.  Use just INSN_UID for determining whether an insn
18586         should be only disconnected from the insn stream.
18587         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18589 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18591         PR tree-optimization/57104
18592         * tsan.c (instrument_expr): Don't instrument accesses to
18593         DECL_HARD_REGISTER VAR_DECLs.
18595 2013-04-30  Richard Biener  <rguenther@suse.de>
18597         * function.h (loops_for_fn): New inline function.
18598         (set_loops_for_fn): Likewise.
18599         * cfgloop.h (place_new_loop): Add struct function parameter.
18600         (get_loop): Likewise.
18601         (get_loops): Likewise.
18602         (number_of_loops): Likewise.
18603         (fel_next): Adjust.
18604         (fel_init): Likewise.
18605         * cfg.c (get_loop_copy): Adjust.
18606         * cfgloop.c (flow_loops_dump): Likewise.
18607         (record_loop_exits): Likewise.
18608         (verify_loop_structure): Likewise.
18609         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18610         (estimate_reg_pressure_cost): Likewise.
18611         (mark_loop_exit_edges): Likewise.
18612         * cfgloopmanip.c (place_new_loop): Likewise.
18613         (add_loop): Likewise.
18614         (duplicate_loop): Likewise.
18615         * graph.c (draw_cfg_nodes): Likewise.
18616         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18617         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18618         (extract_affine_chrec): Likewise.
18619         (build_scop_iteration_domain): Likewise.
18620         * graphite.c (graphite_initialize): Likewise.
18621         * ira-build.c (create_loop_tree_nodes): Likewise.
18622         (more_one_region_p): Likewise.
18623         (rebuild_regno_allocno_maps): Likewise.
18624         (mark_loops_for_removal): Likewise.
18625         (mark_all_loops_for_removal): Likewise.
18626         (remove_unnecessary_regions): Likewise.
18627         (ira_build): Likewise.
18628         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18629         * loop-init.c (fix_loop_structure): Likewise.
18630         (gate_rtl_move_loop_invariants): Likewise.
18631         (gate_rtl_unswitch): Likewise.
18632         (gate_rtl_unroll_and_peel_loops): Likewise.
18633         (rtl_doloop): Likewise.
18634         * lto-streamer-in.c (input_cfg): Likewise.
18635         * lto-streamer-out.c (output_cfg): Likewise.
18636         * modulo-sched.c (sms_schedule): Likewise.
18637         * predict.c (tree_estimate_probability): Likewise.
18638         (tree_estimate_probability_driver): Likewise.
18639         (estimate_loops): Likewise.
18640         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18641         (move_sese_region_to_fn): Likewise.
18642         (debug_loop_num): Likewise.
18643         * tree-chrec.c (chrec_evaluate): Likewise.
18644         (hide_evolution_in_other_loops_than_loop): Likewise.
18645         (chrec_component_in_loop_num): Likewise.
18646         (reset_evolution_in_loop): Likewise.
18647         (evolution_function_is_invariant_rec_p): Likewise.
18648         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18649         * tree-inline.c (copy_loops): Likewise.
18650         (copy_cfg_body): Likewise.
18651         (tree_function_versioning): Likewise.
18652         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18653         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18654         Likewise.
18655         (add_to_evolution_1): Likewise.
18656         (scev_const_prop): Likewise.
18657         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18658         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18659         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18660         (tree_ssa_lim_initialize): Likewise.
18661         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18662         (verify_loop_closed_ssa): Likewise.
18663         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18664         (tree_ssa_loop_im): Likewise.
18665         (tree_ssa_loop_unswitch): Likewise.
18666         (tree_vectorize): Likewise.
18667         (check_data_deps): Likewise.
18668         (tree_ssa_loop_ivcanon): Likewise.
18669         (tree_ssa_loop_bounds): Likewise.
18670         (tree_complete_unroll): Likewise.
18671         (tree_complete_unroll_inner): Likewise.
18672         (tree_parallelize_loops): Likewise.
18673         (tree_ssa_loop_prefetch): Likewise.
18674         (tree_ssa_loop_ivopts): Likewise.
18675         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18676         * tree-vectorizer.c (vectorize_loops): Likewise.
18678 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18680         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18681         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18682         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18683         with EABI_LINK_SPEC.
18685 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18687         PR target/44578
18688         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18689         alternative.
18691 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18693         PR target/57097
18694         * lra-constraints.c (process_alt_operands): Discourage a bit more
18695         using memory for pseudos.  Print cost dump for alternatives.
18696         Modify cost values for conflicts with early clobbers.
18697         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18699 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18701         PR target/57098
18702         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18704 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18706         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18707         from/to S register.
18708         (movdi_aarch64): Support LDR/STR from/to D register.
18710 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18712         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18713         or higher by default.
18715 2013-04-29  Richard Biener  <rguenther@suse.de>
18717         PR middle-end/57075
18718         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18719         even if not adding abnormal edges for calls that can make
18720         abnormal gotos.
18722 2013-04-29  Richard Biener  <rguenther@suse.de>
18724         PR middle-end/57103
18725         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18726         TREE_BLOCK.
18727         (move_stmt_r): Remove redundant checking.
18729 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18731         PR bootstrap/57077
18732         * basic-block.h (apply_scale): New function.
18733         (apply_probability): Use apply_scale.
18734         * gimple-streamer-in.c (input_bb): Ditto.
18735         * lto-streamer-in.c (input_cfg): Ditto.
18736         * lto-cgraph.c (merge_profile_summaries): Ditto.
18737         * tree-optimize.c (execute_fixup_cfg): Ditto.
18738         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18739         (copy_edges_for_bb): Ditto.
18740         (copy_cfg_body): Ditto.
18742 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18744         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18745         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18746         (tail_merge_optimize): Handle current_loops == NULL.
18748 2013-04-26  Jeff Law  <law@redhat.com>
18750         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18751         (simplify_cond_using_ranges): Generalize code to simplify
18752         COND_EXPRs where one argument is a constant and the other
18753         is an SSA_NAME created by an integral type conversion.
18755 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18757         * config/arm/arm.md (store_minmaxsi): Use only when
18758         optimize_insn_for_size_p.
18760 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18762         PR target/57108
18763         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18765 2013-04-29  Richard Biener  <rguenther@suse.de>
18767         PR middle-end/57089
18768         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18769         loop tree make sure to schedule a fixup for the child as well.
18770         (expand_omp_for_generic): Properly add loops.
18771         (expand_omp_for_static_nochunk): Likewise.
18772         (expand_omp_for_static_chunk): Likewise.
18773         (expand_omp_for): For the degenerate case fixup loops.
18774         (expand_omp_sections): Fix default bb placement in loops.
18775         (expand_omp_atomic_pipeline): Properly add loops.
18777 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18779         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18781 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18783         * tree-ssa-tail-merge.c: Update header comment.
18785 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18787         * config/aarch64/arm_neon.h
18788         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18789         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18790         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18791         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18792         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18793         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18794         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18795         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18797 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18799         * config/aarch64/aarch64-simd.md
18800         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18801         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18802         fix_trunc, fixuns_trunc.
18803         (ftrunc<VDQF:mode>2): New.
18804         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18805         (fix_trunc_optab): New.
18807 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18809         * config/aarch64/aarch64-builtins.c
18810         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18811         iceilf, lround, iroundf.
18813 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18815         PR target/54349
18816         * config/i386/i386.h (enum ix86_tune_indices)
18817         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18818         New, split from X86_TUNE_INTER_UNIT_MOVES.
18819         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18820         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18821         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18822         (TARGET_INTER_UNIT_MOVES): Remove.
18823         * config/i386/i386.c (initial_ix86_tune_features): Update.
18824         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18825         (ix86_expand_convert_uns_didf_sse): Use
18826         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18827         (ix86_expand_vector_init_one_nonzero): Ditto.
18828         (ix86_expand_vector_init_interleave): Ditto.
18829         (inline_secondary_memory_needed): Return true for moves from SSE class
18830         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18831         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18832         * config/i386/constraints.md (Yi, Ym): Depend on
18833         TARGET_INTER_UNIT_MOVES_TO_VEC.
18834         (Yj, Yn): New constraints.
18835         * config/i386/i386.md (*movdi_internal): Change constraints of
18836         operand 1 from Yi to Yj and from Ym to Yn.
18837         (*movsi_internal): Ditto.
18838         (*movdf_internal): Ditto.
18839         (*movsf_internal): Ditto.
18840         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18841         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18842         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18843         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18844         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18845         * config/i386/sse.md (movdi_to_sse): Ditto.
18846         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18847         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18848         TARGET_INTER_UNIT_MOVES.
18849         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18850         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18851         instead of TARGET_INTER_UNIT_MOVES.
18852         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18853         operand 1 from Yi to Yj and from Ym to Yn.
18855 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18857         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18858         (float_truncate_hi_): Likewise.
18859         (float_extend_lo_): Likewise.
18860         (float_truncate_lo_): Likewise.
18861         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18862         (aarch64_float_extend_lo_v2df): Likewise.
18863         (vec_unpacks_hi_v4sf): Likewise.
18864         (aarch64_float_truncate_lo_v2sf): Likewise.
18865         (aarch64_float_truncate_hi_v4sf): Likewise.
18866         (vec_pack_trunc_v2df): Likewise.
18867         (vec_pack_trunc_df): Likewise.
18869 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18871         * config/aarch64/aarch64-builtins.c
18872         (aarch64_fold_builtin): Fold float conversions.
18873         * config/aarch64/aarch64-simd-builtins.def
18874         (floatv2si, floatv4si, floatv2di): New.
18875         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18876         * config/aarch64/aarch64-simd.md
18877         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18878         * config/aarch64/iterators.md (FLOATUORS): New.
18879         (optab): Add float, floatuns.
18880         (su_optab): Likewise.
18882 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18884         * config/aarch64/aarch64-builtins.c
18885         (aarch64_builtin_vectorized_function): Use new names for
18886         fcvt builtins.
18887         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18888         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18889         (fcvtzu): Split as...
18890         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18891         (fcvtas): Split as...
18892         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18893         (fcvtau): Split as...
18894         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18895         (fcvtps): Split as...
18896         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18897         (fcvtpu): Split as...
18898         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18899         (fcvtms): Split as...
18900         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18901         (fcvtmu): Split as...
18902         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18903         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18904         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18905         (lfrintnusf, lfrintnudf): Likewise.
18906         * config/aarch64/aarch64-simd.md
18907         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18908         define_insn.
18909         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18910         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18911         (fcvt_pattern): Likewise.
18913 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18915         * config/aarch64/aarch64-simd.md
18916         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18917         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18919 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18921         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18922         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18923         (vrnd<a,m,n,p>_f32): Implement using builtins.
18924         (vrnd<i,x><q>_f<32, 64>): New.
18926 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18928         * config/aarch64/aarch64-builtins.c
18929         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18930         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18931         (frintz): Rename to...
18932         (btrunc): ...this.
18933         (frintp): Rename to...
18934         (ceil): ...this.
18935         (frintm): Rename to...
18936         (floor): ...this.
18937         (frinti): Rename to...
18938         (nearbyint): ...this.
18939         (frintx): Rename to...
18940         (rint): ...this.
18941         (frinta): Rename to...
18942         (round): ...this.
18943         * config/aarch64/aarch64-simd.md
18944         (aarch64_frint<frint_suffix><mode>): Delete.
18945         (<frint_pattern><mode>2): Convert to insn.
18946         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18947         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18948         (frint_pattern): Likewise.
18949         (frint_suffix): Likewise.
18951 2013-04-29  Richard Biener  <rguenther@suse.de>
18953         PR tree-optimization/57081
18954         * loop-init.c: Include tree-flow.h.
18955         (loop_optimizer_finalize): Free number of iteration estimates.
18956         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18958 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18960         PR tree-optimization/57083
18961         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18962         non-singleton shift count range, zero extend low_bound for uns case.
18964         * config/i386/predicates.md (general_vector_operand): New predicate.
18965         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18966         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18967         if they aren't nonimmediate operands.  If their original values
18968         satisfy const_vector_equal_evenodd_p, don't shift them.
18969         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18970         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18971         if not nonimmediate_operand.
18972         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18973         instead of register_operand.
18974         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18976 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18978         * stor-layout.c (finalize_size_functions): Allocate a structure and
18979         reset cfun before dumping the functions.
18981 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18983         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18985         PR target/56866
18986         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18987         use xop_pmacsdqh if uns_p.
18988         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18989         the immediate rotate count.
18991 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18993         * rtl.h (struct rtx_def): Add comment for field jump.
18994         (LRA_SUBREG_P): New macro.
18995         * recog.c (register_operand): Check LRA_SUBREG_P.
18996         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18997         stack frame.
18998         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18999         (lra_final_code_change): Skip subreg change for operators.
19000         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19001         if there are no operand changes.
19002         * lra-constraints.c (curr_insn_set): New.
19003         (match_reload): Set LRA_SUBREG_P.
19004         (emit_spill_move): Ditto.
19005         (check_and_process_move): Use curr_insn_set. Process only single
19006         set insns.  Don't initialize sec_mem_p and change_p.
19007         (simplify_operand_subreg): Use LRA_SUBREG_P.
19008         (reg_in_class_p): New function.
19009         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19010         of #ifdef.  Add code to remove cycling.
19011         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19012         non-null disp.  Reload inner instead of disp when base and index
19013         are null.  Try to put lo_sum into register.
19014         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19015         (check_and_process_move): Move code for move cost check to
19016         simple_move_p.  Remove equiv_substitution.
19017         (simple_move_p): New function.
19018         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19019         curr_insn_set.  Call check_and_process_move only for single set
19020         insns.  Use the new function.  Move call of check_and_process_move
19021         after operand equiv substitution and address process.
19023 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19025         PR go/57045
19026         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19027         with nonlocal goto receivers or returns twice calls, ignore
19028         unininitialized values from abnormal edges to nl goto receiver
19029         or returns twice call.
19031 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19033         PR tree-optimization/57051
19034         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19035         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19036         bitsize.
19038 2013-04-26  Richard Biener  <rguenther@suse.de>
19040         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19041         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19042         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19043         (fixup_loop_arrays_after_move): New function.
19044         (move_sese_region_to_fn): Properly outline the loop tree parts
19045         of the SESE region.
19047 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19049         * config/i386/i386.md (type, unit): Fix long lines.
19051 2013-04-26  Richard Biener  <rguenther@suse.de>
19053         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19054         (lto-streamer-out.o): Likewise.
19055         * cfgloop.c (init_loops_structure): Export, add struct function
19056         argument and adjust.
19057         (flow_loops_find): Adjust.
19058         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19059         (init_loops_structure): Declare.
19060         * lto-streamer-in.c: Include cfgloop.h.
19061         (input_cfg): Input the loop tree.
19062         * lto-streamer-out.c: Include cfgloop.h.
19063         (output_cfg): Output the loop tree.
19064         (output_struct_function_base): Do not drop PROP_loops.
19066 2013-03-26  Richard Biener  <rguenther@suse.de>
19068         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19069         (pass_build_cfg): Provide PROP_loops.
19070         (move_sese_region_to_fn): Remove loops that are outlined into fn
19071         for now.
19072         * tree-inline.c: Include cfgloop.h.
19073         (initialize_cfun): Do not drop PROP_loops.
19074         (copy_loops): New function.
19075         (copy_cfg_body): Copy loop structure.
19076         (tree_function_versioning): Initialize destination loop tree.
19077         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19078         (pass_parallelize_loops): Do IL verification.
19079         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19080         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19081         the CFG make sure we fixup loops as well.
19082         * tree-ssa-tail-merge.c: Include cfgloop.h.
19083         (replace_block_by): When merging loop latches mark loops for fixup.
19084         * lto-streamer-out.c (output_struct_function_base): Drop
19085         PROP_loops for now.
19086         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19087         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19088         * ipa-split.c: Include cfgloop.h.
19089         (split_function): Add the new return block to the loop tree root.
19090         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19091         whether we have removed the forwarder block.
19092         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19093         * cfgloop.h (place_new_loop): Declare.
19094         * cfgloopmanip.c (place_new_loop): Export.
19095         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19096         (tree-switch-conversion.o): Likewise.
19097         (tree-complex.o): Likewise.
19098         (tree-inline.o): Likewise.
19099         (tree-ssa-tailmerge.o): Likewise.
19100         (ipa-split.o): Likewise.
19101         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19102         (tree-ssa-copy.o): Likewise.
19103         * tree-switch-conversion.c: Include cfgloop.h
19104         (process_switch): If we emit a bit-test cascade, schedule loops
19105         for fixup.
19106         * tree-complex.c: Include cfgloop.h.
19107         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19108         * asan.c: Include cfgloop.h.
19109         (create_cond_insert_point): Properly add new basic-blocks to
19110         loops, schedule loop fixup.
19111         * cfgloop.c (verify_loop_structure): Check that looks are not
19112         marked for fixup.
19113         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19114         to loops.
19115         (expand_omp_for_generic): Likewise.
19116         (expand_omp_sections): Likewise.
19117         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19118         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19119         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19120         is initialized, not when loops are present.
19121         * tree-parloops.c (parallelize_loops): Remove checking here.
19122         * passes.c (init_optimization_passes): Schedule a copy-propagation
19123         pass before complete unrolling of inner loops.
19125 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19127         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19128         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19129         (colorize_init): Add argument to _WIN32 version.
19130         * toplev.c: Include diagnostic-color.h.
19131         (process_options): Default to -fdiagnostics-color=auto if
19132         GCC_COLORS env var is in the environment.
19133         * common.opt (fdiagnostics-color=): Add Var and Init.
19134         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19135         env var is in the environment, the default is auto rather than never.
19137         * diagnostic.h (file_name_as_prefix): Add context argument.
19138         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19139         the string as locus.
19140         * langhooks.c (lhd_print_error_function): Adjust caller.
19142 2013-04-25  Lawrence Crowl  <crowl@google.com>
19144         * var-tracking.c (shared_hash_def::htab):
19145         Change type to hash_table.  Update dependent calls and types.
19147 2013-04-25  Lawrence Crowl  <crowl@google.com>
19149         * Makefile.in: Update as needed below.
19151         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19152         Move declaration to after the type's method definitons.
19154         * attribs.c (htab_t scoped_attributes::attribute_hash):
19155         Change type to hash_table.  Update dependent calls and types.
19157         * bitmap.c (htab_t bitmap_desc_hash):
19158         Change type to hash_table.  Update dependent calls and types.
19160         * cselib.c (htab_t cselib_hash_table):
19161         Change type to hash_table.  Update dependent calls and types.
19163         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19164         (hash_string_slot_node): Move implementation into lto-streamer.h
19165         struct string_slot_hasher.
19166         (eq_string_slot_node): Likewise.
19168         * data-streamer-out.c: Update output_block::string_hash_table
19169         dependent calls and types.
19171         * dwarf2cfi.c (htab_t trace_index):
19172         Change type to hash_table.  Update dependent calls and types.
19174         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19175         Change type to hash_table.  Update dependent calls and types.
19176         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19177         (htab_t optimize_external_refs::map): Likewise.
19178         (htab_t output_comp_unit::extern_map): Likewise.
19179         (htab_t output_comdat_type_unit::extern_map): Likewise.
19180         (htab_t output_macinfo::macinfo_htab): Likewise.
19181         (htab_t optimize_location_lists::htab): Likewise.
19182         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19184         * except.c (htab_t ehspec_hash_type):
19185         Change type to hash_table.  Update dependent calls and types.
19186         (assign_filter_values::ttypes): Likewise.
19187         (assign_filter_values::ehspec): Likewise.
19188         (sjlj_assign_call_site_values::ar_hash): Likewise.
19189         (convert_to_eh_region_ranges::ar_hash): Likewise.
19191         * gcse.c (htab_t pre_ldst_table):
19192         Change type to hash_table.  Update dependent calls and types.
19194         * ggc-common.c (htab_t saving_htab):
19195         Change type to hash_table.  Update dependent calls and types.
19196         (htab_t loc_hash): Likewise.
19197         (htab_t ptr_hash): Likewise.
19198         (call_count): Rename ggc_call_count.
19199         (call_alloc): Rename ggc_call_alloc.
19200         (loc_descriptor): Rename make_loc_descriptor.
19201         (add_statistics): Rename ggc_add_statistics.
19203         * ggc-common.c (saving_htab):
19204         Change type to hash_table.  Update dependent calls and types.
19206         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19207         (push_gimplify_context): Likewise.
19208         (pop_gimplify_context): Likewise.
19209         (struct gimple_temp_hash_elt): Added.
19210         (struct gimplify_hasher): Likewise.
19211         (struct gimplify_ctx.temp_htab):
19212         Change type to hash_table.  Update dependent calls and types.
19214         * gimple-fold.c: Include gimplify-ctx.h.
19216         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19217         Change type to hash_table.  Update dependent calls and types.
19218         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19219         avoid potential global name collision.
19221         * gimplify.c: Include gimplify-ctx.h.
19222         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19223         (htab_t gimplify_ctx::temp_htab):
19224         Update dependent calls and types for new type hash_table.
19225         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19226         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19228         * gimplify-ctx.h: New.
19229         (struct gimple_temp_hash_elt): Move from gimplify.c.
19230         (class gimplify_hasher): New.
19231         (struct gimplify_ctx): Move from gimple.h.
19232         (htab_t gimplify_ctx::temp_htab):
19233         Change type to hash_table.  Update dependent calls and types.
19235         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19236         (htab_t ivs_params::newivs_index):
19237         Change type to hash_table.  Update dependent calls and types.
19238         (htab_t ivs_params::params_index): Likewise.
19239         (htab_t print_generated_program::params_index): Likewise.
19240         (htab_t gloog::newivs_index): Likewise.
19241         (htab_t gloog::params_index): Likewise.
19243         * graphite.c: Include graphite-htab.h.
19244         4htab_t graphite_transform_loops::bb_pbb_mapping):
19245         Change type to hash_table.  Update dependent calls and types.
19247         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19248         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19249         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19251         * graphite-dependences.c: Include graphite-htab.h.
19252         (loop_is_parallel_p): Change hash table type of parameter.
19254         * graphite-htab.h: New.
19255         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19256         (extern find_pbb_via_hash): Move from graphite-poly.h.
19257         (extern loop_is_parallel_p): Move from graphite-poly.h.
19258         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19260         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19261         (extern loop_is_parallel_p): Move to graphite-htab.h.
19262         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19264         * haifa-sched.c (htab_t delay_htab):
19265         Change type to hash_table.  Update dependent calls and types.
19266         (htab_t delay_htab_i2): Likewise.
19268         * ira-color.c (htab_t allocno_hard_regs_htab):
19269         Change type to hash_table.  Update dependent calls and types.
19271         * ira-costs.c (htab_t cost_classes_htab):
19272         Change type to hash_table.  Update dependent calls and types.
19274         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19275         Change type to hash_table.  Update dependent calls and types.
19277         * loop-iv.c (htab_t bivs):
19278         Change type to hash_table.  Update dependent calls and types.
19280         * loop-unroll.c (htab_t opt_info::insns_to_split):
19281         Change type to hash_table.  Update dependent calls and types.
19282         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19284         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19285         (struct string_slot_hasher): New.
19286         (htab_t output_block::string_hash_table):
19287         Change type to hash_table.  Update dependent calls and types.
19289         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19290         (htab_t file_name_hash_table):
19291         Change type to hash_table.  Update dependent calls and types.
19293         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19294         calls and types.
19296         * lto-streamer.c (htab_t tree_htab):
19297         Change type to hash_table.  Update dependent calls and types.
19299         * omp-low.c: Include gimplify-ctx.h.
19301         * passes.c (htab_t name_to_pass_map):
19302         Change type to hash_table.  Update dependent calls and types.
19303         (pass_traverse): Rename to passes_pass_traverse.
19305         * plugin.c (htab_t event_tab):
19306         Change type to hash_table.  Update dependent calls and types.
19308         * postreload-gcse.c (htab_t expr_table):
19309         Change type to hash_table.  Update dependent calls and types.
19310         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19312         * sese.c (debug_rename_map_1): Make extern.
19313         (htab_t copy_bb_and_scalar_dependences::rename_map):
19314         Change type to hash_table.  Update dependent calls and types.
19316         * sese.h (extern debug_rename_map): Move to .c file.
19318         * store-motion.c (htab_t store_motion_mems_table):
19319         Change type to hash_table.  Update dependent calls and types.
19321         * trans-mem.c (htab_t tm_new_mem_hash):
19322         Change type to hash_table.  Update dependent calls and types.
19324         * tree-browser.c (htab_t TB_up_ht):
19325         Change type to hash_table.  Update dependent calls and types.
19327         * tree-cfg.c (htab_t discriminator_per_locus):
19328         Change type to hash_table.  Update dependent calls and types.
19330         * tree-complex.c: Include tree-hasher.h
19331         (htab_t complex_variable_components):
19332         Change type to hash_table.  Update dependent calls and types.
19334         * tree-eh.c (htab_t finally_tree):
19335         Change type to hash_table.  Update dependent calls and types.
19337         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19338         struct int_tree_hasher.
19339         (extern int_tree_map_eq): Likewise.
19340         (uid_decl_map_hash): Removed.
19341         (extern decl_tree_map_eq): Likewise.
19343         * tree-hasher.h: New.
19344         (struct int_tree_hasher): New.
19345         (typedef int_tree_htab_type): New.
19347         * tree-inline.c: Include gimplify-ctx.h.
19349         * tree-mudflap.c: Include gimplify-ctx.h.
19351         * tree-parloops.c: Include tree-hasher.h.
19352         (htab_t eliminate_local_variables_stmt::decl_address):
19353         Change type to hash_table.  Update dependent calls and types.
19354         (htab_t separate_decls_in_region::decl_copies): Likewise.
19356         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19357         Change type to hash_table.  Update dependent calls and types.
19359         * tree-sra.c (candidates):
19360         Change type to hash_table.  Update dependent calls and types.
19362         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19363         in tree-flow.h.
19364         (int_tree_map_hash): Likewise.
19366         * tree-ssa-dom.c (htab_t avail_exprs):
19367         Change type to hash_table.  Update dependent calls and types.
19369         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19370         Change type to hash_table.  Update dependent calls and types.
19372         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19373         Change type to hash_table.  Update dependent calls and types.
19375         * tree-ssa-phiopt.c (seen_ssa_names):
19376         Change type to hash_table.  Update dependent calls and types.
19378         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19379         Change type to hash_table.  Update dependent calls and types.
19381         * tree-ssa-uncprop.c (equiv):
19382         Change type to hash_table.  Update dependent calls and types.
19384 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19386         PR rtl-optimization/57003
19387         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19388         call note_stores with kill_clobbered_value callback again after
19389         killing regs_invalidated_by_call.
19391 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19393         * config/aarch64/aarch64-simd.md
19394         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19395         (aarch64_simd_bsl<mode>): Likewise.
19396         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19398 2013-04-25  Marek Polacek  <polacek@redhat.com>
19400         PR tree-optimization/57066
19401         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19403 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19405         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19407 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19409         * config/aarch64/aarch64-builtins.c
19410         (aarch64_fold_builtin): New.
19411         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19412         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19413         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19414         * config/aarch64/arm_neon.h
19415         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19417 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19418             Tejas Belagod  <tejas.belagod@arm.com>
19420         * config/aarch64/aarch64-builtins.c
19421         (aarch64_gimple_fold_builtin): New.
19422         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19423         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19424         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19425         (addvv4sf): Update.
19426         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19428 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19430         * config/aarch64/aarch64.md
19431         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19433 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19435         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19436         (*ngcsi_uxtw): New pattern.
19438 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19439             Julian Brown  <julian@codesourcery.com>
19441         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19442         (TB_DREG): Add T_V4HF.
19443         (v4hf_UP): New macro.
19444         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19445         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19446         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19447         built-ins.
19448         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19449         (arm_vector_mode_supported_p): Handle V4HF.
19450         (arm_mangle_map): Handle V4HFmode.
19451         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19452         * config/arm/arm_neon_builtins.def: Add entries for
19453         vcvtv4hfv4sf, vcvtv4sfv4hf.
19454         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19455         (neon_vcvtv4hfv4sf): Likewise.
19456         * config/arm/neon-gen.ml: Handle half-precision floating point
19457         features.
19458         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19459         * config/arm/arm_neon.h: Regenerate.
19460         * config/arm/neon.ml (type elts): Add F16.
19461         (type vectype): Add T_float16x4, T_floatHF.
19462         (type vecmode): Add V4HF.
19463         (type features): Add Requires_FP_bit feature.
19464         (elt_width): Handle F16.
19465         (elt_class): Likewise.
19466         (elt_of_class_width): Likewise.
19467         (mode_of_elt): Refactor.
19468         (type_for_elt): Handle F16, fix error messages.
19469         (vectype_size): Handle T_float16x4.
19470         (vcvt_sh): New function.
19471         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19472         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19473         (string_of_mode): Handle V4HF.
19474         * doc/arm-neon-intrinsics.texi: Regenerate.
19476 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19478         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19479         format specifier in 'X' case.
19481 2013-04-25  Alan Modra  <amodra@gmail.com>
19483         PR target/57052
19484         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19485         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19486         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19487         Repeat for many other rotate/shift and mask patterns using subregs.
19488         Name lshiftrt insns.
19489         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19490         on WORDS_BIG_ENDIAN.
19492 2013-04-25  Alan Modra  <amodra@gmail.com>
19494         * config.gcc: Support little-endian powerpc-linux targets.
19495         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19496         (LINK_OS_LINUX_SPEC): Define.
19497         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19498         Preserve MASK_LITTLE_ENDIAN.
19499         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19500         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19501         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19502         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19503         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19504         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19505         Correct fp word order for little-endian.  Don't shift toc entries
19506         smaller than a word for little-endian.
19507         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19508         (bswapdi2 splits): Correct low-part subreg for little-endian.
19509         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19510         low/high where such is correct only for be.
19511         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19512         little-endian for -mcall-aixdesc.
19514 2013-04-25  Alan Modra  <amodra@gmail.com>
19516         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19517         replace_equiv_address_nv.
19519 2013-04-25  Alan Modra  <amodra@gmail.com>
19521         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19523 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19525         Revert:
19526         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19527         * rtl.h (struct rtx_def): ...
19529 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19531         PR rtl-optimizations/57046
19532         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19533         for multi-reg splits.
19535 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19537         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19539 2013-04-24  Sterling Augustine  <saugustine@google.com>
19541         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19542         (comp_dir_string, debug_str_dwo_section): New.
19543         (DEBUG_STR_DWO_SECTION): Rename to ...
19544         (DEBUG_DWO_STR_SECTION): ... this.
19545         (DEBUG_NORM_STR_SECTION): Delete.
19546         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19547         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19548         (find_AT_string): Move most logic to ...
19549         (find_AT_string_in_table): ... here.  New.
19550         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19551         add_skeleton_AT_string.  Delete logic.
19552         (output_skeleton_debug_sections): Remove call to
19553         add_top_level_skeleton_die_attrs.
19554         (add_comp_dir_attribute): Move logic to comp_dir_string.
19555         (dwarf2out_init): Initialize debug_str_dwo_section.
19556         (output_indirect_string): Call find_string_form.
19557         (output_indirect_strings): Rewrite.
19558         (prune_unused_types): Empty skeleton_debug_str_hash.
19559         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19560         (dwarf2out_finish):  Call output_indirect_strings.
19562 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19564         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19566 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19568         * rtl.h (struct rtx_def): Add comment for field jump.
19569         (LRA_SUBREG_P): New macro.
19570         * recog.c (register_operand): Check LRA_SUBREG_P.
19571         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19572         stack frame.
19573         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19574         (lra_final_code_change): Skip subreg change for operators.
19575         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19576         if there are no operand changes.
19577         * lra-constraints.c (curr_insn_set): New.
19578         (match_reload): Set LRA_SUBREG_P.
19579         (emit_spill_move): Ditto.
19580         (check_and_process_move): Use curr_insn_set. Process only single
19581         set insns.  Don't initialize sec_mem_p and change_p.
19582         (simplify_operand_subreg): Use LRA_SUBREG_P.
19583         (reg_in_class_p): New function.
19584         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19585         of #ifdef.  Add code to remove cycling.
19586         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19587         non-null disp.  Reload inner instead of disp when base and index
19588         are null.  Try to put lo_sum into register.
19589         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19590         (check_and_process_move): Move code for move cost check to
19591         simple_move_p.  Remove equiv_substitution.
19592         (simple_move_p): New function.
19593         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19594         curr_insn_set.  Call check_and_process_move only for single set
19595         insns.  Use the new function.  Move call of check_and_process_move
19596         after operand equiv substitution and address process.
19598 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19600         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19601         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19602         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19604 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19606         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19608 2013-04-24  Marek Polacek  <polacek@redhat.com>
19610         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19611         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19612         (select_loops_exit_conditions): Likewise.
19613         (number_of_iterations_for_all_loops): Likewise.
19614         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19615         (scev_analysis): Likewise.
19617 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19618             Chao-ying Fu  <fu@mips.com>
19620         * config/mips/micromips.md (jraddiusp): New pattern.
19621         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19622         instruction if possible.
19624 2013-04-24  Alan Modra  <amodra@gmail.com>
19626         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19628 2013-04-24  Julian Brown  <julian@codesourcery.com>
19629             Chung-Lin Tang  <cltang@codesourcery.com>
19631         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19632         dependency behavior in enumeration type DIE generation. Add TODO note
19633         to comments about future DW_FORM_sdata/udata re-work of related code.
19635 2013-04-23  Lawrence Crowl  <crowl@google.com>
19637         * Makefile.in: Update as needed below.
19639         * hash-table.h (class hash_table):
19640         Correct many methods with parameter types compare_type to the correct
19641         value_type.  (Correct code was unlikely to notice the change.)
19642         (hash_table::elements_with_deleted) New.
19643         (class hashtable::iterator): New.
19644         (hashtable::begin()): New.
19645         (hashtable::end()): New.
19646         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19648         * statistics.c (statistics_hashes):
19649         Change type to hash_table.  Update dependent calls and types.
19651         * tree-into-ssa.c (var_infos):
19652         Change type to hash_table.  Update dependent calls and types.
19654         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19655         Change type to hash_table.  Update dependent calls and types.
19657         * tree-ssa-loop-im.c (struct mem_ref.refs):
19658         Change type to hash_table.  Update dependent calls and types.
19660         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19661         Change type to hash_table.  Update dependent calls and types.
19663         * tree-ssa-sccvn.c (vn_tables_s::nary):
19664         Change type to hash_table.  Update dependent calls and types.
19665         (vn_tables_s::phis): Likewise.
19666         (vn_tables_s::references): Likewise.
19668         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19669         (vn_reference_eq): Update parameter and return types.
19671         * tree-ssa-structalias.c (pointer_equiv_class_table):
19672         Change type to hash_table.  Update dependent calls and types.
19673         (location_equiv_class_table): Likewise.
19675         * tree-vect-data-refs.c: Consequential changes for making
19676         peeling a hash_table.
19678         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19679         (destroy_loop_vec_info): Dependent hash_table update.
19681         * tree-vectorizer.h (peeling_htab):
19682         Change type to hash_table.  Update dependent calls and types.
19684 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19686         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19687         to check the register content is equal or not.
19688         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19689         to assign register content record.
19690         * lra-eliminations.c (update_reg_eliminate): Use
19691         lra_update_reg_val_offset to update register content offset.
19692         * lra-int.h (struct lra_reg): Add offset member.
19693         (lra_reg_val_equal_p): New static inline function.
19694         (lra_update_reg_val_offset): New static inline function.
19695         (lra_assign_reg_val): New static inline function.
19696         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19697         to assign register content record.
19698         (initialize_lra_reg_info_element): Initial offset to zero.
19700 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19702         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19703         operands.  Record compression.
19705 2013-04-23  Xinliang David Li  <davidxl@google.com>
19707         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19709 2013-04-23  Richard Biener  <rguenther@suse.de>
19711         PR middle-end/57036
19712         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19713         parameter, only add abnormal goto edges from the copied body
19714         if the call could perform abnormal gotos.
19715         (copy_cfg_body): Adjust.
19717 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19719         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19721 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19723         * coretypes.h (gimple_stmt_iterator): Add struct to make
19724         compatible with C.
19726 2013-04-23  Richard Biener  <rguenther@suse.de>
19728         PR tree-optimization/57026
19729         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19730         from SSA names occuring in abnormal PHI nodes.
19732 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19734         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19736 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19738         * lto/lto.c (print_lto_report_1): Declare early.
19739         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19741 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19743         * common.opt (-flto-report-wpa): Add.
19744         * doc/invoke.texi (-flto-report-wpa): Add.
19745         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19746         (lto_main): dito.
19748 2013-04-22  Xinliang David Li  <davidxl@google.com>
19750         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19751         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19752         * Makefile.in: New dependency
19754             David Daney <ddaney.cavm@gmail.com>
19756         * configure.ac (gcc_cv_as_micromips_support):  Use the
19757         --fatal-warnings option.
19758         * configure: Regenerate.
19760 2013-04-22  Marek Polacek  <polacek@redhat.com>
19762         PR sanitizer/56990
19763         * tsan.c (instrument_expr): Don't instrument expression
19764         in case its size is zero.
19766 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19768         PR target/57032
19769         Revert:
19770         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19772         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19774 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19776         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19777         (gimple_stmt_iterator): New typedef.
19778         * gimple.h (gimple_stmt_iterator): Rename to...
19779         (gimple_stmt_iterator_d): ... This.
19780         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19781         trees be valid for GIMPLE and GENERIC.
19782         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19783         * gimple-fold.c (gimple_fold_call): Call target hook
19784         gimple_fold_builtin.
19785         * hooks.c (hook_bool_gsiptr_false): New.
19786         * hooks.h (hook_bool_gsiptr_false): New.
19787         * target.def (fold_stmt): New.
19788         * doc/tm.texi: Regenerate.
19790 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19792         PR target/57018
19793         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19794         a set sp if no stack realignment.
19796 2013-04-22  Nick Clifton  <nickc@redhat.com>
19798         * config.gcc (tilegx-linux): Extend extra_objs rather than
19799         overwriting it.
19800         (tilepro-linux): Likewise.
19802 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19804         * config/aarch64/aarch64-builtins.c
19805         (CF): Remove.
19806         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19807         (VAR<1-12>): Add MAP parameter.
19808         (BUILTIN_*): Likewise.
19809         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19810         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19811         (aarch64_ushl_n<mode>): Likewise.
19812         (aarch64_sshr_n<mode>): Likewise.
19813         (aarch64_ushr_n<mode>): Likewise.
19814         (aarch64_<maxmin><mode>): Likewise.
19815         (aarch64_sqrt<mode>): Likewise.
19816         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19817         (vshr<q>_n_*): Likewise.
19819 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19821         * config/aarch64/aarch64-builtins.c
19822         (aarch64_simd_builtin_type_mode): Handle SF types.
19823         (sf_UP): Define.
19824         (BUILTIN_GPF): Define.
19825         (aarch64_init_simd_builtins): Handle SF types.
19826         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19827         (frecps): Likewise.
19828         (frecpx): Likewise.
19829         * config/aarch64/aarch64-simd.md
19830         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19831         (aarch64_frecpe<mode>): New.
19832         (aarch64_frecps<mode>): Likewise.
19833         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19834         (v8type): Add frecp<esx>.
19835         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19836         (aarch64_frecps<mode>): Likewise.
19837         * config/aarch64/iterators.md (FRECP): New.
19838         (frecp_suffix): Likewise.
19839         * config/aarch64/arm_neon.h
19840         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19842 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19844         PR target/56995
19845         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19846         (REG_CLASS_NAMES): Idem.
19847         (REG_CLASS_CONTENTS): Idem.
19848         (REGCLASS_HAS_FP_REG): Idem.
19849         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19850         (sh_conditional_register_usage): Idem.
19852 2013-04-21  Jeff Law  <law@redhat.com>
19854         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19855         (ssa_forward_propagate_and_combine): Use it.
19857 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19859         * lra.c: Update the flow chart diagram.
19861 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19863         PR rtl-optimization/56847
19864         * lra-constraints.c (process_alt_operands): Discourage alternative
19865         with non-matche doffsettable memory constraint fro memory with
19866         known offset.
19868 2013-04-19  Richard Biener  <rguenther@suse.de>
19870         PR tree-optimization/56982
19871         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19872         function.
19873         * gimplify.c (gimplify_call_expr): Notice special calls.
19874         (gimplify_modify_expr): Likewise.
19875         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19876         abnormal control flow receivers.
19877         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19878         in the same way as cfun->has_nonlocal_labels.
19879         (gimple_purge_dead_abnormal_call_edges): Likewise.
19880         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19881         receivers start a basic-block.
19883 2013-04-19  Richard Biener  <rguenther@suse.de>
19885         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19886         member ...
19887         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19888         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19889         (SLP_TREE_LOAD_PERMUTATION): Add.
19890         (vect_transform_slp_perm_load): Adjust prototype.
19891         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19892         (vect_free_slp_instance): Likewise.
19893         (vect_create_new_slp_node): Likewise.
19894         (vect_supported_slp_permutation_p): Remove.
19895         (vect_slp_rearrange_stmts): Adjust.
19896         (vect_supported_load_permutation_p): Likewise.  Inline
19897         vect_supported_slp_permutation_p here.
19898         (vect_analyze_slp_instance): Compute load permutations per
19899         slp node instead of per instance.
19900         (vect_get_slp_defs): Adjust.
19901         (vect_transform_slp_perm_load): Likewise.
19902         (vect_schedule_slp_instance): Remove redundant code.
19903         (vect_schedule_slp): Remove hack for PR56270, add it ...
19904         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19905         CSE loads for SLP.  Adjust.
19907 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19909         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19910         spelling in two comments.
19912 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19914         PR target/56797
19915         * config/arm/arm.c (load_multiple_sequence): Require SP
19916         as base register for loads if SP is in the register list.
19918 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19920         PR tree-optimization/56718
19921         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19922         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19923         and made public.  Adjusted all callers.
19924         (ipa_intraprocedural_devirtualization): New function.
19925         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19926         (ipa_intraprocedural_devirtualization): Likewise.
19927         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19929 2013-04-19  Richard Biener  <rguenther@suse.de>
19931         PR tree-optimization/57000
19932         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19934 2013-04-19  Terry Guo  <terry.guo@arm.com>
19936         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19937         Replace with ...
19938         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19939         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19940         (cortex_m4_fmacs): Use new reservations.
19941         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19943 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19945         PR rtl-optimization/56999
19946         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19947         related code.
19948         (lra_coalesce): Remove split_origin_bitmap and related code.
19949         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19950         ranges if necessary.
19952 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19954         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19955         New array.
19956         (ix86_expand_call): Remove clobbered_registers array and use
19957         x86_64_ms_sysv_extra_clobbered_registers instead.
19958         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19959         Declare here.
19960         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19961         predicate.
19962         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19963         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19964         (*call_value_rex64_ms_sysv): Ditto.
19966 2013-04-18  Cary Coutant  <ccoutant@google.com>
19968         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19969         parent instead of die_mark.
19971 2013-04-18  Diego Novillo  <dnovillo@google.com>
19973         * gimple.c (create_gimple_tmp): New.
19974         (get_expr_type): New.
19975         (build_assign): New.
19976         (build_type_cast): New.
19977         * gimple.h (enum ssa_mode): Define.
19978         (gimple_seq_set_location): New.
19979         * asan.c (build_check_stmt): Change some gimple_build_* calls
19980         to use build_assign and build_type_cast.
19982 2013-04-18  Richard Biener  <rguenther@suse.de>
19984         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19985         handle negative step.  Remove redundant checks.
19986         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19987         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19988         for negative step and grouped loads fail to vectorize.
19990 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19992         * emit-rtl.c (reset_insn_used_flags): New function.
19993         (reset_all_used_flags): Use it.
19994         (verify_insn_sharing): New function.
19995         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19997 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19999         PR tree-optimization/56984
20000         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20001         and (x >> M) >= N don't register any assertion if N << M is the
20002         minimum value.
20004 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20006         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20007         do not delete_insn insns that have not yet been emitted, only
20008         unlink them with remove_insn.
20009         * df-scan.c (df_insn_delete): Revert r197492.
20011 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20013         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20014         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20016 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20018         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20019         define_insn_and_split.
20020         (and_scc,ior_scc,negscc): Likewise.
20021         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20023 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20025         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20026         can benefit from using a sequence of LDRD instructions in epilogue
20027         instead of a single LDM instruction.
20029 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20031         PR 45688
20032         * doc/extend.texi: Fix typo.
20034 2013-04-17  Richard Biener  <rguenther@suse.de>
20036         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20037         (vect_build_slp_tree): ... here.
20038         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20039         match.  Remove special-casing of mismatched complex loads.
20040         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20041         re-try the match with swapped commutative operands.
20042         (vect_supported_load_permutation_p): Remove special-casing of
20043         mismatched complex loads.
20044         (vect_analyze_slp_instance): Adjust.
20046 2013-04-17  Richard Biener  <rguenther@suse.de>
20048         PR rtl-optimization/56921
20049         * cfgloop.h (struct loop): Add simple_loop_desc member.
20050         (struct niter_desc): Mark with GTY(()).
20051         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20052         * loop-iv.c (get_simple_loop_desc): Likewise.
20053         (free_simple_loop_desc): Likewise.
20055         Revert
20056         2013-04-16  Richard Biener  <rguenther@suse.de>
20058         PR rtl-optimization/56921
20059         * loop-init.c (pass_rtl_move_loop_invariants): Add
20060         TODO_do_not_ggc_collect to todo_flags_finish.
20061         (pass_rtl_unswitch): Same.
20062         (pass_rtl_unroll_and_peel_loops): Same.
20063         (pass_rtl_doloop): Same.
20065 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20067         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20068         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20069         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20070         references.
20071         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20072         * tree-streamer.c (record_common_node): Adjust reference in comment.
20074 2013-04-17  Terry Guo  <terry.guo@arm.com>
20076         * config/arm/cortex-m4.md: Add a new bypass.
20078 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20080         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20081         New pattern.
20082         (*subs_<optab><mode>_multp2): New pattern.
20083         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20084         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20086 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20088         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20089         (*subs_mul_imm_<mode>): New pattern.
20091 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20093         PR target/56948
20094         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20095         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20096         (vsx_movti_32bit): Change j->wa to O->wa.
20098 2013-04-16  Richard Biener  <rguenther@suse.de>
20100         PR rtl-optimization/56921
20101         * loop-init.c (pass_rtl_move_loop_invariants): Add
20102         TODO_do_not_ggc_collect to todo_flags_finish.
20103         (pass_rtl_unswitch): Same.
20104         (pass_rtl_unroll_and_peel_loops): Same.
20105         (pass_rtl_doloop): Same.
20107 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20109         * config/arm/arm.c (emit_multi_reg_push): New declaration
20110         for an existing function.
20111         (arm_emit_strd_push): New function.
20112         (arm_expand_prologue): Used here.
20113         (arm_emit_ldrd_pop): New function.
20114         (arm_expand_epilogue): Used here.
20115         (arm_get_frame_offsets): Update condition.
20116         (arm_emit_multi_reg_pop): Add a special case for load of a single
20117         register with writeback.
20119 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20121         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20122         description.
20124 2013-04-16  Richard Biener  <rguenther@suse.de>
20126         PR tree-optimization/56756
20127         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20128         (first_mem_ref_loc): New.
20129         (execute_sm): Place the load temporarily before a previous
20130         access instead of in the latch edge to ensure its SSA dependencies
20131         are defined at points dominating the load.
20133 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20135         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20136         correct fix by moving header and footer insn to the footer of
20137         the merged basic block.  Clear BB_END of the merged-away block.
20139         PR middle-end/43631
20140         * emit-rtl.c (make_note_raw): New function.
20141         (link_insn_into_chain): New static inline function.
20142         (add_insn): Use it.
20143         (add_insn_before, add_insn_after): Factor insn chain linking code...
20144         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20145         using link_insn_into_chain.
20146         (note_outside_basic_block_p): New helper function for emit_note_after
20147         and emit_note_before.
20148         (emit_note_after): Use nobb variant of add_insn_after if the note
20149         should not be contained in a basic block.
20150         (emit_note_before): Use nobb variant of add_insn_before if the note
20151         should not be contained in a basic block.
20152         (emit_note_copy): Use make_note_raw.
20153         (emit_note): Likewise.
20154         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20155         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20156         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20157         the moved barrier the tail of the basic block it follows.
20158         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20160 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20162         PR tree-optimization/56962
20163         * gimple-ssa-strength-reduction.c (record_increment): Only set
20164         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20165         either rhs1 or rhs2 is equal to c->base_expr.
20167 2013-04-15  Richard Biener  <rguenther@suse.de>
20169         PR tree-optimization/56933
20170         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20171         member.
20172         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20173         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20174         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20175         dependence check ...
20176         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20177         ... here.
20178         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20179         GROUP_READ_WRITE_DEPENDENCE.
20181 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20183         * emit-rtl.c (reset_all_used_flags): New function.
20184         (verify_rtl_sharing): Call reset_all_used_flags before and after
20185         performing the checks.
20187 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20189         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20190         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20191         * config/arm/constraints.md (De): New constraint.
20192         * config/arm/neon.md (anddi3_neon): Delete.
20193         (neon_vand<mode>): Expand to standard anddi3 pattern.
20194         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20195         Move earlier in the file.
20196         (neon_inv_logic_op2): Likewise.
20197         (arm_anddi_operand_neon): New predicate.
20199 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20201         * configure.ac (gcc_cv_ld_as_needed): Set
20202         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20203         Use -z ignore, -z record on *-*-solaris2*.
20204         (HAVE_LD_AS_NEEDED): Update comment.
20205         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20206         * configure: Regenerate.
20207         * config.in: Regenerate.
20208         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20209         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20210         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20211         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20212         equivalents.  Fix markup.
20213         * doc/tm.texi: Regenerate.
20215 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20217         * config/i386/i386.opt: New option mstack-protector-guard=.
20218         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20219         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20220         TARGET_SSP_TLS_GUARD.
20221         * config/i386/i386.c (ix86_option_override_internal): Set
20222         ix86_stack_protector_guard.
20223         * config/i386/i386.md (stack_protect_set): Enable for
20224         TARGET_SSP_TLS_GUARD only.
20225         (stack_protect_set_<mode>): Ditto.
20226         (stack_protect_test): Ditto.
20227         (stack_protect_test_<mode>): Ditto.
20228         * doc/invoke.texi (i386 Option): Document.
20230 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20232         PR target/56890
20233         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20234         (S_MODES): Set H_MODE bit.
20235         (SF_MODES): Set only S_MODE and SF_MODE bits.
20236         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20237         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20238         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20239         <MODE_FLOAT>: Likewise.
20241 2013-04-15  Joey Ye  <joey.ye@arm.com>
20243         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20245 2013-04-15  Joey Ye  <joey.ye@arm.com>
20247         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20248         for real far jump.
20249         (thumb_far_jump_used_p): Count instruction size and set
20250         far_jump_used.
20252 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20254         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20255         * resource.c (mark_target_live_regs): Reformat conditional block.
20257 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20259         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20260         notes, they are emitted only just before final.
20261         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20263 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20265         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20266         * cfgrtl.c (delete_insn): Call it here instead.
20267         * lra-spills.c (lra_final_code_change): Use delete_insn.
20268         * haifa-sched.c (sched_remove_insn): Likewise.
20269         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20270         returning to the nop pool.
20271         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20272         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20274 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20276         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20277         * doc/tm.texi: Regenerated.
20279 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20281         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20282         QImode checks.
20284 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20286         * df-core.c (df_find_def): Compare register numbers.
20287         (df_find_use): Likewise.
20289 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20291         PR target/56903
20292         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20293         lra_in_progress for return.
20295 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20297         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20298         define_insn into define_insn_and_split and emit movsicc patterns.
20300 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20302         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20304 2013-04-12  Richard Biener  <rguenther@suse.de>
20306         * tree-pass.h (TODO_do_not_ggc_collect): New.
20307         * passes.c (execute_one_ipa_transform_pass): Honor
20308         TODO_do_not_ggc_collect.
20309         (execute_one_pass): Likewise.
20311         Revert
20312         2013-04-10  Richard Biener  <rguenther@suse.de>
20314         * passes.c (init_optimization_passes): Remove reload pass.
20315         * ira.c (do_reload): Merge into ...
20316         (ira): ... this.
20317         (rest_of_handle_reload): Remove.
20318         (pass_reload): Likewise.
20319         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20320         of reload for vzeroupper pass placement.
20322 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20324         PR tree-optimization/56918
20325         PR tree-optimization/56920
20326         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20327         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20328         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20329         use wide_mul_with_sign method.
20331 2013-04-12  Richard Biener  <rguenther@suse.de>
20333         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20334         not be considered a gimple constant.
20336 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20338         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20339         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20340         operations.
20342 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20343             Jakub Jelinek  <jakub@redhat.com>
20345         * opts.c: Include diagnostic-color.h.
20346         (common_handle_option): Handle OPT_fdiagnostics_color_.
20347         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20348         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20349         (diagnostic-color.o): New.
20350         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20351         (diagnostic_color_rule): New enum.
20352         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20353         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20354         the location string.
20355         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20356         either NULL, or color kind.
20357         * diagnostic-color.c: New file.
20358         * diagnostic-color.h: New file.
20359         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20360         arguments.
20361         * doc/invoke.texi (-fdiagnostics-color): Document.
20362         * pretty-print.h (pp_show_color): Define.
20363         (struct pretty_print_info): Add show_color field.
20364         * diagnostic.c: Include diagnostic-color.h.
20365         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20366         macros.  Colorize error:, warning: etc. strings and also the location
20367         string.
20368         (diagnostic_show_locus): Colorize the caret line.
20369         * pretty-print.c: Include diagnostic-color.h.
20370         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20371         inside of %< %> quotes or quoted through q format modifier.
20373 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20375         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20377 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20379         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20380         code in CC_NZ mode.
20381         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20382         pattern.
20384 2013-04-11  Marek Polacek  <polacek@redhat.com>
20386         PR tree-optimization/48184
20387         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20389 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20391         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20392         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20393         (skip_simple_arithmetic): Tidy up.
20394         * tree.h (skip_simple_constant_arithmetic): Declare.
20396 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20398         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20400 2013-04-11  Richard Biener  <rguenther@suse.de>
20402         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20403         generate vector constants.
20405 2013-04-11  Richard Biener  <rguenther@suse.de>
20407         PR tree-optimization/56878
20408         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20409         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20410         New function.
20411         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20412         Prefer to align the DR with the most invariant base address.
20414 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20416         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20417         comment.
20419 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20421         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20422         floating-point vector comparisons against 0.
20424 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20426         PR tree-optimization/56899
20427         * fold-const.c (extract_muldiv_1): Apply distributive law
20428         only if TYPE_OVERFLOW_WRAPS (ctype).
20430 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20432         PR target/56124
20433         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20434         loading has side effect.
20436 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20438         * config/sparc/sparc.c: Include tree-pass.h.
20439         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20440         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20441         head of file.  Change return type.  Split off gate function.
20442         (sparc_gate_work_around_errata): New function.
20443         (pass_work_around_errata): New pass definition.
20444         (insert_pass_work_around_errata) New pass insert definition to
20445         insert pass_work_around_errata just after delayed-branch scheduling.
20446         (sparc_option_override): Insert the pass.
20447         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20449 2013-04-10  David S. Miller  <davem@davemloft.net>
20451         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20452         or -mcpu=hypersparc.
20454         * target.def (cstore_mode): New hook.
20455         * target.h: Include insn-codes.h
20456         * targhooks.c: Likewise.
20457         (default_cstore_mode): New function.
20458         * targhooks.h: Declare it.
20459         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20460         * doc/tm.texi: Rebuild.
20461         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20462         target hook, rather than inspecting the insn_data.
20463         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20464         (TARGET_CSTORE_MODE): Redefine.
20465         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20466         result patterns.
20467         * config/sparc/predicates.md (cstore_result_operand): New special
20468         predicate.
20469         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20470         Use it for operand 0.
20471         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20472         (*snesi_special): Likewise.
20473         (*snesi_zero): Likewise.
20474         (*seqsi_zero): Likewise.
20475         (*sltu_insn): Likewise.
20476         (*sgeu_insn): Likewise.
20477         (*seqdi_special): Make operand 0 and comparison operation be of
20478         DImode.
20479         (*snedi_special): Likewise.
20480         (*snedi_special_vis3): Likewise.
20481         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20482         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20483         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20484         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20485         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20486         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20487         (*sltu_extend_sp64): Likewise.
20488         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20489         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20490         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20491         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20492         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20494 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20496         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20497         (aarch64_start_file): Use the new function.
20499 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20500             Jason Merrill  <jason@redhat.com>
20502         * common.opt: Add -gdwarf.
20503         * opts.c (common_handle_option): Handle it.
20504         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20506 2013-04-10  Richard Biener  <rguenther@suse.de>
20508         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20509         (execute_one_ipa_transform_pass): But unconditionally here.
20510         (execute_one_pass): And here.
20511         (init_optimization_passes): Remove reload pass.
20512         * tree-pass.h (TODO_ggc_collect): Remove.
20513         (pass_reload): Likewise.
20514         * ira.c (do_reload): Merge into ...
20515         (ira): ... this.
20516         (rest_of_handle_reload): Remove.
20517         (pass_reload): Likewise.
20518         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20519         of reload for vzeroupper pass placement.
20520         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20521         and todo_flags_finish of all passes.
20523 2013-04-10  Richard Biener  <rguenther@suse.de>
20525         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20526         first_const_oprnd field, rename first_def_type to first_op_type.
20527         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20528         (vect_get_and_check_slp_defs): Always use the type of the
20529         operand.  Allow mixed vect_external_def, vect_constant_def types.
20530         (vect_get_constant_vectors): Handle mixed vect_external_def,
20531         vect_constant_def types.
20533 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20535         PR tree-optimization/55524
20536         * tree-ssa-math-opts.c
20537         (convert_mult_to_fma): Don't use an fms construct
20538         when we don't have an fms operation, but fnma, and it looks
20539         likely that we'll be able to use the latter.
20541 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20543         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20544         function.
20545         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20546         inline fail caused by overwritable functions.
20548 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20550         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20551         unnecessary bits in the constant power of two case.
20553 2013-04-10  Richard Biener  <rguenther@suse.de>
20555         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20556         broken code swapping operands.
20557         (vect_build_slp_tree): Do not compute load permutations here.
20558         (vect_analyze_slp_instance): Compute load permutations here,
20559         after building the SLP tree.
20561 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20563         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20564         of next/prev_real_insn.
20566 2013-04-09  Jan Hubicka  <jh@suse.cz>
20568         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20569         Drop aliased parameter.
20570         (function_and_variable_visibility): Do not handle alias pairs.
20571         * cgraph.c (varpool_externally_visible_p): Update prototype.
20572         * varpool.c (varpool_add_new_variable): Update.
20574 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20576         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20578 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20580         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20582         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20584 2013-04-09  Marek Polacek  <polacek@redhat.com>
20586         PR tree-optimization/48762
20587         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20589 2013-04-09  Richard Biener  <rguenther@suse.de>
20591         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20592         dealing with cost.
20593         (vect_build_slp_tree): Likewise.
20594         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20595         calculating the cost of a SLP instance.
20596         (vect_analyze_slp_instance): Use it from here, after building
20597         the SLP tree.
20599 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20601         PR middle-end/56883
20602         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20603         expand_omp_for_static_chunk): Use simple_p = true in
20604         force_gimple_operand_gsi calls when assigning to addressable decls.
20606 2013-04-09  Jeff Law  <law@redhat.com>
20608         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20609         when the boolean was created by converting a wider object which
20610         had a boolean range.
20612 2013-04-09  Richard Biener  <rguenther@suse.de>
20614         * tree-vectorizer.h (slp_void_p): Remove.
20615         (slp_tree): Typedef before _slp_tree declaration.
20616         (struct _slp_tree): Use a vector of slp_tree as children.
20617         (vect_get_place_in_interleaving_chain): Remove.
20618         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20619         Move ...
20620         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20621         and make static.
20622         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20623         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20624         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20625         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20626         Use slp_node instead of slp_void_p and adjust.
20628 2013-04-09  Richard Biener  <rguenther@suse.de>
20630         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20631         work that is not necessary.
20633 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20635         PR tree-optimization/56854
20636         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20637         forward into clobber stmts if it would change MEM_REF lhs into
20638         non-MEM_REF.
20640 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20642         * tree.c (type_hash_lookup, type_hash_add): Make static.
20643         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20645 2013-04-09  Richard Biener  <rguenther@suse.de>
20647         * tree.h (unsave_expr_now): Remove.
20648         * tree-inline.c (mark_local_for_remap_r): Remove.
20649         (unsave_expr_1): Likewise.
20650         (unsave_r): Likewise.
20651         (unsave_expr_now): Likewise.
20652         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20653         (propagate_tree_value): Likewise.
20655 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20657         * doc/rtl.texi (sequence): Rewrite documentation to match the
20658         current use of SEQUENCE rtl objects.
20659         * rtl.def (SEQUENCE): Likewise.
20661         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20662         Update documentation.
20663         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20664         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20666         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20668 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20670         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20671         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20672         methods.
20673         (estimate_edge_size_and_time): Add comment to suggest using rounding
20674         methods.
20675         (estimate_node_size_and_time): Ditto.
20676         (remap_edge_change_prob): Use helper rounding divide methods.
20677         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20678         (gimple_mod_pow2_value_transform): Ditto.
20679         (gimple_mod_subtract_transform): Ditto.
20680         (gimple_ic_transform): Ditto.
20681         (gimple_stringops_transform): Ditto.
20682         * stmt.c (conditional_probability): Ditto.
20683         (emit_case_dispatch_table): Ditto.
20684         * lto-cgraph.c (merge_profile_summaries): Ditto.
20685         * tree-optimize.c (execute_fixup_cfg): Ditto.
20686         * cfgcleanup.c (try_forward_edges): Ditto.
20687         * cfgloopmanip.c (scale_loop_profile): Ditto.
20688         (loopify): Ditto.
20689         (duplicate_loop_to_header_edge): Ditto.
20690         (lv_adjust_loop_entry_edge): Ditto.
20691         * tree-vect-loop.c (vect_transform_loop): Ditto.
20692         * profile.c (compute_branch_probabilities): Ditto.
20693         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20694         * lto-streamer-in.c (input_cfg): Ditto.
20695         * gimple-streamer-in.c (input_bb): Ditto.
20696         * ipa-cp.c (update_profiling_info): Ditto.
20697         (update_specialized_profile): Ditto.
20698         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20699         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20700         rounding methods.
20701         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20702         (compute_trg_info): Ditto.
20703         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20704         (purge_dead_edges): Ditto.
20705         * loop-unswitch.c (unswitch_loop): Ditto.
20706         * cgraphclones.c (cgraph_clone_edge): Ditto.
20707         (cgraph_clone_node): Ditto.
20708         * tree-inline.c (copy_bb): Ditto.
20709         (copy_edges_for_bb): Ditto.
20710         (initialize_cfun): Ditto.
20711         (copy_cfg_body): Ditto.
20712         (expand_call_inline): Ditto.
20714 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20716         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20717         TARGET_CYGWIN64 by TARGET_64BIT.
20719 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20721         * config/epiphany/epiphany.md (GPR_1): New constant.
20722         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20723         * config/epiphany/epiphany.c (gen_compare_reg):
20724         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20725         is already in place.
20726         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20727         Don't require being called during rtl expansion; If y operlaps r0,
20728         return 0.
20729         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20730         (epiphany_expand_epilogue): Likewise.
20732         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20733         Don't use CC_FPmode for ORDERED / UNORDERED.
20734         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20736         * config/epiphany/constraints.md (CnL): New constraint.
20737         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20738         * config/epiphany/predicates.md (add_operand): Allow 1024.
20740         * config/epiphany/epiphany.md (logical_op): New code iterator.
20741         (op_mnc): New code attribute.
20742         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20743         (mov_f+1, mov_f+2): New peephole2 patterns.
20745         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20746         (cstoresi4): Also allow re-use of zero result when doing a NE
20747         comparison to a non-zero operand.
20748         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20750         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20751         Use gen_addsi3_i / gen_subsi3_i.
20753 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20755         PR c++/34949
20756         PR c++/50243
20757         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20758         contain anything but clobbers, at most one __builtin_stack_restore,
20759         optionally debug stmts and final resx, and if it has at least one
20760         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20761         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20762         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20763         which isn't defaut definition, remove them.
20764         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20765         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20766         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20767         with MEM_REF LHS with SSA_NAME address.
20769 2013-04-08  Jeff Law  <law@redhat.com>
20771         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20773 2013-04-08  Richard Biener  <rguenther@suse.de>
20775         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20776         extra newline.
20777         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20778         determined vector type.
20779         (vect_analyze_data_refs): Likewise.
20780         (vect_get_new_vect_var): Adjust.
20781         (vect_create_destination_var): Preserve SSA name versions.
20782         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20783         not dump anything here.
20785 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20787         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20788         Add member lr_slot_known.
20789         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20790         if necessary.
20791         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20792         Remove code that sets lr_slot_offset according to what a previous
20793         version of epiphany_emit_save_restore used to do.
20794         (epiphany_emit_save_restore): When doing an lr save or restore,
20795         set/verify lr_slot_known and lr_slot_offset.
20797 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20799         PR target/54338
20800         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20801         in ALL_REGS.
20803 2013-04-08  Richard Biener  <rguenther@suse.de>
20805         * alias.c (find_base_term): Fix thinko in previous change.
20807 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20809         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20810         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20811         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20812         if possible to compute val.
20813         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20814         For QImode integers don't require anything about precision.  Use
20815         const_with_all_bytes_same to find out if the constant doesn't have
20816         repeated bytes in it.
20818 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20820         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20821         within mode size.
20823 2013-04-08  Marek Polacek  <polacek@redhat.com>
20825         PR rtl-optimization/48182
20826         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20827         value to 1.
20829 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20831         PR target/55487
20832         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20833         nuses, make sure we have a label.
20835 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20837         PR target/56843
20838         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20839         (rs6000_emit_swdiv_low_precision): Remove.
20840         (rs6000_emit_swdiv): Rewrite to handle between one and four
20841         iterations of Newton-Raphson generally; modify required number of
20842         iterations for some cases.
20843         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20845 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20847         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20848         set-but-unused variable.
20850         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20851         basic blocks of released function bodies garbage-collectable.
20853         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20854         (struct rtl_opt_pass): Add TODO_df_finish.
20856         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20858 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20860         * config/arm/constraints.md (q): New constraint.
20861         * config/arm/ldrdstrd.md: New file.
20862         * config/arm/arm.md (ldrdstrd.md) New include.
20863         (arm_movdi): Use "q" instead of "r" constraint
20864         for double-word memory access.
20865         (movdf_soft_insn): Likewise.
20866         * config/arm/vfp.md (movdi_vfp): Likewise.
20867         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20868         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20869         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20870         (mem_ok_for_ldrd_strd): Likewise.
20871         (output_move_double): Update assertion.
20873 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20875         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20877 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20879         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20880         define_insn_and_split.
20881         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20883 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20885         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20886         define_insn_and_split.
20887         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20888         (shiftsi3_compare): New pattern.
20889         (rrx): New pattern.
20890         * config/arm/unspecs.md (UNSPEC_RRX): New.
20892 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20894         * config/arm/arm.md (negdi_extendsidi): New pattern.
20895         (negdi_zero_extendsidi): Likewise.
20897 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20899         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20900         define_insn_and_split.
20901         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20902         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20904 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20906         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20907         define_insn_and_split.
20908         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20909         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20911 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20913         * config/arm/arm.md (subsi3_carryin): New pattern.
20914         (subsi3_carryin_const): Likewise.
20915         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20916         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20918 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20920         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20922 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20924         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20925         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20927 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20929         * config/arm/arm.c (arm_expand_builtin): Change fcode
20930         type to unsigned int.
20932 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20934         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20936 2013-04-04  Ian Lance Taylor  <iant@google.com>
20938         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20939         language standard.
20941 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20943         PR middle-end/56729
20944         * df-scan.c (df_insn_delete): Disable failing assert.
20946 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20948         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20949         New function prototype.
20950         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20951         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20952         (arm_builtin_vectorized_function): New function.
20954 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20956         * config/arm/arm_neon_builtins.def: New file.
20957         * config/arm/arm.c (neon_builtin_data): Move contents to
20958         arm_neon_builtins.def.
20959         (enum arm_builtins): Include neon builtin definitions.
20960         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20961         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20963 2013-04-04  Marek Polacek  <polacek@redhat.com>
20965         PR tree-optimization/48186
20966         * predict.c (maybe_hot_frequency_p): Return false if
20967         HOT_BB_FREQUENCY_FRACTION is 0.
20968         (cgraph_maybe_hot_edge_p): Likewise.
20970 2013-04-04  Richard Biener  <rguenther@suse.de>
20972         PR tree-optimization/56826
20973         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20974         more accurately.
20976 2013-04-04  Richard Biener  <rguenther@suse.de>
20978         PR tree-optimization/56213
20979         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20980         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20982 2013-04-04  Richard Biener  <rguenther@suse.de>
20984         PR tree-optimization/56837
20985         * tree-loop-distribution.c (classify_partition): For non-zero
20986         values require that the value has the same precision as its
20987         mode to be useful as memset value.
20989 2013-04-03  Nick Clifton  <nickc@redhat.com>
20991         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20992         (fmssf4): Use fmsf.s on E3V5 architectures.
20993         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20994         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20996 2013-04-03  Jeff Law  <law@redhat.com>
20998         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20999         (lra-eliminations.o): Likewise.
21001 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21003         * gcov-io.c (compute_working_sets): Moved most of body of old
21004         compute_working_sets here from profile.c.
21005         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21006         (gcov_working_set_t): Moved typedef here from basic-block.h
21007         (compute_working_set): Declare.
21008         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21009         (get_working_sets): Renamed from compute_working_set,
21010         replace most of body with call to new compute_working_sets.
21011         (get_exec_counts): Replace call to compute_working_sets
21012         to get_working_sets.
21013         * profile.h (get_working_sets): Renamed from compute_working_set.
21014         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21015         to get_working_sets.
21016         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21017         * gcov-dump.c (dump_working_sets): New function.
21019 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21021         * hwint.c (sext_hwi, zext_hwi): New functions.
21022         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21023         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21024         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21025         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21026         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21027         (sext_hwi, zext_hwi): New functions.
21029 2013-04-03  Jeff Law  <law@redhat.com>
21031         PR tree-optimization/56799
21032         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21033         back test for widening conversion erroneously dropped in prior change.
21035 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21037         PR target/56809
21038         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21039         instead of next_real_insn.
21041 2013-04-03  Marek Polacek  <polacek@redhat.com>
21043         PR sanitizer/55702
21044         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21046 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21048         PR target/56809
21049         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21050         next_real_insn.
21051         (thumb1_output_casesi): Likewise.
21052         (thumb2_output_casesi): Likewise.
21054 2013-04-03  Richard Biener  <rguenther@suse.de>
21056         PR tree-optimization/56817
21057         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21058         Split out ...
21059         (tree_unroll_loops_completely_1): ... new function to manually
21060         walk the loop tree, properly defering outer loops of unrolled
21061         loops to later iterations.
21063 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21065         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21066         (vectorizable_load): Likewise.
21067         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21068         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21070 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21072         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21073         BIT_FIELD_REF.
21075 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21077         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21079 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21081         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21083 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21085         PR tree-optimization/56790
21086         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21087         folding.
21089 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21091         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21092         Handle VEC_MERGE.
21093         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21094         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21095         equal arguments.
21097 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21099         PR c/19449
21100         * tree.h (force_folding_builtin_constant_p): New decl.
21101         * builtins.c (force_folding_builtin_constant_p): New variable.
21102         (fold_builtin_constant_p): Fold immediately also if
21103         force_folding_builtin_constant_p.
21105 2013-04-03  Richard Biener  <rguenther@suse.de>
21107         PR tree-optimization/56812
21108         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21109         DRs of the same interleaving chain are independent.
21111 2013-04-02  Jason Merrill  <jason@redhat.com>
21113         * gdbinit.in (pbb): Use debug fn.
21115 2013-04-02  Lawrence Crowl  <crowl@google.com>
21117         * sese.h (struct ivtype_map_elt_s): Remove unused.
21118         (extern debug_ivtype_map): Remove unused.
21119         (extern eq_ivtype_map_elts): Remove unused.
21120         * sese.c (debug_ivtype_map): Removed unused.
21121         (debug_ivtype_map_1): Removed unused.
21122         (debug_ivtype_elt): Remove unused.
21123         (eq_ivtype_map_elts): Remove unused.
21126 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21128         PR target/52790
21129         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21130         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21131         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21132         function.
21133         (legitimize_pe_coff_symbol): Likewise.
21134         (is_imported_p): New helper-function.
21135         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21136         for Windows x64 targets.
21137         (ix86_expand_prologue): Optimize for pe-coff targets.
21138         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21139         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21140         medium/large code-model.
21141         (legitimize_pic_address): Likewise.
21142         (legitimize_tls_address): Likewise.
21143         (ix86_expand_call): Likewise.
21144         (x86_output_mi_thunk): Likewise.
21145         (get_dllimport_decl): Add new beimport argument.
21146         (construct_plt_address): Don't assert for x64 pe-coff targets.
21147         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21148         targets.
21149         (SYMBOL_FLAG_STUBVAR): New macro.
21150         (SYMBOL_REF_STUBVAR_P): Likewise.
21151         * config/i386/winnt.c (stub_list): New structure.
21152         (stub_head): New local variable.
21153         (i386_pe_record_stub): New function.
21154         (i386_pe_file_end): Emit refptr-stubs.
21156 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21158         PR rtl-optimization/56745
21159         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21160         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21162         PR c++/34949
21163         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21164         and both of them are MEM_REFs, just compare first argument for
21165         equality and attempt to deal even with differing offsets.
21167         PR c++/34949
21168         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21169         of gimple_clobber_p to be MEM_REF.
21170         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21171         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21172         after gimplification.
21173         * asan.c (get_mem_ref_of_assignment): Don't instrument
21174         gimple_clobber_p stmts.
21175         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21176         gimple_clobber_p stmt if they have MEM_REF lhs and
21177         are dead because of another gimple_clobber_p stmt.
21178         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21179         gimple_clobber_p stmts like debug stmts.
21180         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21181         that refer to unused VAR_DECLs or uninitialized values.
21182         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21183         gimple_clobber_p stmts if they refer to removed parameters.
21184         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21185         formatting.
21187 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21189         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21190         using SWI48 mode attribute.
21192 2013-04-02  Wei Mi  <wmi@google.com>
21194         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21195         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21196         *<rotate_insn><mode>3_mask in i386.md.
21198 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21200         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21202 2013-04-02  Richard Biener  <rguenther@suse.de>
21204         PR tree-optimization/56778
21205         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21206         Runtime alias tests are not supported for gather loads.
21207         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21208         stmts referenced from SSA operands before updating SSA form.
21210 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21211             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21213         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21214         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21215         * config/arm/cortex-a53.md: New file.
21216         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21217         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21218         * config/arm/arm.c (arm_issue_rate): Likewise.
21219         * config/arm/arm-tune.md: Regenerate
21220         * config/arm/arm-tables.opt: Regenerate.
21221         * config/arm/arm-cores.def: Add cortex-a53.
21223 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21225         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21226         non-static link.
21228 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21230         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21231         scalar load/store operations using B/H registers.
21232         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21234 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21236         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21237         scalar move.
21238         * config/aarch64/aarch64.c
21239         (aarch64_simd_scalar_immediate_valid_for_move): New.
21240         * config/aarch64/aarch64-protos.h
21241         (aarch64_simd_scalar_immediate_valid_for_move): New.
21242         * config/aarch64/constraints.md (Dh, Dq): New.
21243         * config/aarch64/iterators.md (hq): New.
21245 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21247         * reorg.c (get_branch_condition): Deal with conditional returns.
21248         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21250 2013-04-01  Wei Mi  <wmi@google.com>
21252         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21253         Truncate operand 2 using %b asm operand modifier.
21254         (*<shift_insn><mode>3_mask): Ditto.
21255         (*<rotate_insn><mode>3_mask): Ditto.
21257 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21259         PR middle-end/56798
21260         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21262 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21264         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21265         of next_real_insn.
21266         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21268 2013-03-30  Lawrence Crowl  <crowl@google.com>
21270         * dse.c (clear_alias_sets): Remove never set.
21271         (disqualified_clear_alias_sets): Remove never set.
21272         (clear_alias_mode_pool): Remove never set.
21273         (dse_step0): Remove condition that is never true.
21274         (canon_address): Remove condition that is never true.
21275         (dse_step7): Remove condition that is never true.
21276         (rest_of_handle_dse): Remove condition that is never true.
21277         (rest_of_handle_dse::did_global): Remove never read from above.
21278         (dse_step2_spill): Remove never called from above.
21279         (dse_step5_spill): Remove never called from above.
21281 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21283         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21284         JUMP_TABLE_DATA changes.
21285         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21286         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21287         (Insns) <jump_table_data>: New entry.
21288         * doc/tm.texi: Regenerate.
21290         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21292         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21293         for table jump at the end of a basic block using tablejump_p.
21294         * targhooks.c (default_invalid_within_doloop): Likewise.
21295         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21296         target hook implementation that is identical to the default hook.
21297         (rs6000_invalid_within_doloop): Remove.
21299         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21300         unused variable from tablejump_p call.
21302         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21303         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21304         (INSN_DELETED_P): Likewise.
21305         (emit_jump_table_data): New prototype.
21306         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21307         after 4th as unused.
21308         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21309         * sched-vis.c (print_insn): Likewise.
21310         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21311         insn for compatibility with back ends that use next_active_insn to
21312         identify jump table data.
21313         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21314         (remove_insn): Likewise.
21315         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21316         to be emitted.
21317         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21318         (emit_jump_table_data): New function.
21320         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21321         basic block, a JUMP_TABLE_DATA never is.
21322         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21323         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21324         off from code handling real insns.
21325         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21326         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21327         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21328         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21329         is not a NONDEBUG_INSN_P.
21330         * ira-costs.c (scan_one_insn): Likewise.
21331         * jump.c (mark_all_labels): Likewise.
21332         (mark_jump_label_1): Likewise.
21333         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21334         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21335         a basic block.
21336         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21337         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21338         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21339         JUMP_TABLE_DATA_P insns.
21340         (calculate_elim_costs_all_insns): Likewise.
21341         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21342         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21343         (delete_output_reload): Code style fixups.
21344         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21345         insn flags on this non-insn.
21346         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21347         as scheduling barriers, for pre-change compatibility.
21348         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21349         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21351         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21352         redundant JUMP_TABLE_DATA_P test.
21353         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21354         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21355         (frv_for_each_packet): Likewise.
21356         * config/i386/i386.c (min_insn_size): Likewise.
21357         (ix86_avoid_jump_mispredicts): Likewise.
21358         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21359         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21360         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21361         (mips16_insn_length): Robustify.
21362         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21363         (mips16_split_long_branches): Likewise.
21364         * config/pa/pa.c (pa_combine_instructions): Likewise.
21365         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21366         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21367         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21368         as contributing to pool range lengths.
21369         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21370         Remove redundant JUMP_TABLE_DATA_P test.
21371         (sh_loop_align): Likewise.
21372         (split_branches): Likewise.
21373         (sh_insn_length_adjustment): Likewise.
21374         * config/spu/spu.c (get_branch_target): Likewise.
21376 2013-03-29  Jan Hubicka  <jh@suse.cz>
21378         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21379         gcov streaming; stream hot bb threshold to ltrans.
21380         * predict.c (get_hot_bb_threshold): Break out from ....
21381         (maybe_hot_count_p): ... here.
21382         (set_hot_bb_threshold): New function.
21383         * lto-section-in.c (lto_section_name): Add profile.
21384         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21385         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21386         and data-streamer.h
21387         (histogram_entry): New structure.
21388         (histogram, histogram_pool): New global vars.
21389         (histogram_hash): New structure.
21390         (histogram_hash::hash): New method.
21391         (histogram_hash::equal): Likewise.
21392         (account_time_size): New function.
21393         (cmp_counts): New function.
21394         (dump_histogram): New function.
21395         (ipa_profile_generate_summary): New function.
21396         (ipa_profile_write_summary): New function.
21397         (ipa_profile_read_summary): New function.
21398         (ipa_profile): Decide on threshold.
21399         (pass_ipa_profile): Add ipa_profile_write_summary and
21400         ipa_profile_read_summary.
21401         * Makefile.in (ipa.o): Update dependencies.
21402         * lto-streamer.h (LTO_section_ipa_profile): New section.
21404 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21406         * tree.h (VAR_P): New.
21408 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21410         PR lto/56777
21411         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21413 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21415         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21416         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21417         (control_flow_insn_p): Likewise.
21418         * cfgrtl.c (duplicate_insn_chain): Likewise.
21419         * final.c (get_attr_length_1): Likewise.
21420         (shorten_branches): Likewise.
21421         (final_scan_insn): Likewise.
21422         * function.c (instantiate_virtual_regs): Likewise.
21423         * gcse.c (insert_insn_end_basic_block): Likewise.
21424         * ira-costs.c (scan_one_insn): Likewise.
21425         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21426         * lra.c (check_rtl): Likewise.
21427         * reload1.c (elimination_costs_in_insn): Likewise.
21428         * reorg.c (follow_jumps): Likewise.
21430         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21431         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21432         (thumb_far_jump_used_p): Likewise.
21433         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21434         (workaround_speculation): Likewise.
21435         (add_sched_insns_for_speculation): Likewise.
21436         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21437         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21438         (frv_for_each_packet): Likewise.
21439         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21440         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21441         (final_emit_insn_group_barriers): Likewise.
21442         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21443         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21444         (mips16_insn_length): Likewise.
21445         * config/pa/pa.c (pa_reorg): Likewise.
21446         (pa_combine_instructions): Likewise.
21447         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21448         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21449         (sh_reorg): Likewise.
21450         (split_branches): Likewise.
21451         * config/spu/spu.c (get_branch_target): Likewise.
21453         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21454         JUMP_TABLE_DATA_P.
21456 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21458         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21459         Fix declaration name.
21461 2013-03-28  Lawrence Crowl  <crowl@google.com>
21463         * graphds.h (struct graph.indicies): Remove unused.
21464         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21465         (SCOP_ORIGINAL_PDDRS): Remove unused.
21466         * sese.h (extern insert_loop_close_phis): Removed unused.
21467         (extern insert_guard_phis): Removed unused.
21468         (extern ivtype_map_elt_info): Removed unused.
21469         (new_ivtype_map_elt): Removed unused.
21470         * sese.c (ivtype_map_elt_info): Removed unused.
21472 2013-03-28  Lawrence Crowl  <crowl@google.com>
21474         * Makefile.in: Add several missing include dependences.
21475         (DUMPFILE_H): New.
21476         (test-dump.o): New.  This object is not added to any executable,
21477         but is present for ad-hoc testing.
21478         * bitmap.c
21479         (debug (const bitmap_head_def &)): New.
21480         (debug (const bitmap_head_def *)): New.
21481         * bitmap.h
21482         (extern debug (const bitmap_head_def &)): New.
21483         (extern debug (const bitmap_head_def *)): New.
21484         * cfg.c
21485         (debug (edge_def &)): New.
21486         (debug (edge_def *)): New.
21487         * cfghooks.c
21488         (debug (basic_block_def &)): New.
21489         (debug (basic_block_def *)): New.
21490         * dumpfile.h
21491         (dump_node (const_tree, int, FILE *)): Correct source file.
21492         * dwarf2out.c
21493         (debug (die_struct &)): New.
21494         (debug (die_struct *)): New.
21495         * dwarf2out.h
21496         (extern debug (die_struct &)): New.
21497         (extern debug (die_struct *)): New.
21498         * gimple-pretty-print.c
21499         (debug (gimple_statement_d &)): New.
21500         (debug (gimple_statement_d *)): New.
21501         * gimple-pretty-print.h
21502         (extern debug (gimple_statement_d &)): New.
21503         (extern debug (gimple_statement_d *)): New.
21504         * ira-build.c
21505         (debug (ira_allocno_copy &)): New.
21506         (debug (ira_allocno_copy *)): New.
21507         (debug (ira_allocno &)): New.
21508         (debug (ira_allocno *)): New.
21509         * ira-int.h
21510         (extern debug (ira_allocno_copy &)): New.
21511         (extern debug (ira_allocno_copy *)): New.
21512         (extern debug (ira_allocno &)): New.
21513         (extern debug (ira_allocno *)): New.
21514         * ira-lives.c
21515         (debug (live_range &)): New.
21516         (debug (live_range *)): New.
21517         * lra-int.h
21518         (debug (lra_live_range &)): New.
21519         (debug (lra_live_range *)): New.
21520         * lra-lives.c
21521         (debug (lra_live_range &)): New.
21522         (debug (lra_live_range *)): New.
21523         * omega.c
21524         (debug (omega_pb_d &)): New.
21525         (debug (omega_pb_d *)): New.
21526         * omega.h
21527         (extern debug (omega_pb_d &)): New.
21528         (extern debug (omega_pb_d *)): New.
21529         * print-rtl.c
21530         (debug (const rtx_def &)): New.
21531         (debug (const rtx_def *)): New.
21532         * print-tree.c
21533         (debug_tree (tree): Move within file.
21534         (debug_raw (const tree_node &)): New.
21535         (debug_raw (const tree_node *)): New.
21536         (dump_tree_via_hooks (const tree_node *, int)): New.
21537         (debug (const tree_node &)): New.
21538         (debug (const tree_node *)): New.
21539         (debug_verbose (const tree_node &)): New.
21540         (debug_verbose (const tree_node *)): New.
21541         (debug_head (const tree_node &)): New.
21542         (debug_head (const tree_node *)): New.
21543         (debug_body (const tree_node &)): New.
21544         (debug_body (const tree_node *)): New.
21545         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21546         (debug (vec<tree, va_gc> &)): New.
21547         (debug (vec<tree, va_gc> *)): New.
21548         * rtl.h
21549         (extern debug (const rtx_def &)): New.
21550         (extern debug (const rtx_def *)): New.
21551         * sbitmap.c
21552         (debug_raw (simple_bitmap_def &)): New.
21553         (debug_raw (simple_bitmap_def *)): New.
21554         (debug (simple_bitmap_def &)): New.
21555         (debug (simple_bitmap_def *)): New.
21556         * sbitmap.h
21557         (extern debug (simple_bitmap_def &)): New.
21558         (extern debug (simple_bitmap_def *)): New.
21559         (extern debug_raw (simple_bitmap_def &)): New.
21560         (extern debug_raw (simple_bitmap_def *)): New.
21561         * sel-sched-dump.c
21562         (debug (vinsn_def &)): New.
21563         (debug (vinsn_def *)): New.
21564         (debug_verbose (vinsn_def &)): New.
21565         (debug_verbose (vinsn_def *)): New.
21566         (debug (expr_def &)): New.
21567         (debug (expr_def *)): New.
21568         (debug_verbose (expr_def &)): New.
21569         (debug_verbose (expr_def *)): New.
21570         (debug (vec<rtx> &)): New.
21571         (debug (vec<rtx> *)): New.
21572         * sel-sched-dump.h
21573         (extern debug (vinsn_def &)): New.
21574         (extern debug (vinsn_def *)): New.
21575         (extern debug_verbose (vinsn_def &)): New.
21576         (extern debug_verbose (vinsn_def *)): New.
21577         (extern debug (expr_def &)): New.
21578         (extern debug (expr_def *)): New.
21579         (extern debug_verbose (expr_def &)): New.
21580         (extern debug_verbose (expr_def *)): New.
21581         (extern debug (vec<rtx> &)): New.
21582         (extern debug (vec<rtx> *)): New.
21583         * sel-sched-ir.h
21584         (_list_iter_cond_expr): Make inline instead of static.
21585         * sreal.c
21586         (debug (sreal &)): New.
21587         (debug (sreal *)): New.
21588         * sreal.h
21589         (extern debug (sreal &)): New.
21590         (extern debug (sreal *)): New.
21591         * tree.h
21592         (extern debug_raw (const tree_node &)): New.
21593         (extern debug_raw (const tree_node *)): New.
21594         (extern debug (const tree_node &)): New.
21595         (extern debug (const tree_node *)): New.
21596         (extern debug_verbose (const tree_node &)): New.
21597         (extern debug_verbose (const tree_node *)): New.
21598         (extern debug_head (const tree_node &)): New.
21599         (extern debug_head (const tree_node *)): New.
21600         (extern debug_body (const tree_node &)): New.
21601         (extern debug_body (const tree_node *)): New.
21602         (extern debug (vec<tree, va_gc> &)): New.
21603         (extern debug (vec<tree, va_gc> *)): New.
21604         * tree-cfg.c
21605         (debug (struct loop &)): New.
21606         (debug (struct loop *)): New.
21607         (debug_verbose (struct loop &)): New.
21608         (debug_verbose (struct loop *)): New.
21609         * tree-dump.c: Add header dependence.
21610         * tree-flow.h
21611         (extern debug (struct loop &)): New.
21612         (extern debug (struct loop *)): New.
21613         (extern debug_verbose (struct loop &)): New.
21614         (extern debug_verbose (struct loop *)): New.
21615         * tree-data-ref.c
21616         (debug (data_reference &)): New.
21617         (debug (data_reference *)): New.
21618         (debug (vec<data_reference_p> &)): New.
21619         (debug (vec<data_reference_p> *)): New.
21620         (debug (vec<ddr_p> &)): New.
21621         (debug (vec<ddr_p> *)): New.
21622         * tree-data-ref.h
21623         (extern debug (data_reference &)): New.
21624         (extern debug (data_reference *)): New.
21625         (extern debug (vec<data_reference_p> &)): New.
21626         (extern debug (vec<data_reference_p> *)): New.
21627         (extern debug (vec<ddr_p> &)): New.
21628         (extern debug (vec<ddr_p> *)): New.
21629         * tree-ssa-alias.c
21630         (debug (pt_solution &)): New.
21631         (debug (pt_solution *)): New.
21632         * tree-ssa-alias.h
21633         (extern debug (pt_solution &)): New.
21634         (extern debug (pt_solution *)): New.
21635         * tree-ssa-alias.c
21636         (debug (_var_map &)): New.
21637         (debug (_var_map *)): New.
21638         (debug (tree_live_info_d &)): New.
21639         (debug (tree_live_info_d *)): New.
21640         * tree-ssa-alias.h
21641         (extern debug (_var_map &)): New.
21642         (extern debug (_var_map *)): New.
21643         (extern debug (tree_live_info_d &)): New.
21644         (extern debug (tree_live_info_d *)): New.
21646 2013-03-28  Jan Hubicka  <jh@suse.cz>
21648         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21650 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21652         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21653         record only when desired or required.
21655 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21657         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21658         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21660 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21662         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21663         (*andsi3_compare0_uxtw): New pattern.
21664         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21665         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21667 2013-03-28  Jan Hubicka  <jh@suse.cz>
21669         * data-streamer-in.c (streamer_read_gcov_count): New function.
21670         * gimple-streamer-out.c: Include value-prof.h.
21671         (output_gimple_stmt): Output histogram.
21672         (output_bb): Use streamer_write_gcov_count.
21673         * value-prof.c: Include data-streamer.h
21674         (dump_histogram_value): Add HIST_TYPE_MAX.
21675         (stream_out_histogram_value): New function.
21676         (stream_in_histogram_value): New function.
21677         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21678         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21679         * data-streamer-out.c (streamer_write_gcov_count): New function.
21680         (streamer_write_gcov_count_stream): New function.
21681         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21682         (lto_output_node): Likewise.
21683         (input_node, input_edge): Likewise.
21684         * lto-streamer-out.c (output_cfg): Update streaming.
21685         * lto-streamer-in.c (input_cfg): Likewise.
21686         * data-streamer.h (streamer_write_gcov_count,
21687         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21688         * gimple-streamer-in.c: Include value-prof.h
21689         (input_gimple_stmt): Input histograms.
21690         (input_bb): Update profile streaming.
21692 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21694         * genmodes.c (emit_max_int): New function.
21695         (emit_insn_modes_h): Added call to emit_max_function.
21696         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21697         Added doc.
21698         * machmode.def: Fixed comment.
21700 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21702         * combine.c (try_combine): Removed useless assert.
21703         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21705 2013-03-28  Marek Polacek  <polacek@redhat.com>
21706             Richard Biener  <rguenther@suse.de>
21708         PR tree-optimization/56695
21709         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21710         build signed result of a vector comparison.
21711         * tree-cfg.c (verify_gimple_comparison): Check that a result
21712         of a vector comparison has signed type.
21714 2013-03-28  Richard Biener  <rguenther@suse.de>
21716         PR tree-optimization/37021
21717         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21718         do not restrict gaps between groups.
21719         * tree-vect-stmts.c (vectorizable_load): Properly account for
21720         a gap between groups.
21722 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21724         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21725         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21726         is not enabled.
21728 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21730         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21731         * doc/extend.texi (Named Address Spaces): Ditto.
21732         (Variable Attributes): Ditto.
21734 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21736         * config.build: Add support for cygwin x64 target.
21737         * config.gcc: Likewise.
21738         * config.host: Likewise.
21739         * configure.ac: Likewise
21740         * configure: Regenerated.
21742 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21744         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21745         * config/i386/t-cygwin-w64: New file.
21746         * config/i386/cygwin-w64.h: New file.
21747         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21748         and add support for x64-cygwin target.
21749         (CPP_SPEC): Likewise.
21750         (CXX_WRAP_SPEC_LIST): Undefine before define.
21751         (LIBGCJ_SONAME): Use 15 as version.
21753 2013-03-27  Richard Biener  <rguenther@suse.de>
21755         PR tree-optimization/56716
21756         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21757         dumping for ref nodes.
21759 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21761         PR tree-optimization/55334
21762         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21763         restricted pointers to arrays.
21765 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21767         * Makefile.in (.SUFFIXES): Add .cc.
21768         (.c.o): Apply same recipe for implicit rule .cc.o.
21770 2013-03-27  Richard Biener  <rguenther@suse.de>
21772         PR tree-optimization/37021
21773         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21774         REALPART/IMAGPART_EXPRs around the supported refs.
21775         * tree-ssa-structalias.c (find_func_aliases): Assume that
21776         floating-point values are not used to transfer pointers.
21778 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21780         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21781         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21782         * doc/tm.texi: Regenerate.
21783         * targhooks.h (default_has_ifunc_p): New.
21784         * targhooks.c (default_has_ifunc_p): Ditto.
21785         * config/linux-protos.h: New file.
21786         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21787         hook for linux which disables support of indirect functions in android.
21788         * config/linux-android.c: New file.
21789         * config/t-linux-android.c: Ditto.
21790         * config.gcc: Added new object file linux-android.o.
21791         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21792         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21793         * varasm.c (do_assemble_alias): Likewise.
21794         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21795         doesn't support indirect functions.
21796         * configure: Regenerate.
21798 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21800         PR target/56102
21801         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21802         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21803         mult-word mode.
21805 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21807         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21809 2013-03-27  Terry Guo  <terry.guo@arm.com>
21811         * config/arm/arm-cores.def: Added core cortex-r7.
21812         * config/arm/arm-tune.md: Regenerated.
21813         * config/arm/arm-tables.opt: Regenerated.
21814         * doc/invoke.texi: Added entry for core cortex-r7.
21816 2013-03-27  Walter Lee  <walt@tilera.com>
21818         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21819         double-decrement of next_scratch_regno.
21821 2013-03-27  Walter Lee  <walt@tilera.com>
21823         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21824         input operands.
21825         (insn_v1mulus): Ditto.
21826         (insn_v2muls): Ditto.
21828 2013-03-27  Walter Lee  <walt@tilera.com>
21830         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21831         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21833 2013-03-27  Walter Lee  <walt@tilera.com>
21835         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21836         (*sibcall_value): Ditto.
21838 2013-03-27  Walter Lee  <walt@tilera.com>
21840         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21841         (insn_mnz_v8qi): ... this ...
21842         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21843         vector equivalent.
21844         (insn_v<n>mnz): Replaced by ...
21845         (insn_v1mnz): ... this ...
21846         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21847         equivalent.
21848         (insn_mz_<mode>): Replaced by ...
21849         (insn_mz_v8qi): ... this ...
21850         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21851         vector equivalent.
21852         (insn_v<n>mz): Replaced by ...
21853         (insn_v1mz): ... this ...
21854         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21855         equivalent.
21857 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21859         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21861 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21863         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21864         than fprintf with a non-constant, non-format string.
21866 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21868         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21869         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21870         operand 0 predicate.
21871         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21872         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21873         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21874         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21875         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21876         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21877         (mov<mode>_insv_1): Remove expander.  Merge insn with
21878         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21879         Use general_x64nomem_operand as operand 1 predicate.
21880         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21881         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21882         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21883         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21884         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21885         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21886         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21887         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21888         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21889         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21890         (general_x64nomem_operand): Ditto.
21892 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21894         * config/rtems.opt: Add -pthread option.
21896 2013-03-26  Richard Biener  <rguenther@suse.de>
21898         * alias.c (find_base_term): Avoid redundant and not used recursion.
21899         (base_alias_check): Get the initial base term from the caller.
21900         (true_dependence_1): Compute and pass base terms to base_alias_check.
21901         (write_dependence_p): Likewise.
21902         (may_alias_p): Likewise.
21904 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21906         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21907         PC-relative load in SI modes and above only.
21909 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21911         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21912         * config/arm/iwmmxt.md (WCGR0): Update.
21913         (WCGR1, WCGR2, WCGR3): Likewise.
21915 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21917         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21918         Use x64 and nox64 isa attributes.
21920 2013-03-26  Richard Biener  <rguenther@suse.de>
21922         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21923         alignment computations and rely on get_object_alignment_1
21924         for the !TYPE_P case.
21925         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21927 2013-03-26  Walter Lee  <walt@tilera.com>
21929         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21930         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21932 2013-03-25  Jeff Law  <law@redhat.com>
21934         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21935         check for INTEGRAL_TYPE_P that was missing due to checking in
21936         wrong version of prior patch.
21938 2013-03-25  Walter Lee  <walt@tilera.com>
21940         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21941         TILEGX_INSN_SHUFFLEBYTES1.
21942         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21943         shufflebytes1.
21944         (tilegx_builtins): Ditto.
21945         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21947 2013-03-25  Walter Lee  <walt@tilera.com>
21949         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21950         (floatunssisf2): New pattern.
21951         (floatsidf2): New pattern.
21952         (floatunssidf2): New pattern.
21954 2013-03-25  Walter Lee  <walt@tilera.com>
21956         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21957         tests for constraint J, K, N, P.
21959 2013-03-25  Walter Lee  <walt@tilera.com>
21961         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21962         Use indirect/pcrel encoding.
21963         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21964         Ditto.
21966 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21968         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21969         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21970         * config/mips/mips.c (mips_option_override): Set IMADD default.
21971         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21972         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21973         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21974         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21975         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21977 2013-03-25  Jeff Law  <law@redhat.com>
21979         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21980         slightly to avoid creating and folding useless trees.  Simplify
21981         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21983 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21985         * config/i386/i386.md (*zero_extendsidi2): Merge with
21986         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21987         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21988         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21989         targets.  Clarify comment.
21991 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21993         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21994         pass-through jump functions differently.
21995         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21996         up jump functions.
21998 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22000         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22001         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22002         process it.
22003         (ipa_get_indirect_edge_target): New function.
22004         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22005         ipa_get_indirect_edge_target.  Update all callers.
22006         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22007         ipa_get_indirect_edge_target_1 instead of calling
22008         ipa_get_indirect_edge_target.
22009         (create_specialized_node): Pass aggvlas to
22010         ipcp_discover_new_direct_edges.
22012 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22014         * config/arm/arm.md (f_sels, f_seld): New types.
22015         (*cmov<mode>): New pattern.
22016         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22017         predicate.
22019 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22021         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22022         POSIX-printf for mingw-hosted  builds.
22024 2013-03-25  Richard Biener  <rguenther@suse.de>
22026         PR middle-end/56694
22027         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22028         must-not-throw stmt location.
22030 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22032         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22033         Emit load-acquire versions when acq is true.
22034         (arm_emit_store_exclusive): Add rel parameter.
22035         Emit store-release versions when rel is true.
22036         (arm_split_compare_and_swap): Use acquire-release instructions
22037         instead.
22038         of barriers when appropriate.
22039         (arm_split_atomic_op): Likewise.
22040         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22041         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22042         (VUNSPEC_SLX): Likewise.
22043         (VUNSPEC_LDA): Likewise.
22044         (VUNSPEC_STL): Likewise.
22045         * config/arm/sync.md (atomic_load<mode>): New pattern.
22046         (atomic_store<mode>): Likewise.
22047         (arm_load_acquire_exclusive<mode>): Likewise.
22048         (arm_load_acquire_exclusivesi): Likewise.
22049         (arm_load_acquire_exclusivedi): Likewise.
22050         (arm_store_release_exclusive<mode>): Likewise.
22052 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22054         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22055         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22056         * config/mip/predicates.md (lwsp_swsp_operand,
22057         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22058         sb16_operand, db4_operand, db7_operand, ib3_operand,
22059         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22060         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22061         andi16_operand): New predicates.
22062         * config/mips/mips.md (compression): New attribute.
22063         (enabled): New attribute.
22064         (length): Consider compression in computing length.
22065         (shift_compression): New code attribute.
22066         (*add<mode>3): New operands. Record compression.
22067         (sub<mode>3): Likewise.
22068         (one_cmpl<mode>2): Likewise.
22069         (*and<mode>3): Likewise.
22070         (*ior<mode>3): Likewise.
22071         (unnamed pattern for xor): Likewise.
22072         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22073         (*<optab><mode>3): Likewise.
22074         (*mov<mode>_internal: Likewise.
22075         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22076         (mips_unsigned_immediate_p): New.
22077         (umips_lwsp_swsp_address_p): New.
22078         (m16_based_address_p): New.
22079         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22080         (mips_unsigned_immediate_p): New prototype.
22081         (lwsp_swsp_address_p): New prototype.
22082         (m16_based_address_p): New prototype.
22083         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22084         (mips_signed_immediate_p): New function.
22085         (m16_based_address_p): New function.
22086         (lwsp_swsp_address_p): New function.
22087         (mips_print_operand_punctuation): Recognize short delay slot insns
22088         for microMIPS.add<mode>3"
22090 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22092         PR target/56720
22093         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22094         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22096 2013-03-25  Richard Biener  <rguenther@suse.de>
22098         PR tree-optimization/56689
22099         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22100         any edge.
22102 2013-03-25  Richard Biener  <rguenther@suse.de>
22104         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22105         of bitmap.
22106         (memory_references): Likewise.
22107         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22108         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22109         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22110         (gather_mem_refs_in_loops): Fold into ...
22111         (analyze_memory_references): ... this.  Move initialization
22112         to tree_ssa_lim_initialize.
22113         (fill_always_executed_in): Rename to ...
22114         (fill_always_executed_in_1): ... this.
22115         (fill_always_executed_in): Move contains_call computation to
22116         this new function from ...
22117         (tree_ssa_lim_initialize): ... here.
22118         (tree_ssa_lim): Call fill_always_executed_in.
22120 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22122         * postreload.c (reload_combine): Fix code detecting returns.
22124 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22126         * function.c (emit_use_return_register_into_block): On cc0 targets,
22127         do not emit the sequence between cc0 setter and user.
22129 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22131         * config/i386/predicates.md (local_symbolic_operand): Interpret
22132         dll-imported symbols as none-local.
22134 2013-03-25  Richard Biener  <rguenther@suse.de>
22136         * tree-ssa-loop-im.c (struct depend): Remove.
22137         (struct lim_aux_data): Make depends a vec of gimples.
22138         (free_lim_aux_data): Adjust.
22139         (add_dependency): Likewise.
22140         (set_level): Likewise.
22142 2013-03-25  Richard Biener  <rguenther@suse.de>
22144         PR middle-end/56434
22145         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22146         the pointer returned by calls with ECF_MALLOC set.
22148 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22150         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22152 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22154         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22155         using MMXMODE mode iterator.
22156         (*move<mode>_internal): Merge with *movv2sf_internal and
22157         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22159 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22161         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22162         (record_last_mem_set_info): Likewise.
22164         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22165         of XNEWVEC followed by memset.
22166         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22168 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22170         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22171         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22172         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22173         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22174         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22175         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22176         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22177         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22178         BARRIER_P instead of GET_CODE.
22180 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22182         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22183         inaccuracy in the probing code.
22185         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22186         (ctrapdi4): Likewise.
22188 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22190         * calls.c (expand_call): Add missing guard to code handling return
22191         of non-BLKmode structures in MSB.
22192         * function.c (expand_function_end): Likewise.
22194 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22196         * combine.c (try_combine): Adjust comment.  Do not add the set of
22197         insn #0 if the destination indirectly is set or dies in insn #2.
22198         Tidy up code to distribute a new note.
22200 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22202         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22203         also for alternatives 16 and 17.
22205 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22207         * config/i386/sse.md (*mov<mode>_internal): Merge with
22208         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22209         Emit insn template depending on type attribute.  Use
22210         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22211         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22212         attribute calculation.  Remove unit attribute calculation.
22213         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22214         Set prefix_data16 attribute for DImode ssemov types.
22215         Use Ym instead of y for SSE-MMX conversion alternatives.
22216         Reorder operand constraints.
22218 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22220         * df.h (df_insn_delete): Adjust prototype.
22221         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22222         and let it decide whether mark the basic block dirty.
22223         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22224         * df-scan.c (df_insn_info_delete): New helper function, split
22225         off from df_insn_delete.
22226         (df_scan_free_bb_info): Use it.
22227         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22228         Likewise.
22229         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22230         that the insn is actually an insn and it has a non-NULL basic block.
22231         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22233 2013-03-22  Richard Biener  <rguenther@suse.de>
22235         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22236         dep_ref members.
22237         (mem_ref_alloc): Do not allocate them.
22238         (refs_independent_p): Do not query or maintain a cache.
22240 2013-03-22  Richard Biener  <rguenther@suse.de>
22242         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22243         (gather_mem_refs_in_loops): Do not compute it.
22244         (analyze_memory_references): Do not allocate it.
22245         (tree_ssa_lim_finalize): Do not free it.
22246         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22248 2013-03-22  Richard Biener  <rguenther@suse.de>
22250         * is-a.h (as_a): Use gcc_checking_assert.
22252 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22254         * config/aarch64/aarch64.c (aarch64_print_operand): New
22255         format specifier for printing a constant in hex.
22256         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22257         format specifier for printing second operand.
22259 2013-03-22  Richard Biener  <rguenther@suse.de>
22261         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22262         bitmaps.
22263         (gather_mem_refs_in_loops): Perform store accumulation here.
22264         (create_vop_ref_mapping_loop): Remove.
22265         (create_vop_ref_mapping): Likewise.
22266         (analyze_memory_references): Initialize refs_stored_in_loop.
22267         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22268         (record_indep_loop): Remove.
22269         (record_dep_loop): New function.
22270         (ref_indep_loop_p_1): Adjust to only walk over references
22271         in the loop, not its subloops.
22272         (ref_indep_loop_p): Rename to ...
22273         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22274         maintaining a more fine-grained cache.
22275         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22276         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22278 2013-03-22  Richard Biener  <rguenther@suse.de>
22280         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22281         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22282         aggregate mem_ref_loc.
22283         (free_mem_ref_locs): Inline into ...
22284         (memref_free): ... this and adjust.
22285         (mem_ref_alloc): Adjust.
22286         (mem_ref_locs_alloc): Remove.
22287         (record_mem_ref_loc): Adjust.
22288         (get_all_locs_in_loop): Rewrite into ...
22289         (for_all_locs_in_loop): ... this iterator.
22290         (rewrite_mem_ref_loc): New functor.
22291         (rewrite_mem_refs): Use for_all_locs_in_loop.
22292         (sm_set_flag_if_changed): New functor.
22293         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22294         (ref_always_accessed): New functor.
22295         (ref_always_accessed_p): Use for_all_locs_in_loop.
22297 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22299         * tree-pass.h (PROP_gimple_lvec): New.
22300         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22301         (init_optimization_passes): Move pass_lower_vector.
22302         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22303         PROP_gimple_lvec.
22304         (pass_lower_vector): Provide PROP_gimple_lvec.
22305         (pass_lower_vector_ssa): Likewise.
22306         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22308 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22310         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22312 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22314         * config/i386/i386.md (*movdi_internal): Disparage slightly
22315         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22316         conversion alternatives.
22318 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22320         PR middle-end/48087
22321         * diagnostic.def (DK_WERROR): New kind.
22322         * diagnostic.h (werrorcount): Define.
22323         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22324         promoted to DK_ERROR, increment DK_WERROR counter instead of
22325         DK_ERROR counter.
22326         * toplev.c (toplev_main): Call print_ignored_options even if
22327         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22328         even if just werrorcount is non-zero.
22330         PR debug/55608
22331         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22332         on failure.
22333         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22334         (string_cst_pool_decl): New function.
22335         (optimize_one_addr_into_implicit_ptr): New function.
22336         (resolve_addr_in_expr): Optimize DWARF location expression
22337         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22338         which doesn't live in memory, but has DW_AT_location or
22339         DW_AT_const_value, or refers to a string literal, into
22340         DW_OP_GNU_implicit_pointer.
22341         (optimize_location_into_implicit_ptr): New function.
22342         (resolve_addr): If removing DW_AT_location of a variable because
22343         it was DW_OP_addr of address of the variable, but the variable doesn't
22344         live in memory, try to emit const value attribute for the initializer.
22346 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22348         * tree.h (VECTOR_TYPE_P): New macro.
22349         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22350         TYPE_MODE): Use it.
22351         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22352         VEC_COND_EXPR cannot be lvalues.
22353         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22355 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22357         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22358         Restrict the transformation to equal modes.
22360 2013-03-21  Richard Biener  <rguenther@suse.de>
22362         PR tree-optimization/39326
22363         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22364         (MEM_ANALYZABLE): Adjust.
22365         (record_mem_ref_loc): Move bitmap ops ...
22366         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22367         unanalyzable refs, do not record locations for it.
22368         (analyze_memory_references): Allocate ref zero as shared
22369         unanalyzable ref.
22370         (refs_independent_p): Do not test for unanalyzed mems here.
22371         (ref_indep_loop_p_1): Special-case disambiguation against
22372         the unanalyzed ref.
22373         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22375 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22377         * config/arm/arm-protos.h (tune_params): Add
22378         prefer_neon_for_64bits field.
22379         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22380         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22381         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22382         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22383         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22384         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22385         (arm_option_override): Handle -mneon-for-64bits new option.
22386         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22387         (prefer_neon_for_64bits): Declare new variable.
22388         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22389         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22390         (arch_enabled): Handle new arch types. Remove support for onlya8
22391         and nota8.
22392         (one_cmpldi2): Use new arch names.
22393         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22394         * config/arm/arm.opt (mneon-for-64bits): Add option.
22395         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22396         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22397         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22398         of onlya8.
22399         * doc/invoke.texi (-mneon-for-64bits): Document.
22401 2013-03-21  Richard Biener  <rguenther@suse.de>
22403         PR tree-optimization/39326
22404         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22405         (sort_bbs_in_loop_postorder_cmp): New function.
22406         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22408 2013-03-21  Richard Biener  <rguenther@suse.de>
22410         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22411         (vect_insert_into_interleaving_chain): Likewise.
22412         (vect_drs_dependent_in_basic_block): Inline ...
22413         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22414         split out from ...
22415         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22416         (vect_check_interleaving): Simplify.
22417         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22418         (vect_slp_analyze_data_ref_dependences): ... this new function.
22419         (dr_group_sort_cmp): New function.
22420         (vect_analyze_data_ref_accesses): Compute data-reference groups
22421         here instead of in vect_analyze_data_ref_dependence.  Use
22422         a more efficient algorithm.
22423         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22424         vect_slp_analyze_data_ref_dependences.  Call
22425         vect_analyze_data_ref_accesses earlier.
22426         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22427         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22428         (vect_slp_analyze_data_ref_dependences): New prototype.
22430 2013-03-21  Richard Biener  <rguenther@suse.de>
22432         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22433         ref is stored in the loop.
22434         (find_refs_for_sm): Walk only over all stores.
22435         (store_motion_loop): Allocate from lim_bitmap_obstack.
22436         (store_motion): Likewise.
22438 2013-03-21  Richard Biener  <rguenther@suse.de>
22440         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22441         Update virtual SSA form.
22443 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22445         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22446         * configure: Regenerate.
22447         * config.in: Regenerate.
22448         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22449         if !HAVE_LD_EH_FRAME_CIEV3.
22451 2013-03-21  Richard Biener  <rguenther@suse.de>
22453         * tree-cfg.c (verify_expr_no_block): New function.
22454         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22455         nor DECL_VALUE_EXPR have locations with associated blocks.
22456         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22457         (clear_unused_block_pointer): Remove code dealing with
22458         blocks in DECL_DEBUG_EXPR locations.
22460 2013-03-21  Richard Biener  <rguenther@suse.de>
22462         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22463         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22464         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22465         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22466         instead of DECL_DEBUG_EXPR_IS_FROM.
22467         * gimplify.c (gimplify_modify_expr): Likewise.
22468         * tree-cfg.c (verify_expr_location_1): Likewise.
22469         * tree-complex.c (create_one_component_var): Likewise.
22470         * tree-sra.c (create_access_replacement): Likewise.
22471         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22472         (clear_unused_block_pointer): Likewise.
22473         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22474         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22475         * var-tracking.c (var_debug_decl): Likewise.
22476         (track_expr_p): Likewise.
22477         * tree-inline.c (add_local_variables): Likewise.  Set
22478         DECL_HAS_DEBUG_EXPR_P after copying it.
22479         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22480         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22482 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22484         PR bootstrap/56656
22485         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22486         * configure: Regenerate.
22487         * config.in: Regenerate.
22488         * config/i386/i386.md (*movdf_internal): Use
22489         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22490         movd instead of movq mnemonic for interunit moves.
22491         (*movdi_internal): Ditto.
22493 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22495         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22496         (abd<mode>_3): New pattern.
22497         (aba<mode>_3): New pattern.
22498         (fabd<mode>_3): New pattern.
22500 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22502         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22503         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22504         occurrence of REGISTER_PREFIX as its empty string.
22506 2013-03-20  Jeff Law  <law@redhat.com>
22508         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22509         addititional equivalences for equality comparisons between an SSA_NAME
22510         and a constant where the SSA_NAME was set from a widening conversion.
22512 2013-03-20  Walter Lee  <walt@tilera.com>
22514         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22516 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22518         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22519         depending on type attribute.
22520         (*movti_internal): Ditto.
22521         (*movtf_internal): Ditto.
22522         (*movxf_internal): Ditto.
22523         (*movdf_internal): Ditto.
22524         (*movsf_internal): Ditto.
22526 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22528         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22529         maybe_vex for sselog1 and ssemov types.
22530         (*movdi_internal): Reorder operand constraints.
22531         (*movsi_internal): Ditto.  Set prefix attribute to
22532         maybe_vex for sselog1 and ssemov types.
22533         (*movtf_internal): Set prefix attribute to maybe_vex
22534         for sselog1 and ssemov types.
22535         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22536         DImode ssemov types.  Reorder operand constraints.
22537         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22538         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22539         attribute for SImode ssemov types.  Reorder operand constraints.
22541 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22543         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22544         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22546 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22548         * config/rs6000/predicates.md (indexed_address, update_address_mem
22549         update_indexed_address_mem): New predicates.
22550         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22551         attribute for load/store instructions.
22552         * config/rs6000/dfp.md (movsd_store): Likewise.
22553         (movsd_load): Likewise.
22554         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22555         (unnamed HI->DI extend define_insn): Likewise.
22556         (unnamed SI->DI extend define_insn): Likewise.
22557         (unnamed QI->SI extend define_insn): Likewise.
22558         (unnamed QI->HI extend define_insn): Likewise.
22559         (unnamed HI->SI extend define_insn): Likewise.
22560         (unnamed HI->SI extend define_insn): Likewise.
22561         (extendsfdf2_fpr): Likewise.
22562         (movsi_internal1): Likewise.
22563         (movsi_internal1_single): Likewise.
22564         (movhi_internal): Likewise.
22565         (movqi_internal): Likewise.
22566         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22567         attribute for load/store instructions.
22568         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22569         instructions.
22570         (mov<mode>_softfloat): Likewise.
22571         (mov<mode>_hardfloat32): Likewise.
22572         (mov<mode>_hardfloat64): Likewise.
22573         (mov<mode>_softfloat64): Likewise.
22574         (movdi_internal32): Likewise.
22575         (movdi_internal64): Likewise.
22576         (probe_stack_<mode>): Likewise.
22578 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22580         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22581         floating point, and decimal floating point to reload iterator.
22583         * config/rs6000/constraints.md (wl constraint): New constraints to
22584         return FLOAT_REGS if certain options are used to reduce the number
22585         of separate patterns that exist in the file.
22586         (wx constraint): Likewise.
22587         (wz constraint): Likewise.
22589         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22590         -mdebug=reg, print wg, wl, wx, and wz constraints.
22591         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22592         Initialize the reload functions for 64-bit binary/decimal floating
22593         point types.
22594         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22595         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22596         create the buffer on the stack to overcome not having a 32-bit
22597         load and store.
22598         (rs6000_emit_move): Likewise.
22599         (rs6000_secondary_memory_needed_rtx): Likewise.
22600         (rs6000_alloc_sdmode_stack_slot): Likewise.
22601         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22602         via xxlxor, just like DFmode 0.0.
22604         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22605         define as 1 if we are running on a power7 or newer.
22606         (enum r6000_reg_class_enum): Add new constraints.
22608         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22609         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22610         with other moves by using conditional constraits (wg).  Use LFIWZX
22611         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22612         (movsd splitter): Likewise.
22613         (movsd_hardfloat): Likewise.
22614         (movsd_softfloat): Likewise.
22616         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22617         binary and decimal floating point moves.
22618         (fmove_ok): New attributes to combine binary and decimal floating
22619         point moves, and to combine power6x (mfpgpr) moves along normal
22620         floating moves.
22621         (real_value_to_target): Likewise.
22622         (f32_lr): Likewise.
22623         (f32_lm): Likewise.
22624         (f32_li): Likewise.
22625         (f32_sr): Likewise.
22626         (f32_sm): Likewise.
22627         (f32_si): Likewise.
22628         (movsf): Combine binary and decimal floating point moves.  Combine
22629         power6x (mfpgpr) moves with other moves by using conditional
22630         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22631         (mov<mode> for SFmode/SDmode); Likewise.
22632         (SFmode/SDmode splitters): Likewise.
22633         (movsf_hardfloat): Likewise.
22634         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22635         (movsf_softfloat): Likewise.
22636         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22638         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22639         wx and wz constraints.
22641         * config/rs6000/constraints.md (wg constraint): New constraint to
22642         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22644         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22645         constraint.
22647         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22648         -mdebug=reg, print wg, wl, wx, and wz constraints.
22649         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22650         Initialize the reload functions for 64-bit binary/decimal floating
22651         point types.
22652         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22653         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22654         create the buffer on the stack to overcome not having a 32-bit
22655         load and store.
22656         (rs6000_emit_move): Likewise.
22657         (rs6000_secondary_memory_needed_rtx): Likewise.
22658         (rs6000_alloc_sdmode_stack_slot): Likewise.
22659         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22660         via xxlxor, just like DFmode 0.0.
22662         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22663         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22664         with other moves by using conditional constraits (wg).  Use LFIWZX
22665         and STFIWX for loading SDmode on power7.
22666         (movdd splitters): Likewise.
22667         (movdd_hardfloat32): Likewise.
22668         (movdd_softfloat32): Likewise.
22669         (movdd_hardfloat64_mfpgpr): Likewise.
22670         (movdd_hardfloat64): Likewise.
22671         (movdd_softfloat64): Likewise.
22673         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22674         64-bit binary and decimal floating point moves.
22675         (FMOVE64X): Likewise.
22676         (movdf): Combine 64-bit binary and decimal floating point moves.
22677         Combine power6x (mfpgpr) moves with other moves by using
22678         conditional constraits (wg).
22679         (mov<mode> for DFmode/DDmode): Likewise.
22680         (DFmode/DDmode splitters): Likewise.
22681         (movdf_hardfloat32): Likewise.
22682         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22683         (movdf_softfloat32): Likewise.
22684         (movdf_hardfloat64_mfpgpr): Likewise.
22685         (movdf_hardfloat64): Likewise.
22686         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22687         (movdf_softfloat64): Likewise.
22688         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22689         (reload_<mode>_load): Move to later in the file so they aren't in
22690         the middle of the floating point move insns.
22691         (reload_<mode>_store): Likewise.
22693         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22694         constraint.
22696         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22697         constraint if -mdebug=reg.
22698         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22699         Enable using dd reload support if needed.
22701         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22702         binary and decimal floating point moves in rs6000.md.
22703         (movtd_internal): Likewise.
22705         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22706         decimal floating point moves.
22707         (movtf): Likewise.
22708         (movtf_internal): Likewise.
22709         (mov<mode>_internal, TDmode/TFmode): Likewise.
22710         (movtf_softfloat): Likewise.
22711         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22713         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22714         movdi_internal64, using wg constraint for move direct operations.
22715         (movdi_internal64): Likewise.
22717         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22718         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22719         the various virtual registers. Use GPR/FPR first/last values,
22720         instead of hard coding the register numbers.  Print which modes
22721         have reload functions registered.
22722         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22723         settings before/after setting cpu, target and subtarget settings.
22724         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22725         and for secondary reload failures in rs6000_secondary_reload_inner.
22726         (rs6000_secondary_reload_fail): Likewise.
22727         (rs6000_secondary_reload_inner): Likewise.
22729         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22730         macros for first/last GPR and FPR registers.
22731         (LAST_GPR_REGNO): Likewise.
22732         (FIRST_FPR_REGNO): Likewise.
22733         (LAST_FPR_REGNO): Likewise.
22735         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22736         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22737         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22738         (vcond<mode><mode>): Likewise.
22739         (vcondu<mode><mode>): Likewise.
22740         (vector_gtu<mode>): Likewise.
22741         (vector_gte<mode>): Likewise.
22742         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22743         to prevent the compiler from converting DImode operations to TImode.
22744         (ior<mode>3): Likewise.
22745         (and<mode>3): Likewise.
22746         (one_cmpl<mode>2): Likewise.
22747         (nor<mode>3): Likewise.
22748         (andc<mode>3): Likewise.
22750         * config/rs6000/constraints.md (wt constraint): New constraint
22751         that returns VSX_REGS if TImode is allowed in VSX registers.
22753         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22754         constant under VSX.
22756         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22757         similar to TImode, but it is restricted to being in the GPRs.
22759         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22760         TImode to occupy a single VSX register.
22762         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22763         -mvsx-timode for power7/power8.
22764         (power7 cpu): Likewise.
22765         (power8 cpu): Likewise.
22767         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22768         sure that TFmode/TDmode take up two registers if they are ever
22769         allowed in the upper VSX registers.
22770         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22771         registers.
22772         (rs6000_init_hard_regno_mode_ok): Likewise.
22773         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22774         constraint.  Print if LRA is turned on.
22775         (rs6000_option_override_internal): Give an error if -mvsx-timode
22776         and VSX is not enabled.
22777         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22778         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22779         to reg+offset addressing.  Use PTImode when checking offset
22780         addresses for validity.
22781         (reg_offset_addressing_ok_p): Likewise.
22782         (rs6000_legitimate_offset_address_p): Likewise.
22783         (rs6000_legitimize_address): Likewise.
22784         (rs6000_legitimize_reload_address): Likewise.
22785         (rs6000_legitimate_address_p): Likewise.
22786         (rs6000_eliminate_indexed_memrefs): Likewise.
22787         (rs6000_emit_move): Likewise.
22788         (rs6000_secondary_reload): Likewise.
22789         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22790         reloads to fpr registers to continue to use reg+offset addressing,
22791         but 64-bit reloads to altivec registers need reg+reg addressing.
22792         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22793         it.  Treat LO_SUM like a PLUS operation.
22794         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22795         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22796         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22797         registers to share a register with a smaller sized type, since VSX
22798         puts scalars in the upper 64-bits.
22799         (print_operand): Add support for PTImode.
22800         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22801         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22802         registers, but don't have arithmetic support.
22803         (rs6000_memory_move_cost): Add test for VSX.
22804         (rs6000_opt_masks): Add -mvsx-timode.
22806         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22807         for TImode.
22808         (VSs): Likewise.
22809         (VSr): Use wt constraint for TImode.
22810         (VSv): Drop TImode support.
22811         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22812         (vsx_movti_64bit): Likewise.
22813         (vsx_movti_32bit): Likewise.
22814         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22815         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22816         one '?' on the appropriate output constraint.  Do not allow TImode
22817         logical operations on 32-bit systems.
22818         (vsx_ior<mode>3): Likewise.
22819         (vsx_xor<mode>3): Likewise.
22820         (vsx_one_cmpl<mode>2): Likewise.
22821         (vsx_nor<mode>3): Likewise.
22822         (vsx_andc<mode>3): Likewise.
22823         (vsx_concat_<mode>): Likewise.
22824         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22826         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22827         OPTION_MASK_VSX_TIMODE.
22828         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22829         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22831         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22832         (TI2 iterator): New iterator for TImode, PTImode.
22833         (wd mode attribute): Add values for vector types.
22834         (movti_string): Replace TI move operations with operations for TImode
22835         and PTImode.  Add support for TImode being allowed in VSX registers.
22836         (mov<mode>_string, TImode/PTImode): Likewise.
22837         (movti_ppc64): Likewise.
22838         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22839         (TI mode splitters): Likewise.
22841         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22842         constraint.
22844 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22846         PR tree-optimization/56355
22847         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22848         Also handle integers with undefined overflow.
22850 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22851             Maciej W. Rozycki  <macro@codesourcery.com>
22852             Tom de Vries  <tom@codesourcery.com>
22853             Nathan Sidwell  <nathan@codesourcery.com>
22854             Iain Sandoe  <iain@codesourcery.com>
22855             Nathan Froyd  <froydnj@codesourcery.com>
22856             Chao-ying Fu  <fu@mips.com>
22858         * doc/extend.texi: (micromips, nomicromips, nocompression):
22859         Document new function attributes.
22860         * doc/invoke.texi (minterlink-compressed, mmicromips,
22861         m14k, m14ke, m14kec): Document new options.
22862         (minterlink-mips16): Update documentation.
22863         * doc/md.texi (ZC, ZD): Document new constraints.
22864         * configure.ac (gcc_cv_as_micromips): Check if linker
22865         supports the .set micromips directive.
22866         * configure: Regenerate.
22867         * config.in: Regenerate.
22868         * config/mips/mips-tables.opt: Regenerate.
22869         * config/mips/micromips.md: New file.
22870         * constraints.md (ZC, ZD): New constraints.
22871         * config/mips/predicates.md (movep_src_register): New predicate.
22872         (movep_src_operand): New predicate.
22873         (non_volatile_mem_operand): New predicate.
22874         * config/mips/mips.md (multimem): New type.
22875         (length): Differentiate between 17-bit and 18-bit branch offsets.
22876         (MOVEP1, MOVEP2): New mode iterator.
22877         (mov_<load>l): Use ZC constraint.
22878         (mov_<load>r): Likewise.
22879         (mov_<store>l): Likewise.
22880         (mov_<store>r): Likewise.
22881         (*branch_equality<mode>_inverted): Add microMIPS support.
22882         (*branch_equality<mode>): Likewise.
22883         (*jump_absolute): Likewise.
22884         (indirect_jump_<mode>): Likewise.
22885         (tablejump_<mode>): Likewise.
22886         (<optab>_internal): Likewise.
22887         (sibcall_internal): Likewise.
22888         (sibcall_value_internal): Likewise.
22889         (prefetch): Use constraint ZD.
22890         * config/mips/mips.opt (minterlink-compressed): New option.
22891         (minterlink-mips16): Now an alias for minterlink-compressed.
22892         (mmicromips): New option.
22893         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22894         (compare_and_swap_12): Likewise.
22895         (sync_add<mode>): Likewise.
22896         (sync_<optab>_12): Likewise.
22897         (sync_old_<optab>_12): Likewise.
22898         (sync_new_<optab>_12): Likewise.
22899         (sync_nand_12): Likewise.
22900         (sync_old_nand_12): Likewise.
22901         (sync_new_nand_12): Likewise.
22902         (sync_sub<mode>): Likewise.
22903         (sync_old_add<mode>): Likewise.
22904         (sync_old_sub<mode>): Likewise.
22905         (sync_new_add<mode>): Likewise.
22906         (sync_new_sub<mode>): Likewise.
22907         (sync_<optab><mode>): Likewise.
22908         (sync_old_<optab><mode>): Likewise.
22909         (sync_new_<optab><mode>): Likewise.
22910         (sync_nand<mode>): Likewise.
22911         (sync_old_nand<mode>): Likewise.
22912         (sync_new_nand<mode>): Likewise.
22913         (sync_lock_test_and_set<mode>): Likewise.
22914         (test_and_set_12): Likewise.
22915         (atomic_compare_and_swap<mode>): Likewise.
22916         (atomic_exchange<mode>_llsc): Likewise.
22917         (atomic_fetch_add<mode>_llsc): Likewise.
22918         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22919         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22920         (umips_save_restore_pattern_p): Likewise.
22921         (umips_load_store_pair_p): Likewise.
22922         (umips_output_load_store_pair): Likewise.
22923         (umips_movep_target_p): Likewise.
22924         (umips_12bit_offset_address_p): Likewise.
22925         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22926         (mips_base_mips16): Rename this...
22927         (mips_base_compression_flags): ...to this. Update all uses.
22928         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22929         (mips_mips16_decl_p): Delete.
22930         (mips_nomips16_decl_p): Delete.
22931         (mips_get_compress_on_flags): New function.
22932         (mips_get_compress_off_flags): New function.
22933         (mips_get_compress_mode): New function.
22934         (mips_get_compress_on_name): New function.
22935         (mips_get_compress_off_name): New function.
22936         (mips_insert_attributes): Support multiple compression types.
22937         (mips_merge_decl_attributes): Likewise.
22938         (umips_12bit_offset_address_p): New function.
22939         (mips_start_function_definition): Emit .set micromips directive.
22940         (mips_call_may_need_jalx_p): New function.
22941         (mips_function_ok_for_sibcall): Add microMIPS support.
22942         (mips_print_operand_punctuation): Support short delay slots and
22943         compact jumps.
22944         (umips_swm_mask, umips_swm_encoding): New.
22945         (umips_build_save_restore): New function.
22946         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22947         (was_mips16_p): Remove.
22948         (old_compression_mode): New.
22949         (mips_set_compression_mode): New function.
22950         (mips_set_current_function): Add microMIPS support.
22951         (mips_option_override): Likewise.
22952         (umips_save_restore_pattern_p): New function.
22953         (umips_output_save_restore): New function.
22954         (umips_load_store_pair_p_1): New function.
22955         (umips_load_store_pair_p): New function.
22956         (umips_output_load_store_pair_1): New function.
22957         (umips_output_load_store_pair): New function.
22958         (umips_movep_target_p) New function.
22959         (mips_prepare_pch_save): Add microMIPS support.
22960         * config/mips/mips.h (TARGET_COMPRESSION): New.
22961         (TARGET_CPU_CPP_BUILTINS): Update macro
22962         to use new compression flags and to support microMIPS.
22963         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22964         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22965         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22966         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22967         (ASM_SPEC): Support mmicromips and mno-micromips.
22968         (M16STORE_REG_P): New macro.
22969         (MIPS_CALL): Support TARGET_MICROMIPS.
22970         (MICROMIPS_J): New macro.
22971         (mips_base_mips16): Rename this...
22972         (mips_base_compression_flags): ...to this.
22973         (UMIPS_12BIT_OFFSET_P): New macro.
22974         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22975         (MULTILIB_DIRNAMES): Likewise.
22976 2013-03-20  Richard Biener  <rguenther@suse.de>
22978         PR tree-optimization/56661
22979         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22980         the result does not have to be distinct.
22982 2013-03-20  Richard Biener  <rguenther@suse.de>
22984         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22985         remap_gimple_op_r.
22987 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22988             Steven Bosscher  <steven@gcc.gnu.org>
22990         PR rtl-optimization/56605
22991         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22993 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22995         PR bootstrap/56656
22996         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22997         that require movd instead of movq.
22999 2013-03-20  Richard Biener  <rguenther@suse.de>
23001         * tree-ssa-structalias.c (struct variable_info): Add pointer
23002         to the first field of an aggregate with sub-vars.  Make
23003         this and the pointer to the next subfield its ID.
23004         (vi_next): New function.
23005         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23006         storedanything_id, integer_id): Increment by one.
23007         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23008         get_call_clobber_vi): Adjust.
23009         (solution_set_expand): Simplify and speedup.
23010         (solution_set_add): Inline into ...
23011         (set_union_with_increment): ... this.  Adjust accordingly.
23012         (do_sd_constraint): Likewise.
23013         (do_ds_constraint): Likewise.
23014         (do_complex_constraint): Simplify.
23015         (build_pred_graph): Adjust.
23016         (solve_graph): Likewise.  Simplify and speedup.
23017         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23018         get_constraint_for_component_ref, get_constraint_for_1,
23019         first_vi_for_offset, first_or_preceding_vi_for_offset,
23020         create_function_info_for, create_variable_info_for_1,
23021         create_variable_info_for, intra_create_variable_infos): Adjust.
23022         (init_base_vars): Push NULL for ID zero.
23023         (compute_points_to_sets): Adjust.
23025 2013-03-20  Richard Biener  <rguenther@suse.de>
23027         * cfgloop.c (verify_loop_structure): Streamline and avoid
23028         ICEing on corrupt loop tree.
23029         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23030         loop tree.
23032 2013-03-20  Richard Biener  <rguenther@suse.de>
23034         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23035         check whether an SSA update is needed.
23037 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23039         * config/mips/constraints.md (T): Rename to...
23040         (Yf): ...this.
23041         (U): Rename to...
23042         (Yd): ...this.
23043         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23044         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23046 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23048         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23049         (*subsi3_carryin_uxtw): Likewise.
23051 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23053         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23054         (*rorsi3_insn_uxtw): Likewise.
23056 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23058         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23059         (*extrsi5_insn_uxtw): Likewise.
23061 2013-03-19  Richard Biener  <rguenther@suse.de>
23063         PR tree-optimization/56273
23064         * passes.c (init_optimization_passes): Move second VRP after DOM.
23066 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23068         * config/i386/i386.md (*movti_internal): Merge from
23069         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23070         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23071         nox64 isa attributes.
23073 2013-03-18  Richard Biener  <rguenther@suse.de>
23075         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23076         (unite): Likewise.
23077         (merge_node_constraints): Likewise.
23078         (build_succ_graph): Likewise.
23079         (valid_graph_edge): Inline into single caller.
23080         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23081         and cache varinfo.
23082         (scc_visit): Fix formatting and variable use.
23083         (do_sd_constraint): Use gcc_checking_assert.
23084         (do_ds_constraint): Likewise.
23085         (do_complex_constraint): Likewise.
23086         (condense_visit): Likewise.  Cleanup.
23087         (dump_pred_graph): New function.
23088         (perform_var_substitution): Dump the pred-graph before
23089         variable substitution.
23090         (find_equivalent_node): Use gcc_checking_assert.
23091         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23093 2013-03-18  Richard Biener  <rguenther@suse.de>
23095         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23096         Remove cond_expr_stmt_list argument and do not gimplify the
23097         built expression.
23098         (vect_loop_versioning): Adjust.
23099         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23100         Cleanup to use less temporaries.
23101         (vect_create_data_ref_ptr): Cleanup.
23103 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23105         PR tree-optimization/56635
23106         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23107         require types_compatible_p types.
23109 2013-03-18  Nick Clifton  <nickc@redhat.com>
23111         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23112         spurious backslash.
23114         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23115         Add missing line to comment describing function.
23117 2013-03-18  Richard Biener  <rguenther@suse.de>
23119         PR tree-optimization/56210
23120         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23121         Handle string / character search functions.
23122         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23124 2013-03-18  Richard Biener  <rguenther@suse.de>
23126         PR middle-end/56483
23127         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23128         and implement properly.
23129         * gimple.h (gimple_cond_single_var_p): Remove.
23131 2013-03-18  Richard Biener  <rguenther@suse.de>
23133         * tree-data-ref.h (find_data_references_in_loop): Declare.
23134         * tree-data-ref.c (get_references_in_stmt): Use a stack
23135         vector pre-allocated in the callers.
23136         (find_data_references_in_stmt): Adjust.
23137         (graphite_find_data_references_in_stmt): Likewise.
23138         (create_rdg_vertices): Likewise.
23139         (find_data_references_in_loop): Export.
23140         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23141         Compute dependences here...
23142         (vect_analyze_data_refs): ...not here.  When we encounter
23143         a non-vectorizable data reference in basic-block vectorization
23144         truncate the data reference vector.  Do not bother to
23145         fixup data-dependence information for gather loads.
23146         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23147         of data references, as reported.
23149 2013-03-18  Richard Biener  <rguenther@suse.de>
23151         PR tree-optimization/3713
23152         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23153         has_constants and expr.
23154         (stmt_has_constants): Properly valueize SSA names when deciding
23155         whether the stmt has constants.
23157 2013-03-18  Richard Biener  <rguenther@suse.de>
23159         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23160         whole function when there is nothing to do.
23161         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23162         * tree-vectorizer.c (vectorize_loops): Update virtual and
23163         loop-closed SSA once.
23164         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23166 2013-03-18  Richard Biener  <rguenther@suse.de>
23168         PR middle-end/56113
23169         * domwalk.c (bb_postorder): New global static.
23170         (cmp_bb_postorder): New function.
23171         (walk_dominator_tree): Replace scheme imposing an order for
23172         visiting dominator sons by one sorting them at the time they
23173         are pushed on the stack.
23175 2013-03-18  Richard Biener  <rguenther@suse.de>
23177         PR tree-optimization/39326
23178         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23179         (struct mem_ref): Replace mem member with ao_ref typed member.
23180         (MEM_ANALYZABLE): Adjust.
23181         (memref_eq): Likewise.
23182         (mem_ref_alloc): Likewise.
23183         (gather_mem_refs_stmt): Likewise.
23184         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23185         (execute_sm_if_changed_flag_set): Adjust.
23186         (execute_sm): Likewise.
23187         (ref_always_accessed_p): Likewise.
23188         (refs_independent_p): Likewise.
23189         (can_sm_ref_p): Likewise.
23191 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23193         PR c/56566
23194         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23195         return 1 even for !unsignedp.
23197 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23199         * config/i386/i386.md (isa): Add x64 and nox64.
23200         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23201         (*pushtf): Enable *roF alternative for x64 isa only.
23202         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23203         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23204         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23205         (*movtf_internal): Merge from *movtf_internal_rex64 and
23206         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23207         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23208         nox64 isa attributes.
23209         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23210         nox64 isa attributes.
23211         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23213 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23215         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23217 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23219         PR target/56640
23220         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23221         class names.  Remove trailing comma after "ALL_REGS".
23223 2013-03-16  Jan Hubicka  <jh@suse.cz>
23225         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23226         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23227         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23228         of cgraph_get_create_node.
23229         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23231 2013-03-16  Jason Merrill  <jason@redhat.com>
23233         PR debug/49090
23234         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23235         with DW_AT_default_value.
23237 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23239         * BASE-VER: Set to 4.9.0.
23241 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23243         PR target/56619
23244         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23245         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23246         Document _x* TSX intrinsics.
23248 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23249             David Holsgrove  <david.holsgrove@xilinx.com>
23251         * configure.ac: Add MicroBlaze TLS support detection.
23252         * configure: Regenerate.
23253         * config/microblaze/microblaze-protos.h
23254         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23255         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23256         * config/microblaze/microblaze.c (microblaze_address_type): Add
23257         ADDRESS_TLS and tls_reloc address types.
23258         (microblaze_address_info): Add tls_reloc.
23259         (TARGET_HAVE_TLS): Define.
23260         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23261          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23262          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23263         load_tls_operand,  microblaze_call_tls_get_addr,
23264         microblaze_legitimize_tls_address): New functions.
23265         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23266         (get_base_reg): Use microblaze_tls_symbol_p.
23267         (microblaze_classify_address): Handle TLS.
23268         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23269         label_mentioned_p and microblaze_tls_referenced_p.
23270         (microblaze_legitimize_address): Handle TLS.
23271         (microblaze_address_insns): Handle ADDRESS_TLS.
23272         (pic_address_needs_scratch): Handle TLS.
23273         (print_operand_address): Handle TLS.
23274         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23275         (microblaze_expand_move): Handle TLS.
23276         (microblaze_legitimate_constant_p): Check
23277         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23278         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23279         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23280         (PIC_OFFSET_TABLE_REGNUM): Set.
23281         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23282         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23283         (addsi3, movsi_internal2, movdf_internal): Update constraints
23284         * config/microblaze/predicates.md (arith_plus_operand): Define
23285         (move_operand): Redefine as move_src_operand,
23286         check microblaze_tls_referenced_p.
23288 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23290         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23291         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23293 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23295         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23296         CC mode for AND.
23298 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23300         PR tree-optimization/53265
23301         * common.opt (Waggressive-loop-optimizations): New option.
23302         * tree-ssa-loop-niter.c: Include tree-pass.h.
23303         (do_warn_aggressive_loop_optimizations): New function.
23304         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23305         if number_of_latch_executions returned constant.
23306         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23307         early.  If number_of_latch_executions returned constant, set
23308         nb_iterations_upper_bound back to it.
23309         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23310         field.
23311         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23312         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23314         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23315         (MULTILIB_OSDIRNAMES): Set.
23316         * genmultilib: If defaultosdirname doesn't start with :: , set
23317         defaultosdirname2 instead, clear it and emit two . multilib_raw
23318         entries instead of just one.
23320 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23322         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23323         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23324         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23325         (SUBTARGET_OVERRIDE_OPTIONS): New.
23327 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23329         PR target/49880
23330         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23331         (musermode): Convert to Var(TARGET_USERMODE).
23332         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23333         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23334         * config/sh/sh.c (sh_option_override): Use
23335         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23336         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23337         condition.
23338         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23339         TARGET_SH4.
23340         (udivsi3_i4_single, divsi3_i4_single): Use
23341         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23343 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23345         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23346         default setting.
23348 2013-03-13  Richard Biener  <rguenther@suse.de>
23350         PR tree-optimization/56608
23351         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23352         calls when vectorizing basic-blocks.
23354 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23356         PR plugins/45078
23357         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23358         tm_file.
23360 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23362         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23364 2013-03-11  Jan Hubicka  <jh@suse.cz>
23366         PR lto/56557
23367         * lto-streamer-out.c (output_symbol_p): Skip references from
23368         constructors of external variables.
23370 2013-03-11  Jan Hubicka  <jh@suse.cz>
23372         PR middle-end/56571
23373         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23374         from pseudos.
23375         * emit-rtl.c (verify_rtx_sharing): Likewise.
23376         (copy_insn_1): Likewise.
23377         * rtl.c (copy_rtx): Likewise.
23379 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23381         PR target/56591
23382         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23383         output_operand_lossage message.
23385 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23387         PR target/56470
23388         * arm.c (shift_op): Validate RTL pattern on the fly.
23389         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23390         the RTL.
23392 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23394         PR target/56347
23395         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23396         new call patterns that clobber %fr12.
23397         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23398         split and postreload patterns.
23399         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23400         registers %fr12 and %fr12R as call used.
23402 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23404         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23405         (canon_address, record_store, replace_read, check_mem_read_rtx,
23406         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23407         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23408         rest_of_handle_dse): Likewise.
23410 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23412         PR middle-end/56524
23413         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23414         Add base_optabs.
23415         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23416         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23417         (save_optabs_if_changed): Replace with...
23418         (init_tree_optimization_optabs): ...this.
23419         * optabs.c (save_optabs_if_changed): Rename to...
23420         (init_tree_optimization_optabs): ...this.  Take the optimization node
23421         as argument.  Do nothing if the base optabs are already correct.
23422         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23423         to recompute optabs.
23424         * function.h (function): Remove optabs field.
23425         * function.c (invoke_set_current_function_hook): Call
23426         init_tree_optimization_optabs.  Use the result to initialize
23427         this_fn_optabs.
23429 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23431         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23432         if GTMA_HAS_NO_INSTRUMENTATION.
23433         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23434         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23435         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23436         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23437         GTMA_HAS_NO_INSTRUMENTATION.
23439 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23441         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23442         libasan_preinit.o.
23444 2013-03-08  Marek Polacek  <polacek@redhat.com>
23445             Jakub Jelinek  <jakub@redhat.com>
23447         PR tree-optimization/56478
23448         * predict.c (is_comparison_with_loop_invariant_p): Change the
23449         type of loop_step to tree.
23450         (predict_loops): Adjust.
23451         (predict_iv_comparison): Perform the computations on double_ints.
23453 2013-03-08  Richard Biener  <rguenther@suse.de>
23455         PR tree-optimization/56570
23456         * tree-cfg.c (verify_expr_location_1): Verify locations for
23457         DECL_DEBUG_EXPR.
23458         * tree-sra.c (create_access_replacement): Strip locations
23459         from DECL_DEBUG_EXPRs.
23461 2013-03-08  Richard Biener  <rguenther@suse.de>
23463         * tree-inline.c (expand_call_inline): Do not associate
23464         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23465         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23467 2013-03-08  Richard Biener  <rguenther@suse.de>
23469         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23470         or block changes with -Og.  Fix for location / block encoding
23471         changes and PHI arguments with locations.
23473 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23475         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23476         for all counters.
23477         (struct output_info): Likewise.
23478         (register_overhead): Remove bad gcc_assert.
23479         (bitmap_find_bit): If there is only a single bitmap element, do not
23480         count a miss as a search.
23481         (print_statistics): Update for counter type changes.
23482         (dump_bitmap_statistics): Likewise.  Print headers such that they
23483         are properly lined up with the printed counters.
23485 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23487         PR tree-optimization/56559
23488         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23489         check that it has only a single use.
23491 2013-03-07  Richard Biener  <rguenther@suse.de>
23493         * doc/invoke.texi (fwhole-program): Discourage use in combination
23494         with -flto.
23496 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23498         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23500         PR tree-optimization/56539
23501         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23502         instead of GSI_CONTINUE_LINKING as last argument to
23503         force_gimple_operand_gsi.  Adjust function comment.
23505         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23506         aarch64-cores.def.
23508         PR middle-end/56548
23509         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23510         promoted mode, convert the result back to the original mode.
23512 2013-03-06  Richard Biener  <rguenther@suse.de>
23514         PR middle-end/56294
23515         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23516         (insert_updated_phi_nodes_compare_uids): New function.
23517         (update_ssa): Sort symbols_to_rename after UID before
23518         traversing it to insert PHI nodes.
23520 2013-03-06  Richard Biener  <rguenther@suse.de>
23522         PR middle-end/50494
23523         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23524         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23526         Revert
23527         2013-02-13  Richard Biener  <rguenther@suse.de>
23529         PR lto/50494
23530         * varasm.c (output_constant_def_1): Get the decl representing
23531         the constant as argument.
23532         (output_constant_def): Wrap output_constant_def_1.
23533         (make_decl_rtl): Use output_constant_def_1 with the decl
23534         representing the constant.
23535         (build_constant_desc): Optionally re-use a decl already
23536         representing the constant.
23537         (tree_output_constant_def): Adjust.
23539 2013-03-06  Joey Ye  <joey.ye@arm.com>
23541         PR lto/50293
23542         * gcc.c (convert_white_space): New function.
23543         (main): Handles white space in function name.
23545 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23547         PR target/56529
23548         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23549         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23550         to SH_DIV_CALL_TABLE for TARGET_SH2.
23551         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23552         list.
23553         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23554         call-table options.
23556 2013-03-05  Sterling Augustine  <saugustine@google.com>
23557             Cary Coutant  <ccoutant@google.com>
23559         PR debug/55364
23560         * dwarf2out.c (resolve_addr): Don't call
23561         remove_loc_list_addr_table_entries a second time for the same
23562         expression.
23564 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23566         PR debug/56510
23567         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23568         (avoid_complex_debug_insns): New function.
23569         (expand_debug_locations): Call it.
23571         PR rtl-optimization/56484
23572         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23573         lifetimes of hard registers on small register class machines.
23575 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23577         * config/microblaze/microblaze-protos.h: Rename
23578         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23579         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23580         fast_interrupt.
23581         (microblaze_fast_interrupt_function_p): New function.
23582         (microblaze_is_interrupt_handler): Rename to
23583         microblaze_is_interrupt_variant and add fast_interrupt check.
23584         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23585         (save_restore_insns): Likewise.
23586         (compute_frame_size): Likewise.
23587         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23588         (microblaze_globalize_label): Likewise.
23589         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23590         * config/microblaze/microblaze.md: Use wrapper
23591         microblaze_is_interrupt_variant.
23593 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23595         * sdbout.c (sdbout_one_type): Switch to current function's section
23596         supporting cold/hot.
23598 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23600         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23601         -mxl-reorder.
23603 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23605         PR middle-end/56461
23606         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23607         if VALGRIND_GET_VBITS is defined, temporarily make object
23608         memory all defined, and restore previous valgrind addressability
23609         and definability afterwards.  Free this_object at the end.
23611         PR middle-end/56461
23612         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23613         right before calling lra_create_live_ranges, also call it
23614         when clearing live_p.  Only call lra_clear_live_ranges
23615         at the end if live_p.
23617         PR middle-end/56461
23618         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23620 2013-03-05  Richard Biener  <rguenther@suse.de>
23622         PR tree-optimization/56521
23623         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23624         value-id.
23626 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23628         PR c++/55135
23629         * except.h (remove_unreachable_eh_regions): New prototype.
23630         * except.c (remove_eh_handler_splicer): New function, split out
23631         of remove_eh_handler.
23632         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23633         warning about running it on many EH regions one at a time.
23634         (remove_unreachable_eh_regions_worker): New function, walk the
23635         EH tree in depth-first order and remove non-marked regions.
23636         (remove_unreachable_eh_regions): New function.
23637         * tree-eh.c (mark_reachable_handlers): New function, split out
23638         from remove_unreachable_handlers.
23639         (remove_unreachable_handlers): Use mark_reachable_handlers and
23640         remove_unreachable_eh_regions.
23641         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23642         and remove_unreachable_eh_regions.
23644 2013-03-05  Richard Biener  <rguenther@suse.de>
23646         PR middle-end/56525
23647         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23648         not freeing them until the end.
23650 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23652         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23654 2013-03-05  Richard Biener  <rguenther@suse.de>
23656         PR tree-optimization/56270
23657         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23658         of loads after scheduling an SLP instance.
23660 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23662         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23663         tic6x.exp.
23664         (check_gcc_parallelize): Run guality.exp as a separate job from
23665         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23666         struct-layout-1.exp with stackalign.exp.
23668         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23670         PR middle-end/56461
23671         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23672         load_index sbitmap even if some bit in it isn't set.
23674         PR middle-end/56461
23675         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23676         (discover_iteration_bound_by_body_walk): Change queues to
23677         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23678         spelling in comment.  Call safe_push on queues[bound_index] directly.
23679         Release queues[queue_index] in every iteration unconditionally.
23680         Release bounds vector.
23682         PR middle-end/56461
23683         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23684         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23685         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23686         inner_phis vector.
23688 2013-03-05  Richard Biener  <rguenther@suse.de>
23690         PR lto/56515
23691         * tree-inline.c (remap_blocks_to_null): New function.
23692         (expand_call_inline): When expanding a call stmt without
23693         an associated block inline remap all callee blocks to NULL.
23695 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23697         PR rtl-optimization/56494
23698         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23699         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23700         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23702         PR middle-end/56461
23703         * sel-sched-ir.c (free_sched_pools): Release
23704         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23705         if succs_info_pool.max_top isn't -1.
23707         PR bootstrap/56509
23708         * opts.c (opts_obstack, opts_concat): Moved to...
23709         * opts-common.c (opts_obstack, opts_concat): ... here.
23711 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23713         PR middle-end/56461
23714         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23716 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23718         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23719         all appropriate places.
23721 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23723         PR tree-optimization/56424
23724         * ipa-split.c (split_function): Do not set the RSO flag if result is
23725         not by reference and its type is a register type.
23727 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23729         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23730         (microblaze_legitimate_pic_operand): Likewise
23731         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23732         new function microblaze_legitimate_pic_operand
23733         * config/microblaze/microblaze-protos.h
23734         (microblaze_legitimate_pic_operand): Declare.
23736 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23738         * config/microblaze/predicates.md (call_insn_simple_operand):
23739         New predicate for supported rtx code types.
23740         * config/microblaze/microblaze.md (call_internal1): Use
23741         call_insn_simple_operand predicate.
23743 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23745         PR middle-end/56461
23746         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23747         partitions.ordered_remove.
23749         PR middle-end/56461
23750         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23751         vec_oprnds0.create (1) for modifier == NONE.
23753         PR middle-end/56461
23754         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23755         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23756         vec_oprnds1 right before pushing anything to it for
23757         scalar_shift_arg.
23759         PR middle-end/56461
23760         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23761         set nbbs to 0 instead of having separate code path.
23762         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23763         instead of false as last argument if returning NULL.
23765 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23767         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23768         the attribute is now called "target" instead of "option".
23769         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23770         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23771         attribute/pragma name for TARGET_OPTION_VALID_P and
23772         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23773         * doc/tm.texi: Regenerated.
23775 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23777         * config/microblaze/microblaze.c:
23778         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23779         * config/microblaze/microblaze.h: Add -mxl-reorder to
23780         DRIVER_SELF_SPECS.
23781         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23782         instructions emitted if TARGET_REORDER.
23783         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23784         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23785         separately.
23787 2013-03-01  Xinliang David Li  <davidxl@google.com>
23789         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23790         walk length.
23792 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23794         PR middle-end/56461
23795         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23796         vector even when returning true.  Fix up function comment formatting.
23798         PR middle-end/56461
23799         * ira-build.c (ira_loop_nodes_count): New variable.
23800         (create_loop_tree_nodes): Initialize it.
23801         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23803         PR middle-end/56461
23804         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23805         method on dr_chain and result_chain.
23806         * tree-vect-stmts.c (vectorizable_store): Only call
23807         result_chain.create if j == 0.
23809         PR middle-end/56461
23810         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23811         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23812         before overwriting it.
23814 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23816         * doc/extended.texi (C Extensions): Change order in @menu
23817         to match @node.
23818         (Other MIPS Built-in Functions): Move last MIPS entry before
23819         "picoChip Built-in Functions".
23820         (SH Built-in Functions): Move after RX Built-in Functions.
23821         * doc/gcc.texi (Introduction): Change order in @menu
23822         to match @node.
23823         * doc/md.texi (Constraints): Ditto.
23824         * gty.texi (Type Information): Ditto.
23825         (User-provided marking routines for template types): Make
23826         subsection.
23827         * doc/invoke.texi (AArch64 Options): Move before
23828         "Adapteva Epiphany Options".
23830 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23831             Jakub Jelinek  <jakub@redhat.com>
23833         PR sanitizer/56454
23834         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23835         no_address_safety_analysis attribute.
23836         * doc/extend.texi (no_address_safety_attribute): Rename to
23837         no_sanitize_address attribute, mention no_address_safety_analysis
23838         attribute as deprecated alias.
23840 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23842         PR middle-end/56461
23843         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23844         type to vec<vec<tree> > *.
23845         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23846         to be vec<tree> instead of vec<tree> *, set vec_defs
23847         to vNULL and call vec_defs.create (number_of_vects), adjust other
23848         uses of vec_defs.
23849         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23850         vectorizable_condition): Adjust vect_get_slp_defs callers.
23852 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23854         * config/aarch64/aarch64.c
23855         (aarch64_float_const_representable): Remove unused variable.
23857 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23859         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23861 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23863         * config/aarch64/aarch64-builtins.c
23864         (aarch64_init_simd_builtins): Make static.
23866 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23868         * config/aarch64/aarch64.c
23869         (aarch64_simd_make_constant): Make static.
23871 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23873         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23874         with no initialization to the RHS of debug statements.
23876 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23878         PR tree-optimization/56294
23879         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23880         Adjust dumping.
23881         (get_access_replacement): Do not call create_access_replacement.
23882         Assert a replacement exists.
23883         (get_repl_default_def_ssa_name): Create the replacement declaration
23884         itself.
23886 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23888         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23889         final_end_function.
23891 2013-02-28  Marek Polacek  <polacek@redhat.com>
23893         PR rtl-optimization/56466
23894         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23895         if we're changing a loop.
23896         (peel_loops_completely): Likewise.
23898 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23900         PR c++/55813
23901         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23903 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23905         PR target/56445
23906         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23907         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23908         INTX_FTYPE_FX, FX_FTYPE_INTX.
23909         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23911 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23913         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23914         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23915         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23916         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23917         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23918         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23919         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23920         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23921         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23922         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23923         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23924         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23925         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23926         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23927         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23928         (avrxmega6): Increase max flash segments from 5 to 6.
23929         * config/avr/t-multilib: Regenerate.
23930         * config/avr/avr-tables.opt: Regenerate.
23931         * doc/avr-mmcu.texi: Regenerate.
23933 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23935         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23936         (avr_device_to_arch): Rename to avr_device_to_ld.
23937         (avr_device_to_as): New prototype.
23938         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23939         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23940         * config/avr/driver-avr.c (avr_device_to_as): New.
23941         (avr_device_to_arch): Rename to avr_device_to_ld.
23943 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23945         PR middle-end/56461
23946         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23947         method on dr_chain and result_chain.
23949         PR middle-end/56461
23950         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23951         pointer_set_destroy on not_executed_last_iteration.
23953         PR middle-end/56461
23954         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23956         PR middle-end/56461
23957         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23958         FOR_EACH_DEFINED_FUNCTION when freeing state.
23960         PR middle-end/56461
23961         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23962         pool_free.
23963         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23964         overwriting it.
23966         PR middle-end/56461
23967         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23968         known_aggs[i].items and release known_aggs vector.
23970         PR middle-end/56461
23971         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23973 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23975         * config/microblaze/microblaze.c (microblaze_emit_compare):
23976         Use xor for EQ/NE comparisions.
23977         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23978         (cbranchsf4): Adjust operator to comparison_operator.
23980 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23982         PR middle-end/56461
23983         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23984         vector.
23985         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23986         vec_safe_push, always update *slot.
23987         (redirect_edge_var_map_clear): Use vec_free.
23988         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23989         (free_var_map_entry): Use vec_free.
23990         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23991         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23993 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23995         PR middle-end/45472
23996         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23997         when the may_trap_p bit of the exprs being merged differs.
23998         Reorder tests for speculativeness in the logical and operator.
24000 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24002         * incpath.c (add_standard_paths): Use reconcat instead of concat
24003         where appropriate and avoid leaking memory.
24005         * opts.h: Include obstack.h.
24006         (opts_concat): New prototype.
24007         (opts_obstack): New declaration.
24008         * opts.c (opts_concat): New function.
24009         (opts_obstack): New variable.
24010         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24011         (finish_options): Use opts_concat instead of concat
24012         and XOBNEWVEC instead of XNEWVEC.
24013         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24014         generate_option): Likewise.
24015         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24016         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24018         PR target/56455
24019         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24020         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24022 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24024         PR middle-end/56461
24025         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24027 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24029         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24030         (arm_block_move_unaligned_straight): Likewise.
24031         (arm_adjust_block_mem): Likewise.
24033 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24035         PR target/48901
24036         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24037         temp, cond and label.
24038         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24040         PR target/52500
24041         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24042         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24044         PR target/52501
24045         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24046         of prologue/epilogue functions.
24048         PR target/52550
24049         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24050         Remove unused variable cfa_offset.
24051         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24053         PR target/54639
24054         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24055         type promotion to unsigned.
24057         PR target/54640
24058         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24059         for HOST_WIDE_INT of 32 bit / same size as int.
24060         (arm_block_move_unaligned_straight): Likewise.
24061         (arm_adjust_block_mem): Likewise.
24063         PR target/54662
24064         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24065         ALL_CFLAGS.
24067 2013-02-26  Marek Polacek  <polacek@redhat.com>
24069         PR tree-optimization/56426
24070         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24072 2013-02-26  Richard Biener  <rguenther@suse.de>
24074         PR target/56444
24075         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24076         unused variable loops.
24078 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24080         PR tree-optimization/56448
24081         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24082         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24083         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24084         later operands of the references, or even first operand for
24085         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24087         PR tree-optimization/56443
24088         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24089         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24090         to type_for_mode langhook.
24092 2013-02-25  Matt Turner  <mattst88@gmail.com>
24094         * doc/invoke.texi: Document r4700.
24096 2013-02-25  Richard Biener  <rguenther@suse.de>
24098         PR tree-optimization/56175
24099         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24100         split out from ...
24101         (simplify_bitwise_binary): ... here.  Also guard the conversion
24102         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24104 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24106         Revert:
24107         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24108             Maciej W. Rozycki  <macro@codesourcery.com>
24109             Tom de Vries  <tom@codesourcery.com>
24110             Nathan Sidwell  <nathan@codesourcery.com>
24111             Iain Sandoe  <iain@codesourcery.com>
24112             Nathan Froyd  <froydnj@codesourcery.com>
24113             Chao-ying Fu  <fu@mips.com>
24115         * doc/extend.texi: (micromips, nomicromips, nocompression):
24116         Document new function attributes.
24117         * doc/invoke.texi (minterlink-compressed, mmicromips,
24118         m14k, m14ke, m14kec): Document new options.
24119         (minterlink-mips16): Update documentation.
24120         * doc/md.texi (ZC, ZD): Document new constraints.
24121         * configure.ac (gcc_cv_as_micromips): Check if linker
24122         supports the .set micromips directive.
24123         * configure: Regenerate.
24124         * config.in: Regenerate.
24125         * config/mips/mips-tables.opt: Regenerate.
24126         * config/mips/micromips.md: New file.
24127         * constraints.md (ZC, AD): New constraints.
24128         * config/mips/predicates.md (movep_src_register): New predicate.
24129         (movep_src_operand): New predicate.
24130         (non_volatile_mem_operand): New predicate.
24131         * config/mips/mips.md (multimem): New type.
24132         (length): Differentiate between 17-bit and 18-bit branch offsets.
24133         (MOVEP1, MOVEP2): New mode iterator.
24134         (mov_<load>l): Use ZC constraint.
24135         (mov_<load>r): Likewise.
24136         (mov_<store>l): Likewise.
24137         (mov_<store>r): Likewise.
24138         (*branch_equality<mode>_inverted): Add microMIPS support.
24139         (*branch_equality<mode>): Likewise.
24140         (*jump_absolute): Likewise.
24141         (indirect_jump_<mode>): Likewise.
24142         (tablejump_<mode>): Likewise.
24143         (<optab>_internal): Likewise.
24144         (sibcall_internal): Likewise.
24145         (sibcall_value_internal): Likewise.
24146         (prefetch): Use constraint ZD.
24147         * config/mips/mips.opt (minterlink-compressed): New option.
24148         (minterlink-mips16): Now an alias for minterlink-compressed.
24149         (mmicromips): New option.
24150         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24151         (compare_and_swap_12): Likewise.
24152         (sync_add<mode>): Likewise.
24153         (sync_<optab>_12): Likewise.
24154         (sync_old_<optab>_12): Likewise.
24155         (sync_new_<optab>_12): Likewise.
24156         (sync_nand_12): Likewise.
24157         (sync_old_nand_12): Likewise.
24158         (sync_new_nand_12): Likewise.
24159         (sync_sub<mode>): Likewise.
24160         (sync_old_add<mode>): Likewise.
24161         (sync_old_sub<mode>): Likewise.
24162         (sync_new_add<mode>): Likewise.
24163         (sync_new_sub<mode>): Likewise.
24164         (sync_<optab><mode>): Likewise.
24165         (sync_old_<optab><mode>): Likewise.
24166         (sync_new_<optab><mode>): Likewise.
24167         (sync_nand<mode>): Likewise.
24168         (sync_old_nand<mode>): Likewise.
24169         (sync_new_nand<mode>): Likewise.
24170         (sync_lock_test_and_set<mode>): Likewise.
24171         (test_and_set_12): Likewise.
24172         (atomic_compare_and_swap<mode>): Likewise.
24173         (atomic_exchange<mode>_llsc): Likewise.
24174         (atomic_fetch_add<mode>_llsc): Likewise.
24175         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24176         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24177         (umips_save_restore_pattern_p): Likewise.
24178         (umips_load_store_pair_p): Likewise.
24179         (umips_output_load_store_pair): Likewise.
24180         (umips_movep_target_p): Likewise.
24181         (umips_12bit_offset_address_p): Likewise.
24182         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24183         (mips_base_mips16): Rename this...
24184         (mips_base_compression_flags): ...to this. Update all uses.
24185         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24186         (mips_mips16_decl_p): Delete.
24187         (mips_nomips16_decl_p): Delete.
24188         (mips_get_compress_on_flags): New function.
24189         (mips_get_compress_off_flags): New function.
24190         (mips_get_compress_mode): New function.
24191         (mips_get_compress_on_name): New function.
24192         (mips_get_compress_off_name): New function.
24193         (mips_insert_attributes): Support multiple compression types.
24194         (mips_merge_decl_attributes): Likewise.
24195         (umips_12bit_offset_address_p): New function.
24196         (mips_start_function_definition): Emit .set micromips directive.
24197         (mips_call_may_need_jalx_p): New function.
24198         (mips_function_ok_for_sibcall): Add microMIPS support.
24199         (mips_print_operand_punctuation): Support short delay slots and
24200         compact jumps.
24201         (umips_swm_mask, umips_swm_encoding): New.
24202         (umips_build_save_restore): New function.
24203         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24204         (was_mips16_p): Remove.
24205         (old_compression_mode): New.
24206         (mips_set_compression_mode): New function.
24207         (mips_set_current_function): Add microMIPS support.
24208         (mips_option_override): Likewise.
24209         (umips_save_restore_pattern_p): New function.
24210         (umips_output_save_restore): New function.
24211         (umips_load_store_pair_p_1): New function.
24212         (umips_load_store_pair_p): New function.
24213         (umips_output_load_store_pair_1): New function.
24214         (umips_output_load_store_pair): New function.
24215         (umips_movep_target_p) New function.
24216         (mips_prepare_pch_save): Add microMIPS support.
24217         * config/mips/mips.h (TARGET_COMPRESSION): New.
24218         (TARGET_CPU_CPP_BUILTINS): Update macro
24219         to use new compression flags and to support microMIPS.
24220         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24221         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24222         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24223         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24224         (ASM_SPEC): Support mmicromips and mno-micromips.
24225         (M16STORE_REG_P): New macro.
24226         (MIPS_CALL): Support TARGET_MICROMIPS.
24227         (MICROMIPS_J): New macro.
24228         (mips_base_mips16): Rename this...
24229         (mips_base_compression_flags): ...to this.
24230         (UMIPS_12BIT_OFFSET_P): New macro.
24231         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24232         (MULTILIB_DIRNAMES): Likewise.
24234 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24236         PR rtl-optimization/56131
24237         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24238         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24239         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24241 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24243         * doc/invoke.texi (-fsanitize=): Move from optimization
24244         to debugging options.
24246 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24248         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24250 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24251             Alexander Monakov  <amonakov@ispras.ru>
24253         PR middle-end/56077
24254         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24255         flush pending lists also on non-jumps.  Adjust comment.
24257 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24258             Maciej W. Rozycki  <macro@codesourcery.com>
24259             Tom de Vries  <tom@codesourcery.com>
24260             Nathan Sidwell  <nathan@codesourcery.com>
24261             Iain Sandoe  <iain@codesourcery.com>
24262             Nathan Froyd  <froydnj@codesourcery.com>
24263             Chao-ying Fu  <fu@mips.com>
24265         * doc/extend.texi: (micromips, nomicromips, nocompression):
24266         Document new function attributes.
24267         * doc/invoke.texi (minterlink-compressed, mmicromips,
24268         m14k, m14ke, m14kec): Document new options.
24269         (minterlink-mips16): Update documentation.
24270         * doc/md.texi (ZC, ZD): Document new constraints.
24271         * configure.ac (gcc_cv_as_micromips): Check if linker
24272         supports the .set micromips directive.
24273         * configure: Regenerate.
24274         * config.in: Regenerate.
24275         * config/mips/mips-tables.opt: Regenerate.
24276         * config/mips/micromips.md: New file.
24277         * constraints.md (ZC, AD): New constraints.
24278         * config/mips/predicates.md (movep_src_register): New predicate.
24279         (movep_src_operand): New predicate.
24280         (non_volatile_mem_operand): New predicate.
24281         * config/mips/mips.md (multimem): New type.
24282         (length): Differentiate between 17-bit and 18-bit branch offsets.
24283         (MOVEP1, MOVEP2): New mode iterator.
24284         (mov_<load>l): Use ZC constraint.
24285         (mov_<load>r): Likewise.
24286         (mov_<store>l): Likewise.
24287         (mov_<store>r): Likewise.
24288         (*branch_equality<mode>_inverted): Add microMIPS support.
24289         (*branch_equality<mode>): Likewise.
24290         (*jump_absolute): Likewise.
24291         (indirect_jump_<mode>): Likewise.
24292         (tablejump_<mode>): Likewise.
24293         (<optab>_internal): Likewise.
24294         (sibcall_internal): Likewise.
24295         (sibcall_value_internal): Likewise.
24296         (prefetch): Use constraint ZD.
24297         * config/mips/mips.opt (minterlink-compressed): New option.
24298         (minterlink-mips16): Now an alias for minterlink-compressed.
24299         (mmicromips): New option.
24300         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24301         (compare_and_swap_12): Likewise.
24302         (sync_add<mode>): Likewise.
24303         (sync_<optab>_12): Likewise.
24304         (sync_old_<optab>_12): Likewise.
24305         (sync_new_<optab>_12): Likewise.
24306         (sync_nand_12): Likewise.
24307         (sync_old_nand_12): Likewise.
24308         (sync_new_nand_12): Likewise.
24309         (sync_sub<mode>): Likewise.
24310         (sync_old_add<mode>): Likewise.
24311         (sync_old_sub<mode>): Likewise.
24312         (sync_new_add<mode>): Likewise.
24313         (sync_new_sub<mode>): Likewise.
24314         (sync_<optab><mode>): Likewise.
24315         (sync_old_<optab><mode>): Likewise.
24316         (sync_new_<optab><mode>): Likewise.
24317         (sync_nand<mode>): Likewise.
24318         (sync_old_nand<mode>): Likewise.
24319         (sync_new_nand<mode>): Likewise.
24320         (sync_lock_test_and_set<mode>): Likewise.
24321         (test_and_set_12): Likewise.
24322         (atomic_compare_and_swap<mode>): Likewise.
24323         (atomic_exchange<mode>_llsc): Likewise.
24324         (atomic_fetch_add<mode>_llsc): Likewise.
24325         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24326         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24327         (umips_save_restore_pattern_p): Likewise.
24328         (umips_load_store_pair_p): Likewise.
24329         (umips_output_load_store_pair): Likewise.
24330         (umips_movep_target_p): Likewise.
24331         (umips_12bit_offset_address_p): Likewise.
24332         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24333         (mips_base_mips16): Rename this...
24334         (mips_base_compression_flags): ...to this. Update all uses.
24335         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24336         (mips_mips16_decl_p): Delete.
24337         (mips_nomips16_decl_p): Delete.
24338         (mips_get_compress_on_flags): New function.
24339         (mips_get_compress_off_flags): New function.
24340         (mips_get_compress_mode): New function.
24341         (mips_get_compress_on_name): New function.
24342         (mips_get_compress_off_name): New function.
24343         (mips_insert_attributes): Support multiple compression types.
24344         (mips_merge_decl_attributes): Likewise.
24345         (umips_12bit_offset_address_p): New function.
24346         (mips_start_function_definition): Emit .set micromips directive.
24347         (mips_call_may_need_jalx_p): New function.
24348         (mips_function_ok_for_sibcall): Add microMIPS support.
24349         (mips_print_operand_punctuation): Support short delay slots and
24350         compact jumps.
24351         (umips_swm_mask, umips_swm_encoding): New.
24352         (umips_build_save_restore): New function.
24353         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24354         (was_mips16_p): Remove.
24355         (old_compression_mode): New.
24356         (mips_set_compression_mode): New function.
24357         (mips_set_current_function): Add microMIPS support.
24358         (mips_option_override): Likewise.
24359         (umips_save_restore_pattern_p): New function.
24360         (umips_output_save_restore): New function.
24361         (umips_load_store_pair_p_1): New function.
24362         (umips_load_store_pair_p): New function.
24363         (umips_output_load_store_pair_1): New function.
24364         (umips_output_load_store_pair): New function.
24365         (umips_movep_target_p) New function.
24366         (mips_prepare_pch_save): Add microMIPS support.
24367         * config/mips/mips.h (TARGET_COMPRESSION): New.
24368         (TARGET_CPU_CPP_BUILTINS): Update macro
24369         to use new compression flags and to support microMIPS.
24370         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24371         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24372         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24373         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24374         (ASM_SPEC): Support mmicromips and mno-micromips.
24375         (M16STORE_REG_P): New macro.
24376         (MIPS_CALL): Support TARGET_MICROMIPS.
24377         (MICROMIPS_J): New macro.
24378         (mips_base_mips16): Rename this...
24379         (mips_base_compression_flags): ...to this.
24380         (UMIPS_12BIT_OFFSET_P): New macro.
24381         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24382         (MULTILIB_DIRNAMES): Likewise.
24384 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24386         PR target/52555
24387         * target-globals.c (save_target_globals): For init_reg_sets and
24388         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24390 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24392         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24393         * config/aarch64/t-aarch64
24394         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24396 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24398         PR inline-asm/56148
24399         * lra-constraints.c (process_alt_operands): Reload operand
24400         conflicting with earlier clobber only if no more other conflicting
24401         operands.
24403 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24405         PR sanitizer/56393
24406         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24407         if not linking a shared library.
24409 2013-02-22  Seth LaForge  <sethml@google.com>
24411         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24413 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24415         * config/arm/arm.md (split for extendsidi): Update condition.
24416         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24417         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24418         (qhs_zextenddi_cstr): Likewise.
24420 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24422         PR middle-end/56420
24423         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24424         avoid signed wrapping.
24425         (expand_mult): Handle properly multiplication by
24426         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24427         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24428         in the compiler if coeff is HOST_WIDE_INT_MIN.
24429         (expand_divmod): Don't make ext_op1 static, change it's type to
24430         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24432         PR rtl-optimization/50339
24433         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24434         field.
24435         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24436         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24437         into splitting_ashiftrt field.
24438         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24439         ASHIFTRT.
24440         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24441         choices.
24443 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24445         PR middle-end/56108
24446         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24447         are sure to go irrevocable.
24449 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24451         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24452         scalars are valid operands.
24454 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24456         PR tree-optimization/56310
24457         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24458         only matching indices and non-negative final offsets.
24459         (intersect_aggregates_with_edge): Pass src_idx to
24460         agg_replacements_to_vector.  Pass src_idx insstead of index to
24461         intersect_with_agg_replacements.
24463 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24465         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24466         instead of hard-wired defaults.
24468 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24470         * doc/invoke.texi (MIPS Options): Update documentation of the
24471         floating-point multiply-accumulate instruction restrictions.
24473 2013-02-21  Kostya Serebryany  <kcc@google.com>
24475         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24476         asan_shadow_offset on x86_64 linux.
24478 2013-02-21  Richard Biener  <rguenther@suse.de>
24480         PR tree-optimization/56415
24481         Revert
24482         2013-02-11  Richard Biener  <rguenther@suse.de>
24484         PR tree-optimization/56273
24485         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24486         first VRP run.
24488 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24490         PR bootstrap/56258
24491         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24492         instead of @itemx.
24494         PR inline-asm/56405
24495         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24496         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24498 2013-02-20  Jan Hubicka  <jh@suse.cz>
24500         PR tree-optimization/56265
24501         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24502         when target is referenced for first time.
24504 2013-02-20  Richard Biener  <rguenther@suse.de>
24506         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24507         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24508         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24509         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24510         not return anything.
24511         (rename_ssa_copies): Do not remove unused locals.
24512         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24513         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24514         * passes.c (execute_function_todo): Do not schedule unused locals
24515         removal if cleanup_tree_cfg did something.
24516         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24517         about the number of removed locals.
24519 2013-02-20  Richard Biener  <rguenther@suse.de>
24521         PR tree-optimization/56398
24522         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24524 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24526         PR tree-optimization/55334
24527         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24528         restricted pointers to arrays.
24530 2013-02-20  Richard Biener  <rguenther@suse.de>
24531             Jakub Jelinek  <jakub@redhat.com>
24533         PR tree-optimization/56396
24534         * tree-ssa-ccp.c (n_const_val): New static variable.
24535         (get_value): Return NULL for SSA names we don't have a lattice
24536         entry for.
24537         (ccp_initialize): Initialize n_const_val.
24538         * tree-ssa-copy.c (n_copy_of): New static variable.
24539         (init_copy_prop): Initialize n_copy_of.
24540         (get_value): Return NULL_TREE for SSA names we don't have a
24541         lattice entry for.
24543 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24545         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24547 2013-02-20  Richard Biener  <rguenther@suse.de>
24549         * genpreds.c (write_lookup_constraint): Do not compare first
24550         letter of the constraint again.
24552 2013-02-20  Richard Biener  <rguenther@suse.de>
24554         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24555         and ceil_log2.
24556         (get_use_iv_cost): Terminate hashtable walk when coming across
24557         an empty entry.
24559 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24561         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24562         reassociation for avx2 targets.
24564 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24566         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24567         Add version check for v8.10.a to enable microblaze_has_clz
24568         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24569         version and TARGET_PATTERN_COMPARE check
24570         * config/microblaze/microblaze.md: New clzsi2 instruction
24572 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24574         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24575         function before branching.
24577 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24579         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24580         DUMP_INSN_RTX_UID.
24581         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24583 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24585         PR middle-end/55889
24586         * sel-sched.c: Include ira.h.
24587         (implicit_clobber_conflict_p): New function.
24588         (moveup_expr): Use it.
24589         * Makefile.in (sel-sched.o): Depend on ira.h.
24591 2013-02-19  Richard Biener  <rguenther@suse.de>
24593         PR tree-optimization/56384
24594         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24595         (vn_hash_type): Split out from ...
24596         (vn_hash_constant_with_type): ... here.
24597         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24598         (vn_phi_eq): Compare types from vn_phi_s structure.
24599         (vn_phi_lookup): Populate vn_phi_s type.
24600         (vn_phi_insert): Likewise.
24602 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24604         PR tree-optimization/56350
24605         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24606         if haven't found reduction or nested cycle operand, rather than
24607         asserting we must find it.
24609         PR tree-optimization/56381
24610         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24611         to fold_build3.
24613 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24614             Jakub Jelinek  <jakub@redhat.com>
24616         PR target/52555
24617         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24618         (swap_optab_enable): Same.
24619         (init_all_optabs): Use argument instead of global.
24620         * tree.h (struct tree_optimization_option): New field target_optabs.
24621         * expr.h (init_all_optabs): Add argument to prototype.
24622         (TREE_OPTIMIZATION_OPTABS): New.
24623         (save_optabs_if_changed): Protoize.
24624         * optabs.h: Declare this_fn_optabs.
24625         * optabs.c (save_optabs_if_changed): New.
24626         Declare this_fn_optabs.
24627         (init_optabs): Add argument to init_all_optabs() call.
24628         * function.c (invoke_set_current_function_hook): Handle per
24629         function optabs.
24630         * function.h (struct function): New field optabs.
24631         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24632         optimization_current_node has changed.
24633         * target-globals.h (save_target_globals_default_opts): Protoize.
24634         * target-globals.c (save_target_globals_default_opts): New.
24636 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24638         PR target/56347
24639         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24640         registers %fr12 and %fr12R as call used.
24642         PR target/56214
24643         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24644         and HImode, require all displacements to be an integer multiple of
24645         their mode size.
24646         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24647         only allow QImode and HImode when reload is in progress and strict is
24648         true.  Likewise for symbolic addresses.  Use base14_operand to check
24649         displacements in REG+BASE addresses.
24651 2013-02-18  Richard Biener  <rguenther@suse.de>
24653         PR tree-optimization/56366
24654         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24655         handle sign-conversion of outer-loop initial induction value.
24657 2013-02-18  Richard Biener  <rguenther@suse.de>
24659         PR middle-end/56349
24660         * cfghooks.c (merge_blocks): If we merge a latch into another
24661         block adjust references to it.
24662         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24663         (verify_loop_structure): Verify that a recorded latch is in fact
24664         a latch.
24666 2013-02-18  Richard Biener  <rguenther@suse.de>
24668         PR tree-optimization/56321
24669         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24670         order SSA name release and virtual operand unlinking.
24672 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24674         * config/microblaze/microblaze.md (save_stack_block): Define.
24675         (restore_stack_block): Likewise.
24677 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24679         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24680         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24681         * config/microblaze/microblaze.c (microblaze_option_override):
24682         Bail out early for PIC modes when target does not support PIC.
24684 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24686         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24687         Replace with a microblaze version.
24688         (microblaze_trampoline_init): Adapt for microblaze.
24689         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24690         microblaze.
24692 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24693             Dodji Seketeli  <dodji@redhat.com>
24695         PR asan/56330
24696         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24697         (instrument_mem_region_access): Do not forget to always put
24698         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24699         0) statement, even for cases where either 'base' or 'base + len'
24700         are not instrumented -- because they have been previously
24701         instrumented.  Simplify the logic by putting all the statements
24702         instrument 'base + len' inside a sequence, and then insert that
24703         sequence right before the current insertion point.  Then, to
24704         instrument 'base + len', just get an iterator on that statement.
24705         And do not forget to update the pointer to iterator the function
24706         received as argument.
24708 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24710         PR rtl-optimization/56348
24711         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24713 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24715         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24716         (clean_graph_dump_file): Pass base to start_graph_dump.
24718 2013-02-14  Richard Henderson  <rth@redhat.com>
24720         PR target/55941
24721         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24723 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24725         * collect2-aix.h: Define F_LOADONLY.
24727 2013-02-14  Richard Biener  <rguenther@suse.de>
24729         PR lto/50494
24730         * varasm.c (output_constant_def_1): Get the decl representing
24731         the constant as argument.
24732         (output_constant_def): Wrap output_constant_def_1.
24733         (make_decl_rtl): Use output_constant_def_1 with the decl
24734         representing the constant.
24735         (build_constant_desc): Optionally re-use a decl already
24736         representing the constant.
24737         (tree_output_constant_def): Adjust.
24739 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24741         Fix an asan crash
24742         * asan.c (instrument_builtin_call):  Really put the length of the
24743         second source argument into src1_len.
24745 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24747         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24748         argument.  If it is false, don't create edge from then_bb to
24749         fallthru_bb.
24750         (insert_if_then_before_iter): Pass true to it.
24751         (build_check_stmt): Pass false to it.
24752         (transform_statements): Flush hash table only on extended basic
24753         block boundaries, rather than at the beginning of every bb.
24754         Don't flush hash table on nonfreeing_call_p calls.
24755         * tree-flow.h (nonfreeing_call_p): New prototype.
24756         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24758 2013-02-13  David S. Miller  <davem@davemloft.net>
24760         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24762 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24764         PR target/56184
24765         * ira.c (max_regno_before_ira): Move from ...
24766         (ira): ... here.
24767         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24768         vec_safe_length.
24770 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24772         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24774 2013-02-13  Richard Biener  <rguenther@suse.de>
24776         PR lto/56295
24777         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24778         globals in MEM_REFs.
24780 2013-02-13  Richard Biener  <rguenther@suse.de>
24782         * loop-init.c (loop_optimizer_init): Clear loop state when
24783         re-initializing preserved loops.
24784         * loop-unswitch.c (unswitch_single_loop): Return whether
24785         we unswitched the loop.  Do not verify loop state here.
24786         (unswitch_loops): When we unswitched a loop discover new loops.
24788 2013-02-13  Kostya Serebryany  <kcc@google.com>
24790         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24791         on x86_64 linux.
24792         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24794 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24796         Avoid instrumenting duplicated memory access in the same basic block
24797         * Makefile.in (asan.o): Add new dependency on hash-table.h
24798         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24799         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24800         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24801         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24802         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24803         (get_mem_ref_of_assignment): New functions.
24804         (get_mem_refs_of_builtin_call): Extract from
24805         instrument_builtin_call and tweak a little bit to make it fit with
24806         the new signature.
24807         (instrument_builtin_call): Use the new
24808         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24809         of is_gimple_builtin_call.
24810         (instrument_derefs, instrument_mem_region_access): Insert the
24811         instrumented memory reference into the hash table.
24812         (maybe_instrument_assignment): Renamed instrument_assignment into
24813         this, and change it to advance the iterator when instrumentation
24814         actually happened and return true in that case.  This makes it
24815         homogeneous with maybe_instrument_assignment, and thus give a
24816         chance to callers to be more 'regular'.
24817         (transform_statements): Clear the memory reference hash table
24818         whenever we enter a new BB, when we cross a function call, or when
24819         we are done transforming statements.  Use
24820         maybe_instrument_assignment instead of instrumentation.  No more
24821         need to special case maybe_instrument_assignment and advance the
24822         iterator after calling it; it's now handled just like
24823         maybe_instrument_call.  Update comment.
24825 2013-02-13  Richard Biener  <rguenther@suse.de>
24827         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24828         Fix loop discovery code.
24830 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24832         PR inline-asm/56148
24833         * lra-constraints.c (process_alt_operands): Match early clobber
24834         operand with itself.  Check conflicts with earlyclobber only if
24835         the operand is not reloaded.  Prefer to reload conflicting operand
24836         if earlyclobber and matching operands are the same.
24838 2013-02-12  Richard Biener  <rguenther@suse.de>
24840         PR lto/56297
24841         * lto-streamer-out.c (write_symbol): Do not output symbols
24842         for hard register variables.
24844 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24846         PR target/54222
24847         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24848         (umulsidi3_insn, mulsidi3_insn): New insns.
24850 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24852         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24853         (struct tune_params): Add vec_costs field.
24854         * config/arm/arm.c (arm_builtin_vectorization_cost)
24855         (arm_add_stmt_cost): New functions.
24856         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24857         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24858         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24859         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24860         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24861         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24862         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24864 2013-02-12  Richard Biener  <rguenther@suse.de>
24866         PR lto/56295
24867         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24868         decls again if possible.
24870 2013-02-12  Richard Biener  <rguenther@suse.de>
24872         PR middle-end/56288
24873         * tree-ssa.c (verify_ssa_name): Fix check, move
24874         SSA_NAME_IN_FREE_LIST check up.
24876 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24877             Steven Bosscher   <steven@gcc.gnu.org>
24879         PR rtl-optimization/56151
24880         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24881         equal to op0 or op1, and last_insn pattern is CODE operation
24882         with MEM dest and one of the operands matches that MEM.
24884 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24886         * doc/extend.texi: Document Function Multiversioning and "default"
24887         parameter string to target attribute.
24888         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24889         target attribute parameter is "default".
24890         (ix86_compare_version_priority): Remove checks for target attribute.
24891         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24892         Remove check for target attribute equal to NULL. Add assert.
24893         (ix86_generate_version_dispatcher_body): Change error to sorry.
24895 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24896             Jack Howarth  <howarth@bromo.med.uc.edu>
24897             Patrick Marlier  <patrick.marlier@gmail.com>
24899         PR libitm/55693
24900         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24901         define ENDFILE_SPEC as TM_DESTRUCTOR.
24902         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24904 2013-02-11  Alexander Potapenko  <glider@google.com>
24905             Jack Howarth  <howarth@bromo.med.uc.edu>
24906             Jakub Jelinek  <jakub@redhat.com>
24908         PR sanitizer/55617
24909         * config/darwin.c (cdtor_record): Rename ctor_record.
24910         (sort_cdtor_records): Rename sort_ctor_records.
24911         (finalize_dtors): New routine to sort destructors by
24912         priority before use in assemble_integer.
24913         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24915 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24917         PR rtl-optimization/56275
24918         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24919         offset is non-negative and less than cmode size before
24920         calling simplify_subreg.
24922 2013-02-11  Richard Biener  <rguenther@suse.de>
24924         PR tree-optimization/56264
24925         * cfgloop.h (fix_loop_structure): Adjust prototype.
24926         * loop-init.c (fix_loop_structure): Return the number of
24927         newly discovered loops.
24928         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24929         are discovered, do a full loop-closed SSA rewrite.
24931 2013-02-11  Richard Biener  <rguenther@suse.de>
24933         PR tree-optimization/56273
24934         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24935         first VRP run.
24936         (check_array_ref): Fix missing newline in dumps.
24937         (search_for_addr_array): Likewise.
24939 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24941         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24943 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24945         PR target/56256
24946         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24948 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24950         PR rtl-optimization/56246
24951         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24952         reload pseudo.
24953         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24954         constraints are satisfied.
24956 2013-02-08  Jeff Law  <law@redhat.com>
24958         PR debug/53948
24959         * emit-rtl.c (reg_is_parm_p): New function.
24960         * regs.h (reg_is_parm_p): New prototype.
24961         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24962         callee-clobbered registers.
24964 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24966         PR target/56043
24967         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24968         If there is no implicit builtin declaration, just return NULL.
24970 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24972         * config/i386/sse.md (FMAMODEM): New mode iterator.
24973         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24974         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24976 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24978         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24979         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24980         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24982 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24984         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24985         (microblaze*-*-elf): Likewise.
24986         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24987         LINK_SPEC.
24988         * config/microblaze/microblaze-c.c: Add builtin defines for
24989         _LITTLE_ENDIAN and _BIG_ENDIAN.
24990         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24991         add to TARGET_DEFAULT flags.
24992         Expand ASM_SPEC and LINK_SPEC.
24993         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24994         * config/microblaze/microblaze.md: Update extendsidi2 and
24995         movdi_internal instructions to use low-order / high-order reg
24996         print_operands.
24997         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24998         options and inversemask / mask of LITTLE_ENDIAN.
24999         * config/microblaze/t-microblaze: Expand multilib options to
25000         include mlittle-endian (le) and update exceptions patterns.
25002 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25004         PR rtl-optimization/56195
25005         * lra-constraints.c (get_reload_reg): Don't reuse regs
25006         if they have smaller mode than requested, if they have
25007         wider mode than requested, try to return a SUBREG.
25009         PR tree-optimization/56250
25010         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25011         if type is unsigned and code isn't MULT_EXPR.
25013 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25015         PR tree-optimization/56064
25016         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25017         bits according to mode.
25018         * fixed-value.h (fixed_from_double_int)
25019         (const_fixed_from_double_int): Adjust comments.
25021 2013-02-08  Richard Biener  <rguenther@suse.de>
25023         PR lto/56231
25024         * lto-streamer.h (struct data_in): Remove current_file, current_line
25025         and current_col members.
25026         * lto-streamer-out.c (lto_output_location): Stream changed bits
25027         en-block for efficiency.
25028         * lto-streamer-in.c (clear_line_info): Remove.
25029         (lto_input_location): Cache current file, line and column
25030         globally via local statics.  Read changed bits en-block.
25031         (input_function): Do not call clear_line_info.
25032         (lto_read_body): Likewise.
25033         (lto_input_toplevel_asms): Likewise.
25035 2013-02-08  Michael Matz  <matz@suse.de>
25037         PR tree-optimization/52448
25038         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25039         (nt_call_phase): New static.
25040         (add_or_mark_expr): Only mark accesses with newer phase than any
25041         call seen.
25042         (nonfreeing_call_p): New.
25043         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25044         (nt_fini_block): Keep blocks marked as visited.
25045         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25047 2013-02-08  Richard Biener  <rguenther@suse.de>
25049         * ira.c (ira): Free broken dominator information.
25051 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25053         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25055 2013-02-08  Marek Polacek  <polacek@redhat.com>
25057         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25059 2013-02-08  Richard Biener  <rguenther@suse.de>
25061         PR middle-end/56181
25062         * cfgloop.h (flow_loops_find): Adjust.
25063         (bb_loop_header_p): Declare.
25064         * cfgloop.c (bb_loop_header_p): New function split out from ...
25065         (flow_loops_find): ... here.  Adjust function signature,
25066         support incremental loop structure update.
25067         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25068         * cfgloopmanip.c (fix_loop_structure): Move ...
25069         * loop-init.c (fix_loop_structure): ... here.
25070         (apply_loop_flags): Split out from ...
25071         (loop_optimizer_init): ... here.
25072         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25073         in incremental mode, only remove dead loops here.
25075 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25077         PR target/54222
25078         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25079         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25080         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25081         (*round<mode>3.libgcc): New insns for fixed-modes.
25082         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25083         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25084         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25085         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25086         implementations.  Define to __builtin_avr_absFX,
25087         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25088         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25089         __builtin_avr_countlsFX, respectively.
25090         * config/avr/avr-c.c (target.h): Include it.
25091         (enum avr_builtin_id): New enum.
25092         (avr_resolve_overloaded_builtin): New static function.
25093         (avr_register_target_pragmas): Use it to set
25094         targetm.resolve_overloaded_builtin.
25095         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25096         tree nodes used by DEF_BUILTIN.
25097         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25098         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25099         <AVR_BUILTIN_xxBITS>: Same.
25101 2013-02-08  Richard Biener  <rguenther@suse.de>
25103         * cfgloop.c (verify_loop_structure): Properly handle
25104         a loop exiting to another loop header.
25105         * ira-int.h (ira_loops): Remove.
25106         * ira.c (ira_loops): Remove.
25107         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25108         (do_reload): Use loop_optimizer_finalize.
25109         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25110         number_of_loops to access the loop tree.
25111         (more_one_region_p): Likewise.
25112         (finish_loop_tree_nodes): Likewise.
25113         (rebuild_regno_allocno_maps): Likewise.
25114         (mark_loops_for_removal): Likewise.
25115         (mark_all_loops_for_removal): Likewise.
25116         (remove_unnecessary_regions): Likewise.
25117         (ira_build): Likewise.
25118         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25120 2013-02-08  Richard Biener  <rguenther@suse.de>
25122         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25123         * ipa-pure-const.c (analyze_function): Avoid calling
25124         mark_irreducible_loops twice.
25125         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25127 2013-02-07  David S. Miller  <davem@davemloft.net>
25129         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25130         on 'reg'.
25131         * var-tracking.c (vt_add_function_parameter): Test the presence of
25132         HAVE_window_save properly and do not remap argument registers when
25133         we have a leaf function.
25135 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25137         PR bootstrap/56227
25138         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25139         instead of "ll".
25140         * config/i386/i386.c (ix86_print_operand): Ditto.
25142 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25144         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25146 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25148         PR rtl-optimization/56225
25149         * lra-constraints.c (process_alt_operands): Check that reload hard
25150         reg can hold value for strict_low_part.
25152 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25154         PR debug/56154
25155         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25156         dwarf2out_end_function.
25157         (in_first_function_p, maybe_at_text_label_p,
25158         first_loclabel_num_not_at_text_label): New variables.
25159         (dwarf2out_var_location): In the first function find out
25160         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25161         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25162         functions.
25164 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25166         PR rtl-optimization/56178
25167         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25168         SUBREG of a register.  Tidy up related block of code.
25169         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25170         note if the source is a register or a SUBREG of a register.
25172 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25174         PR target/56228
25175         * config/rs6000/rs6000.md (ptrm): New mode attr.
25176         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25177         call_value_indirect_aix<pttrsize>,
25178         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25179         m in constraints.
25181 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25183         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25184         if -bnortl. Convert to strcmp and strncmp.
25186 2013-02-07  Alan Modra  <amodra@gmail.com>
25188         PR target/54009
25189         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25190         addresses won't wrap when offsetting.
25191         (rs6000_secondary_reload): Provide secondary reloads needed for
25192         wrapping LO_SUM addresses.
25194 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25196         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25197         MACH, just __MACH__.
25199 2013-02-06  Richard Biener  <rguenther@suse.de>
25201         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25202         instead of calling fix_loop_structure.
25204 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25206         PR middle-end/56217
25207         * omp-low.c (use_pointer_for_field): Return false if
25208         lower_send_shared_vars doesn't generate any copy-out code.
25210 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25212         PR rtl-optimization/56131
25213         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25214         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25215         of the label is NULL.  Add comment.
25217 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25219         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25221         PR sanitizer/55374
25222         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25223         (STATIC_LIBTSAN_LIBS): Likewise.
25224         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25225         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25226         is defined, don't add anything else beyond that.
25227         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25228         (LINK_COMMAND_SPEC): Use them.
25230         PR tree-optimization/56205
25231         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25232         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25233         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25235 2013-02-05  Richard Biener  <rguenther@suse.de>
25237         PR tree-optimization/53342
25238         PR tree-optimization/53185
25239         * tree-vectorizer.h (vect_check_strided_load): Remove.
25240         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25241         not disallow peeling for vectorized strided loads.
25242         (vect_check_strided_load): Make static and simplify.
25243         (vect_analyze_data_refs): Adjust.
25244         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25245         correctly when vectorizing strided loads.
25247 2013-02-05  Richard Biener  <rguenther@suse.de>
25249         * doc/install.texi: Refer to ISL, not PPL.
25251 2013-02-05  Jan Hubicka  <jh@suse.cz>
25253         PR tree-optimization/55789
25254         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25256 2013-02-05  Jan Hubicka  <jh@suse.cz>
25258         PR tree-optimization/55789
25259         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25260         the dead call anyway.
25262 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25264         PR sanitizer/55374
25265         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25267 2013-02-04  Alexander Potapenko  <glider@google.com>
25268             Jack Howarth  <howarth@bromo.med.uc.edu>
25269             Jakub Jelinek  <jakub@redhat.com>
25271         PR sanitizer/55617
25272         * config/darwin.c (sort_ctor_records): Stabilized qsort
25273         on constructor priority by using original position.
25274         (finalize_ctors): New routine to sort constructors by
25275         priority before use in assemble_integer.
25276         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25278 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25280         PR libstdc++/54314
25281         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25282         about visibility on artificial decls.
25283         * config/sol2.c (solaris_assemble_visibility): Likewise.
25285 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25287         PR target/56186
25288         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25289         argument and improve checking of return-argument types for 16-byte
25290         modes.
25291         (ix86_function_value_1): Add additional valtype argument on call
25292         of function_value_64.
25293         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25294         handling infunction_value_64 function.
25296 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25298         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25300 2013-02-04  Richard Biener  <rguenther@suse.de>
25302         PR tree-optimization/56188
25303         * tree-ssa-structalias.c (label_visit): Consider case with
25304         initially non-empty points-to set.
25305         (perform_var_substitution): Dump node mapping and clean up.
25307 2013-02-04  Richard Guenther  <rguenther@suse.de>
25309         PR lto/56168
25310         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25311         node prevail as last resort.
25312         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25313         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25315 2013-02-04  Richard Biener  <rguenther@suse.de>
25317         PR tree-optimization/56113
25318         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25319         Merge into ...
25320         (equiv_class_lookup_or_add): ... this.
25321         (label_visit): Adjust and fix error in previous patch.
25322         (perform_var_substitution): Adjust.
25324 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25326         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25327         * config/sh/sh4-300.md: Likewise.
25328         * config/sh/sh4a.md: Likewise.
25329         * config/sh/constraints.md: Likewise.
25330         * config/sh/sh.md: Likewise.
25331         * config/sh/netbsd-elf.h: Likewise.
25332         * config/sh/predicates.md: Likewise.
25333         * config/sh/sh-protos.h: Likewise.
25334         * config/sh/ushmedia.h: Likewise.
25335         * config/sh/linux.h: Likewise.
25336         * config/sh/sh.c: Likewise.
25337         * config/sh/superh.h: Likewise.
25338         * config/sh/elf.h: Likewise.
25339         * config/sh/sh4.md: Likewise.
25340         * config/sh/sh.h: Likewise.
25342 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25344         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25345         constraint to match_test floating_point_store_memory_operand().
25346         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25347         (base14_operand): New.
25348         (floating_point_store_memory_operand): New.
25349         (integer_store_memory_operand): Revise to use base14_operand and
25350         reg_plus_base_memory_operand.
25351         (move_dest_operand): Allow symbolic_memory_operands.
25352         (symbolic_memory_operand): Check for LO_SOM.
25353         (symbolic_operand): Change default case to break.
25354         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25355         CONST_DOUBLE values to be reloaded by putting them into memory when
25356         the destination is a floating point register.
25357         (movdf): Remove code to handle CONST_DOUBLE.
25358         (movsf): Likewise.
25359         (reload_indf_r1): New.
25360         (reload_insf_r1): New.
25361         Consistently use "Q" and "T" constraints with integer and floating
25362         point move instructions, respectively.
25363         (movdi): Remove FAIL.
25364         Change predicate for source operand unamed DImode move from
25365         general_operand to move_src_operand.
25366         (umulsidi3): Change predicate for destination operand to
25367         register_operand.
25368         Likewise for similar unamed patterns.
25369         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25370         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25371         (hppa_legitimize_address): Simplify mask calculation.
25372         (pa_emit_move_sequence): Revised handling of secondary reloads from
25373         REG+D addresses for floating point loads and stores.  Directly handle
25374         loading CONST0_RTX (mode) to a floating point register.
25375         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25376         to floating point registers.  Don't restrict secondary reloads to
25377         floating point registers to integer modes.  Revise some comments and
25378         cleanup some code.
25379         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25380         (pa_legitimate_address_p): New.
25381         (pa_legitimize_reload_address): New.
25382         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25383         (STRICT_REG_OK_FOR_BASE_P): New.
25384         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25385         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25387 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25388             Andrew Dixie  <andrewd@gentrack.com>
25390         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25391         flag set.
25393 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25395         * expmed.c (extract_bit_field_1): Pass the full width of the
25396         structure to get_best_reg_extraction_insn.
25398 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25400         PR target/54601
25401         * configure.ac (use_cxa_atexit): Add AIX.
25402         * configure: Regenerate.
25404         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25406 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25408         PR debug/54793
25409         * final.c (need_profile_function): New variable.
25410         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25411         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25412         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25413         notes, targetm.asm_out.function_prologue doesn't emit anything,
25414         HAVE_prologue and profiler should be emitted before prologue,
25415         set need_profile_function instead of emitting it.
25416         (final_scan_insn): If need_profile_function, emit
25417         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25418         NOTE_INSN_FUNCTION_BEG note.
25420 2013-02-01  Richard Henderson  <rth@redhat.com>
25422         * config/rs6000/rs6000.md (smulditi3): New.
25423         (umulditi3): New.
25425         * config/alpha/alpha.md (umulditi3): New.
25427 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25429         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25430         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25432 2013-02-01  Richard Biener  <rguenther@suse.de>
25434         PR tree-optimization/56113
25435         * tree-ssa-structalias.c (label_visit): Reduce work for
25436         single-predecessor nodes.
25438 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25440         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25441         range isn't testing for zero.
25443 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25445         PR middle-end/56113
25446         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25448 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25449             Nick Clifton  <nickc@redhat.com>
25451         * config/v850/constraints.md (Q): Define as a memory constraint.
25452         * config/v850/predicates.md (label_ref_operand): New predicate.
25453         (e3v5_shift_operand): New predicate.
25454         (ior_operator): New predicate.
25455         * config/v850/t-v850: Add e3v5 multilib.
25456         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25457         (v850_gen_movdi): Prototype.
25458         * config/v850/v850.c: Add support for e3v5 architecture.
25459         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25460         TARGET_V850E_UP.
25461         (construct_save_jarl): Add e3v5 long JARL support.
25462         (v850_adjust_insn_length): New function.  Adjust length of call
25463         insns when using e3v5 instructions.
25464         (v850_gen_movdi): New function: Generate instructions to move a
25465         DImode value.
25466         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25467         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25468         (TARGET_USE_FPU): Enable for e3v5.
25469         (CONST_OK_FOR_W): New macro.
25470         (ADJUST_INSN_LENGTH): Define.
25471         * config/v850/v850.md (UNSPEC_LOOP): Define.
25472         (attr cpu): Add v850e3v5.
25473         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25474         (movdi): New pattern.
25475         (movdi_internal): New pattern.
25476         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25477         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25478         (cstoresf4): Likewise.
25479         (cstoredf4): Likewise.
25480         (insv): New pattern.
25481         (rotlso3_a): New pattern.
25482         (rotlsi3_b): New pattern
25483         (rotlsi3_v850e3v5): New pattern.
25484         (doloop_begin): New pattern.
25485         (fix_loop_counter): New pattern.
25486         (doloop_end): New pattern.
25487         (branch_normal): Add e3v5 long branch support.
25488         (branch_invert): Likewise.
25489         (branch_z_normal): Likewise.
25490         (branch_z_invert): Likewise.
25491         (branch_nz_normal): Likewise.
25492         (branch_nz_invert): Likewise.
25493         (call_internal_short): Add e3v5 register-indirect JARL support.
25494         (call_internal_long): Likewise.
25495         (call_value_internal_short): Likewise.
25496         (call_value_internal_long): Likewise.
25497         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25498         (mloop): New option.
25499         * config.gcc: Add support for configuring v840e3v5 target.
25500         * doc/invoke.texi: Document new v850 specific command line options.
25502 2013-01-31  Paul Koning  <ni1d@arrl.net>
25504         PR debug/55059
25505         PR debug/54508
25506         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25507         children if parent is a class.
25508         (prune_unused_types_prune): Don't add DW_AT_declaration.
25510 2013-01-31  Richard Biener  <rguenther@suse.de>
25512         PR tree-optimization/56157
25513         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25514         match up operand with SLP child.
25516 2013-01-31  Jason Merrill  <jason@redhat.com>
25518         PR debug/54410
25519         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25520         parameters the first time.
25521         (gen_scheduled_generic_parms_dies): Check completeness here.
25523 2013-01-31  Richard Biener  <rguenther@suse.de>
25525         PR middle-end/53073
25526         * common.opt (faggressive-loop-optimizations): New flag,
25527         enabled by default.
25528         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25529         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25530         infer_loop_bounds_from_undefined by it.
25532 2013-01-31  Richard Biener  <rguenther@suse.de>
25534         PR tree-optimization/56150
25535         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25536         visit virtual operands.
25537         (find_uses_to_rename_bb): Likewise.
25539 2013-01-31  Richard Biener  <rguenther@suse.de>
25541         PR tree-optimization/56150
25542         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25543         mixed store non-store stmts.
25545 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25547         PR sanitizer/55374
25548         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25549         LIBASAN_EARLY_SPEC is defined.
25550         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25551         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25552         before %o.
25553         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25555         PR c++/55742
25556         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25557         invalid args instead of ICEing on it.
25558         (ix86_valid_target_attribute_tree): Return error_mark_node if
25559         ix86_valid_target_attribute_inner_p failed.
25560         (ix86_valid_target_attribute_p): Return false only if
25561         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25562         target("default") attribute.
25563         (sorted_attr_string): Change argument from const char * to tree,
25564         merge in all target attribute arguments rather than just one.
25565         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25566         instead of free.  Avoid using strcat.
25567         (ix86_mangle_function_version_assembler_name): Mangle
25568         target("default") as if no target attribute is present.  Adjust
25569         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25570         instead of xmalloc and XDELETEVEC instead of free.
25571         (ix86_function_versions): Don't return true if one of the decls
25572         doesn't have target attribute.  If they don't and one of the decls
25573         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25574         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25575         (ix86_supports_function_versions): Remove.
25576         (make_name): Fix up formatting.
25577         (make_dispatcher_decl): Remove resolver_name and its initialization.
25578         Avoid leaking memory.
25579         (is_function_default_version): Return true if there is
25580         target("default") attribute rather than no target attribute at all.
25581         (make_resolver_func): Avoid leaking memory.
25582         (ix86_generate_version_dispatcher_body): Likewise.
25583         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25584         * target.def (supports_function_versions): Remove.
25585         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25586         * doc/tm.texi: Regenerated.
25588 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25590         PR rtl-optimization/56144
25591         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25592         for values with side effects.
25594 2013-01-30  Richard Biener  <rguenther@suse.de>
25596         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25597         (sparseset_pop): Likewise.
25598         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25599         to be able to use quick_push in the worker loop.
25601 2013-01-30  Marek Polacek  <polacek@redhat.com>
25603         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25605 2013-01-30  Richard Biener  <rguenther@suse.de>
25607         PR lto/56147
25608         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25610 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25612         PR tree-optimization/56064
25613         * fixed-value.c (fixed_from_double_int): New function.
25614         * fixed-value.h (fixed_from_double_int): New prototype.
25615         (const_fixed_from_double_int): New static inline function.
25616         * fold-const.c (native_interpret_fixed): New static function.
25617         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25618         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25619         (native_encode_fixed): New static function.
25620         (native_encode_expr) <FIXED_CST>: Use it.
25621         (native_interpret_int): Move double_int worker code to...
25622         * double-int.c (double_int::from_buffer): ...this new static method.
25623         * double-int.h (double_int::from_buffer): Prototype it.
25625 2013-01-30  Richard Biener  <rguenther@suse.de>
25627         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25628         New pointer-map and obstack.
25629         (init_alias_vars): Allocate pointer-map and obstack.
25630         (delete_points_to_sets): Free them.
25631         (find_what_var_points_to): Cache result.
25632         (find_what_p_points_to): Adjust for changed interface of
25633         find_what_var_points_to.
25634         (compute_points_to_sets): Likewise.
25635         (ipa_pta_execute): Likewise.
25637 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25639         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25640         * configure: Regenerate.
25641         * config.in: Regenerate.
25642         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25643         #nobits/#progbits if supported.
25645 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25647         PR target/56121
25648         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25649         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25650         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25652 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25654         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25655         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25657 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25659         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25660         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25662 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25664         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25665         declaration.
25666         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25667         * config/arm/cortex-a7.md: New bypasses using
25668         arm_mac_accumulator_is_result.
25670 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25672         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25673         (cortex_a7_neon_mla): Likewise.
25674         (cortex_a7_fpfmad): New reservation.
25675         (cortex_a7_fpmacs): Use ffmas and update required units.
25676         (cortex_a7_fpmuld): Update required units and latency.
25677         (cortex_a7_fpmacd): Likewise.
25678         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25679         (cortex_a7_neon). Likewise.
25680         (bypass) Update participating units.
25682 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25684         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25685         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25686         from fmac to ffma.
25687         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25688         (vfp_fmul): Use ffmad.
25689         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25690         (cortex_r4_fmacd): Use ffmad.
25691         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25692         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25693         (cortex_a9_fmacd): Use ffmad.
25694         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25695         (cortex_a8_vfp_macd): Use ffmad.
25696         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25697         (cortex_a5_fpmacd): Use ffmad.
25698         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25699         (cortex_a15_vfp_macd): Use ffmad.
25700         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25702 2013-01-29  Jason Merrill  <jason@redhat.com>
25704         PR libstdc++/54314
25705         * varasm.c (default_assemble_visibility): Don't warn about
25706         visibility on artificial decls.
25708 2013-01-29  Richard Biener  <rguenther@suse.de>
25710         PR tree-optimization/56113
25711         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25712         the bitmap leader.
25713         (label_visit): Free duplicate bitmaps and record the leader instead.
25714         (perform_var_substitution): Adjust.
25716 2013-01-29  Richard Biener  <rguenther@suse.de>
25718         PR tree-optimization/55270
25719         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25720         the CFG, schedule loops for fixup.
25722 2013-01-29  Nick Clifton  <nickc@redhat.com>
25724         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25725         SP_REG.
25727 2013-01-28  Leif Ekblad  <leif@rdos.net>
25729         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25730         * config/i386/i386.h (TARGET_RDOS): New macro.
25731         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25732         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25733         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25734         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25735         DEFAULT_LARGE_SECTION_THRESHOLD.
25736         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25737         * config/i386/rdos.h: New file.
25738         * config/i386/rdos64.h: New file.
25740 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25742         PR other/54814
25743         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25744         TEST_HARD_REG_BIT.
25746 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25748         PR rtl-optimization/56117
25749         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25750         call cselib_lookup_from_insn on the MEM before calling
25751         add_insn_mem_dependence.
25753 2013-01-28  Richard Biener  <rguenther@suse.de>
25755         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25756         to a stmt that didn't have one.
25757         (copy_phis_for_bb): Likewise for PHI arguments.
25758         (copy_debug_stmt): Likewise for debug stmts.
25760 2013-01-28  Richard Biener  <rguenther@suse.de>
25762         PR tree-optimization/56034
25763         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25764         (partition_builtin_p): Adjust.
25765         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25766         it is the last partition.
25767         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25768         up the vertex for the definition.
25769         (classify_partition): Classify whether a partition is a
25770         PKIND_REDUCTION, thus has uses outside of the loop.
25771         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25772         Merge all PKIND_REDUCTION partitions into the last partition.
25773         (tree_loop_distribution): Seed partitions from reductions as well.
25775 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25777         PR tree-optimization/56125
25778         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25779         pow(x,c) into sqrt(x) * powi(x, n/2) or
25780         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25781         optimizing for size.
25782         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25783         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25784         integer.
25786         PR tree-optimization/56094
25787         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25788         to UNKNOWN_LOCATION while gimplifying expr.
25790 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25792         PR target/56114
25793         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25794         operand 0 in movabs insn template for -masm=intel asm alternative.
25795         (*movabs<mode>_2): Ditto for operand 1.
25797 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25799         PR target/54663
25800         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25801         of microblaze-c.o
25803 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25805         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25806         tm_file.
25808 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25810         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25811         Undef to avoid warning.
25813 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25815         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25816         * configure: Regenerate.
25818 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25820         PR tree-optimization/56098
25821         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25822         for stmts with volatile ops.
25823         (cond_store_replacement): Don't optimize if assign has volatile ops.
25824         (cond_if_else_store_replacement_1): Don't optimize if either
25825         then_assign or else_assign have volatile ops.
25826         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25827         volatile ops.
25829 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25831         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25833 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25835         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25836         missing ':' in asm example.
25838 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25840         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25841         entries into lane and laneq entries.
25842         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25843         Remove AdvSIMD scalar modes.
25844         (aarch64_sq<r>dmulh_laneq<mode>): New.
25845         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25846         modes.
25847         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25848         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25849         * config/aarch64/iterators.md (VCOND): New.
25850         (VCONQ): New.
25852 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25854         PR target/54222
25855         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25856         Add NULL LIBNAME argument to existing definitions.
25857         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25858         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25859         * config/avr/avr.c (DEF_BUILTIN): Same.
25860         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25861         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25862         implementation is available (DECL_ASSEMBLER_NAME is set).
25863         (avr_fold_absfx): New static function.
25864         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25865         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25866         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25867         AVR_BUILTIN_ABSLLK.
25868         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25869         (abshk, absk, abslk, absllk): Provide as static inline functions.
25871 2013-01-25  Marek Polacek  <polacek@redhat.com>
25873         PR tree-optimization/56035
25874         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25876 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25878         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25879         (*movtf_internal_rex64): Add (!o,C) alternative
25880         (*movxf_internal_rex64): Ditto.
25881         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25883 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25885         * doc/invoke.texi: fix typo.
25886         * doc/objc.texi: fix typo.
25888 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25890         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25891         for the first two alternatives.
25893 2013-01-24  Diego Novillo  <dnovillo@google.com>
25895         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25896         (ggc-zone.o): Remove.
25897         * configure.ac: Remove option --with-gc.
25898         * configure: Re-generate.
25899         * doc/install.texi: Remove documentation for --with-gc.
25900         * gengtype.c (write_enum_defn): Remove.  Update all users.
25901         (write_Types_process_field): Remove generation of gt_e_* argument.
25902         (output_type_enum): Remove.  Update all users.
25903         (write_enum_defn): Remove.  Update all users.
25904         (enum alloc_zone): Remove.  Update all users.
25905         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25906         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25907         Update all callers.
25908         (struct ptr_data): Remove field TYPE.  Update all users.
25909         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25910         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25911         Update all users.
25912         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25913         (struct alloc_zone): Remove.
25914         (ggc_internal_alloc_zone_stat): Remove.
25915         (ggc_internal_cleared_alloc_zone_stat): Remove.
25916         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25917         (ggc_pch_count_object): Remove last argument.  Update all users.
25918         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25919         (struct alloc_zone): Remove.
25920         * ggc-zone.c: Remove.
25921         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25922         (struct alloc_zone): Remove.
25923         (ggc_alloc_typed_stat): Remove.
25924         (ggc_alloc_typed): Remove.
25925         (ggc_splay_alloc): Remove first argument.
25926         (rtl_zone): Remove.  Update all users.
25927         (tree_zone): Remove.  Update all users.
25928         (tree_id_zone): Remove.  Update all users.
25929         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25930         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25931         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25932         * tree-ssanames.c: Remove references to zone allocator in comments.
25934 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25936         * config/avr/avr.c (avr_out_fract): Make register numbers that
25937         might be outside of source operand signed.
25939 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25941         * config/i386/constraints.md (Yf): New constraint.
25942         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25943         of f constraint to conditionaly disable x87 register preferences.
25944         (*movdf_internal): Ditto.
25945         (*movsf_internal): Ditto.
25947 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25949         PR inline-asm/55934
25950         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25951         that have operands with impossible constraints.
25952         Add a FIXME for a speed-up opportunity.
25953         * lra-constraints.c (process_alt_operands): Verify that a class
25954         selected from constraints on asms is valid for the operand mode.
25955         (curr_insn_transform): Remove incorrect comment.
25957 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25959         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25960         TOC operand is a valid symbol ref in the constant pool.
25962 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25964         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25966 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25968         PR target/54222
25969         * config/avr/stdfix.h: New file.
25970         * t-avr (stdfix-gcc.h): New rule to build it.
25971         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25973 2013-01-23  Kostya Serebryany  <kcc@google.com>
25975         * config/darwin.h: remove dependency on
25976         CoreFoundation (asan on Mac OS).
25978 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25980         PR target/49069
25981         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25982         instead of cmpdi_operand for first comparison operand.
25983         Don't assert that comparison operands aren't both constants.
25985 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25987         * doc/install.texi (Downloading the Source): Update references to
25988         downloading separate components.
25990 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25992         * doc/extend.texi (__int128): Improve grammar.
25994 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25996         PR target/56028
25997         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25998         alternative to (o,r).
25999         (*movdi_internal_rex64): Remove (!o,n) alternative.
26000         (DImode immediate->memory splitter): Remove.
26001         (DImode immediate->memory peephole2): Remove.
26002         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26003         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26004         alternative to (!o,*r).
26005         (*movtf_internal_sse): New pattern.
26006         (*movxf_internal_rex64): New pattern.
26007         (*movxf_internal): Disable for TARGET_64BIT.
26008         (*movdf_internal_rex64): Remove (!o,F) alternative.
26010 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26012         PR middle-end/56074
26013         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26014         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26015         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26016         stmt locations where LOCATION_LOCUS of the stmt location is
26017         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26019         PR target/55686
26020         * config/i386/i386.md (UNSPEC_STOS): New.
26021         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26022         *strsetqi_1): Add UNSPEC_STOS.
26024 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26026         PR c++/56067
26027         * doc/invoke.texi: Remove left over -Wsynth example.
26029 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26031         PR tree-optimization/56051
26032         * fold-const.c (fold_binary_loc): Don't fold
26033         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26034         a narrowing conversion, or widening conversion from signed
26035         to unsigned.
26037 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26039         PR rtl-optimization/56023
26040         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26041         dependent on debug instruction.
26043 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26045         PR middle-end/56022
26046         * function.c (allocate_struct_function): Call
26047         invoke_set_current_function_hook earlier.
26049 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26051         * reload1.c (init_reload): Only initialize reload_obstack
26052         during the first call.
26054 2013-01-21  Marek Polacek  <polacek@redhat.com>
26056         * cfgloop.c (verify_loop_structure): Fix up grammar.
26058 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26060         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26061         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26063 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26065         PR target/56058
26066         * config/arm/marvell-pj4.md: Update copyright year.
26067         Fix up use of alu to alu_reg and simple_alu_imm.
26069 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26071         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26073 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26075         PR target/55433
26076         * lra-constraints.c (curr_insn_transform): Don't reuse original
26077         insn for secondary memory move when memory mode should be different.
26079 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26081         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26082         atomic_storedi_1): New patterns.
26084 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26086         btver2 pipeline descriptions.
26087         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26088         descriptions.
26089         * config/i386/i386.md (btver2_decode): New type attributes.
26090         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26091         type attributes.
26092         * config/i386/btver2.md: New file describing btver2 pipelines.
26094 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26096         PR tree-optimization/52631
26097         * tree-ssa-sccvn (visit_use): Before looking up the original
26098         statement, try looking up the simplified expression.
26100 2013-01-19  Anthony Green  <green@moxielogic.com>
26102         * config/moxie/moxie.c (moxie_expand_prologue): Set
26103         current_function_static_stack_size.
26105 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26107         PR tree-optimization/56029
26108         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26109         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26111 2013-01-18  Sharad Singhai  <singhai@google.com>
26113         PR tree-optimization/55995
26114         * dumpfile.c (dump_loc): Print location only if available.
26115         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26117 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26119         PR target/55433
26120         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26121         secondary memory move.
26122         (inherit_reload_reg): Use rclass instead of cl for
26123         check_secondary_memory_needed_p.
26125 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26127         PR middle-end/56015
26128         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26129         the case where writing real complex part of target modifies op1.
26131 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26133         * config/aarch64/aarch64-simd.md
26134         (aarch64_vcond_internal<mode>): Handle unordered cases.
26135         * config/aarch64/iterators.md (v_cmp_result): New.
26137 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26138             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26140         * config/arm/marvell-pj4.md: New file.
26141         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26142         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26143         (generic_vfp): Likewise.
26144         * config/arm/arm-cores.def: Add marvell-pj4.
26145         * config/arm/arm-tune.md: Regenerate.
26146         * config/arm/arm-tables.opt: Regenerate.
26147         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26148         * doc/invoke.texi: Document marvell-pj4.
26150 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26152         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26154 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26156         PR debug/54114
26157         PR debug/54402
26158         PR debug/49888
26159         * var-tracking.c (negative_power_of_two_p): New.
26160         (global_get_addr_cache, local_get_addr_cache): New.
26161         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26162         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26163         heading comment.
26164         (vt_stack_offset_p): Remove.
26165         (vt_canon_true_dep): Always canonicalize loc's address.
26166         (clobber_overlapping_mems): Make sure we have a MEM.
26167         (local_get_addr_clear_given_value): New.
26168         (val_reset): Clear local cached entries.
26169         (compute_bb_dataflow): Create and release the local cache.
26170         Disable duplicate MEMs clobbering.
26171         (emit_notes_in_bb): Clobber MEMs likewise.
26172         (vt_emit_notes): Create and release the local cache.
26173         (vt_initialize, vt_finalize): Create and release the global
26174         cache, respectively.
26175         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26177 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26179         PR libmudflap/53359
26180         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26181         not found in the symtab.
26183 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26185         PR debug/56006
26186         PR rtl-optimization/55547
26187         PR rtl-optimization/53827
26188         PR debug/53671
26189         PR debug/49888
26190         * alias.c (offset_overlap_p): New, factored out of...
26191         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26192         the conservative special case for symbolic constants.  Don't
26193         adjust zero sizes on alignment.
26195 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26197         PR rtl-optimization/52573
26198         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26199         REG_UNUSED for the same register.
26201 2013-01-17  Richard Biener  <rguenther@suse.de>
26202             Marek Polacek  <polacek@redhat.com>
26204         PR rtl-optimization/55833
26205         * loop-unswitch.c (unswitch_loops): Move loop verification...
26206         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26207         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26208         Set it to true when we're removing a loop from hierarchy tree in
26209         an irreducible region.
26210         (fix_bb_placements): Adjust caller.
26211         (fix_loop_placements): Likewise.
26213 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26215         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26216         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26217         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26218         Remove ID.  Adjust comments.
26219         * config/avr/avr-c.c (avr_builtin_name): Remove.
26220         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26221         * config/avr/avr.c (avr_tolower): New static function.
26222         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26223         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26224         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26225         default expansion.
26227 2013-01-17  Jan Hubicka  <jh@suse.cz>
26229         PR tree-optimization/55273
26230         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26232 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26234         PR target/55981
26235         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26236         store through atomic_store<mode>_1.
26237         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26239 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26241         PR tree-optimizations/55264
26242         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26243         for virtual methods.
26244         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26245         virtual methods before inlining is over.
26246         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26247         virtual functions.
26248         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26249         non-virtual.
26251 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26253         PR rtl-optimization/56005
26254         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26255         pending reads for prefetch.
26257 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26259         * config/aarch64/aarch64.md
26260         (*cstoresi_neg_uxtw): New pattern.
26261         (*cmovsi_insn_uxtw): New pattern.
26262         (*<optab>si3_uxtw): New pattern.
26263         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26264         (*<optab>si3_insn_uxtw): New pattern.
26265         (*bswapsi2_uxtw): New pattern.
26267 2013-01-16  Richard Biener  <rguenther@suse.de>
26269         * tree-inline.c (tree_function_versioning): Remove set but
26270         never used variable.
26272 2013-01-16  Richard Biener  <rguenther@suse.de>
26274         PR tree-optimization/55964
26275         * tree-flow.h (rename_variables_in_loop): Remove.
26276         (rename_variables_in_bb): Likewise.
26277         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26278         (copy_loop_before): Adjust and delete update-ssa status.
26279         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26280         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26281         (rename_variables_in_loop): Remove.
26282         (slpeel_update_phis_for_duplicate_loop): Likewise.
26283         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26284         use available cfg machinery instead of duplicating it.
26285         Update PHI nodes and perform poor-mans SSA update here.
26286         (slpeel_tree_peel_loop_to_edge): Adjust.
26288 2013-01-16  Richard Biener  <rguenther@suse.de>
26290         PR tree-optimization/54767
26291         PR tree-optimization/53465
26292         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26293         (vrp_visit_phi_node): For PHI arguments coming via backedges
26294         drop all symbolical range information.
26295         (execute_vrp): Compute backedges.
26297 2013-01-16  Richard Biener  <rguenther@suse.de>
26299         * doc/install.texi: Update CLooG and ISL requirements to
26300         0.18.0 and 0.11.1.
26302 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26304         PR target/55301
26305         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26306         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26307         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26309 2013-01-16  DJ Delorie  <dj@redhat.com>
26311         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26312         (UNSPECV_SP_SWITCH_E): New.
26313         (sp_switch_1): Change to an unspec.
26314         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26315         replace $r15.
26317 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26319         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26320         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26321         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26322         (expand_mem_thread_fence): Ditto.
26323         (expand_mem_signal_fence): Ditto.
26324         (expand_atomic_load): Ditto.
26325         (expand_atomic_store): Ditto.
26327 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26329         PR rtl-optimization/55547
26330         PR rtl-optimization/53827
26331         PR debug/53671
26332         PR debug/49888
26333         * alias.c (memrefs_conflict_p): Set sizes to negative after
26334         AND adjustments.
26336 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26338         PR target/55940
26339         * function.c (thread_prologue_and_epilogue_insns): Always
26340         add crtl->drap_reg to set_up_by_prologue.set, even if
26341         stack_realign_drap is false.
26343 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26345         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26346         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26347         *call): Fix indention.
26349 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26351         PR target/55876
26352         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26353         Update comment.
26355 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26357         PR rtl-optimization/55153
26358         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26360 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26362         PR tree-optimization/55920
26363         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26364         accesses as grp_to_be_debug_replaced.
26366 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26368         PR tree-optimization/55920
26369         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26370         there is non-useless type conversion needed from debug rhs to lhs,
26371         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26373 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26374             Mikael Pettersson  <mikpe@it.uu.se>
26376         PR target/43961
26377         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26378         Thumb.
26379         (ASM_OUTPUT_CASE_LABEL): Remove.
26380         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26381         * final.c (shorten_branches): Update alignment of labels before
26382         jump tables if CASE_VECTOR_SHORTEN_MODE.
26384 2013-01-15  Richard Biener  <rguenther@suse.de>
26386         PR bootstrap/55961
26387         * system.h: Do not include gmp.h for building host tools.
26389 2013-01-15  Richard Biener  <rguenther@suse.de>
26391         PR middle-end/55882
26392         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26393         account for bitpos when computing alignment.
26395 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26397         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26398         (ix86_target_macros_internal): Likewise.
26400         * config/i386/i386.c (m_CORE2I7): Removed.
26401         (m_CORE_HASWELL): New macro.
26402         (m_CORE_ALL): Likewise.
26403         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26404         (initial_ix86_arch_features): Likewise.
26405         (processor_target_table): Initializations for Core avx2.
26406         (cpu_names): New names "core-avx2".
26407         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26408         PROCESSOR_CORE_HASWELL.
26409         (ix86_issue_rate): New case.
26410         (ia32_multipass_dfa_lookahead): Likewise.
26411         (ix86_sched_init_global): Likewise.
26413         * config/i386/i386.h (TARGET_HASWELL): New macro.
26414         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26415         (processor_type): New PROCESSOR_HASWELL.
26417 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26419         PR tree-optimization/55955
26420         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26421         *SHIFT_EXPR and *ROTATE_EXPR codes.
26423         PR tree-optimization/48766
26424         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26425         -ftrapv disable -fwrapv.
26427 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26429         PR target/55974
26430         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26431         etc. to 1 and not to __flash.
26432         Use LL suffix for __INT24_MAX__ with -mint8.
26433         Use ULL suffix for __UINT24_MAX__ with -mint8.
26435 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26437         * config/avr/avr-arch.h
26438         (struct base_arch_s): Use typedef avr_arch_t instead.
26439         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26440         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26441         * config/avr/avr.c: Same.
26442         * config/avr/avr-devices.c: Same.
26443         * config/avr/driver-avr.c: Same.
26444         * config/avr/gen-avr-mmcu-texi.c: Same.
26445         * config/avr/avr-mcus.def: Adjust comment.
26447 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26449         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26450         * config/aarch64/iterators.md (VALLDI): New.
26452 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26453             Andi Kleen  <ak@linux.intel.com>
26455         PR target/55948
26456         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26457         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26458         memmodel flag.
26460 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26462         * config/avr/avr-stdint.h: Remove trailing blanks.
26463         * config/avr/avr-log.h: Same.
26464         * config/avr/avr-arch.h: Same.
26465         * config/avr/avr-devices.c: Same.
26466         * config/avr/avr-dimode.md: Same.
26467         * config/avr/predicates.md: Same.
26468         * config/avr/avr-c.c: Same.  And fix typo.
26470         * config/avr/avr-protos.h: Same.  And:
26471         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26472         (init_cumulative_args): Rename to avr_init_cumulative_args.
26473         (expand_prologue): Rename to avr_expand_prologue.
26474         (expand_epilogue): Rename to avr_expand_epilogue.
26475         (adjust_insn_length): Rename to avr_adjust_insn_length.
26476         (notice_update_cc): Rename to avr_notice_update_cc.
26477         (final_prescan_insn): Rename to avr_final_prescan_insn.
26478         * config/avr/avr.c: Same.
26479         * config/avr/avr.h: Same.
26480         * config/avr/avr.md: Remove trailing blanks.
26481         (prologue): Use avr_expand_prologue.
26482         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26484 2013-01-14  Richard Biener  <rguenther@suse.de>
26486         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26487         verify_location, collect_subblocks): New functions.
26488         (verify_gimple_in_cfg): Verify that locations only reference
26489         BLOCKs in the functions BLOCK tree.
26491 2013-01-14  Richard Biener  <rguenther@suse.de>
26493         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26494         PHI argument.
26495         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26496         unshare reference.
26497         (insert_out_of_ssa_copy_on_edge): Likewise.
26498         (rewrite_close_phi_out_of_ssa): Likewise.
26499         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26500         debug expressions.
26501         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26502         propagated constants.
26503         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26504         can not be shared.
26506 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26508         * config/avr/avr-modes.def: Add GPL copyright notice.
26510 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26512         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26513         MEMMODEL_MASK to determine memory model.
26514         (atomic_store<mode>): Ditto from operands[2].
26515         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26517 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26519         PR fortran/55935
26520         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26521         (fold_gimple_assign): Don't call unshare_expr here.
26522         (fold_ctor_reference): Call unshare_expr.
26524 2013-01-13  Terry Guo  <terry.guo@arm.com>
26526         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26527         * doc/fragments.texi: Document MULTILIB_REUSE.
26528         * gcc.c (multilib_reuse): New internal spec.
26529         (set_multilib_dir): Also search multilib from multilib_reuse.
26530         * genmultilib (tmpmultilib3): Refactor code.
26531         (tmpmultilib4): Ditto.
26532         (multilib_reuse): New multilib argument.
26534 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26536         * Makefile.in: Update copyright.
26538 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26540         PR middle-end/55890
26541         * calls.c (expand_call): Check if arg_nr is valid.
26543 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26545         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26546         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26547         documentation.  Add missing '__' in front of
26548         __builtin_ia32_packssdw256.
26550 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26552         PR target/55719
26553         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26554         NO_REGS for larl operands.
26555         (s390_reload_larl_operand): Use s390_load_address instead of
26556         emit_move_insn.
26558 2013-01-11  Richard Biener  <rguenther@suse.de>
26560         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26561         (verify_node_sharing): ... here.
26562         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26564 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26566         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26567         Substitute TREECHECKING.
26568         * configure: Regenerate.
26569         * Makefile.in (TREECHECKING): New.
26571 2013-01-11  Richard Guenther  <rguenther@suse.de>
26573         PR tree-optimization/44061
26574         * tree-vrp.c (extract_range_basic): Compute zero as
26575         value-range for __builtin_constant_p of function parameters.
26577 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26579         Update copyright years.
26581 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26583         PR rtl-optimization/55672
26584         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26585         const to be eliminable.
26587 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26589         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26590         * configure: Regenerate.
26592 2013-01-10  Richard Biener  <rguenther@suse.de>
26594         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26596 2013-01-10  Richard Biener  <rguenther@suse.de>
26598         PR bootstrap/55792
26599         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26600         locations for virtual PHI arguments.
26601         (rewrite_update_phi_arguments): Likewise.
26603 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26605         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26606         on to assembler.
26608 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26610         PR tree-optimization/55921
26611         * tree-complex.c (expand_complex_asm): New function.
26612         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26614 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26616         PR target/55718
26617         * config/s390/s390.c (s390_symref_operand_p)
26618         (s390_loadrelative_operand_p): Merge the two functions.
26619         (s390_check_qrst_address, print_operand_address): Add parameters
26620         to s390_loadrelative_operand_p invokation.
26621         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26622         (s390_reload_larl_operand, s390_secondary_reload): Use
26623         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26624         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26626 2013-01-09  Mike Stump  <mikestump@comcast.net>
26628         * dse.c (record_store): Remove unnecessary assert.
26630 2013-01-09  Jan Hubicka  <jh@suse.cz>
26632         PR tree-optimization/55569
26633         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26634         * cfgloop.h (scale_loop_profile): Likewise.
26636 2013-01-09  Jan Hubicka  <jh@suse.cz>
26638         PR lto/45375
26639         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26640         functions.
26641         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26643 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26645         PR middle-end/55114
26646         * expr.h (maybe_emit_group_store): Declare.
26647         * expr.c (maybe_emit_group_store): New function.
26648         * builtins.c (expand_builtin_int_roundingfn): Call it.
26649         (expand_builtin_int_roundingfn_2): Likewise.
26651 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26653         PR rtl-optimization/55829
26654         * lra-constraints.c (match_reload): Add code for absent output.
26655         (curr_insn_transform): Add code for reloads of matched inputs
26656         without output.
26658 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26660         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26661         attribute of movddup insn to DF.
26662         (*vec_interleave_lowv2df): Ditto.
26663         (vec_dupv2df): Ditto.
26665 2013-01-09  Jan Hubicka  <jh@suse.cz>
26667         PR tree-optimiation/55875
26668         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26669         EVERY_ITERATION parameter.
26670         (number_of_iterations_exit): Check if exit is executed every iteration.
26671         (idx_infer_loop_bounds): Similarly here.
26672         (n_of_executions_at_most): Simplify
26673         to only test for cases where statement is dominated by the
26674         particular bound; handle correctly the "postdominance" test.
26675         (scev_probably_wraps_p): Use max loop iterations info
26676         as a global bound first.
26678 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26679             Nick Clifton  <nickc@redhat.com>
26681         * config/v850/v850.md (cbranchsf4): New pattern.
26682         (cstoresf4): New pattern.
26683         (cbranchdf4): New pattern.
26684         (cstoredf4): New pattern.
26685         (movsicc): Disallow floating point comparisons.
26686         (cmpsf_le_insn): Fix order of operators.
26687         (cmpsf_lt_insn): Likewise.
26688         (cmpsf_eq_insn): Likewise.
26689         (cmpdf_le_insn): Likewise.
26690         (cmpdf_lt_insn): Likewise.
26691         (cmpdf_eq_insn): Likewise.
26692         (cmpsf_ge_insn): Use LE comparison.
26693         (cmpdf_ge_insn): Likewise.
26694         (cmpsf_gt_insn): Use LT comparison.
26695         (cmpdf_gt_insn): Likewise.
26696         (cmpsf_ne_insn): Delete pattern.
26697         (cmpdf_ne_insn): Delete pattern.
26698         * config/v850/v850.c (v850_gen_float_compare): Use
26699         gen_cmpdf_eq_insn for NE comparison.
26700         (v850_float_z_comparison_operator)
26701         (v850_float_nz_comparison_operator): Move from here ...
26702         * config/v850/predicates.md: ... to here.  Move GT and GE
26703         comparisons into v850_float_z_comparison_operator.
26704         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26705         Delete prototype.
26706         (v850_float_nz_comparison_operator): Likewise.
26708 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26710         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26711         with calls to gen_insvsi/gen_insvdi.
26713 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26715         * config/i386/i386.c (initial_ix86_tune_features): Set up
26716         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26718 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26719             Jakub Jelinek  <jakub@redhat.com>
26721         PR tree-optimization/48189
26722         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26723         If nitercst is 0, don't predict the exit edge.
26725 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26727         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26728         in asm_fprintf with reg_names.
26729         (aarch64_print_operand_address): Likewise.
26730         (aarch64_return_addr): Likewise.
26731         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26733 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26735         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26736         (INT_U6_BITS): Likewise.
26737         * config/pa/predicates.md (uint6_operand): New predicate.
26738         (shift5_operand, shift6_operand): Likewise.
26739         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26740         arith32_operand.
26741         (lshrdi3): Use shift6_operand.
26742         (shrpsi4, shrpdi4): New insn patterns.
26743         (extzv): Delete expander.
26744         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26745         predicates in unamed zero extract patterns.  Tighten common constraint.
26746         (extv): Delete expander.
26747         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26748         predicates in unamed sign extract patterns.  Tighten common constraint.
26749         (insv): Delete expander.
26750         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26751         predicates in unamed insert patterns.  Tighten common constraint.
26752         Change uint32_operand predicate to uint6_operand predicate in unamed
26753         DImode pattern to insert constant values of type 1...1xxxx.
26755 2013-01-04  Jan Hubicka  <jh@suse.cz>
26757         PR tree-optimization/55823
26758         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26759         issue.
26761 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26762             Uros Bizjak  <ubizjak@gmail.com>
26764         PR rtl-optimization/55845
26765         * df-problems.c (can_move_insns_across): Stop scanning at
26766         volatile_insn_p source instruction or give up if
26767         across_from .. across_to range contains any volatile_insn_p
26768         instructions.
26770 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26772         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26773         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26774         Declare.
26775         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26776         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26778 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26780         PR fortran/55341
26781         * asan.c (asan_clear_shadow): New function.
26782         (asan_emit_stack_protection): Use it.
26784 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26786         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26787         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26788         with tab instead of space.
26790 2013-01-08  Nick Clifton  <nickc@redhat.com>
26792         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26793         register bank 0 at the start of an interrupt handler.
26794         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26795         MDBH registers.
26797 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26799         * config/aarch64/aarch64-simd.md
26800         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26801         (aarch64_simd_bsl): Likewise.
26802         (aarch64_vcond_internal<mode>): Likewise.
26803         (vcond<mode><mode>): Likewise.
26804         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26805         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26807 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26809         * config/aarch64/aarch64-builtins.c
26810         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26812 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26814         PR debug/55579
26815         * tree-sra.c (analyze_access_subtree): Return true also after
26816         potentially creating a debug-only replacement.
26818 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26820         PR middle-end/55890
26821         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26823         PR tree-optimization/54120
26824         * tree-vrp.c (range_fits_type_p): Don't allow
26825         src_precision < precision from signed vr to unsigned_p
26826         if vr->min or vr->max is negative.
26827         (simplify_float_conversion_using_ranges): Test can_float_p
26828         against CODE_FOR_nothing.
26830 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26831             Richard Biener  <rguenther@suse.de>
26833         PR middle-end/55851
26834         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26835         types instead of just INTEGER_TYPE types.
26837 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26839         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26840         TF_SIZE): Define.
26842 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26844         PR target/42661
26845         * config/mips/mips.opt: Change mad to mmad to match documentation.
26847 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26849         PR target/55897
26850         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26851         .progmemx.data now.
26853 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26855         PR target/55897
26856         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26857         (avr_addrspace_t): Add .section_name field.
26858         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26859         array size.
26860         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26861         NULL entry.  Put __memx into .progmemx.data.
26862         (progmem_section_prefix): Remove.
26863         (avr_asm_init_sections): No need to initialize progmem_section.
26864         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26865         section name prefix.
26866         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26867         retrieve the progmem section.
26868         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26869         boundary to run over avr_addrspace[].
26870         (avr_register_target_pragmas): Ditto.
26872 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26874         * varasm.c (output_constant_def_contents): For asan_protect_global
26875         protected strings, adjust DECL_ALIGN if needed, before testing for
26876         anchored symbols.
26877         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26878         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26879         normal decls.
26880         (output_object_block): For asan protected decls, emit asan padding
26881         after their contents.
26882         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26883         (asan_finish_file): Test it here instead.
26885 2013-01-07  Nick Clifton  <nickc@redhat.com>
26886             Matthias Klose  <doko@debian.org>
26887             Doug Kwan  <dougkwan@google.com>
26888             H.J. Lu  <hongjiu.lu@intel.com>
26890         PR driver/55470
26891         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26893         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26895         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26897         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26899         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26901 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26903         PR target/54461
26904         * doc/install.texi (Cross-Compiler-Specific Options): Document
26905         --with-avrlibc.
26907 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26909         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26910         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26911         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26912         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26913         vqmovun_high_s64): Fix source operand number and update copyright.
26915 2013-01-07  Richard Biener  <rguenther@suse.de>
26917         PR middle-end/55890
26918         * gimple.h (gimple_call_builtin_p): New overload.
26919         * gimple.c (validate_call): New function.
26920         (gimple_call_builtin_p): Likewise.
26921         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26922         Use gimple_call_builtin_p.
26923         (find_func_clobbers): Likewise.
26924         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26925         (strlen_optimize_stmt): Likewise.
26927 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26929         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26930         (vld1q_dup_*): Likewise.
26931         (vld1_*): Likewise.
26932         (vld1q_*): Likewise.
26933         (vld1_lane_*): Likewise.
26934         (vld1q_lane_*): Likewise.
26936 2013-01-07  Richard Biener  <rguenther@suse.de>
26938         * lto-streamer.h (LTO_minor_version): Bump to 2.
26940 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26942         * config/aarch64/aarch64-protos.h
26943         (aarch64_const_double_zero_rtx_p): Rename to...
26944         (aarch64_float_const_zero_rtx_p): ...this.
26945         (aarch64_float_const_representable_p): New.
26946         (aarch64_output_simd_mov_immediate): Likewise.
26947         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26948         move immediate case.
26949         * config/aarch64/aarch64.c
26950         (aarch64_const_double_zero_rtx_p): Rename to...
26951         (aarch64_float_const_zero_rtx_p): ...this.
26952         (aarch64_print_operand): Allow printing of new constants.
26953         (aarch64_valid_floating_const): New.
26954         (aarch64_legitimate_constant_p): Check for valid floating-point
26955         constants.
26956         (aarch64_simd_valid_immediate): Likewise.
26957         (aarch64_vect_float_const_representable_p): New.
26958         (aarch64_float_const_representable_p): Likewise.
26959         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26960         (aarch64_output_simd_mov_immediate): New.
26961         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26962         (*movdf_aarch64): Likewise.
26963         * config/aarch64/constraints.md (Ufc): New.
26964         (Y): call aarch64_float_const_zero_rtx.
26965         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26967 2013-01-07  Richard Biener  <rguenther@suse.de>
26969         PR tree-optimization/55888
26970         PR tree-optimization/55862
26971         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26972         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26973         not if it is contained therein.
26975 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26977         * config/avr/t-avr: Typo.
26979 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26981         PR55243
26982         * config/avr/t-avr: Don't automatically rebuild
26983         $(srcdir)/config/avr/t-multilib
26984         $(srcdir)/config/avr/avr-tables.opt
26985         $(srcdir)/doc/avr-mmcu.texi
26986         (avr-mcus): New phony target to build them on request.
26987         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26988         * avr/avr-mcus.def: Adjust comments.
26990 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26992         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26994 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26996         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26998 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27000         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27002 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27004         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27005         to generate profiling.
27006         * config/rs6000/aix64.h (LIB_SPEC): Same.
27008 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27010         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27011         New function.
27012         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27014 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27016         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27017         unconditionally.
27018         (ix86_expand_move): Ditto.
27019         (ix86_zero_extend_to_Pmode): Ditto.
27020         (ix86_expand_call): Ditto.
27021         (ix86_expand_special_args_builtin): Ditto.
27022         (ix86_expand_builtin): Ditto.
27024 2013-01-04  Richard Biener  <rguenther@suse.de>
27026         PR tree-optimization/55862
27027         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27028         translating them through PHI nodes.
27030 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27032         PR tree-optimization/55755
27033         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27034         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27036 2013-01-04  Marek Polacek  <polacek@redhat.com>
27038         PR middle-end/55859
27039         * opts.c (default_options_optimization): Clarify error message.
27041 2013-01-04  Richard Biener  <rguenther@suse.de>
27043         PR middle-end/55863
27044         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27045         reassociation.
27047 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27049         PR target/53789
27050         * config/pa/pa.md (movsi): Revert previous change.
27051         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27052         references.
27054 2013-01-03  Richard Henderson  <rth@redhat.com>
27056         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27057         after eliminating TLS symbols.
27059 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27061         PR bootstrap/50167
27062         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27063         * graphite-poly.c (debug_gmp_value): Likewise.
27065 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27067         PR target/55712
27068         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27069         selected code model, define __code_mode_small__, __code_model_medium__,
27070         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27071         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27072         xchg temporary register with %k.  Declare temporary register as
27073         early clobbered.
27074         [__x86_64__]: For medium and large code models, preserve %rbx register.
27076 2013-01-03  Richard Biener  <rguenther@suse.de>
27078         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27079         (dump_subscript): Adjust.
27080         (finalize_ddr_dependent): Do not dump redundant info.
27081         (analyze_siv_subscript): Adjust.
27082         (subscript_dependence_tester): Likewise.
27083         (compute_affine_dependence): Likewise.
27085 2013-01-03  Richard Biener  <rguenther@suse.de>
27087         Revert
27088         2013-01-03  Richard Biener  <rguenther@suse.de>
27090         PR tree-optimization/55857
27091         * tree-vect-stmts.c (vectorizable_load): Do not setup
27092         re-alignment for invariant loads.
27094         2013-01-02  Richard Biener  <rguenther@suse.de>
27096         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27097         invariant load do not generate a vector load from the scalar location.
27099 2013-01-03  Richard Biener  <rguenther@suse.de>
27101         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27102         for not vectorizing.
27103         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27104         not build INDIRECT_REFs, call get_name once only.
27105         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27106         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27108 2013-01-03  Richard Biener  <rguenther@suse.de>
27110         PR tree-optimization/55857
27111         * tree-vect-stmts.c (vectorizable_load): Do not setup
27112         re-alignment for invariant loads.
27114 2013-01-03  Richard Biener  <rguenther@suse.de>
27116         PR lto/55848
27117         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27118         prefer a built-in decl.
27120 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27122         * gcc.c (process_command): Update copyright notice dates.
27123         * gcov.c (print_version): Likewise.
27124         * gcov-dump.c (print_version): Likewise.
27126         PR rtl-optimization/55838
27127         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27128         iv0.step, iv1.step and step.
27130 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27131             Marc Glisse  <marc.glisse@inria.fr>
27133         PR tree-optimization/55832
27134         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27135         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27136         integer_{one,zero}_node.
27138 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27140         PR debug/54402
27141         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27142         * var-tracking.c (reverse_op): Don't add reverse ops to
27143         VALUEs that have already
27144         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27146 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27148         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27150 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27152         * dumpfile.c (dump_loc): Print filename with location.
27153         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27154         new location_t parameter to emit complete unroll message with
27155         new dump framework.
27156         (canonicalize_loop_induction_variables): Compute loops location
27157         and pass to try_unroll_loop_completely.
27158         * loop-unroll.c (report_unroll_peel): New function.
27159         (peel_loops_completely): Use new dump format with location
27160         for main dumpfile message, and invoke report_unroll_peel on success.
27161         (decide_unrolling_and_peeling): Ditto.
27162         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27163         by report_unroll_peel.
27164         (decide_peel_completely): Ditto.
27165         (decide_unroll_constant_iterations): Ditto.
27166         (decide_unroll_runtime_iterations): Ditto.
27167         (decide_peel_simple): Ditto.
27168         (decide_unroll_stupid): Ditto.
27169         * cfgloop.c (get_loop_location): New function.
27170         * cfgloop.h (get_loop_location): Declare.
27172 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27174         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27175         NULL.
27177 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27179         PR middle-end/55198
27180         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27181         BLKmode objects when EXPAND_MEMORY is specified.
27183 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27185         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27186         in loop predicate.
27187         (fold_builtin_cpu): Do not share cpu model decls across statements.
27189 2013-01-02  Jason Merrill  <jason@redhat.com>
27191         PR c++/55804
27192         * tree.c (build_array_type_1): Revert earlier change.
27194 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27196         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27197         "cortex-a57".
27198         * config/aarch64/aarch64-tune.md: Re-generate.
27200 2013-01-02  Richard Biener  <rguenther@suse.de>
27202         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27203         invariant load do not generate a vector load from the scalar location.
27205 2013-01-02  Richard Biener  <rguenther@suse.de>
27207         PR bootstrap/55784
27208         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27209         * configure: Regenerate.
27211 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27213         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27214         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27215         (expand_builtin_int_roundingfn_2): Keep the original target around
27216         for the fallback case.
27218 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27220         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27221         to be clear for sign changes.
27223 2013-01-01  Jan Hubicka  <jh@suse.cz>
27225         * ipa-inline-analysis.c: Fix formatting.
27227 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27229         PR tree-optimization/55831
27230         * tree-vect-loop.c (get_initial_def_for_induction): Use
27231         gsi_after_labels instead of gsi_start_bb.
27233 Copyright (C) 2013 Free Software Foundation, Inc.
27235 Copying and distribution of this file, with or without modification,
27236 are permitted in any medium without royalty provided the copyright
27237 notice and this notice are preserved.